<%BANNER%>

The Mechanisms of Lightning Leader Propagation and Ground Attachment

Permanent Link: http://ufdc.ufl.edu/UFE0044602/00001

Material Information

Title: The Mechanisms of Lightning Leader Propagation and Ground Attachment
Physical Description: 1 online resource (544 p.)
Language: english
Creator: Hill, Jonathan D
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: lightning
Electrical and Computer Engineering -- Dissertations, Academic -- UF
Genre: Electrical and Computer Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Lightning data were collected at the International Center for Lightning Research and Testing at Camp Blanding, Florida from 2009 to 2011.  Data were obtained for 12 natural negative cloud-to-ground lightning discharges and 46 rocket-and-wire triggered lightning discharges.  The mechanisms and characteristics of upward and downward leader propagation and downward leader attachment to ground were examined using data from high-speed framing cameras, electric field derivative (dE/dt) sensors, plastic and lanthanum bromide (LaBr3) energetic radiation (x-ray) scintillation detectors, channel-base currents, a Lightning Mapping Array (LMA), and a C-band dual-polarimetric radar.  The dE/dt and energetic radiation measurements form a 10-station time-of-arrival (TOA) network used to determine the locations and emission times of sources within about 750 m of ground.  High-speed video images of a stepped leader are analyzed to determine characteristics of optical phenomena, such as space stems/leaders, associated with the sequence of electrical breakdowns that occur during the formation of a leader step.  Observations are compared to those obtained for dart-stepped leader steps in triggered lightning.  dE/dt waveforms of "chaotic" dart leaders preceding triggered and natural lightning strokes are analyzed using TOA techniques and their characteristics are compared to dart and dart-stepped leader processes.  "Chaotic" dart leaders are found to emit copious x-rays in a nearly continuous manner prior to the return stroke.  The initial stage (IS) processes of nine triggered lightning discharges are mapped in three-dimensions with the LMA.  The geometrical and electrical properties of IS branching are determined and compared to channel-base currents.  The LMA and radar are used to examine the effects of hydrometeor structure on the propagating IS channels.  Vertically-propagating IS channels are observed to turn horizontal at 3-6 km altitude, near the 0° C level, often propagating for many kilometers along the tops of high-reflectivity rain-shafts.  The propagation characteristics and attachment processes of triggered and natural lightning dart-stepped leaders are analyzed.  Properties of the dE/dt pulses following the final downward leader step are discussed with respect to the measured channel-base current.  Upward leader lengths, speeds, and durations are calculated.  A total of 30 x-ray sources are TOA-located and compared spatially and temporally to corresponding dE/dt source locations.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jonathan D Hill.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Uman, Martin A.
Local: Co-adviser: Rakov, Vladimir A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044602:00001

Permanent Link: http://ufdc.ufl.edu/UFE0044602/00001

Material Information

Title: The Mechanisms of Lightning Leader Propagation and Ground Attachment
Physical Description: 1 online resource (544 p.)
Language: english
Creator: Hill, Jonathan D
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: lightning
Electrical and Computer Engineering -- Dissertations, Academic -- UF
Genre: Electrical and Computer Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Lightning data were collected at the International Center for Lightning Research and Testing at Camp Blanding, Florida from 2009 to 2011.  Data were obtained for 12 natural negative cloud-to-ground lightning discharges and 46 rocket-and-wire triggered lightning discharges.  The mechanisms and characteristics of upward and downward leader propagation and downward leader attachment to ground were examined using data from high-speed framing cameras, electric field derivative (dE/dt) sensors, plastic and lanthanum bromide (LaBr3) energetic radiation (x-ray) scintillation detectors, channel-base currents, a Lightning Mapping Array (LMA), and a C-band dual-polarimetric radar.  The dE/dt and energetic radiation measurements form a 10-station time-of-arrival (TOA) network used to determine the locations and emission times of sources within about 750 m of ground.  High-speed video images of a stepped leader are analyzed to determine characteristics of optical phenomena, such as space stems/leaders, associated with the sequence of electrical breakdowns that occur during the formation of a leader step.  Observations are compared to those obtained for dart-stepped leader steps in triggered lightning.  dE/dt waveforms of "chaotic" dart leaders preceding triggered and natural lightning strokes are analyzed using TOA techniques and their characteristics are compared to dart and dart-stepped leader processes.  "Chaotic" dart leaders are found to emit copious x-rays in a nearly continuous manner prior to the return stroke.  The initial stage (IS) processes of nine triggered lightning discharges are mapped in three-dimensions with the LMA.  The geometrical and electrical properties of IS branching are determined and compared to channel-base currents.  The LMA and radar are used to examine the effects of hydrometeor structure on the propagating IS channels.  Vertically-propagating IS channels are observed to turn horizontal at 3-6 km altitude, near the 0° C level, often propagating for many kilometers along the tops of high-reflectivity rain-shafts.  The propagation characteristics and attachment processes of triggered and natural lightning dart-stepped leaders are analyzed.  Properties of the dE/dt pulses following the final downward leader step are discussed with respect to the measured channel-base current.  Upward leader lengths, speeds, and durations are calculated.  A total of 30 x-ray sources are TOA-located and compared spatially and temporally to corresponding dE/dt source locations.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jonathan D Hill.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Uman, Martin A.
Local: Co-adviser: Rakov, Vladimir A.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044602:00001


This item has the following downloads:


Full Text

PAGE 1

1 THE MECHANISMS OF LIGHTNING LEADER PROPAGATION AND GROUND ATTACHMENT By JONATHAN DUSTIN HILL A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Jonathan Dustin Hill

PAGE 3

3 To my loving wife, Ashley, and our wonderful son Brady

PAGE 4

4 ACKNOWLEDGMENTS I would first like to express my heartfelt gratitude to my advisor, Dr. Martin Uman, for making this work possible and for providing me with excellent academic and professional guidance throughout the progress of my Ph.D studies. I credit Dr. Uman with teaching me, by example, the proper way of conducting research. His knowledge of light ning and enthusiasm for the subject are truly inspirational. I would also like to thank my co chair, Dr. Vladimir Rakov, for his support and valuable contributions over the past six years. I extend special appreciation to Dr. Douglas Jordan, who has not only provided me with invaluable direction and assistance in my experimental work, but perhaps more importantly, has been a wonderful friend and mentor. I can't thank you enough. Dr. Joseph Dwyer Dr. Hamid Rassoul and Dr. Michael Biggerstaff also dese rve significant credit for their contributions to this work. I also thank Robert Olsen III, Michael Stapleton, and George Schnetzer for their willingness to share their expertise in the design and fabrication of electronics and antenna systems. Special g ratitude is owed to Dr. Carlos Mata, Angel Mata, and Kevin Bivona for allowing the ICLRT team to utilize and benefit from their outstanding repertoire of instrumentation, and for lending their expertise in constructing robust data acquisition and transmiss ion systems. Over the past six years, a number of undergraduate students have worked at the ICLRT to support the data collection effort. They all deserve a tremendous amount of credit for their dedication and hard work. These individuals include Nolan B rooks, Adam Dunhoft, Paul Anderson, and Julia Jordan. Your efforts are appreciated more than you know. I would also like to thank graduate students Terry Ngin, John Pilkey, William Gamerota and Neal Dupree for their dedication and hard work in maintaini ng and operating the measurement network at the ICLRT. Dr. Joseph Howard and Dr. Christopher Biagi deserve special credit for their contributions to this work, and also for being great friends

PAGE 5

5 and research partners. Finally, I would like to thank to my b eautiful wife Ashley Hill, my father Skip Hill, and my mother Carol Flowers for their unbelievable support and encouragement during the past six years. It has been a true pleasure and an incredible experience to work in the Lightning Research Group at the University of Florida.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 9 LIST OF FIGURES ................................ ................................ ................................ ....................... 12 ABSTRACT ................................ ................................ ................................ ................................ ... 22 1 INTRODUCTION AND LITERATURE REVIEW ................................ .............................. 24 1.1 Introduction ................................ ................................ ................................ .................. 31 1.2 The Global Electric Circuit and Thundercloud Electrical Structure ............................ 33 1.3 The Lightning Discharge ................................ ................................ ............................. 35 1.4 Downward Negative Cloud to Ground Lightning ................................ ....................... 37 1.5 Rocket Triggered L ightning ................................ ................................ ........................ 42 2 DESCRIPTION OF EXPERIMENT ................................ ................................ ...................... 56 2.1 ICLRT Measurement Network Infrastructure ................................ ............................. 57 2.2 ICLRT Measurement Control System ................................ ................................ ......... 58 2.2.1 HAL ................................ ................................ ................................ ................... 59 2.2.2 PICs ................................ ................................ ................................ .................... 60 2.2.2.1 2001 Edition PIC ................................ ................................ ...................... 62 2.2.2.2 2006 Edition PIC ................................ ................................ ...................... 63 2.2.2.3 2011 Edition PIC ................................ ................................ ...................... 65 2.3 ICLRT Fiber Optic Data Transmission System ................................ .......................... 67 2.4 ICLRT Digital Storage Oscilloscopes ................................ ................................ ......... 68 2.4.1 LeCroy Waverunner I (LT344) ................................ ................................ .......... 69 2.4.2 LeCroy Waverunner II (LT374) ................................ ................................ ........ 70 2.4.3 LeCroy Waverunner 44 Xi ................................ ................................ ................. 70 2.4.4 Yokogawa DL716 ................................ ................................ .............................. 71 2.4.5 Yokogawa DL750 ................................ ................................ .............................. 72 2.5 HBM Digitization System ................................ ................................ ............................ 73 2.6 Storm Detection and Data Acquisition System Arming/Disarming Process ............... 76 2.7 ICLRT Triggering System ................................ ................................ ........................... 83 2.8 ICLRT GPS Time System ................................ ................................ ........................... 85 2.9 ICLRT High Speed Cameras ................................ ................................ ....................... 86 2.10 ICLRT Sti ll Cameras and HD Video Cameras ................................ ..................... 91 2.11 ICLRT Rockets and Rocket Launching System ................................ ................... 95 2.11.1 Tower Launching Facility ................................ ................................ .............. 96 2.11.2 Field (Ground) Launching Facility (2010) ................................ .................... 98 2.11.3 Field Launching Facility (2011) ................................ ................................ .. 100 2.12 ICLRT Channel Base Current Measurements ................................ .................... 101 2.12.1 2009 Channel Base Current Measurements ................................ ................. 101

PAGE 7

7 2.12.2 2010 Channel Base Current Measurements ................................ ................. 103 2.12.3 2011 Channel Base Current Measurements ................................ ................. 106 2.13 ICLRT dE/dt Measurements ................................ ................................ ............... 108 2.14 ICLRT Energetic Radiation Measurements ................................ ........................ 115 2.14.1 NaI Detectors ................................ ................................ ............................... 117 2.14.2 LaBr 3 Detectors ................................ ................................ ............................ 119 2.14.3 Plastic Detectors ................................ ................................ ........................... 120 2.15 Lightning Mapping Array (LMA) ................................ ................................ ...... 123 2.16 SMART Radar ................................ ................................ ................................ .... 127 2.17 2009 Measurement Description ................................ ................................ .......... 128 2.18 2010 Measurement Descript ion ................................ ................................ .......... 129 2.19 2011 Measurement Description ................................ ................................ .......... 130 3 TIME OF ARRIVAL MEASUREMENTS ................................ ................................ ......... 181 3.1 ICLRT TOA Network ................................ ................................ ................................ 187 3.2 Determination of Sensor Locations ................................ ................................ ........... 189 3.3 Measurement of Time Delays ................................ ................................ .................... 191 3.4 DSO Time Base Synchronization ................................ ................................ .............. 195 3.5 dE/dt Signal Arrival Time Selection (DSO dE/dt Network) ................................ ..... 198 3.6 dE/dt Signal Arrival Time Selection (HBM Network) ................................ .............. 201 3.7 X ray Signal Arrival Time Selection ................................ ................................ ......... 203 3.8 Calculation of Source Locations and Emission Times ................................ .............. 204 3.9 Spatial and Timing Errors of the ICLRT TOA Networks ................................ ......... 209 4 DAT A DOCUMENTATION, DATA, AND GENERAL DATA STATISTICS ................ 222 4.1 Data Cataloguing and Documentation ................................ ................................ ....... 222 4.2 Synopsis of Collected Ligh tning Data ................................ ................................ ....... 226 4.3 General Statistics for Triggered Lightning Return Strokes (2009 2011) .................. 229 4.4 General Statistics for Triggere d Lightning UPL/ICC Currents (2009 2011) ............ 232 5 HIGH SPEED VIDEO OBSERVATIONS OF A NATURAL LIGHTNING STEPPED LEADER ................................ ................................ ................................ ............................... 255 5.1 Liter ature Review ................................ ................................ ................................ ....... 255 5.2 Experimental Setup and Data Processing Techniques ................................ ............... 259 5.3 Results ................................ ................................ ................................ ........................ 26 1 5.4 Discussion of Results ................................ ................................ ................................ 267 6 "CHAOTIC" DART LEADERS IN TRIGGERED AND NATURAL LIGHTNING ......... 286 6.1 Lit erature Review ................................ ................................ ................................ ....... 286 6.2 Experimental Description and Data Processing Techniques ................................ ..... 291 6.3 Background and Overall Characteristics of "Chaotic" Dart Leaders ......................... 294 6.4 Data and Analysis ................................ ................................ ................................ ...... 297 6.4.1 Triggered Lightning Flash UF 10 13 ................................ ............................... 298 6.4.2 Triggered Lightning Flash UF 10 24 ................................ ............................... 303

PAGE 8

8 6.4.3 High Speed Video Images of Triggered Lightning "Chaotic" Dart Leaders ... 308 6.4.4 Discussion of Results for Triggered Lightning "Chaotic" Dart Leaders ......... 310 6.4.5 Natural Lightning Flash MSE 11 01 ................................ ................................ 314 6.4.6 Discussion of Results for Natural Lightning "Chaotic" Dart Leaders ............. 317 7 LIGHTNING MAPPING ARRAY OBSERVATIONS OF THE INITIAL STAGE OF TRIGGERED LIGHTNING DISCH ARGES ................................ ................................ ....... 341 7.1 Background and Experimental Setup ................................ ................................ ......... 341 7.2 The Events of August 5, 2011 ................................ ................................ .................... 345 7.2.1 Flash UF 11 24 ................................ ................................ ................................ 346 7.2.2 Flash UF 11 25 ................................ ................................ ................................ 349 7.2.3 Flash UF 11 26 ................................ ................................ ................................ 354 7.3 Flash UF 11 32 (August 18, 2011) ................................ ................................ ............ 363 7.4 LMA Observations of Additional 2011 Triggered Flashes ................................ ....... 367 7.5 Discussion of 2011 Triggered Lightning LMA Observations ................................ ... 367 8 PROPAGATION CHARACTERISTICS AND ATTACHMENT PROCESSES OF DART STEPPED LEADERS IN TRIGGERED AND NATURAL LIGHNTING VIA DE/DT AND X RAY TOA MEASUREMENTS ................................ ................................ 417 8.1 Background and Literature Review ................................ ................................ ........... 418 8.2 Analysis of Triggered and Natural Lightning D art Stepped Leaders ........................ 430 8.2.1 Tenth Return Stroke of Flash UF 11 15 ................................ .......................... 430 8.2.2 First Return Stroke of Flash UF 11 25 ................................ ............................ 439 8.2.3 Second Stroke of Flash UF 11 32 ................................ ................................ .... 444 8.2.4 Fourth Stroke of Flash UF 11 35 ................................ ................................ ..... 451 8.2.5 Second Stroke of Flash MSE 11 01 ................................ ................................ 458 8.3 Discussion of Results ................................ ................................ ................................ 464 9 SUMMARY OF RESULTS AND RECOMMENDATIONS FO R FUTURE RESEARCH ................................ ................................ ................................ ......................... 520 9.1 Summary of Experimental Results ................................ ................................ ............ 520 9.2 Recommendations for Future Research ................................ ................................ ..... 530 LIST OF REFERENCES ................................ ................................ ................................ ............. 532 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 544

PAGE 9

9 LIST OF TABLES Table page 2 1 The locations of all Campbell Scientific field mills. ................................ ....................... 143 2 2 The locations of the seven LMA stations ................................ ................................ ........ 166 2 3 2009 ICLRT measurement specifications ................................ ................................ ........ 171 2 4 2009 ICLRT photographic setup specifications. ................................ ............................. 172 2 5 2 010 ICLRT measurement specifications. ................................ ................................ ....... 174 2 6 2010 ICLRT photographic setup specifications. ................................ ............................. 175 2 7 2011 ICLRT measurement s pecifications. ................................ ................................ ....... 177 2 8 2011 HBM digitization system measurement specifications. ................................ .......... 178 2 9 2011 ICLRT photographic setup specifi cations. ................................ ............................. 180 3 1 The spatial coordinates of all TOA sensors at the ICLRT from the 2009 and 2011 site surveys.. ................................ ................................ ................................ ........................... 211 3 2 Measure d time delays from each TOA sensor to the DSO input. ................................ .... 213 3 3 Measured time delays from each energetic radiation detector to the DSO input. ........... 214 4 1 A list of natural lightning discharges that terminated on or very near the ICLRT from 2009 2011. ................................ ................................ ................................ ....................... 237 4 2 2009 triggered lightning event characteristics.. ................................ ............................... 238 4 3 2010 triggered lightning event characteristics.. ................................ ............................... 239 4 4 2011 triggered lightning event characteristics.. ................................ ............................... 240 4 5 A list of dart stepped leaders preceding triggered lightning return strokes between 2009 2011. ................................ ................................ ................................ ....................... 241 4 6 A list of "chaotic" dart leaders preceding trigger ed lightning return strokes between 2009 2011. ................................ ................................ ................................ ....................... 242 4 7 Triggered lightning return stroke peak current statistics for studies at the Kennedy Space Center in Florida, Saint Privat d'Allier in Fr ance, Fort McClellan in Alabama, and Camp Blanding in Florida from 1985 2011. ................................ ............................. 244

PAGE 10

10 4 8 Measured channel base current statistics during the initial stage (IS) of triggered lightning discharges a t Fort McClellan in Alabama, and Camp Blanding in Florida. ..... 254 5 1 Previous optically obtained stepped leader statistics ................................ ....................... 270 5 2 Measured stepped leader statistics assuming a range of 1 km. ................................ ........ 274 5 3 Measured space stem/leader statistics assuming a range of 1 km. ............................... 278 6 1 General information on the four "chaotic" dart leaders recorded between June and August 2010. ................................ ................................ ................................ .................... 321 6 2 Calculated three dimensional TOA locations, emission times, velocities be tween successive located pulses, and associated errors for individual pulses occurring during three bursts for event UF 10 13 on June 21, 2010 ................................ ............... 324 6 3 Calculated three dimensional TOA location s, emission times, velocities between successive located pulses, and associated errors for individual pulses occurring during four bursts for event UF 10 24 on August 13, 2010 ................................ ............. 330 7 1 A list o f working LMA stations on each day during summer 2011 where triggered lightning data were collected ................................ ................................ ........................... 372 7 2 Measured geometrical and electrical parameters of the UPL/ICC for nine triggered lig htning flashes during summer 2011. ................................ ................................ ............ 414 8 1 dE/dt TOA source locations within 20 s of the return stroke for the dart stepped leader preceding the te nth return stroke of flash UF 11 15 on July 7, 2011. ................... 478 8 2 dE/dt TOA source locations within 25 s of the return stroke for the dart stepped leader preceding the tenth return stroke of flash UF 11 25 on August 5, 2011 ............... 486 8 3 dE/dt TOA source locations within 14 s of the return stroke for the dart stepped leader preceding the tenth return stroke of flash UF 11 32 on August 18, 2 011. ............ 492 8 4 dE/dt and X ray TOA source locations for the dart stepped leader preceding the second return stroke of flash UF 11 32 on August 18, 2011. ................................ .......... 497 8 5 dE/dt TOA source locations within 12 s of the return stroke for the dart stepped leader preceding the fourth return stroke of flash UF 11 35 on August 18, 2011. ......... 501 8 6 dE/dt and X ray TOA source locations for the dart stepped leader preceding the second return stroke of flash UF 11 35 on August 18, 2011. ................................ .......... 507 8 7 dE/dt TOA source locations within 23 s of the return stroke for the dart stepped leader preceding the second return stroke of natural flash MSE 11 01 on July 7, 2011. ................................ ................................ ................................ ................................ 512

PAGE 11

11 8 8 dE/dt and x ray TOA source locations for the dart s tepped leader preceding the second return stroke of flash MSE 11 01 on July 7, 2011. ................................ .............. 515 8 9 Measured and calculated statistics of the seven triggered lightning dart stepped leaders and one na tural lightning dart stepped leader ................................ ...................... 517 8 10 Measured statistics of the 30 x ray source locations ................................ ....................... 518

PAGE 12

12 LIST OF FIGU RES Figure page 1 1 The four classifications of cloud to ground lightning. ................................ ...................... 49 1 2 Six stage sequential decomposition of the negative cloud to ground lightning discharge. ................................ ................................ ................................ ........................... 50 1 3 High speed video frames of a negative stepped leader. ................................ ..................... 51 1 4 Elec tric field waveforms for natural flash MSE 11 01 ................................ ...................... 52 1 5 Six stage sequential decomposition of rocket triggered lightning discharge. ................... 53 1 6 Still photographs of flash UF 09 25 triggered on June 29, 2009 and flash UF 11 24 on August 5, 2011 ................................ ................................ ................................ .............. 54 1 7 Waveforms of the initial stage (IS) of flash UF 10 13. ................................ ..................... 55 2 1 Perspective aerial view of the ICLRT ................................ ................................ .............. 134 2 2 Photograph of a n example of one of 25 individual ground based measurement stations at the ICLRT ................................ ................................ ................................ ....... 135 2 3 Serial command packet structure from HAL ................................ ................................ ... 136 2 4 Photogaphs of t hree generations of ICLRT PIC controllers ................................ ............ 137 2 5 Photograph nside the rear door of the Launch Control trailer. ................................ ........ 138 2 6 Photograph of f ront rack space in Launch Control ................................ .......................... 139 2 7 Photographs of HBM GEN16t digitization system and HBM 7600 Isolated Digitizer. .. 140 2 8 Screen capture of the NLDN display o n the LTS2005 software during a storm on May 15, 2012. ................................ ................................ ................................ ................ 1 41 2 9 Photograph and l ocations of the 8 Campbell Scientific field mills overlaid on an aerial photograph of the ICLRT. ................................ ................................ ...................... 142 2 10 The field installation of the Northeast Optical detector (NEO)....................... ................ 1 44 2 11 Photograph of ICLRT Master Trigger Box in Launch Control. ................................ ...... 145 2 12 Photographs of Photron SA1.1, Phantom V7.3, and Cordin 550 high speed cameras. ... 146 2 13 A block diagram of the full ICLRT instrument triggering topology.. ............................. 147

PAGE 13

13 2 14 A erial perspective view of the ICLRT showing the locations of the still cameras. ......... 148 2 15 An example of a fiberglass r ocket used at the ICLRT to trigger lightning. .................... 149 2 16 A photograph of the Tower Launching facility.. ................................ ............................. 150 2 17 1/20 th scale m odel experiment of the lightning protection system (LPS) over Pad 39B at Kennedy Space Center. ................................ ................................ ................................ 151 2 18 Photograph of the 2010 Field Launching facility. ................................ ........................... 152 2 19 Block diagram of the rocket launching system for th e Field Launcher (2010, 2011). .... 153 2 20 The 2011 Field Launcher with intercepting wire ring and down con ductors. ............... 154 2 21 A p hotograph and electrical schematic of the Tower Launcher current measure ment box (2009 configuration). ................................ ................................ ................................ 155 2 22 E lectrical schematic s of Initial Stage and Return Stroke current measurement box es (2010 configuration) ................................ ................................ ................................ ........ 156 2 23 A photograph and electrical schematic of the 2011 current measurement box ............... 157 2 24 A photograph and schematic of the dE/dt fl at plate antenna installation. ....................... 158 2 25 A vertical cross section of th e dE/dt flat plate antenna installation. ................................ 159 2 26 Photographs and a schematic of the dE/dt measurement electronics box installation s .. 160 2 27 P hotograph s and a schematic of a TERA box ................................ ................................ 161 2 28 A photograph of the LaBr 3 detector TERA box (Station 17). ................................ ......... 162 2 29 Plot of the single photon responses of the NaI (green trace) and LaBr 3 (blue trace) detector s to a Cs 137, 662 keV source. ................................ ................................ ............ 163 2 30 P hotograph of the field installa tion of the plasti c detectors. ................................ ............ 164 2 31 P hotograph s and a schematic of the of the plastic scintillator s ................................ ..... 165 2 32 Plan view of the seven LMA st ation locations. ................................ ............................... 167 2 33 P hotograph s of the Blast Wall LMA antenna and the LMA electronics enclosure ......... 168 2 34 A photograph of the C band dual polarimetric SMART Radar ................................ ...... 169 2 35 A erial photograph s of the ICLRT 2009 measurement network ................................ ....... 170 2 36 Aerial ph otographs of the ICLRT 2010 measurement network ................................ ....... 173

PAGE 14

14 2 37 Aerial photographs of the ICLRT 2011 measurement network ................................ ...... 176 2 38 20 11 inverte d electric field antenna ................................ ................................ ................ 179 3 1 Line diagram of nominal and field delay measurement configurations for all TOA measurements. ................................ ................................ ................................ .................. 212 3 2 Line diagram of plastic scintillator PMT/PMT base measurement delay configuration. ................................ ................................ ................................ ................... 215 3 3 Line diagram of the DSO channel connectivity for all TOA measurements. .................. 216 3 4 A plot of the 10 dE/dt waveforms from a triggered lightning dart step ped leader on August 18, 2011. ................................ ................................ ................................ .............. 217 3 5 An expanded view of the 10 dE/dt waveforms shown in Figure 4 4. ............................. 218 3 6 A plot of the 10 energetic radiation waveforms (plastic and LaBr 3 detectors) from the same triggered lightning dart stepped leader shown i n Figure 3 4. ................................ 219 3 7 An expanded view of the 10 energetic radiation waveforms shown i n Figure 3 6. ....... 220 3 8 A plot of the 10 dE/dt waveforms recorded on the HBM Genesis system for the same dart stepped leader pulse shown in Figure 3 5.. ................................ .............................. 221 4 1 Distributions of triggered lightning return stroke peak currents at Cam p Blanding from 2009 2011 ................................ ................................ ................................ ............... 243 4 2 Distribution of triggered lightning return stroke peak currents versus preceding leader type. ................................ ................................ ................................ ...................... 245 4 3 Histogram of the occurrences of dart stepped leaders and "chaotic" dart leaders versus triggered lightning subsequent return stroke order ................................ ............... 246 4 4 Distributions of triggered lightni ng flash multiplicities at Camp Blanding from 2009 2011 ................................ ................................ ................................ ................................ .. 247 4 5 Distributions of triggered lightning UPL/ICC time durations for 46 flashes triggered at Camp Blanding from 2009 2011 ................................ ................................ ................ 248 4 6 Distributions of triggered lightning UPL/ICC charge transfers for 46 flashes triggered at Camp Blanding from 2009 2011 ................................ ................................ 249 4 7 Distributions of triggered lightning UPL/ICC average current amplitudes for 46 flashes triggered at Camp Blanding from 2009 2011 ................................ ..................... 250 4 8 Distributions of the time duration between the initi ation of the sustained UPL and the ICV for 37 triggered lightning flashes at Camp Blanding from 2009 2011 .................... 251

PAGE 15

15 4 9 Distributions of calculated action integrals from the UPL initiation to the I CV for 37 flashes triggered at Camp Blanding from 2009 2011. ................................ ................... 250 9 4 10 Distributions of the average current amplitude from the UPL to the ICV for 37 triggered lightning flashes at Camp Blandi ng from 2009 2011 ................................ .. 251 50 5 1 Two frames of a dart stepped leader preceding the eighth return stroke of flash UF 08 18 on September 17, 2008. ................................ ................................ ......................... 271 5 2 Ten consecutive frames (~42 s) of a dart stepped leader preceding the fifth return stroke of a triggered lightning discharge (flash UF 09 25 on June 29, 2009). ................ 272 5 3 A full fr ame time integrated image of a natural stepped leader ................................ ..... 273 5 4 A five step sequential artist sketch of the step formation process inferred from the 3.3 s frame data.. ................................ ................................ ................................ ............ 275 5 5 First example of space stem/leaders of a natural negative stepped leader step Five consecutive 3.33 s frames are shown (about 16.7 s total). ................................ ......... 276 5 6 Second example of space stem/leaders of a natural negative stepped leader step. Five consecutive 3.33 s frames are shown (about 16.7 s total). ................................ .. 277 5 7 Space stems #1 and #2. Five consecutive 3.33 s frames ar e shown (about 16.7 s total) ................................ ................................ ................................ ................................ 279 5 8 Space stems #3 and # 4 Five consecutive 3.33 s frames ar e shown (about 16.7 s total). ................................ ................................ ................................ ................................ 280 5 9 Space stems #5 and #6. Five consecutive 3.33 s frames ar e shown (about 16.7 s total). ................................ ................................ ................................ ................................ 281 5 10 Space stems #7 and #8. Five consecutive 3.33 s frames a re shown (about 16.7 s total). ................................ ................................ ................................ ................................ 282 5 11 Space stems #9 and #10. Five consecutive 3.33 s frames ar e shown (about 16.7 s total). ................................ ................................ ................................ ................................ 283 5 12 Space stems #11 and #15. Five consecutive 3.33 s frames ar e shown (about 16.7 s total). ................................ ................................ ................................ ................................ 284 5 13 Space stem #16. Five consecutive 3.33 s frames ar e shown ( about 16.7 s total). ...... 285 6 1 LaBr 3 /PMT detector response to a CS 137 662 keV source. ................................ ........... 319 6 2 Electric field derivative (dE/dt ) and LaBr 3 energetic radiation records of three leader types preceding t riggered lightning return strokes. ................................ .......................... 320 6 3 dE/dt record s of the "chaotic" dart leader on June 21, 2010 (flash UF 10 13 ) ................ 322

PAGE 16

16 6 4 Measurement criteria for burst width, pulse width, and pulse amplitude. ....................... 323 6 5 TOA source locations for 75 pulses l ocated during the "chaotic" dart leader on June 21, 2010 (flash UF 10 13) ................................ ................................ ................................ 325 6 6 Best fit estimate for vertical velocity of the "chaotic" dart leader on June 21, 2010 (flash UF 10 13) ................................ ................................ ................................ ............... 326 6 7 dE/dt waveform measured 183 m from the lightning channel base, and channel base current waveforms of three different sensitivities (flash UF 10 13) ................................ 327 6 8 LaBr 3 energetic radiation record for the "chaotic" dart leader on June 21, 2010 (flash UF 10 13) ................................ ................................ ................................ ......................... 328 6 9 dE/dt records of the "chaotic" dart leader on August 1 3, 2010 (flash UF 10 24) ........... 329 6 10 TOA source location estimates for 45 pulses located during the "chaotic" dart leader on August 13, 2010 (flash UF 10 24) ................................ ................................ .............. 331 6 11 Best fit estimate for vertical velocity of the "chaotic" dart leader on August 13, 2010 (flash UF 10 24) ................................ ................................ ................................ ............... 332 6 12 d E/dt waveform measured 179 m from the lightning channel base and, channel base current waveforms of three different sensitivities (flash UF 10 24) ................................ 333 6 13 LaBr 3 energetic radiation record for the "chaotic" dart leader on August 13, 2010 (flash UF 10 24) ................................ ................................ ................................ ............... 334 6 14 Comparison of high speed video images recorded during a "chaotic" dart leader and a typical dart leader. ................................ ................................ ................................ ......... 335 6 15 High speed video images of even t UF 10 24 on August 13, 2010. ................................ 336 6 16 dE/dt waveform of the "chaotic" dart leader preceding the third return stroke of flash MSE 11 0 1 on July 7, 2011. ................................ ................................ ............................ 337 6 17 dE/dt and energetic radiation waveform s of the "chaotic" dart leader preceding the third return stroke of flash MSE 11 01 on July 7, 2011 ................................ .................. 338 6 18 dE/dt waveform of the "chaotic" dart leader preceding the fourth return stroke of flash MSE 11 01 on July 7, 2011. ................................ ................................ .................... 339 6 19 dE/dt and energeti c radiation waveform s of the "chaotic" dart leader preceding the fourth return stroke of flash MSE 11 01 on July 7, 2011 ................................ ................ 340 7 1 Still photographs of flashes UF 11 24 UF 11 25, and UF 11 26 o n August 5, 2011 ..... 373 7 2 Three dimensional plot of the LMA source locations for f lash UF 11 24 on August 5, 2011 ................................ ................................ ................................ ............................. 374

PAGE 17

17 7 3 Fou r projection views of the LMA source locations for flash UF 11 24 on August 5, 2011 ................................ ................................ ................................ ................................ .. 375 7 4 Three dimensional view of the LMA source locations beginning at 19:33:19.900 (UT) associated with t he initial UPL and subsequent IS branches of flash UF 11 24 on August 5, 2011. ................................ ................................ ................................ ........... 376 7 5 LMA source altitude locationsfor flash UF 11 24 on August 5, 2011 overlaid on a 910 ms window of the mea sured channel base current (II Low measurement). ............. 377 7 6 RHI scans taken by the SMART radar at the time of flash UF 11 24 on August 5, 2011 ................................ ................................ ................................ ................................ .. 37 8 7 7 Histogram (bin size equal to 30 sources) of the altitude distribution of LMA sources of flash UF 11 24 on August 5, 2011. ................................ ................................ ............. 379 7 8 Three dimensional plot of the LMA sourc e locations for fla sh UF 11 25 on August 5, 2011. ................................ ................................ ................................ ........................... 380 7 9 Four projection views of the LMA source locations for flash UF 11 25 on August 5, 2011 ................................ ................................ ................................ ................................ .. 381 7 10 Three dimensional view of the LMA source locations beginning at 19:43:34.377 (UT) associated with the initial UPL and subsequent IS branches of fla sh UF 11 25 on August 5, 2011. ................................ ................................ ................................ ........... 382 7 11 LMA source altitude locations for flash UF 11 25 on August 5, 2011 overlaid on a 590 ms window of the measured channel base curr ent (II Very Low measurement). .. 383 7 12 RHI scans taken by the SMART radar at the time of flash UF 11 25 on August 5, 2011 ................................ ................................ ................................ ................................ .. 384 7 13 Histogram (bin size equal to 30 sources) of the altitude distribution of LMA sources of flash UF 11 25 on August 5, 2011. ................................ ................................ ............. 385 7 14 LMA and HBM dE/dt TOA easting coordinates for 33 commonly located precursor current pulses during the wire ascent of fla sh UF 11 25 on August 5, 2011. .................. 386 7 15 LMA and HBM dE/dt TOA northing coordinates for 33 commonly located precursor current pulses during the wire ascent of flash UF 11 25 on August 5, 2011. .................. 387 7 16 LMA and HBM dE/dt TOA altitude coordinates for 33 commonly located precursor current pulses during the wire ascent of fla sh UF 11 25 on August 5, 2011. .................. 388 7 17 Three dimensional plot of the LMA source locations for fla sh UF 11 26 on August 5, 2011. ................................ ................................ ................................ ............................. 389

PAGE 18

18 7 18 Four projection views of the LMA source locations for flash UF 11 26 on August 5, 2011. ................................ ................................ ................................ ................................ 390 7 19 Three dimensional view of the LMA source locations beginning at 19:49:58.507 (UT) associated with the initial UPL and subsequent IS branches of flash UF 11 26 on August 5, 201 1. ................................ ................................ ................................ ........... 391 7 20 LMA source altitude locations of flash UF 11 26 overlaid on a 600 ms waveform of the measured channel base current (II Low measurement). ................................ ............ 392 7 21 A 35 ms window of the channel base current waveform (II Low measurement) shown in Figure 7 20 from 110 145 ms. ................................ ................................ ......... 393 7 22 A 75 ms window of the channel base curr ent waveform (II Low measurement) shown in Figure 7 20 from 225 300 ms during the current polarity reversal of flash UF 11 26.. ................................ ................................ ................................ ........................ 394 7 23 A 300 ms waveform s of the the electric field and the c hannel base current (II Low measurement), including the polarity reversal.. ................................ ............................... 395 7 24 RHI scans taken by the SMART radar at the time of flash UF 11 26 on August 5, 2011 ................................ ................................ ................................ ................................ .. 396 7 25 Histogram (bin size equal to 30 sources) of the altitude distribution of LMA sources of flash UF 11 26 on August 5, 2011. ................................ ................................ ............. 397 7 26 Three dim ensional plot of the LMA source locations for flash UF 11 32 on August 18, 2011.. ................................ ................................ ................................ .......................... 398 7 27 Four projection views of the LMA source locations for flash UF 11 32 on August 18, 2011 ................................ ................................ ................................ ................................ .. 399 7 28 Three dimensional view of the LMA source locations beginning at 20:37:29.870 (UT) associated with the initial UPL and subsequent IS branches of flash UF 11 32 on August 18, 2011 ................................ ................................ ................................ .......... 400 7 29 LMA source altitude locations for flash UF 11 32 on August 18, 2011 overlaid on a 200 ms window of the measured channel base current. ................................ .................. 401 7 30 Histogram (bin size equal to 30 sources) of the altitude distribution of LMA sources of flash UF 11 32 on August 18, 2011. ................................ ................................ ........... 402 7 31 Four projection views of the LMA source locations for flash UF 11 32 on August 18, 2011 ................................ ................................ ................................ ................................ .. 403 7 32 Three dimensional plot of the LMA source locations for flash UF 11 11 on June 23, 2011. ................................ ................................ ................................ ................................ 404

PAGE 19

19 7 33 Four projection views of the LMA source locations for flash UF 11 11 on June 23, 2011 ................................ ................................ ................................ ................................ .. 405 7 34 Three dimensional plot of the LMA source locations for flash UF 11 28 on Augu st 12, 2011. ................................ ................................ ................................ ........................... 406 7 35 Four projection views of the LMA source locations for flash UF 11 28 on August 12, 2011. ................................ ................................ ................................ ................................ 407 7 36 Three dimensional plot of the LMA source locations for flas h UF 11 33 on August 18, 2011. ................................ ................................ ................................ ........................... 408 7 37 Four projection views of the LMA source locations for flash UF 11 33 on August 18, 2011 ................................ ................................ ................................ ................................ .. 409 7 38 Three dimensional plot of the LMA source locations for flash UF 11 34 on August 18, 2011 ................................ ................................ ................................ ............................ 410 7 39 Four projection views of the LMA source locations for flash UF 11 34 on August 18, 2011 ................................ ................................ ................................ ................................ .. 411 7 40 Three dimensional plot of the LMA source locations for flash UF 11 35 on August 18, 2011 ................................ ................................ ................................ ............................ 412 7 41 Four projection views of the LMA source locations for flash UF 11 35 on August 18, 2011. ................................ ................................ ................................ ................................ 413 7 42 Histograms (bin size equal to 30 sources) of the alti tude distributions of LMA sources for nine triggered lightning flashes during summer 2011. ................................ 415 7 43 Cumulative LMA source altitude distribution for the nine triggered lightning flashes duri ng summer 2011 ................................ ................................ ................................ ....... 416 8 1 dE/dt TOA source locations for the dart stepped leader preceding the second stroke of triggered flash UF 11 15 on July 7, 2011. ................................ ................................ ... 476 8 2 A 26 s plot of the dE/dt measured at Station 3 in association with the dart stepped leader preceding the tenth stroke of flash UF 11 15 ................................ ........................ 477 8 3 dE/dt waveform measured at Station 3 plotted versus the II Very Low and II High channel base current for the second stroke of flash UF 11 15 ................................ ......... 479 8 4 A 3 s pl ot of the measured dE/dt at NASA Station 1 plotted against the numerical dI/dt for the tenth stroke of flash UF 11 15. ................................ ................................ .... 480 8 5 26 s electric field and dE/dt waveforms of the tenth stroke of flash UF 11 15. ............ 481 8 6 Photron high speed video frames (60 s total) of the dart stepped leader preceding the tenth stroke of flash UF 11 15. ................................ ................................ .................. 482

PAGE 20

20 8 7 60 s x ray and dE/dt waveforms during the dart stepped leader preceding the tenth stroke of flash UF 11 15. ................................ ................................ ................................ 483 8 8 dE/dt TOA source locations for the dart stepped l eader preceding the first stroke of triggered flash UF 11 25 on August 5, 2011. ................................ ................................ .. 484 8 9 A 34 s plot of the dE/dt measured at Station 3 in association with the dart stepped leader preceding the first stroke of flash UF 11 25. ................................ ........................ 485 8 10 dE/dt waveform measured at Station 3 plotted versus the II Very Low and II High channel base current for the first stroke of flash UF 11 25 ................................ ............. 487 8 11 A 3 s plot of the measured dE/dt at NASA Station 1 plotted against the numerical dI/dt for the first stroke of flash UF 11 25. ................................ ................................ ...... 488 8 12 20 s electric field and dE/dt waveforms of the first stroke of flash UF 11 25 .............. 489 8 13 dE/dt TOA source locations for the dart stepped leader preceding the first stroke of trig gered flash UF 11 32 on August 18, 2011.. ................................ ............................... 490 8 14 An 18 s plot of the dE/dt measured at Station 3 in association with the dart stepped leader preceding the second stroke of flash UF 11 32. ................................ .................... 491 8 15 dE/dt waveform measured at Station 3 plotted versus the II Very Low and II High channel base current for the second stroke of flash UF 11 32. ................................ ........ 493 8 16 A 3 s plot of the measured dE/dt at NASA Station 1 plotted against the numerical dI/dt for the second stroke of flash UF 11 32. ................................ ................................ 494 8 17 12 s electric fiel d and dE/dt waveforms of the second stroke of flash UF 11 32 .......... 495 8 18 30 s x ray and dE/dt waveforms during the dart stepped leader preceding the second stroke of flash UF 11 32. ................................ ................................ ..................... 496 8 19 Three projection views of the dE/dt and x ray source locations for the dart stepped leader preceding the second stroke of flash UF 11 32.. ................................ .................. 498 8 20 dE/dt TOA source locations for the dart stepped leader preceding the fourth stroke of triggered flash UF 11 35 on August 18, 2011. ................................ ............................... 499 8 21 An 18 s plot of the dE/dt me asured at Station 3 in association with the dart stepped leader preceding the tenth stroke of flash UF 11 35. ................................ ....................... 500 8 22 dE/dt waveform measured at Station 3 plotted versus the II Very Low and II High channel base current for the fourth stroke of flash UF 11 35. ................................ ......... 502 8 23 A 3 s plot of the measured dE/dt at NASA Station 1 plotted against the numerical dI/dt for the tenth stroke of flash UF 11 35. ................................ ................................ .... 503

PAGE 21

21 8 24 A still photograph of flash UF 11 35 taken from the Launch Control trailer. ................. 504 8 25 16 s electric f ield and dE/dt waveforms of the tenth stroke of flash UF 11 35 ............ 505 8 26 90 s x ray and dE/dt waveforms during the dart stepped leader preceding the fourth stroke of flash UF 11 35. ................................ ................................ ................................ 506 8 27 Three projection views of the dE/dt and x ray source locations for the dart stepped leader preceding the fourth stroke of flash UF 11 35.. ................................ .................... 508 8 28 dE/dt TOA source locations for the dart stepped leader preceding the second stroke of natural flash MSE 11 01 on July 7, 2011. ................................ ................................ ... 509 8 29 The easting versus altitude proj ection of the dE/dt source locations for the second stroke of flash MSE 11 01 overlaid on a still photograph taken from IS2 ...................... 510 8 30 A 35 s plot of the dE/dt measured at Station 3 in associa tion with the dart stepped leader preceding the second stroke of natural flash MSE 11 01. ................................ .... 511 8 31 35 s electric field and dE/dt waveforms of the second stroke of natural flash MSE 11 01 ................................ ................................ ................................ ................................ 513 8 32 275 s x ray and dE/dt waveforms during the dart stepped leader preceding the second stroke of natural flash MSE 11 01. ................................ ................................ ...... 514 8 33 Three projection views of the dE/dt and x ray source locations for the dart stepped leader preceding the second stroke of flash MSE 11 01. ................................ ................. 516 8 34 Distribution of x ray source altit udes for the 30 sources that were TOA located in flashes UF 11 32, UF 11 35, and MSE 11 01. ................................ ................................ 519

PAGE 22

22 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment o f the Requirements for the Degree of Doctor of Philosophy THE MECHANISMS OF LI GHTNING LEADER PROPA GATION AND GROUND ATTACHMENT By Jonathan Dustin Hill August 2012 Chair: Martin A. Uman Cochair: Vladimir A. Rakov Major: Electrical and Computer Engineerin g Lightning data were collected at the International Center for Lightning Research and Testing at Camp Blanding, Florida from 2009 to 2011. Data were obtained for 12 natural negative cloud to ground lightning discharges and 46 rocket and wire triggered lightning discharges. The mechanisms and characteristics of upward and downward le ader propagation and downward leader attachment to ground were examined using data from high speed framing cameras, electric field derivative (dE/dt) sensors, plastic and la nthanum bromide (LaBr 3 ) energetic radiation (x ray) scintillation detectors, channel base currents, a L ightning Mapping Array (LMA), and a C band dual polarimetric radar. The dE/dt and energetic radiation measurements form a 10 station time of arrival (TO A) network used to determine the locations and emission times of sources within about 75 0 m of ground. High speed video images of a stepped leader are analyzed to determine characteristics of optical phenomena, such as space stems/leaders, associated with the sequence of electrical breakdowns that occur during the formation of a leader step. Observations are compared to those obtained for dart stepped leader steps in triggered lightning. dE/dt waveforms of "chaotic" dart leaders preceding triggered and n atural lightning strokes are analyzed using TOA techniques and their characteristics are

PAGE 23

23 compared to dart and dart stepped leader processes. "Chaotic" dart leaders are found to emit copious x rays in a nearly continuous manner prior to the return stroke. The initial stage (IS) processes of nine triggered lightning discharges are mapped in three dimensions with the LMA. The geometrical and electrical properties of IS branching are determined and compared to channel base currents. The LMA and radar are us ed to examine the effects of hydrometeor structure on the propagating IS channels. Vertically propagating IS channels are observed to turn horizontal at 3 6 km altitude, near the 0 C level, often propagating for many kilometers along the tops of high ref lectivity rain shafts. The propagation characteristics and attachment processes of triggered and natural lightning dart stepped leaders are analyzed. Properties of the dE/dt pulses following the final downward leader step are discussed with respect to th e measured channel base current. Upward leader lengths, speeds, and durations are calculated. A total of 30 x ray source s are TOA located and compared spatially and temporally to corresponding dE/dt source locations

PAGE 24

24 CHAPTER 1 INTRO DUCTION AND LITERATU RE REVIEW The experimental data and analyses presented in this dissertation represent the culmination of six years of dedicated lightning research by the author. The general goals of the work are to provide advancement in knowledge of two fundamental, yet poorly understood lightning processes: 1) the propagation of lightning leaders through both virgin air and pre conditioned channels, and 2) the attachment of lightning leaders to ground or ground based objects. The mechanisms by which these processes occur are studied by analyzing the sub microsecond electromagnetic emissions of propagating lightning discharges, and when available, correlated high speed video observations of the discharge processes. Data presented in this dissertation primarily encompass two classes of measurements: 1) wideband (DC 30 MHz) electric field derivative (dE/dt) and energetic radiation (x ray) measurements of the emissions from negatively charged, descending lightning leaders, within a 500 m radius of the di scharge and within about 500 m of the ground, preceding both natural and triggered lightning return strokes, and 2) measurements of the narrowband (66 72 MHz) radiation emitted by upward propagating, positively charged leaders (UPLs) during the initial sta ge (IS) of rocket triggered lightning discharges, the measurements covering a horizontal area of about 100 km 2 and extending in altitude to about 12 km. Both classes of measurements incorporate time of arrival (TOA) techniques to map, in three dimensions, the spatial locations of the tips of propagating lightning leaders. Chapter 2 of this dissertation provides a comprehensive description of the experimental setup for data collected between 2009 2011 at the International Center for Lightning Research and Testing (ICLRT). Detailed descriptions are provided of the measurement control systems, data transmission systems, and digitization systems. The electric field derivative (dE/dt) and

PAGE 25

25 energetic radiation (x ray) measurements that provide much of the data presented in this dissertation are thoroughly described. The author was personally responsible for designing, implementing, and maintaining three separate TOA networks, each containing up to ten individual stations with both dE/dt and energetic radiation measurements. Chapter 2 also describes new photographic systems implemented at the ICLRT including high speed cameras, high definition movie cameras, and an automated network of digital still cameras. The seven station Lightning Mapping Array (LMA) inst alled at the ICLRT prior to summer 2011 is described in detail. Specific measurement setup and photographic setup parameters are given for each year of data collection. In Chapter 3, in depth discussions are provided of the operation and methodology of the TOA networks at the ICLRT. Discussions include 1) the precise determination of sensor locations, 2) the determination of fiber optic and cabling delays between the sensor output and the digital storage oscilloscope (DSO) input, 3) determination of ti me delays through photomultiplier tubes (PMT) mounted to plastic scintillation detectors, 4) correlation of DSO times bases, 5) selection of signal arrival times (both for dE/dt and energetic radiation sources), 6) determination of three dimensional source locations and emission times using a non linear least squares optimization technique, and 7) spatial location errors of the TOA networks. Chapter 4 discusses standard procedures for cataloguing and documenting collected lightning data at the ICLRT. Lis ts are provided of recorded lightning events (both natural and triggered lightning discharges) from 2009 2011. Parameters of natural lightning events such as the approximate ground strike location, peak current, and multiplicity are given based on records obtained from the National Lightning Detection Network (NLDN). Parameters of triggered lightning initial stage currents are given including the cumulative duration of the upward positive

PAGE 26

26 leader (UPL) and initial continuous current (ICC), the charge trans fer of the UPL/ICC, and the average current amplitude of the UPL/ICC. The time and characteristics of the current between the initiation of the sustained UPL and the explosion of the triggering wire (the initial current variation or "ICV") is also given f or events where the quantity can be measured. For each triggered flash, the maximum return stroke peak current and the flash multiplicity are given. Triggered lightning return stroke peak currents are compared to the preceding leader type. Chapter 4 als o provides a statistical analysis of the previously mentioned parameters for triggered lightning events between 2009 2011 with data presented in histogram format for each individual year of study and for the entire dataset. Chapter 5 discusses high spee d video observations of a natural lightning stepped leader recording during summer 2010. The leader was photographed at a frame rate of 300,000 frames/s, or an exposure time of 3.33 s. This video represents the fastest images recorded of a natural light ing discharge to date. Stepped leader parameters were measured from the video including step length, interstep interval, leader speed, and the occurrence of space stems/leaders for a total of eight channel branches. The leader step formation process was analyzed for 82 individual steps. In Chapter 6, observations are presented for four "chaotic" dart leaders recorded preceding triggered lightning return strokes during summer 2010, two of which are analyzed in significant detail. There are no previous reports in the literature of "chaotic" dart leaders associated with triggered lightning return strokes. dE/dt and energetic radiation waveforms are characterized and compared to waveforms from more well documented dart and dart stepped leaders associated with triggered lightning return strokes. The sub microsecond features of the dE/dt waveforms within 10 s of the return stroke are analyzed using the TOA technique.

PAGE 27

27 Energetic radiation waveforms within 13 s of the return stroke are analyzed and single p hoton energies are calculated. High speed video images of two triggered lightning "chaotic" dart leaders are analyzed and compared to those of dart leaders and dart stepped leaders preceding triggered lightning return strokes. Similar analyses to those d escribed above are performed for two natural "chaotic" dart leaders that preceded the third and fourth strokes of a four stroke flash that occurred during July, 2011. Observations of the initial stage processes of nine triggered lightning discharges duri ng summer 2011 are analyzed in Chapter 7 using a combination of data from a seven station Lightning Mapping Array (LMA), channel base currents, and vertical scan radar images taken with the University of Oklahoma C band dual polarimeteric SMART radar. The initial stage processes of four triggered flashes are analyzed and discussed in detail and statistics are presented for all nine flashes. The progression of the initial UPL channel and subsequent IS branches are fully analyzed including their respective path lengths and propagation speeds. The measured channel base current is examined at the times of the IS branches. LMA and radar data are examined to determine how the hydrometeor structure at ascending altitudes above the triggering site affects the pr opagation of positively charged IS channels, and in once case, a negative charged upward leader that initiated a naturally appearing bi level intracloud discharge during the IS process of a triggered lightning. In Chapter 8, the propagation characteristi cs and attachment processes of seven triggered lightning dart stepped leaders and one natural lightning dart stepped leader are analyzed. Five events are discussed in detail (four triggered and one natural) using dE/dt and x ray TOA source locations, chan nel base currents, and when available, photographic data from high speed and DSLR still cameras. The analyses primarily focus on the time period within 20 s of the return

PAGE 28

28 stroke, or within about 100 m of the ground. The dE/dt and channel base current wa veforms are time aligned and compared to determine the physical significance in regard to the attachment process of TOA located dE/dt pulses following the final downward dart stepped leader step. In each case, an u pper boundar y is imposed on the altitud e of the junction height between the upward and downward leaders from the timing comparison of the fast transition peak of the measured dE/dt and numerical derivative of the h igh level channel base current. X ray source locations are calculated for two tr iggered lightning dart stepped leaders and one natural lightning dart stepped leader. The spatial and temporal relationships of the x ray sources and the corresponding dE/dt sources are compared. The successful determination of x ray source locations as a function of source altitude is discussed. Chapter 9 contains a summary of all results and the author's recommendations for future research. The following list of peer reviewed journal papers have been published as a result of the work presented in this dissertation. Hill, J. D., M. A. Uman, and D. M. Jordan (2011), High speed video observations of a lightning stepped leader J. Geophys. Res. 116 D16117, doi:10.1029/2011JD015818. Hill, J. D., M. A. Uman, D. M. Jordan, J. R. Dwyer, and H. Rassoul (2012 ), leaders in triggered lightning: Electric fields, X rays, and source locations J. Geophys. Res. 117 D03118, doi:10.1029/2011JD016737. Hill, J. D., J. Pilkey, M. A. Uman, D. M. Jordan, W. Rison, and P. R. Krehbiel (2012), Geometrical an d electrical characteristics of the initial stage in Florida triggered lightning Geophys. Res. Lett. 39 L09807, doi:10.1029/2012GL051932. Several additional journal papers are in draft stage or are planned based on the dissertation material. The follo wing is a list of significant new ex perimental systems described in this dissertation.

PAGE 29

29 The first 10 station, three dimensional dE/dt TOA system was constructed that is capable of locating wideband sources radiated by lightning leaders with sub meter accura cy in the lateral directions and meter level accuracy in the vertical direction. The first 10 station, three dimensional energetic radiation TOA system was constructed consisting of eight plastic scintillation detectors and two lanthanum bromide (LaBr 3 ) scintillation detectors that is capable of locating energetic radiation (x ray) sources associated with propagating lightning leaders. The following is a list of significant new research contributions described in this dissertation. For 46 triggered flas hes from 2009 2011, the geometric mean (GM) of the UPL/ICC duration was 387 ms, the GM UPL/ICC charge transfer was 50 C, and the GM UPL/ICC average current amplitude was 130 A. The GM UPL/ICC duration was 27 39% larger than previously reported statistics, the GM UPL/ICC charge transfer was 64 85% larger than previously reported statistics, and the GM UPL/ICC average current amplitude was 31 35% higher than previously reported statistics. The GM triggered lightning return stroke peak current for 120 str okes preceded by dart leaders was 9.5 kA, the GM peak current for 18 strokes preceded by dart stepped leaders was 17.5 kA, and the GM peak current for 17 strokes preceded by "chaotic" dart leaders was 17.7 kA. The GM peak currents for triggered lightning return strokes associated with dart stepped and "chaotic" dart leaders were about 84% and 86% higher than those associated with dart leaders. A high speed video of a stepped leader preceding a natural first return stroke was recorded at 300,000 frames/s ( 3.33 s frame integration), the highest time resolution photographs of a stepped leader recorded to date. Step lengths and interstep intervals for 82 leader steps were found to be shorter than those reported in the literature and obtained with streak phot ographic techniques. Sixteen instances of the optical phenomena referred to as "space stems" or "space leaders" were imaged in the high speed video of the stepped leader, the first photographs of these phenomena associated with natural lightning stepped leaders. The high speed photographs demonstrate that space stems/leaders play an important role in the leader step formation process. The lengths of the space stems/leaders (average of 3.9 m) and the separations of the space/stems leaders from the tip o f the previous leader channel (average of 2.1 m) were found to be comparable to those measured for dart stepped leader steps in triggered lightning flashes. Following the step formation of 28 leader steps in the high speed video of the stepped leader, u pward propagating luminosity waves were photographed in either one, two, or three successive 3.33 s frames with average speed of 7.5 x 10 6 m/s. This characteristic

PAGE 30

30 had not previously been documented in stepped leader steps, and may have led to possible o verestimation of step length from prior streak photographic records. The first "chaotic" dart leaders associated with triggered lightning return strokes were recorded. dE/dt TOA calculations of the continuous high frequency pulses recorded within 10 s of the return stroke for two events revealed that successive source locations often "bounce" up and down over relative heights of several tens of meters. The calculated velocities between successive source locations often exceed the speed of light, indic ating the points are likely being radiated nearly simultaneously from multiple points close to the propagating leader tip. "Chaotic" dart leaders associated with triggered lightning return strokes were found to emit copious energetic radiation (x rays) up to 13 s prior to the return stroke with some single photons having energies in excess of 2 MeV. Two natural "chaotic" dart leaders exhibited similar dE/dt characteristics as those observed in association with triggered return strokes, but for consider ably longer durations (up to 100 s). Energetic radiation (x rays) was detected associated with natural "chaotic" dart leaders up to 45 s prior to the return stroke with single photon energies up to about 1.76 MeV. The first Lightning Mapping Array (L MA) VHF images of the triggered lightning initial stage process in Florida were acquired. For nine flashes, the IS transitions from vertical to horizontal propagation between 3 6 km, near the 0 C level of 4 5 km and several kilometers below the likely ce nter of the primary negative cloud charge region. The channel base current is found not to change significantly at the time of upward IS branches. LMA source locations were obtained for positive polarity precursor current pulses on the ascending trigg ering wire with amplitudes less than 10 A, in contrast to expectations found in the literature. It was demonstrated via LMA source locations that classical rocket and wire triggered lightning can initiate a more or less naturally appearing bi level intrac loud discharge, which resulted in a 57 ms duration current polarity reversal measured at the lightning channel base. The leader burst process following the final downward dart stepped leader step is found to coincide with the initial significant backgro und change in channel base current, suggesting the process is involved with the initial interaction of the streamers zones of the downward and upward leaders. Slow front pulses are found to coincide with large increases in the channel base current to near the maximum upward leader current.

PAGE 31

31 From timing comparison of the dE/dt and dI/dt fast transition peaks, it was found that the junction point between the downward and upward leaders could occur no more than 8 m above the intercepting wire for seven trigge red lightning dart stepped leaders. A total of 30 x ray source locations were determined for two triggered dart stepped leaders and one natural dart stepped leader. X ray sources were located within 30 m of the locations of the associated dE/dt pulse p eaks for the triggered events and within 40 m of the associated dE/dt pulse peaks for the natural event. The total separation between the x ray and dE/dt sources was dominated by vertical displacement, with the x ray sources occurring beneath the dE/dt so urces in at least 26 of the 30 cases. X ray sources followed the general path of the leader channel and were distributed randomly in the lateral directions about the channel. X ray sources followed the emission time of the dE/dt pulse peaks in all 30 c ases by averages of 150 ns (GM 85 ns), 290 ns (GM 170 ns), and 280 ns (GM 150 ns) for flashes UF 11 32, UF 11 35, and MSE 11 01, respectively. 1.1 Introduction Much of the data presented in this dissertation are associated with triggered lightning. Ligh tning, by its very nature, is a difficult and unpredictable natural phenomena to study. The ability to artificially initiate lightning in a reasonably controlled method, including controlling where the discharge terminates, provides researchers with the a bility to study lightning discharge processes at very close range (within 500 m) in a highly repeatable manner, a nearly impossible task with natural lightning. Prior knowledge of the discharge termination point also allows the lightning current to be mea sured at the channel base. Triggered lightning discharge processes are not a direct proxy for similar phenomena associated with natural lightning, but the processes are sufficient similar (particular when comparing triggered lightning leader/return stroke sequences to subsequent leader/return stroke sequences in natural lightning) that the knowledge obtained of leader propagation mechanisms and attachment processes for triggered lightning can be extended to natural lightning with some degree of confidence. Lightning related experiments were conducted at the International Center for Lightning Research and Testing (ICLRT), a facility jointly operated by the University of Florida and the

PAGE 32

32 Florida Institute of Technology. The ICLRT occupies a land area of ab out 1 km 2 on the Camp Blanding Army National Guard base east of Starke, FL. Since 1993, experiments have been performed at the ICLRT related to atmospheric electricity, lightning physics, and lightning protection. The majority of the published literature detailing the experimental results at the ICLRT over the past 19 years has focused on triggered lightning. Measurements of the close electromagnetic environment (channel base current, electric and magnetic fields and their time derivatives) in triggered lightning at the ICLRT have been described by Rub i nstein et al. [1995], Lalande et al. [1998], Uman et al. [2000, 2002], Crawford et al. [2001], Kodali et al. [2005], Rakov et al. [2005], Jerauld [2004, 2007], and Schoene et al. [2003a, 2009], among other s. Optical properties of dart and dart stepped leaders preceding triggered lightning return strokes are reported by Wang et al. [1999a, 1999b ] and Biagi et al. [2009, 2010], and optical properties of triggered lightning return strokes are given by Olsen e t al. [2004] and Wang et al. [2005]. Parameters of the IS process in triggered lightning are reported by Wang et al. [1999 c ], Rakov et al. [2003], Miki et al. [2005], Olsen et al. [2006], Yoshida et al. [2010, 2012], Biagi et al. [201 2 ], and Hill et al. [ 2012]. TOA measurements of dart stepped and "chaotic" dart leaders preceding triggered lightning return strokes and their subsequent attachment processes to ground are given in Howard et al. [2010] and Hill et al. [2012]. The close electromagnetic envir onment due to natural lightning discharges terminating within or very near the ICLRT network of sensors has also been studied, though not to the extent of the triggered lightning work described above. Optical and TOA measurements of stepped leader steppin g processes are given in Wang et al. [2000], Howard et al. [2008, 2010], and Hill et al. [2011]. An analysis of close (< 1 km) electric and magnetic fields and their time derivatives for stepped leaders preceding natural first strokes is given by Jerauld et al. [2008].

PAGE 33

33 In 2001, Moore et al. [2001] observed bursts of energetic radiation in the 1 2 ms prior to the return stroke in three natural negative cloud to ground flashes that terminated near South Baldy Peak at Langmuir Lab in New Mexico. The photons, some with energies in excess of 1 MeV, were detected using sodium iodide (NaI) scintillation detectors. This work prompted the installation of similar energetic radiation detectors in 2002 at the ICLRT to study the energetic radiation emission of both tr iggered and natural lightning discharges. Dwyer et al. [2003] was the first to show that negative dart and dart stepped leaders preceding triggered lightning return strokes produce large bursts of energetic radiation (x rays). Dwyer et al. [2004] showed that the x rays emitted by leaders associated with triggered lightning return strokes had typical energies of about 250 keV and that the x rays were emitted in short bursts with typical durations less than 1 s. Later, Dwyer et al. [2005] reported similar observations of energetic radiation emitted by natural stepped leaders as those initially reported by Moore et al. [2001], but also showed that the x rays were specifically related temporally to the formation of discrete stepped leader steps. The x ray b ursts associated with stepped leader steps shared similar characteristics with those recorded is association with dart and dart stepped leaders preceding triggered lightning return strokes. Since 2005, further studies of the x ray emission from negative l ightning leaders associated with both natural and triggered lightning return strokes have been conducted by Howard et al. [2008, 2010], Saleh et al. [2009], Dwyer et al. [2011], and Hill et al. [2012]. 1.2 The Global Electric Circuit and Thundercloud Ele ctrical Structure Before describing the processes that occur during natural and triggered lightning discharges, it is useful to provide a brief overview of the so called "global electric circuit" and to discuss the more macro scale effects of lightning. U man [1974] modeled the Earth atmosphere system as a lossy spherical capacitor with the boundaries consisting of the Earth's surface and the electrosphere. The electrosphere is defined as the region of the atmosphere just above 60 km

PAGE 34

34 where the free electro ns are the major contributor to the conductivity [e.g., Roble and Tzur, 1986; Reid, 1986]. In fair weather conditions, the electric field near the surface of the Earth is about +100 V/m (downward pointed electric field vector according to the atmospheric electricity sign convention), and decreases with increasing altitude [e.g., Gringel et al., 1986; Reid 1986]. If one evaluates the line integral of the electric field between the Earth's surface and the electrosphere (at 60 km), the potential difference between the electrosphere and the Earth is about 300 kV. Considering the electric field is highest at low altitude, most of the potential drop occurs within 20 km of the Earth's surface. According to the model proposed by Uman [1974], the Earth's surface is negatively charged with a total charge magnitude of about 5 x 10 5 C. An equal positive charge is distributed throughout the atmosphere, though more than 90% of the positive charge is thought to be found within 5 km of the Earth's surface [e.g., MacGor man and Rust, 1998]. Since the atmospheric medium between the two lossy capacitor boundaries is somewhat conductive, there exists a continuous leakage current of the order of 1 kA (a current density of 2 pA/m 2 ). This current, were there no charging mecha nism to replenish the neutralized charge, would fully neutralize the charge on the Earth and in the atmosphere in a time of about 10 minutes. Considering the Earth/electrosphere capacitor is observed to remain charged, there is, by necessity, a mechanism that replenishes the neutralized charge. Wilson [1920] suggested that thunderstorms (and hence, lightning processes) serve to regulate the negative charge on the Earth's surface. According to Rakov and Uman [2003], there are about 2000 thunderstorms occu rring at any one time on Earth, covering about 10% of the Earth's total surface area. These thunderstorms balance the fair weather leakage current via the transportation of negative charge to ground by cloud to ground lightning and corona current, and pos itive

PAGE 35

35 charge to ground via precipitation. This view of charge regulation on the Earth's surface is the so called "classical" view of atmospheric electricity. The primary lightning producing clouds are of the cumulonimbus type, commonly referred to as "t hunderclouds". The thundercloud electrification process is outside the scope of this dissertation, but, in general, electrification occurs in the temperature range between 0 C and 40 C where liquid water and ice particles coexist, forming a mixed phase region. Individual hydrometeors become charged and subsequently become separated from hydrometeors with opposite charge by gravity, eventually leading to large volume regions of separated, but equal polarity charges. Remote measurements of the clo ud charge structure of cumulonimbus clouds [e.g., Krehbiel, 1986] and in situ (inside the cloud) measurements [e.g., Byrne et al., 1983; Marshall et al., 1989] have shown that the basic charge structure of typical cumulonimbus clouds can be approximated as a vertical tripole structure. In Florida thunderclouds, the tripole is composed of a net positive charge region centered at about 12 km altitude, a net negative charge region centered at about 7 km altitude, and a smaller, net positive charge region at a bout 2 km altitude. The charge magnitudes of the upper positive and negative charge regions are typically about 40 C. The lower positive charge region (which may not be present at all in some thunderclouds) has an order of magnitude less charge than the upper positive charge region. More recent in situ measurements of thundercloud charge structure [e.g., Marshall and Rust, 1991; Stolzenburg et al., 1998 b ] suggest that the tripole model may be a gross simplification, and that as many as 10 separate charge layers may exist in a given thundercloud. 1.3 The Lightning Discharge The term "lightning", without further clarification, encompasses a broad spectrum of discharge processes. To the layman, the term "lightning" generally refers to the classical cloud to ground discharge associated with a thundercloud, this type of discharge being the most

PAGE 36

36 commonly encountered and viewed from the Earth's surface. Lightning discharges can occur not only associated with thunderclouds, but also in dust storms, volcanic er uptions, and nuclear explosions [e.g., Rakov and Uman, 2003]. Lightning discharges associated with thunderclouds can be further classified according to their origination and termination points. About 75% of all lightning discharges never come into contac t with the Earth's surface. These events, which are collectively referred to as "cloud discharges", can be classified as 1) intracloud discharges (lightning occurring within a single thundercloud), 2) intercloud discharges (lightning propagating between a djacent or nearby thunderclouds), or 3) air discharges (lightning occurring between a thundercloud and clear air). Cloud discharge processes have traditionally been very difficult to study and differentiate from one another via measurements from ground ba sed electric field systems (the waveforms of the three types of cloud discharges are very similar), though with the advent of three dimensional VHF mapping networks, the differences in cloud discharge processes are easier to distinguish. The remaining 25% of lightning discharges fall into the cloud to ground category. Cloud to ground lightning discharges can also be further classified into four groups by the direction of initial leader propagation and the polarity of net charge transport to ground. All f our types of discharges effectively transport cloud charge to the Earth. The four types of cloud to ground lightning discharges are shown graphically in Figure 1 1. The four classifications of cloud to ground lightning are 1) downward negative lightning, 2) downward positive lightning, 3) upward negative lightning, and 4) upward positive lightning. About 90% of all cloud to ground lightning discharges are of the downward negative type. Downward positive lightning comprises most of the remaining 10% of c loud to ground discharges. Upward negative and upward positive lightning discharges occur rarely in comparison to downward lightning discharges of either polarity, and typically only initiate from

PAGE 37

37 h igh objects (over 100 m or so) or from more reasonably si zed objects located on mountains at relatively high altitude [e.g., Rakov and Uman, 2003]. In Figure 1 1, the charge on the propagating leader and the charges in the thunderclouds above are shown. The direction of propagation of each leader is shown with a red arrow. The downward negative cloud to ground lightning discharge is the most thoroughly studied of the four types of lightning shown in Figure 1 1. The discharge process of the negative cloud to ground lightning is described in Section 1 4. The i nitial stage of natural upward negative lightning from tall structures is very similar to that of the initial stage of triggered lightning discharges. A complete description of the sequence of events that occurs in a triggered lightning discharge is provi ded in Section 1.5. 1 4 Downward Negative Cloud to Ground Lightning In Figure 1 2, the downward negative cloud to ground lightning discharge process is decomposed into a six stage sequential process. The different electrical breakdown components of the discharge are color coded according to the key at the bottom left. Time progresses from left to right. In the first stage, a negatively charged stepped leader (shown in bright green) propagates towards the lower positive charge region at the base of the cloud (assuming the tripolar thundercloud charge structure discussed in Section 1 3) and then exits the cloud and propagates towards the ground. The stepped leader travels through virgin air in a discontinuous manner, moving forward in a series of discret e "steps". The stepped leader channel typically exhibits extensive branching as it propagates towards the ground (the first stage of Figure 1 2). Optical properties of stepped leader stepping have been studied via streak photography by Schonland et al. [ 1935], Schonland [1956], Berger [1967], and Orville and Idone [1982], through the use of photodiode photographic systems [e.g., Chen et al., 1999; Krider, 1974; Lu et al., 2008], and through electric field measurements [e.g., Kitigawa, 1957; Krider and Rad da, 1975; Krider et al., 1977; Thomson, 1980; Beasley et al., 1982; Cooray and Lundquist, 1985]. A

PAGE 38

38 thorough literature review of the properties of stepped leader stepping including step length, interstep intervals, and leader propagation speed recorded in prior studies is given in Section 5 1. From prior studies, typical values for step length, interstep interval, and stepped leader propagation speed are 5 40 m, 5 50 s, and 2 x 10 5 m/s, respectively. The actual leader step formation process occurs on a time scale of about 1 s and is unresolved in previous studies. Knowledge of how negative stepped leader steps may form is garnered from long laboratory spark experiments [e.g., Gorin et al., 1976; Les Renardieres Group, 1978; Ortega et al., 1994; Reess e t al. 1995; Bazelyan and Raizer, 1998; Gallimberti et al., 2002]. A thorough review of the long spark discharge process is given in Section 5 1. As the stepped leader approaches ground (within several hundred meters) the potential difference between t he leader tip and ground (which can be some tens of megavolts [e.g., Bazelyan et al. 1978]) initiates one or more upward positively charged leaders (UPLs) from the ground or ground based objects (shown in red in the second stage of Figure 1 2). The UPLs p ropagate towards the descending stepped leader tip at a speed of the order of 10 5 m/s with a typical current of about 100 A. The UPLs deposit positive charge along their propagation paths. One or more of the UPLs may make contact with the downward propag ating stepped leader. UPLs that successfully connect with the stepped leader channel are called "upward connecting leaders". It is not uncommon for several unconnected UPLs to exist in the presence of a descending stepped leader. The connection(s) of the downward negative and upward positive leaders are commonly referred to as the attachment process. After the attachment process occurs, an initially bi directional and then upward moving, positively charged wave is initiated that travels towards the cloud at a velocity of the order of 10 8 m/s [e.g., Schonland et al. 1935; Orville and Idone, 1982; Mach and Rust, 1989]. This wave is commonly referred to as the return

PAGE 39

39 stroke, and is shown in blue in the third stage of Figure 1 2. The first return stroke has a typical peak current of about 30 kA [e.g., Rakov and Uman, 2003]. The return stroke is the most studied lightning process and is also the responsible for the visual and auditory features typically associated with lightning. The deleterious effects of lightning are also typically related to the return strokes process and at times, the continuing current that follows. The measured parameters of natural return stroke current waveforms most often cited in the literature were determined by Karl Berger and co workers in the course of their pioneering lightning research efforts [e.g., Rakov and Uman, 2003]. Berger and co workers measured direct lightning currents at the tops of two 70 m telecommunications towers on the summit of Mount San Salvatore in Lugano Switzerland. The measured current waveform parameters are reported in Berger [1955a,b, 1962, 1967a,b, 1972, 1980], Berger and Volgelsanger [1965, 1969], Berger and Garbagnati [1984] and Berger et al. [1975]. The return stroke neutralizes most of the ne gative charged deposited along the channel by the downward moving stepped leader, the net effect being the lowering of negative charge to ground. After the return stroke wave has reached the cloud, a conditioned channel between the cloud and ground remain s as shown in the fourth stage of Figure 1 2. In about 80% of negative cloud to ground lightning discharges, after a time period of often several tens of milliseconds, a second negatively charged leader follows the path of the first leader/return stroke s equence to ground [e.g., Rakov and Uman, 2003]. This process is shown in the fifth stage of Figure 1 2 in aqua. These subsequent leaders are most often dart leaders. As a result of propagation through a pre conditioned channel, dart leaders usually prop agate at a speed of the order of 10 7 m/s, about two orders of magnitude faster than the stepped leader preceding the first return stroke [e.g., Orville and Idone, 1982; Jordan et al., 1992]. At times, leaders preceding subsequent return strokes propagate in a stepped manner

PAGE 40

40 similar to stepped leaders preceding first strokes. These leaders are referred to as dart stepped leaders. Unlike stepped leaders which propagate through virgin air, dart stepped leaders do tend to follow the general path of the previ ous leader/return stroke sequence without branching. Compared to typical stepped leaders, dart stepped leaders generally propagate about an order of magnitude faster, have shorter step lengths, and have shorter interstep intervals. Schonland et al. [1956 ], using streak photographic techniques, found that the average speeds of six natural dart stepped leaders ranged from 0.5 to 1.7 x 10 6 m/s with typical step lengths of about 10 m and typical interstep intervals of about 10 s. Orville and Idone [1982] al so used streak photographic techniques to record four natural dart stepped leaders. They reported average overall propagation speeds from 2.1 to 4.6 x 10 6 m/s, typical step lengths from 10 to 20 m, and typical interstep intervals of the order of 4 to 10 s. From electric field records, Krider et al. [1977] found that dart stepped leaders have interstep intervals of 6.5 s and 7.8 s within 200 s of the return stroke in Florida and Arizona respectively. Davis [1999] used dE/dt TOA measurements to estimat e the speeds of seven natural dart stepped leaders to be 3.5 x 10 6 m/s within 1 km of the ground. Davis [1999] also reported average dart stepped leader interstep intervals to be 4.1 s near ground. In natural lightning, dart stepped leaders occur prior to the second return stroke over five times more frequently than all higher order strokes combined [e.g., Rakov and Uman, 2003]. Subsequent return strokes can also be preceded by "chaotic" dart leaders, which exhibit irregular electric field signatures wi thin the several hundred microseconds prior to the return stroke that are clearly different from the relatively smooth signatures radiated by dart leaders. A thorough literature review of "chaotic" leader processes is given in Section 6 1. "Chaotic" lea der processes have been described by Weidman [1982], Bailey et al. [1988], Willett et al.

PAGE 41

41 [1990], Rakov and Uman [1990], Davis [1999], Gomes et al. [2004], Makela et al. [2007], Lan et al. [2011], and Hill et al. [2012]. When the subsequent channel reach es the ground, a second upward moving return stroke wave is initiated as shown in the sixth stage of Figure 1 2. Subsequent return strokes have typical peak currents of the order of 10 15 kA [e.g., Rakov and Uman, 2003]. The dart leader/return stroke seq uence often repeats 2 4 times in a given flash, and can occur more than 20 times. The typical duration of the negative cloud to ground lightning discharge is several hundreds of milliseconds, but can continue for more than a second. In Figure 1 3, the s tepped leader, upward positive leader, and return stroke processes are shown in a series of three high speed photographs of a natural lightning discharge that terminated within the ICLRT measurement network on June 30, 2010 (flash MSE 10 01). The top phot ograph was taken from the Office Trailer, a distance of about 305 m northwest of the ground strike location. The middle and bottom photographs were taken from the Launch Control facility, a distance of about 220 m northeast of the ground strike location. The top photograph in Figure 1 3 shows an extensively branched stepped leader propagating towards the ground. The middle photograph shows a small section of the same descending stepped leader from a nearly orthogonal direction. The middle photograph als o shows an upward positive leader with length of about 15 m propagating from the ground in response to the stepped leader overhead. The bottom photograph shows the return stroke that resulted from the connection of the negative stepped leader and the upwa rd positive connecting leader. An additional unconnected upward positive leader is imaged in the same frame propagating from a telephone pole to the north of the strike location. Several branch components (stepped leader branches re illuminated by the re turn stroke current wave) are visible in the bottom photograph.

PAGE 42

42 Electric field waveforms from a four stroke natural flash (MSE 11 01 on July 7, 2011) that struck ground in the southwest quadrant of the ICLRT are shown in Figure 1 4. The waveforms were r ecorded by a flat plate electric field sensor located about 350 m from the ground strike points. The first return stroke was preceded by a stepped leader, the second by a dart stepped leader, and the third and fourth by "chaotic" dart leaders. The full f lash electric field waveform (800 ms) is shown at the top in Figure 1 4. Expanded views of the first three leader/return stroke sequences, labeled "A", "B", and "C" in the top plot of Figure 1 4, are shown at the bottom of Figure 1 4. From the plots of F igure 1 4, the different leader durations (and hence, propagation speeds) of stepped leaders, dart stepped leaders, and "chaotic" dart (or dart) leaders are evident. The stepped leader preceding the first stroke had duration of about 20.2 ms, the dart ste pped leader prior to the second stroke had duration of about 1.25 ms, and the "chaotic" dart leader preceding the third stroke had duration of about 215 s. Assuming the leaders originated at an altitude of 7 km and propagated in a straight line to ground the average speeds of the three leaders would have been about 3.5 x 10 5 m/s, 5.6 x 10 6 m/s, and 3.3 x 10 7 m/s, respectively. These calculated leader speeds are likely somewhat underestimated considering the normal tortuous nature of negative leader chan nels propagating between the cloud and ground. 1.5 Rocket Triggered Lightning In 1960, it was first shown that lightning could be artificially initiated by launching a small rocket trailing a thin grounded wire toward a region of high negative charge co ncentration produced by an overhead or nearby thunderstorm [e.g., Newman, 1965; Newman et al., 1967]. Rocket triggered lightning studies have since been conducted abroad in France, Japan, Brazil, and China, and in the United States at Langmuir Lab in New Mexico, at Fort McClellan in Alabama, and at the Kennedy Space Center and Camp Blanding in Florida. There are two techniques for artificially initiating lightning, 1) "classical" triggering, and 2) "altitude"

PAGE 43

43 triggering. Only the "classical" triggering technique will be discussed in this dissertation. In Figure 1 5 a six stage sequential illustration of the "classical" triggering process is shown. The different discharge processes are color coded according to the key at the bottom left and time progre sses from left to right. In the first stage of Figure 1 5, a rocket is launched trailing a thin grounded wire. At the ICLRT, rockets are launched when the quasi static electric field measured at ground falls below 5 kV/m, typically during the dissipatin g stages of convective thunderstorms when the natural lightning flash rate is about one flash per minute. The rocket accelerates to a typical peak velocity of about 150 m/s [e.g., Biagi et al., 2011 a ]. During the ascent of the triggering wire, electrical breakdown occurs at the wire tip called precursor pulses [e.g., Lalande et al ., 1998; Willett et al ., 1999 ; Biagi et al. 2011b 2012]. The precursors are essentially small sparks emanating from the top of the triggering wire that fail to evolve into a p ropagating leader channel. Precursors exhibit a damped oscillatory signature in the measured channel base current waveform, with the period of the oscillation lengthening as the wire ascends. Precursor pulses often have current amplitudes from about 10 1 50 A. After a typical time duration of 1 3 seconds, and at an average altitude between 200 400 m, an upward positively charged leader (UPL) propagates from the ascending triggering wire towards the cloud charge overhead. This process is shown in red in t he second stage of Figure 1 5. The UPL propagates upward, often in a stepped manner [e.g. Laroche et al., 1988; Idone, 1992; Biagi et al., 2011 b ]. UPL propagation speeds have been shown to increase with height from as low as about 10 4 m/s within 100 m o f ground to 3.3 x 10 6 m/s at about 1.5 km altitude [e.g., Yoshida et al., 2010]. After a normal duration of a 10 ms or so, the UPL current increases sufficiently to cause the triggering wire to explode, shown in bright green in the third stage of Figure 1 5. The explosion of the triggering wire often produces a pronounced drop in magnitude in the measured

PAGE 44

44 channel base current called the initial current variation (ICV). Sometimes, the channel base current drops to a level apparently near zero (within the resolution of the measurement system) and remains at that level for a millisecond or more. When the current fall to zero, there are frequently super imposed return stroke like pulses that occur during the zero current interval. These pulses, which were c lassified by Olsen et al. [2006] as attempted reconnection pulses (ARPs), are unsuccessful attempts to reestablish the current flow between the UPL and ground. Streak images of ARPs given by Olsen et al. [2006] reveal that a downward leader like process i nitiates from the top of the triggering wire, and subsequently produces the return stroke like current signature when it reaches ground. The ARPs have typical amplitudes of about 100 A and rise times less than 1 s. The zero current interval is interrupt ed (and current flow is reestablished between the UPL and ground) by a successful reconnection pulse (RP) that often has amplitude of a 1 kA or more. Olsen et al. [2006] classified ICVs with pronounced zero current intervals as Type I events. More freque ntly, the current measured at the triggered lightning channel base does not drop to zero during the ICV ( Olsen et al. [2006] classify this type of ICV as Type II events), but instead falls to a level of often a few tens of amperes for several hundred micro seconds. Often, the end of the ICV period in Type II events is marked by a large current pulse with amplitude from several hundred amperes to more than 1 kA. Properties of the ICV and subsequent reestablishment of current flow have also been discussed in Rakov et al. [2003]. The triggered lightning channel is typically straight and vertical below the top of the triggering wire and is tortuous above the triggering wire. Following the explosion of the triggering wire, a relatively slowly varying curren t flows for often several hundreds of milliseconds [e.g., Wang et al., 1999; Miki et al., 2005]. This slowly varying current, commonly

PAGE 45

45 referred to as the Initial Continuous Current (ICC), usually has average current amplitude of the order of 100 A and can contain superimposed current pulses with amplitudes of up to several kilo amperes. The ICC typically transfers several tens of coulombs of negative charge to ground [e.g., Wang et al., 1999; Miki et al., 2005] and can transfer up to several hundred coulo mbs of negative charge to ground [e.g., Hill et al., 2012]. The precursor current pulses, UPL, and ICC together comprise the initial stage (IS) of the triggered lightning discharge. Following the cessation of the ICC, the current at the lightning channel base usually decays to a level at or near zero for several tens of milliseconds [e.g., Rakov and Uman, 2003]. As shown in the fourth stage of Figure 1 5, a conditioned channel remains, similar to that created by a stepped leader and first return stroke s equence in negative natural cloud to ground lightning. Oftentimes, a subsequent leader will traverse the conditioned path between the cloud and ground created by explosion of the triggering wire (shown propagating downward in aqua in stage 5 of Figure 1 5 ), initiating a subsequent return stroke (shown propagating upward in blue in stage 6 of Figure 1 5). Leaders preceding triggered lightning return strokes can be dart leaders, "chaotic" dart leaders, or dart stepped leaders. Dart leaders in triggered lig htning are similar to those in natural lightning with typical propagation velocities of the order of 10 7 m/s [e.g., Idone et al., 1984 b ; Jordan et al., 1992]. Dart leaders usually exhibit a relatively smooth electric field signature without impulsive comp onents indicative of a stepping process. "Chaotic" dart leaders observed preceding triggered lightning return strokes propagate at similar speeds as typical dart leaders, but exhibit significant high frequency, large amplitude variations in the observed e lectric field signature. A thorough discussion of the "chaotic" dart leader process associated with triggered lightning return strokes is given in Chapter 6. Dart stepped leaders are also observed preceding triggered lightning return strokes, and appear to have similar characteristics as those observed prior to

PAGE 46

46 natural lightning strokes. Wang et al. [1999] used the ALPS photodiode system to measure the speed of a dart stepped leader preceding a triggered lightning return stroke to increase from 2 x 10 6 m/s to 8 x 10 6 m/s as the leader descended from 200 m to 40 m in altitude. Biagi et al. [2010] used high speed video frames to measure the speed of a dart stepped leader preceding a triggered lightning flash to be between 2.7 x 10 6 m/s and 3.1 x 10 6 m/s. Howard et al. [2010] estimated the speed of a dart stepped leader preceding a triggered lightning return stroke to be 4.8 x 10 6 m/s using dE/dt TOA locations. Finally, Idone and Orville [1984 a ] used streak photographic techniques to estimate the steps le ngths and interstep intervals of two dart stepped leaders in triggered lightning discharges to be from 5 to 10 m and 2 to 8 s respectively. Triggered lightning return strokes are similar to natural lightning subsequent strokes, with typical propagation velocities of the order of 10 8 m/s [e.g., Idone et al., 1984 b ; Willett et al., 1988] and average peak currents of 10 15 kA [e.g., Rakov and Uman, 2003]. The leader/return sequence shown in the fifth and sixth stages of Figure 1 5 often repeats 2 4 times f or triggered lightning discharges in Florida, and has been observed to occur over 20 times for triggered lightning events in New Mexico [e.g., Idone and Orville 1984]. Still photographs of two triggered lightning discharges at the ICLRT are shown in Fi gure 1 6, the first ( Figure 1 6A ) taken of a five stroke flash (UF 09 25) on June 29 2009, and the second (Figure 1 6B ) of a one stroke flash (UF 11 24) on August 5, 2011. The photographs are each six second time exposures. The image of flash UF 09 25, taken from the Launch Control trailer about 50 m from the triggered lightning channel, shows the exploded triggering wire in the center of the frame. The relatively long continuing current duration of the ICC appears on the still photograph as the explode d triggering wire being blown to the left in the frame by the wind. The five leader/return strokes sequences occur to the left of the luminosity due to the exploded

PAGE 47

47 triggering wire and also progress towards the left side of the frame with increasing time. The fifth return stroke (annotated at far left), which occurred more than 1 s following the first stroke, was preceded by a dart stepped leader and exhibited higher than average peak current (32.9 kA). The still photograph at right of flash UF 11 24 was taken from a distance of about 450 m from the triggered lightning channel. The exploded triggering wire is clearly evident as the straight section of channel (green colored due to the vaporized copper) that reaches a height of about 180 m. Above the top of the triggering wire, the UPL channel, which propagates through virgin air, exhibits considerable tortuosity as it travels towards the negative cloud charge overhead. In normal triggered lightning events, subsequent leader/return strokes sequences (one in this case) follow the general path of the UPL and the exploded triggering wire below. In Figure 1 7, example waveforms are shown of the IS of flash UF 10 13 on June 21, 2010. The plot at the top of Figure 1 7 shows a 950 ms channel base current wave form that includes the final 217 ms of the wire ascent and the full 684 ms duration of the UPL/ICC process. Expanded views are shown in the three plots at the bottom of Figure 1 7 of A) a precursor current pulse measured on the ascending triggering wire, B) the beginning of the sustained upward positive leader (UPL), and C) the ICV period. The precursor current pulse shown at the bottom left exhibits the damped oscillatory nature previously described due to the current wave bouncing up and down the ascend ing triggering wire. The polarity changes for subsequent peaks are due to the current reflection coefficient (treating the triggering wire as a transmission line) at the top of the wire (approximated as an open circuit) and at the ground (approximated as a short circuit). In this case, the resolved oscillations occur for a time of about 30 s. The beginning of the sustained UPL in the bottom middle plot of Figure 1 7 is characterized by a series of bipolar precursor like pulses (which are the initial UPL steps from

PAGE 48

48 the top of the triggering wire) that occur every 30 50 s. Over a time span of about 300 s these bipolar pulses transition to unipolar pulses as the UPL ascends farther above the top of the triggering wire, an effect likely due to the increas ed leader channel resistance as the UPL continues to extend upward [e.g., Lalande et al., 1998]. An expansion of the ICV process of flash UF 10 13 is shown at bottom left in Figure 1 7. The ICV occurred about 26 ms following the initiation of the sustai ned UPL. This flash exhibited a Type I ICV [e.g., Olsen et al., 2006] in which the channel base current magnitude dropped to a level at or near zero. In this case, the zero current interval had duration of about 784 s and contained three super imposed a ttempted reconnection pulses with current amplitudes from 87 92 A. The current was reestablished between the ascending UPL and ground by a reconnection pulse with amplitude of about 246 A.

PAGE 49

49 Figure 1 1. The four classifications of cloud to ground lig htn ing. A ) downward negative lightning, B ) upward negative lightning, C ) downward positive lightning, and D ) upward positive lightning. The direction of leader propagation is annotated with a red arrow in each case. The discharges are named for the dire ction of propagation of the leader and for the polarity of the net cloud charge transferred to ground. Adapted from Rakov and Uman [2003] and Jerauld [2007].

PAGE 50

50 Figure 1 2. Six stage sequential decomposition of the negative cloud to ground lightning di scharge. The different electrical breakdown stages are color coded according to the key at the bottom left. Time progresses from left to right.

PAGE 51

51 Figure 1 3. High speed video frames of a negative stepped leader ( A B ), an upward positive con necting leader ( B ), a first return stroke ( C ), an unconnected upward positive leader ( C ), and several branch components ( C ). The flash (MSE 10 01) occurred on June 30, 2010 and was imaged by high speed cameras in the Office Trailer and Launch Control. Ph otos courtesy of the author. A B C

PAGE 52

52 Figure 1 4. A full flash electric field waveform for natural flash MSE 11 01. The waveform was recorded about 350 m from the lightning ground strike point. Expanded views of the first three leader/return stroke sequences are shown at bottom, for A) the stepped leader preceding the first stroke, B) the dart stepped leader preceding the second stroke, and C) the "chaotic" dart leader preceding the third stroke.

PAGE 53

53 Figure 1 5. Six stage sequential decomposition of rocket triggered lightning discharge. The different electrical breakdown stages are color coded according to the key at the bottom left. Time progresses from left to right.

PAGE 54

54 A B Figure 1 6. Triggered lightning still photographs. A) photograph o f flash UF 09 25 triggered on June 29, 2009. The exploded triggering wire and the fifth leader/return stroke sequence are annotated. B) photograph of flash UF 11 24 on August 5, 2011. The exploded triggering wire, the top of the triggering wire (180 m a ltitude), and the path of the subsequent UPL are annotated. Photos courtesy of the author.

PAGE 55

55 Figure 1 7. At top, a 950 ms waveform of the initial stage (IS) of flash UF 10 13 Expanded views are provided at bottom of A) a precursor current pulse on th e ascending triggering wire, B) the beginning of the sustained upward positive leader (UPL), and C) the initial current variation (ICV) due to the explosion of the triggering wire with three attempted reconnection pulses (ARPs) and the successful reconnect ion pulse (RP).

PAGE 56

56 CHAPTER 2 DESCRIPTION OF EXPER IMENT The data presented in this dissertation were collected at the International Center for Lightning Research and Testing (ICLRT), primarily during the summers of 2009, 2010, and 2011. The ICLRT is loc ated about four miles east of Starke, FL on the southwestern corner of the Camp Blanding Army National Guard base and occupies a land area of about 1 km 2 The typical lightning measurement network at the ICLRT consists of about 80 individual sensors locat ed at 25 separate stations for measuring the electric field, magnetic field, optical, and energetic radiation emissions of both natural and triggered lightning discharges. In addition, an extensive array of high speed cameras, high definition movie camera s, and DSLR still cameras are used to image the triggered lightning channel to an altitude of about 450 m and also to capture natural lightning occurring anywhere in the immediate vicinity of the ICLRT. In this chapter, the general architecture of the ICL RT lightning measurement network pertaining to the control of individual measurements, the transmission and storage of waveform and photographic data, the triggering of various data acquisition systems, and the time stamping of acquired data will be descri bed. The physical construction and technical characteristics of the measurement systems specifically relating to the data presented in this work will be discussed in detail. The ICLRT measurement network is comprised of two primary subsets of sensors: 1 ) the Multiple Station Experiment (MSE), and 2) the Thunderstorm Energetic Radiation Array (TERA). The MSE network includes flat plate electric field and electric field derivative (dE/dt) antennas and coaxial loop magnetic field antennas. The TERA networ k includes both lead shielded and un shielded Sodium Iodide (NaI) energetic radiation detectors, plastic energetic radiation detectors, and Lanthanum Bromide (LaBr 3 ) energetic radiation detectors. A subset of the MSE and TERA networks is a 10 station time of arrival (TOA) network that contains, at

PAGE 57

57 present, 10 dE/dt sensors, 8 plastic energetic radiation detectors, 8 NaI energetic radiation detectors, and 2 LaBr 3 energetic radiation detectors. 2.1 ICLRT Measurement Network Infrastructure A perspective ae rial view of the ICLRT taken in January, 2012 is shown in Figure 2 1. The primary buildings and the two rocket launching facilities are labeled. Each of the 25 ground stations is connected to the centrally located Launch Control facility, a shielded meta l trailer, via an armored fiber optic cable containing either 4, 6, or 8 individual 62.5/125 m (core/cladding) multi mode fibers. The optical fibers serve as conduits for both measurement control (Section 2.2) and data transmission (Section 2.3) and are impervious to the lightning electromagnetic environment. Optical fiber runs at the ICLRT range from a few tens of meters to about 800 m in length. All optical fibers are laid in sections of PVC rain gutter (visible in Figure 2 1 as straight white lines c onnecting various points on the site to Launch Control) for additional protection. Waveform data transmitted to Launch Control via the multi mode fiber optic links are converted to electrical signals and subsequently digitized and temporarily stored on a n etwork of digital storage oscilloscopes (DSOs). Characteristics of the different DSOs used at the ICLRT are described in Section 2.4. A second, independent digitization system is located in the Office Trailer at the far northwest corner of the ICLRT. Da ta from a select set of measurements are digitized in the field and transmitted over 9/125 m single mode fiber optic links to the Office Trailer where the digital signals are recorded to disk. This system will henceforth be referred to as the HBM digitiz ation system and will be described in detail in Section 2.5. High speed and high definition video measurements are conducted from the Office Trailer, Launch Control, Optical Building, and from the Blast Wall observation station. Small wooden instrument ation stations (labeled IS1 IS4 in Figure 2 1) house DSLR still cameras. The

PAGE 58

58 high speed cameras, high definition cameras, and DSLR still cameras used at the ICLRT will be discussed in Sections 2.9 and 2.10. A typical measurement station (Station 5) is shown in Figure 2 2. This station contains a total of four ICLRT measurements: 1) a flat plate dE/dt sensor, 2) a flat plate electric field sensor, 3) a NaI energetic radiation detector, and 4) a plastic energetic radiation detector. The flat plate ante nnas (electric field and dE/dt) appear identical on the surface but their accompanying electronics, which are located in a shielded metal box underground, differ. The NaI energetic radiation detector is housed in the large white box and the plastic energe tic radiation detector is housed in the adjacent flat metal box. The two LaBr 3 energetic radiation detectors (not shown in Figure 2 2) are enclosed boxes similar to those housing the NaI energetic radiation detectors. A detailed description of the dE/dt measurements is given in Section 2.13 and a similar description of all energetic radiation detectors is given in Section 2.14. 2.2 ICLRT Measurement Control System With such a large number of measurements in the field, the ability to monitor and control remotely each measurement in real time becomes very necessary. Despite the best efforts of ICLRT personnel to maintain the measurement network, problems arise in the form of power failures, connectivity failures, water intrusion, and animal related cablin g damage. The measurement control system at the ICLRT has continuously evolved over the past 10 years to meet the needs of the experiments being conducted. The backbone of the control system is a central control computer (referred to as "HAL"). HAL comm unicates with a network of micro controllers placed at each individual measurement through a custom National Instruments Labview virtual instrument. The microcontrollers at each station (referred to as "PICs") have the ability to report such parameters as battery voltage and ambient temperature, generate precision calibration signals for calibrating fiber optic link gain or attenuation, switch power to external

PAGE 59

59 electronics, and attenuate incoming signals. When HAL performs a "check" of each measurement by sending a status command to the appropriate PIC, the response from the PIC provides the operator with relative assurance that the accompanying measurement is configured correctly and is operating suitably to collect high quality data. The individual comp onents of the measurement control system will be described in detail in the following sections. 2.2.1 HAL The control computer "HAL" runs on the 32 bit Windows XP Professional platform and the primary control software is the National Instruments Labview 8.5 software package. Labview is a graphical programming language that is ideally suited for instrument control. The software also allows the user to easily create custom "front panels", which are essentially Graphical User Interfaces (GUIs), for initiat ing tasks by simple button clicks and also to display charts and graphs of real time incoming data. There are two primary Labview control programs that run continuously on HAL: 1) Mercury Fields and 2) Mercury PICs and Scopes While the programs contain much of the same general architecture, Mercury Fields is used exclusively for controlling the network in an automated fashion when lightning data are being collected while Mercury PICs and Scopes is used primarily for everyday troubleshooting tasks and fo r monitoring the network status. These Labview programs were initially written by Rob Olsen III and have been significantly modified by the author. HAL communicates with the PIC controllers in the field through the use of a simple sequence of serial co mmands. Every PIC controller has a unique address (or at times multiple addresses) that serve as identifiers for communication with HAL. Prior to the middle of summer 2006, HAL broadcast its control commands and queries via a 900 MHz RF link. In 2006, t he RF control system was replaced with a fiber optic control system for better reliability. The capabilities of the PIC controllers used at the ICLRT are described in detail in Section 2.2.2.

PAGE 60

60 HAL also has the ability to communicate with all the DSOs at the ICLRT. Depending on the DSO model, the communication is over an internal Ethernet network or GPIB. The DSOs have built in software to respond to a set of commands and queries that allow their configurations to be changed remotely. The remote control of the ICLRT DSO network is described in Section 2.6. 2.2.2 PICs In Figure 2 3, a graphical representation of the serial command and receive byte structure is given. When a user prompts HAL to send a command to a PIC controller in the field, HAL genera tes a 5 byte command (Figure 2 3A). The first three bytes contain a three letter "workgroup". Prior to 2011, all PIC controllers at the ICLRT were pre programmed to respond only to the "RTL" workgroup. The purpose of the workgroup is to allow the indepe ndent control of multiple groups of PIC controllers that perhaps share the same unique PIC address. Byte 4 in the command contains the unique hexadecimal address of the PIC in the field where communication is desired. Finally, byte 5 contains the actual hexadecimal command to be executed by the designated PIC controller. In Figure 2 3B, the bit structure of the command byte described above is further annotated. The most significant bit controls the output power of the PIC controller, which may be used t o switch 12 V power to fiber optic transmitters, amplifiers, photomultiplier tubes (PMTs), and solar power control switches. Bits 6 7 control the generation of a precision amplitude (either 1 V or 0.1 V peak to peak into a 50 load impedance) 100 Hz squa re wave that is used to calibrate the gain or attenuation of the fiber optic link (deviation from the ideal unity gain) back to Launch Control. The five least significant bits control networks of onboard attenuator networks. The attenuator networks are d esigned to reduce an incoming signal by either 20 dB, 14 dB, 10 dB, 6 dB, 3dB, or any combination thereof with the characteristic

PAGE 61

61 impedance of 50 looking into either side of the resistor network. The attenuators are used when input signals are larger th an dynamic range of the fiber optic transmitters. As an example, a command sent from HAL such as 'RTLE580' would prompt PIC 'E5' to switch on its output power with no calibration wave or attenuation. The command 'RTLE5A4' would prompt PIC 'E5' to switch output power with a 1 V calibration square wave attenuated by 10 dB. In addition to a typical command instruction, a PIC controller may receive a status command. The status command is generally used to check the integrity of the communications link betw een HAL and any PIC in the field and to monitor the ambient conditions at the measurement in question. The status of any PIC controller is obtained by sending a hexadecimal '7E' in the one byte command. After a PIC controller receives and executes a comm and instruction or receives a status command, it replies with a 17 byte response. The byte structure of this response is shown graphically in Figure 2.3C. Starting from the most significant byte, bytes 1 3 contain the workgroup of the PIC in question (no rmally RTL for field measurements), bytes 4 6 contain the PIC address in decimal format, bytes 7 9 contain the executed command in decimal format, bytes 10 14 contain the supply battery voltage to millivolt accuracy, and bytes 16 17 contain the ambient tem perature at the location of the PIC in degrees Celsius. The serial byte stream is sent back to HAL (either over an RF or fiber optic link) where the Labview software parses the received serial stream and displays the information for the user to view. Wh en a user desires to send commands to multiple PIC controllers on the site, the Labview software on HAL executes the command/receive task sequentially for each PIC. If the user desires for all PIC controllers on the site to response to the same command (n ot a status query), the user can send the specified command with the one byte PIC address set to 'FF'. In this case,

PAGE 62

62 all PIC controllers in the specified workgroup will execute the desired command and will not provide a subsequent response (doing so would result in HAL receiving the status responses from all PICS nearly simultaneously). An image of the three versions of PIC controllers presently in use at the ICLRT is shown in Figure 2 4. From left to right, the devices were deployed in the field in 200 1, 2006, and 2011, respectively, and will be referred to in this document accordingly. The three versions of PIC controllers will be described in the following sections. 2.2.2.1 2001 Edition PIC The 2001 PIC controller was designed primarily by Mike Stap leton and was the first fully functional, self contained control unit placed in the field at each measurement. Each 2001 PIC controller was equipped with two 16 place switches for setting its unique hexadecimal address (00 FE). The 2001 PIC communicates strictly over 200 m plastic fiber (a low grade optical fiber capable of low data rate transmission over short distances). In the original ICLRT measurement control architecture, which was utilized through the middle of summer 2006, HAL transmitted it's c ontrol signals via a 900 MHz RF transceiver. Each measurement station was equipped with an RF PIC which received the broadcast signal and subsequently relayed the signal over plastic fiber. At stations with multiple measurements, the optical control sign al was again fanned out with a plastic fiber fan out board that replicated the control signal on the necessary number of ports. The broadcast control signal was received by every PIC controller on the site. However, only the PIC with the unique hexadecim al address included in the received signal from HAL could actually respond to the command. The 2001 PIC response to commands from HAL were transmitted back over a second plastic fiber to the RF PIC and were then broadcast. HAL subsequently parsed the PIC controller response from the RF transceiver as described in Section 2.1.

PAGE 63

63 The 2001 PIC controllers were equipped with a single channel, and hence a single PIC address. The actual printed circuit boards (PCBs) were boxed in a milled enclosure (Figure 2. 4, left) with two BNC feed through connectors for the signal input and output. The PIC input was connected directly to the sensor or to the output of amplification electronics when applicable. When the PIC was set to apply no attenuation, the signal path from the input to output BNC terminals was a short circuit. When attenuation was applied, the input signal was switched through the appropriate attenuator network(s) before being routed to the output BNC terminal. Similarly, when a command was received to output a calibration square wave, the output BNC terminal was switched to the output pin of an operational amplifier circuit that generated the precision amplitude square wave. The 2001 PIC controllers were powered with a single 12 V battery. 2.2.2.2 2006 Edition PIC Several major changes to the measurement network at the ICLRT prompted the revision of the original 2001 PIC controller. First, the RF control system had become unreliable. ICLRT staff typically spent a significant portion of each day t rying to coerce the RF PICs to function appropriately. Second, the TERA network was expanded to 24 stations in 2006, many of which contained two individual measurements (lead shielded and un shielded NaI energetic radiation detectors), and hence there was a need for a two channel controller. The 2006 edition PICs were designed by Mike Stapleton and Rob Olsen III. The most significant change to the PIC topology was the addition of a 62.5 m fiber transceiver for control purposes. Instead of the control c ommands from HAL being broadcast over an RF link, the commands were first fed to a newly designed 62.5 m optical fan out board with 24 individual output ports. The optical fan out board was placed in the rear of the Launch Control facility. With the new control system, each station had a dedicated control fiber. A 2006 PIC was placed in the TERA box at each

PAGE 64

64 measurement station. The control signal from HAL was received directly by the 2006 PIC and was then re transmitted out of the 2006 PIC over plastic fiber. Similar to the architecture used with the 2001 PICs, the signal transmitted over plastic fiber either passed to a plastic fiber fan out board (in the case of multiple 2001 PICs at a station) or was connected directly to a single 2001 PIC. The 200 6/2001 PICs were essentially arranged in a master slave configuration. The responses of any 2001 PICs at a station were transmitted through the 2006 PIC and sent to Launch Control over the dedicated 62.5 m control fiber. The upgraded control system prov ided a major improvement in the reliability of the measurement network. As previously mentioned, the 2006 PICs also incorporated a second channel primarily to support the additional NaI measurements. The 2006 PICs were not mounted in a standard enclosu re, but were instead mounted in a field replaceable unit (FRU) inside each TERA box. The FRU was essentially an aluminum housing that contained mounting brackets for the PIC controller and fiber optic transmitters in addition to power and ground buses for external electronics. Details of the TERA box construction are given in Section 2.14. Unlike the 2001 PICs, the 2006 PICs had their four BNC inputs and outputs (to support two individual channels) mounted directly to the PCB. Like the 2001 PICs, the 20 06 PICs have identical 16 place switches for setting the hexadecimal addresses of the device. The 2006 PICs two channels correspond to consecutive even and odd hexadecimal addresses (i.e., if the PIC address switch is set to '70', the Channel 1 address is '70' and the Channel 2 address is '71'). Having two channels on the PIC allows the user to switch power, calibration signals, and attenuations to the two measurements completely independently. The attenuator networks function identically on the 2001 and 2006 edition PICs, though the discrete passive elements are surface mount instead of

PAGE 65

65 wired components. The 2006 PIC's were similarly powered with a single 12 V battery, most often housed inside the TERA box. 2.2.2.3 2011 Edition PIC Through four years of use, the 2006 edition PICs performed reliably. Starting in 2010, some of the 2006 PICs began to experience random failures to respond to commands from HAL. Also, the power switching circuitry on a handful of 2006 PIC controllers began to fail. Testin g of the controllers revealed that poor solder joints and bad vias between the multi layer PCBs were in large part responsible for the difficulties. Several other issues with the 2006 edition PICs were also discovered. First, the two input channels share d a common ground plane, creating a ground loop in the situation where both channels were used. This posed a problem when the 2006 PICs were used in triggered lightning channel base current measurements when currents of the order of 30 kA were flowing in close proximity to the PIC controller. Also, the 2006 PICs could not be operated in slave mode (i.e., they could not receive control commands on their plastic fiber ports from another 2006 PIC). This posed a problem in situations where the ability to dai sy chain a series of PIC controllers would be advantageous. The idea of a somewhat simplified, yet more rugged and potentially more usage flexible PIC controller was conceived during the offseason prior to summer 2011. The 2011 edition PICs were primaril y designed by Rob Olsen III with significant input from Doug Jordan and the author. The base edition of the 2011 PIC controller was designed with the ability to control and switch power to a single measurement channel. The base controller also had the a bility to control the shutter actuations of up to two Nikon DSLR still cameras (Section 2.10). There are no onboard attenuators on the 2011 PIC controller. The decision was made to remove the attenuator networks in order to make the PIC PCB more compact and to reduce the number of mechanical components on the board (i.e., relays to switch input signals through the attenuator

PAGE 66

66 networks). At the present time, only a set of two channel base current measurements utilize any attenuation, and those attenuation values are constant. While the general control topology was kept the same on the 2011 PIC, a second 62.5 m fiber transceiver was added to the board to complement the existing 62.5 m transceiver and the plastic fiber transmitter and receiver. Unlike the 2006 PIC where control signals could only re transmitted from 62.5 m fiber to plastic fiber and vice versa, the 2011 PIC allows a control signal received on either of the 62.5 m fiber ports or the plastic fiber receive port to be re transmitted out of a ll of the other ports. This allows the 2011 PICs to be slaved off of each other over strictly glass fiber, alleviating the need to have a dedicated control fiber originating in Launch Control for each 2011 PIC. The 2011 PICs can also slave off of a 2006 edition PIC and can act as a master for any number of 2001 edition PICS. The 2011 PICs were designed with a large header port where daughter PCBs could easily be connected. Daughter boards have thus far been designed for replicating the two channel fun ctionality of the 2006 PIC and for utilizing the 2011 PIC as a controller for the field rocket launcher. The 2011 "channel" boards have the same square wave calibration sources as the 2006 PICs, but again, do not have onboard attenuator networks. The PIC addresses for the 2011 PICs are hardcoded in software instead of being set by onboard mechanical hexadecimal switches. The 2011 PICs are also hardcoded to respond to three different workgroups RTL, CMM, and CMS. The later two groups are used for contro lling DSLR still cameras and will be discussed in section 2.10. The 2011 PICs and all daughter boards were baked to withstand large temperature fluctuations, professionally populated with all surface mount components, and conformal coated.

PAGE 67

67 The devices are expected to provide an excellent level of performance and reliability for many years. 2.3 ICLRT Fiber Optic Data Transmission System Each measurement station at the ICLRT is connected to the Launch Control facility with an armored fiber optic bundle containing either 4, 6, or 8 individual 62.5 m fibers. The fiber bundles are stripped at both ends (typically about 6 m of the armor jacked is stripped at Launch Control and about 1.5 m in the field), and the individual fibers are then terminated by ICLR T personnel using a Corning terminating kit with an ST type connector. No polishing or epoxy is involved in the termination process. As discussed in Section 2.2, one of the fibers at each station is used as a control link between HAL and all PICs at the particular station. As a matter of convention, the blue jacketed fiber is typically used as the control fiber and is connected to the optical fan out board in the rear of Launch Control. The remaining fibers in each bundle are used to transmit waveform d ata from each measurement location to Launch Control. Opticomm MMV 120C fiber optic links are used for the transmission of data. An Opticomm fiber optic link consists of a transmitter/receiver pair with the transmitter placed in either a TERA box (for e nergetic radiation measurements) or in a shielded underground enclosure (for electric field, dE/dt, and magnetic field measurements) and the receiver placed in one of six 14 slot racks in the rear of Launch Control (Figure 2 5). The Opticomm fiber optic l ink has a nominal bandwidth from DC to 30 MHz and the transmission is FM modulated. The carrier frequency is 70 MHz. The optical transmission wavelength is 1310 nm. The Opticomm link has a linear dynamic range of +/ 1 V, and itactually saturates at abo ut +/ 1.2 V. The input impedance to the Opticomm tran to noise ratio is greater than 67 dB using RS 250C standards with a fiber length of 1 km. In

PAGE 68

68 practice, well terminated fibers have peak to peak noise of about 30 mV, giving a signal to noise ransmitter for impedance matching purposes. The PIC controllers at each measurement switch 12 V power to the Opticomm transmitters on command from HAL. The power inputs to the Opticomm transmitters are polarity insensitive. When storms conditions are no t present, the Opticomm transmitters are not powered. The Opticomm receiver racks in the rear of Launch Control can house up to 14 individual receiver cards. The receivers demodulate the optical FM signals from the field and output electrical signals ca pable of driving a 50 load impedance. Any DC offset introduced by the fiber link can be zeroed out for by turning an external potentiometer on the rear of the Opticomm receiver card. At times, the external potentiometer does not have the necessary rang e to compensate for the DC offset introduced by the fiber link. In this case, the back plate of the receiver card can be removed to access an internal potentiometer with sufficient ability to correct for large DC offsets. A second internal potentiometer sets the gain or attenuation of the fiber link. Ideally, the internal gain setting is configured where the 1 V peak to peak calibration square wave generated by the PIC controller at each measurement produces an identical 1 V signal on a DSO input in Laun ch Control. In practice, the PIC generated calibration waves produce signals in Launch Control ranging from about 900 mV to 1.1 V. This deviation from unity gain is accounted for in the amplitude calibration of all field measurements. 2.4 ICLRT Digital Storage Oscilloscopes A variety of different DSOs are used at the ICLRT for recording waveforms of different time duration and frequency content. The DSOs in use are manufactured by either LeCroy or

PAGE 69

69 Yokogawa. In general, LeCroy DSOs are operated in a se gmented memory mode to digitize relatively short time duration signals at high input bandwidth and sampling rate, while the Yokogawa digitizers record longer time duration signals at lower input bandwidth and sampling rate. All ICLRT DSOs save their reco rded data to internal hard disks. The data are copied to external computers after each storm using either the ScopeXplorer software (proprietary FTP software offered by LeCroy) or with a generic FTP program (typically FTP Commander). As discussed in Sect ion 2.2, all DSOs at the ICLRT are controlled by HAL over either Ethernet (LeCroy) or GPIB (Yokogawa) connectivity. When storm conditions are present at the ICLRT, all DSOs are commanded by HAL to begin an "arming" process. The details of the arming proc ess will be described in detail in Section 2.6. All DSOs in Launch Control are powered through five individual APC 3000 VA battery backup units that are mounted in the bottom section of each rack (Figure 2 5). The 62.5 m fiber optic fan out boards for m easurement control and the Opticomm receiver racks are also powered through the same battery backup units. 2.4.1 LeCroy Waverunner I (LT344) The LeCroy Waverunner I is a 4 channel, 8 bit digitizer with maximum input bandwidth of 500 MHz. During the data collection period for this work, there were four LeCroy Waverunner I DSO's installed in Launch Control (Scope 12, Scope 13, Scope 14, and Scope 17). Scope 12 and Scope 13 were used to digitize the outputs of eight unshielded NaI detectors located at eigh t of the TOA stations. Scope 14 and Scope 17 were used to correlate the time bases of other LeCroy DSOs. The details of the DSO time base correlation are given in Section 4.4. The scopes were configured to record at 250 MS/s (4 ns sampling resolution) i n segmented memory mode. The onboard memory of the LeCroy Waverunner I is 1 Mpts/channel, providing a record length of 2 ms per channel with two memory segments. The memory segments were

PAGE 70

70 triggered to record on an external trigger threshold of 1 V with 50 % pre trigger. The triggering scheme at the ICLRT is discussed in Section 2.7. The input bandwidth of each channel was limited to 25 MHz with an internal anti aliasing filter. Each channel can be configured with an input impedance of 50 or 1 M 2.4 .2 LeCroy Waverunner II (LT374) A total of two LeCroy Waverunner II DSOs were used at the ICLRT during the period of data collection for this work (Scope 20 and Scope 21), though these DSOs were removed from the network in 2009. When the DSOs were in use, their function was to digitize the outputs of eight dE/dt sensors at the then eight TOA stations. The LeCroy Waverunner II is a 4 channel, 8 bit digitizer with maximum input bandwidth of 350 MHz. The functionality of the LeCroy Waverunner II is essentia lly identical to that of the Waverunner I, but it has significantly greater memory depth per channel. The two Waverunner II DSOs were configured to record at 250 MS/s in segmented memory mode. The memory depth per channel on the Waverunner II DSO is 4 Mp ts, providing a total acquisition of eight individual memory segments of 2 ms length. Similarly, the segments were triggered to record with an external trigger threshold of 1 V with 50% pre trigger. The input bandwidth of each channel was limited to 25 M Hz with an internal anti aliasing filter. Each channel can be configured to have an input impedance of 50 or 1 M 2.4.3 LeCroy Waverunner 44 Xi At the beginning of summer 2009, a total of eight LeCroy Waverunner 44 Xi scopes were purchased. The LeCr oy Waverunner 44 Xi offered significant improvements over the Waverunner I and Waverunner II DSOs previously used at the ICLRT for recording high bandwidth signals. Similar to the previous generation of LeCroy DSOs, the Waverunner 44 Xi

PAGE 71

71 can digitize up to four channels with 8 bit amplitude resolution. The Waverunner 44 Xi runs on the Windows XP platform and offers a 10.4" full touch screen display. The DSO supports a maximum input bandwidth of 400 MHz. Most importantly, the Waverunner 44 Xi has a memory depth of 12.5 Mpts/channel. The Waverunner 44 Xi DSOs are configured to record at 250 MS/s in segmented memory mode, providing a full record of 10 memory segments at 5 ms length. The Waverunner 44 Xi DSOs were used to record the outputs of ten dE/dt sen sors (Scope 18, Scope 20, and Scope 21), the outputs of eight plastic energetic radiation detectors (Scope 28 and Scope 29), the outputs of two LaBr 3 energetic radiation detectors (Scope 18), and three channel base current measurements (Scope 26). With th e exception of Scope 26, the Waverunner 44 Xi's were triggered to record on an external trigger threshold of 1 V with 50% pre trigger. The input bandwidth of each channel was limited to 20 MHz with an internal anti aliasing filter. Each channel can be co nfigured with an input impedance of 50 or 1 M 2.4.4 Yokogawa DL716 The Yokogawa DL716 was the first generation of Yokogawa digitizer used at the ICLRT. The Yokogawa DL716 offers 16 input channels, each capable of recording continuously for 1.6 s at a sampling rate of 10 MS/s. The DL716 digitizes with 12 bit amplitude resolution. The input bandwidth for each channel is 4 MHz and the input impedance is 1 M The Yokogawa DL716 has traditionally been used to digitize electric and magnetic field wavefo rms, channel base current waveforms, and NaI energetic radiation waveforms. Recently, the two Yokogawa DL716 DSOs have been used almost exclusively to digitize NaI energetic radiation measurements. While the large memory depth of the DSO is very convenie nt, the resulting cumulative file is also large, about 530 MB. The approximate time to save the file to internal hard disk is about 14 minutes. This long save time is very problematic during the typical short

PAGE 72

72 duration Florida thunderstorms. As a result, data are often missed on both DL716 DSOs when lightning events (either natural or triggered) occur more frequently than the waveform save time. The Yokogawa DL716's are externally triggered with 50% pre trigger. 2.4.5 Yokogawa DL750 The Yokogawa DL750 is the successor of the DL716 model discussed in Section 2.4.4. Like the DL716, the DL750 supports up to 16 input channels. The DL750 digitizes with 12 bit amplitude resolution. The DL750 has somewhat greater memory depth than the DL716. Each channel is capable of recording a full 2 s record at a sampling rate of 10 MS/s. The input bandwidth of each channel is 3 MHz and the input impedance is 1 M The primary advantage of the DL750 over the DL716 is the time to save the full memory depth (about 670 Mb) is only 7 minutes. While the save time of the Yokogawa DL750 does dictate the frequency with which attempts to trigger lightning can be conducted, the time period is generally manageable. There are five DL750 DSOs installed in Launch Control (Scopes 22 25, 30). Scope 22, Scope 24, and Scope 25 digitize the channel base current measurements along with the electric and magnetic field sensor outputs. Scope 23 digitizes the outputs of a set of NaI energetic radiation detectors in addition to the outputs of two PIN photodiode optical detectors (these detectors also serve as a trigger mechanism for natural lightning events on and near the ICLRT and will be discussed in Section 2.7). Scope 30 digitizes a set of unshielded NaI detectors during the ascent of the triggering wire and subsequent upward positive leader (UPL). Scope 23 and Scope 24 trigger on the same external trigger as the network of LeCroy DSOs and record with 50% pre trigger. Scope 22, Scope 25, and Scope 30 trigger directly on the button pu sh which initiates the launching of a rocket to trigger lightning. Scope 22 and Scope 30 trigger a pre defined amount of time after the button push (usually 0.5 1.5 s) with no pre trigger. Scope 25 triggers two

PAGE 73

73 seconds following Scope 22 and Scope 30 wit h 200 ms of pre trigger. Together, Scope 22 and Scope 25 provide a continuous 3.8 s record length with 200 ms of record overlap. The trigger pulse for Scope 22, Scope 25, and Scope 30 is generated by a National Instruments DAQ card (model NI USB 6212) th at is controlled with Labview by HAL. Scope 22 and Scope 30 trigger on the rising edge of the 2 s wide pulse and Scope 25 triggers on the falling edge. The DAQ card generates the pulse on a digital output pin, and the pulse is then fed to a high current buffer circuit that is capable of driving a 50 load impedance. An image of the front racks that hold most of the DSOs in Launch Control is shown in Figure 2 6. The different DSO models discussed in the above section are annotated. 2.5 HBM Digitization System In 2009, Dr. Carlos Mata, an employee of Arctic Slope Research Corporation (ASRC, a NASA sub contractor) who received his Ph.D at UF under Dr. V.A. Rakov, brought a new digitization system to the ICLRT similar to a system that would eventually be d eployed at Pad 39B at Kennedy Space Center (KSC). The digitization system at KSC was intended to record electromagnetic measurements on and around the launch pad, in addition to the outputs of a variety of sensors that measure the local atmospheric condit ions. The system was manufactured by HBM (formerly Nicolet). At the ICLRT, the ASRC group tested the HBM system (model GEN16t Transient Recorder) by recording a variety of electromagnetic and current measurements during 2009 and 2010. Prior to Summer 20 11, Dr. Mata agreed to allow the author to utilize twelve channels of the HBM system to build a redundant dE/dt network and to install two redundant channel base current measurement. The channel base current and dE/dt sensor configurations at the ICLRT wil l be described in detail in Sections 2.12 and 2.13, but the

PAGE 74

74 digitization of these measurements will be described here and compared to that of the existing DSO digitization system. A measurement channel on the HBM digitization system consists of a model 7 600 Isolated Digitizer, a duplex 9/125 m single mode fiber link (LC type terminations), a GEN series 4 channel receiver card, and a storage medium. Annotated images of the GEN16t Transient Recorder and the 7600 Isolated Digitizer are shown in Figure 2 7. The 7600 Isolated Digitizer is placed in a shielded enclosure in the field directly at the sensor location. The digitizer is powered by a 12 V battery. In the case of the flat plate dE/dt antenna, the output of the antenna is connected to the digitizer input with a short length of RG 223 U coaxial cable. The digitizer has through resistor is placed at the input for all measurements. The front end of the digitizer is configured by the Perception software tha t controls the GEN16t Transient Recorder. Control commands are transmitted over one of the two single mode fibers that connect the digitizer to the Transient Recorder. All digitizers at the ICLRT record at 100 MS/s, the maximum possible acquisition rate of the system, with 14 bit resolution. The digitizer has a 3 dB point of 25 MHz when operated in wideband mode. The input dynamic range of the digitizer is capable of being adjusted from +/ 20 mV to +/ 100 V. The digitized output of the sensor is tra nsmitted to the Transient Recorder over the second single mode fiber. The operational wavelength is 1310 nm. The duplex single mode fiber is connected to one port of a 4 channel GEN receiver card. Each receiver card has 900 MS (1.8 GB) of onboard mem ory. The Perception software offers a wide variety of digital filters that can be applied to the received data, though no filters were used ms in duration and indiv idually triggered. The options for triggering the data acquisition are

PAGE 75

75 essentially limitless. The system can be configured to trigger on various characteristics of any input channel as well as external trigger sources. Typically, the system was configur ed to trigger signal (Section 2.7) as possible inputs. The redundant trigger sources insure that no data are missed due to a faulty measurement. The data collec ted by the GEN series receiver cards is transferred in real time to a solid state hard disk on the control computer via a 1000Base T Ethernet link. The quality of the data obtained on the HBM digitization system is substantially better than that recorde d on the ICLRT DSO network, particularly for data sampled at higher frequency. The dE/dt channels that are recorded on the HBM Transient Recorder will from this point forward be referred to as the "HBM dE/dt network". The HBM dE/dt network has six prima ry advantages over the original DSO dE/dt network, 1) as a result of the digital transmission, the recorded signals have extremely low noise (~ 2 mV peak to peak noise compared to ~ 30 mV peak to peak noise on the analog Opticomm fiber optic links), 2) the HBM digitizers have higher bandwidth (25 MHz versus 20 MHz on the Opticomm fiber optic links), 3) the HBM digitizers have far greater resolution (14 bit digitization versus 8 bit digitization on the LeCroy DSOs), 4) the front end dynamic range of the HBM Genesis digitizers can be easily controlled to provide a gain up to 50 over the original DSO dE/dt network without external amplification electronics, 5) the HBM Genesis recorder can be configured to record longer data segments than the LeCroy DSOs and can also be triggered to record selectively on any combination of the input channels, and 6) the HBM Genesis recorder automatically measures and removes cabling delays for each field digitizer every time the system is armed. Prior to Summer 2011, dE/dt wavef orms had not been recorded at the ICLRT during the ascent of the

PAGE 76

76 triggering wire, upward positive leader, and subsequent explosion of the triggering wire. As a result of the trigger flexibility and longer record length, dE/dt waveforms could be obtained o n the HBM Genesis system not only immediately surrounding return strokes, but also during the full initial stage process of triggered lightning discharges. The only limitation of the HBM digitization system compared to the DSO digitization system is that the maximum sampling rate is 100 MS/s (10 ns sampling time resolution). This was initially a concern for TOA calculations, which for the original DSO dE/dt network were based on waveforms sampled at 250 MS/s (4 ns sampling time resolution). However, TOA results indicate the lower noise and higher resolution of the HBM dE/dt network more than compensate for the increase in sampling time resolution. 2.6 Storm Detection and Data Acquisition System Arming/Disarming Process ICLRT personnel constantly monitor the atmospheric conditions within a large radius of the measurement network in order to be well prepared to collect lightning data. The output of the S band National Weather Service (NWS) Weather Surveillance Radar 1988 Doppler (WSR 88D) [e.g., Crum and Alberty, 1993] located near Jacksonville, FL, 68 km to the north northeast of the ICLRT, is viewed on the NWS website ( www.nws.noaa.gov ). The looped radar image allows ICLRT personnel to view the larger horizontal s cale structure of approaching cloud systems. The center of the 0.5 elevation radar beam from the WSR 88D is at an altitude of about 900 meters at the range of the ICLRT from the Jacksonville radar site. Lightning data are also constantly monitored via a n internet data stream of the National Lightning Detection Network (NLDN), a national network of TOA and MDF (magnetic direction finding) sensors that locates both cloud to ground and cloud lightning [e.g., Jerauld et al. 2005; Cummins et al., 2006; Cummi ns and Murphy, 2009; Nag et al. 2011]. The NLDN data are provided by Vaisala, Inc. The NLDN output is displayed on a computer in Launch Control using the LTS2005 software package. An image of the NLDN display during a storm on May 15, 2012 is shown in

PAGE 77

77 F igure 2 8. The spatial locations of lightning discharges are shown on the LTS2005 display within about 30 s of the actual time of the discharge. For each recorded lightning, the type of discharge (cloud to ground or cloud), the polarity of the discharge (positive or negative), the Latitude and Longitude of the discharge, the number of return strokes (multiplicity), and the estimated peak current (kA) is given. The time of the discharge is also recorded to 100 ms accuracy. Microsecond level timing of rec orded discharges can be requested directly from Vaisala. The combination of near real time radar and lightning data allows ICLRT personnel to be prepared to collect lightning data well in advance of approaching storms. When storms system are within abou t 20 km of the ICLRT, a network of eight Campbell Scientific CS 110 field mills are used to measured the quasi static electric field at ground level. The location of the field mills are shown overlaid on an aerial view of the ICLRT at top in Figure 2 9 an d an image of a field mill is shown at bottom in Figure 2 9. The absolute spatial locations of the eight field mills in Lat/Long are given in Table 2 1. The relative coordinates of the field mills to the ICLRT coordinate origin (located on the far southw est corner of the site) are also provided, as are the distances between the field mills and the two launching facilities. The field mills measure the vertical component of the electric field at a sampling rate of 5 Hz via a grounded reciprocating shutter that covers and uncovers a sensing plate. The amplitude of the output of the field mills provides a measure of the magnitude and polarity of the charge at altitude. While the 5 Hz sampling rate prevents the field mills from accurately reproducing transie nt lightning electric field waveforms, the total field changes (deposition and subsequent neutralization of charge) of lightning discharges are well resolved. Hence, the field mill output provides both a measure of the quasi static charge structure and a measure of the frequency of discharge activity of nearby or overhead storms. The field mills are used as the primary metric

PAGE 78

78 for knowing when to prepare to collect lightning data. During storm conditions, the quasi static electric field measured by the fi eld mills is also the primary metric for determining when to launch a rocket. At the ICLRT, an electric field of 5 kV/m (atmospheric electricity sign convention, electric field vector pointing upward towards the negative cloud charge overhead) is used as a minimum threshold for attempting to trigger lightning. Most triggered lightning attempts are conducted during the dissipating stages of convective storms when the natural lightning frequency is relatively low and but significant negative charge remains at altitude. Rockets are launched when the electric field drops below 5 kV/m following a natural discharge altitude. The outputs of the Campbell Scientif ic field mills are serialized and transmitted over 62.5 m fiber links. Each field mill requires two fibers one for instrument control and one for data transmission. The outputs of the Launch Control and Office Trailer field mills are recorded by the NLD N computer in Launch Control while the remaining six field mills are recorded by a separate computer in the Optical Building. The field mills are powered with 12 V batteries that are continuously charged with 80 W solar panels. In clear weather conditio ns, the majority of the instruments in the field are not connected to power. The PIC controllers are constantly powered to maintain communication with each measurement, but fiber optic transmitters, amplifiers, photo multiplier tubes, and other peripheral electronics are powered off. When lightning is detected within about 20 km of the ICLRT and/or when the quasi static electric field measured on the field mills diverges from the fair weather field (~100 200 V/m), the measurement network is commanded to b egin an arming Mercury Fields

PAGE 79

79 position. When the arming command is received, HAL executes a sequence of procedures that are described below in chronological order: 1) All DSOs are commanded to load a pre defined calibration panel. The panel is saved o n the internal hard at this time so that the functionality of their external buttons is turned off. The DSOs are configured to sample at 100 KS/s for 10 ms per time division for a total duration of 100 ms. The LeCroy DSOs and Yokogawa DL750 DSOs also create directories on their int ernal hard disks with the corresponding date as the directory name. 2) All PIC controllers in the field are sent a command to switch on their output power and the internally generated calibration signal (1 V peak to load). The commands are sent to the PIC controllers sequentially, the order of wh ich is byte serial command structure for each PIC controller that was described in Section 2.2.2. The actually 2 byte command to switch on the PIC power and calibration signal is a 3) When all PIC controllers have switched on their calibration signals, all DSOs are sent a of the 100 Hz calibration square wave. The LeCroy Waverunner I and Waverunner II DSOs save the files in the previously created date directory with naming convention

PAGE 80

80 sequential file save. The LeCroy 44 Xi DSOs save the calibration files in the appropriate (1 file corresponding to the SC prefix. The file numbers save the calibration files to the main hard disk (not a directory) with the naming and increment with each sequential file save. Finally, the Yokogawa DL750 DSOs save the files to the appropriate date each sequential file save. 4) All PIC controllers in the field are sent a command to switch off their calibrati on square waves, leave their output power port on, and switch on the appropriate attenuation setting (if applicable). Again, the commands are sent to the PIC controllers sequentially via a s, the command sent power port with no applied attenuation. The output power of the PIC powers the Opticomm transmitters and any peripheral electronics. For measuremen ts that are attenuated (such as low sensitivity channel base current measurements), a command such 20 dB and 16 dB attenuators, effectively reducing the amplitude of the input signal to the Opticomm transmitter by about 98.4%.

PAGE 81

81 5) After all PIC controllers have executed the previous commands, all DSOs are commanded to load their data acquisition panel. For LeCroy Waverunner I and Croy 44 Xi DSOs, the defined setup configurations suitable for digitizing lightning electromagnetic waveforms. For LeCroy DSOs, the sampling rate is set to 250 MS/s and the instrument is configure d to record in segmented memory mode for periods of either 2 ms or 5 ms. For Yokogawa DSOs, the data panels MS/s and the instruments are configured to record a single record with duration of 1.6 s waveform data when a trigger is received ( Section 2.7) and the n automatically re arm, file number (same number scheme as described above). Fo r 44 Xi models, the file name atically saves waveform data when a trigger is received and then, after saving, subsequently re arms. The file type is the file number.

PAGE 82

82 The DSOs and PIC controllers remain in the state described above throughout the duration of a storm. When storm conditions two methods. When the network is armed initially, a 20 minute timer is started. If the timer start the method to disarm). The full disarming process is described below in c hronological order: 1) All DSOs are sent a command to load their calibration panels (save file names described in Sequence 1 of the arming process). 2) All PIC controllers are sequentially commanded to turn off attenuators (if applicable), leave output power on and switch on their calibration square wave signals. This is 3) All DSOs are sent a command to save a screenshot of the calibration square wave. 4) All PIC controllers are sent a command turn off their output power and calibration square wave. Again, this is accomplished by HAL reading a text file 5) control mode, which enables the functionalities of their front panel buttons. the front panel of the Mercury Fields s the same as the normal disarming routine except that the calibration sequence (Sequence 2 and 3

PAGE 83

83 proceeding through a calibration routine, or if the network is armed and there is no lightning activity, eliminating the need for a calibration sequence. 2.7 ICLRT Triggering System At the ICLRT, many different instruments (DSO's, high speed cameras, etc.) are focused on recording the properties of the same lightning event. In order to synchronize the recordings of multiple instruments, they often need to share a common trigger source. The majority of instruments at the ICLRT are configured to record (tri gger) on either natural lightning occurring on or very near the site or on any triggered lightning discharges. To detect natural lightning terminating on or very near the ICLRT, two optical detectors (PIN photodiodes) are installed at the far northeast and southwest corners of the measurement network. These detectors are referred to as NEO (Northeast Optical) and SWO (Southwest Optical). A thorough description of the optical sensor construction and circuitry is given in Jerauld [2007]. An image of the Southwest optical detector is shown in Figure 2 10. The outputs of both optical detectors are transmitted back to Launch Control over the typical Opticomm fiber optic links discussed in Section 2.3. Both signals first pass to a comparator circuit that c hecks if the signals each have amplitude greater than 100 mV. If that condition is satisfied, a logic level pulse is generated on each channel that is wide enough to compensate for differences in propagation delays and fiber optic cabling delays between t he two input channels, ensuring that the two high going pulses will overlap. Both pulses are then fed to a two input AND gate. The output of the AND gate feeds a high current buffer that is capable of driving a 50 load impedance (amplitude of 3.3 V).

PAGE 84

84 A second trigger source is the measured current at the base of the triggered lightning channel. As discussed in Section 2.4.3, three of the channel base current measurements are digitized on Scope 26, a LeCroy Waverunner 44 Xi. Scope 26 is configured to trigger directly on the measured current from the least sensitive measurement (II High). This particular measurement is set to record current amplitudes up to about 63 kA. The triggering criteria on Scope 26 is set to a window trigger centered at grou nd with upper and lower boundaries of +/ 100 mV. This allows the DSO to trigger on either polarity current with a trigger threshold magnitude of about 6 kA. Like the other Waverunner 44 Xi DSOs, Scope 26 is configured to record in segmented memory mode (10 segments at 5 ms length). The LeCroy 44 Xi also has the ability to output a trigger pulse anytime the input trigger criteria is met. Hence, every time a current pulse arrives at the scope input with magnitude greater than 6 kA, an output trigger puls e is generated assuming the prior segment has finished recording. The amplitude of the output trigger pulse is 400 mV into a 50 load impedance. The output pulses of both the AND gate that is triggered by the optical detectors and Scope 26 that is trigge red by the channel base current are subsequently routed to the ICLRT master trigger box (Figure 2.11). The master trigger box has a total of 16 individual trigger inputs, each with a selectable trigger threshold of 1.53 V or 0.256 V, and a selectable trig ger polarity. For all experiments discussed in this document, only the two inputs described above were used. If the amplitude criteria are met at either of the trigger inputs, one shot circuits generate logic level pulses that are then fed to an OR gate. If any input on the OR gate is true (active high), the master trigger box outputs a 100 s pulse on all 16 of its output channels. The output channels are capable of driving 50 load impedances. The 16 trigger outputs are used to drive the external t rigger inputs of all LeCroy and Yokogawa DSOs in Launch Control with the exception

PAGE 85

85 of Scope 22, Scope 25, and Scope 30 ( Section 2.4.5). The trigger signal is also routed to two S.I. Tech Model 2817 TTL Bit Driver transmitters. One of these drivers is co nnected to the Office Trailer and the other is connected to the Optical Building, both with single mode fibers. A corresponding S.I. Tech Model 2817 TTL Bit Driver receiver is mounted in each location. This connectivity allows high speed cameras or other independent instruments in either location to trigger on the same source as the primary ICLRT data acquisition network. The ICLRT master trigger box was designed by Mike Stapleton and Dr. Douglas Jordan and was installed at the beginning of summer 2009. Under normal conditions, the first several return strokes in cloud to ground natural lightning will be sufficiently bright to trigger the optical detectors. The explosion of the triggering wire and subsequent ICC process during triggered lightning disc harges is typically insufficiently bright to trigger the optical detectors. Triggered lightning return strokes often meet the trigger criteria both optically and in peak current magnitude (6 kA). 2.8 ICLRT GPS Time System Two independent GPS time receiv er cards are installed at the ICLRT, one in Launch Control and one in the Office Trailer. The GPS time receiver cards are manufactured by Symmetricom and are model bc637PCI U. The receiver card in the Office Trailer is mounted in the computer that contro ls the Photron high speed camera and the card in Launch Control is mounted in the NLDN computer immediately next to HAL. The cards receive the GPS time signals from external antennas (Symmetricom model number 812573 50) mounted above the roof lines of bot h buildings. The receiver cards operate in Synchronized Generator mode and according to the manufacturer, are capable of supplying precise timing with only 100's of nanoseconds of error over the course of thousands of years. The receiver card is equipped with time code outputs for providing GPS time to peripheral instruments. The high speed cameras at

PAGE 86

86 the ICLRT (Photron and Phantom models) are connected to the analog IRIG B time code format from the receiver cards. The receiver cards also have an "event input, which takes in a logic level pulse and produces an interrupt that can be read by the host computer. The event input of the receiver card in Launch Control is connected through a logic OR gate to one output of the master trigger box described in S ection 2.7. A custom C++ program written by Dr. Douglas Jordan recognizes the interrupt generated by the receiver card and writes the corresponding GPS time (to the nearest 1 s) to a text file. Time stamps are thus generated coincident with any return s troke in a natural or triggered lightning flash that satisfies the trigger criteria described in Section 2.7. The other port of the logic OR gate that drives the event input on the receiver card is connected to the output of a one shot circuit that produc es at 50 s pulse. The one shot circuit receives the 2 s pulse from the DAQ card that triggers Scope 22, Scope 25, and Scope 30 a specified time after the button is pushed to fire a rocket. This arrangement provides a GPS time stamp coincident with the b eginning of the record on Scope 22 and Scope 30. 2.9 ICLRT High Speed Cameras High speed cameras have been used at the ICLRT since summer 2008. In 2009, we purchased a Photron SA1.1 and were also loaned a Phantom V7.3 from NASA. In 2010, a set of two P hantom V310 high speed cameras were also loaned from NASA to complement the existing instruments. In 2011, a Cordin high speed camera was installed in the Optical Building. Though no data from this camera will be presented in this dissertation, a brief d iscussion of the camera and its capabilities is given in the section below. The high speed photographic data presented in this dissertation were primarily acquired with the Photron SA1.1. Images of the Photron SA1.1, Phantom V7.3, and Cordin 550 high spe ed cameras are shown in Figure 2 12. The Photron has a CMOS sensor, similar to those in high end DSLR still cameras, with a 1024 x 1024 pixel array. The pixels are 20 m square. Images are recorded with 12 bit gray

PAGE 87

87 scale amplitude resolution. The pixel area decreases with increasing frame rate due to data read out constraints from the sensor. The camera is capable of recording images at frame rates up to 1 Mframe/sec (1 s resolution), but in that case the recorded pixel region is too small to be of val ue for lightning measurements. The camera is mounted on its side on a tripod in the Office Trailer, a distance of 430 m from the Tower Launcher and 300 m from the Field Launcher. The Photron is capable of reading out horizontally oriented pixel arrays fa ster than vertical pixel arrays (i.e., a vertical by horizontal pixel area of 64 x 128 can be recorded at a frame rate of 300 kilo frames per second (kfps) while a vertical by horizontal pixel area of 128 x 64 can only be recorded at 108 kfps), hence the s ide mounted configuration. The high speed images presented in this dissertation were recorded in 2010 2011. During those summers, the Photron was operated primarily at a frame rate of 300,000 kfps, providing images of 320 x 32 pixels (vertical x horizont al). The Photron is equipped with a Nikon F mount lens mounting system. The camera does not have the ability to electronically control the lens aperture, so only Nikkor AF D model lenses can be used (with a manual aperture control ring). From 2010 2011, the camera was operated with either a 14 mm f/2.8D or 20 mm f/2.8D fixed focal length lens. The wide angle lenses afford the greatest total spatial area (primarily altitude) within the small pixel region, though the spatial resolution is obviously poorer With a 20 mm lens, the spatial resolution of the Photron camera was 0.43 m/pixel at the Tower Launcher (430 m) and 0.3 m/pixel at the Field Launcher (300 m). The total viewing area (vertical x horizontal) was 138 m x 14 m at the Tower Launcher and 96 m x 10 m at the Field launcher. With a 14 mm lens, the spatial resolution was 0.61 m/pixel at the Tower Launcher and 0.43 m/pixel at the Field Launcher. The total viewing area was 195 m x 20 at the Tower Launcher and 138 m x 14 m at the Field launcher.

PAGE 88

88 The Photron SA1.1 typically operated in a partitioned memory mode with three individually triggered memory segments. At the normal operating frame rate of 300 kfps, a memory segment had time duration of about 2.485 s with a total of 745,517 frames. The P hotron was triggered on the output of the master trigger box (Section 2.7), which for typical triggered lightning events corresponded to the first return stroke following the ICC period. The camera was configured to record with 50% pre trigger. By operat ing the camera with partitioned memory, the image files did not have to be saved until all three segments were triggered. Despite 1000Base T Ethernet connectivity, a single memory partition, which occupies of the order of 15 GB of disk space, can take an impractical amount of time to transfer to the host computer during storm conditions. Image files were saved with the MRAW file format, which is essentially an archive file format of raw image data. Camera configuration and file saving was handled through the proprietary Photron FASTCAM software. The Phantom V7.3 is a lower performance camera than the Photron SA1.1, and was used primarily to view a wide field, vertically oriented image of the full triggered lightning channel. During normal triggered lig htning events, the camera captured the full ascent of the rocket, the explosion of the triggering wire, the full ICC process, and all subsequent leader/return stroke sequences. The Phantom V7.3 has a 800 x 600 pixel CMOS sensor. The images are recorded w ith 14 bit grayscale amplitude resolution. The pixels are 22 m square. The Phantom usually operated from either the Office Trailer (2009, 2010) or from the Blast Wall observation station (2011). In 2009 and 2010, camera configuration and file saving wa s handled from a dedicated laptop provided by NASA. The camera connected directly to the laptop with a 100Base T Ethernet connection. In 2011, the camera was connected to a Transition Networks SISTG1040 111 media converter. The device allows a 1000Base T Ethernet connection to be routed over a

PAGE 89

89 single mode fiber pair. The fiber pair terminated in the Optical Building at a second media converter, the output of which was connected to a control computer (the same computer that records the outputs of six of the Campbell Scientific field mills). The Phantom V7.3 was typically operated at a frame rate of 8 kfps, or a frame interval time of 125 s (exposure time was actually 120 s). The pixel region was typically set to 800 x 150 (vertical x horizontal). Dur ing 2009 2010, when the camera was connected directly to an IRIG B time code input, the camera was triggered manually with a push button switch closure. In 2011, when the camera operated remotely at the Blast Wall observation station with no time code inp ut, the trigger was provided by a S.I. Tech Model 2817 TTL Bit Driver connected to a single mode fiber link. The 2 s trigger pulse from the DAQ card in Launch Control that triggered Scope 22, Scope 25, and Scope 30 (and generated a GPS time stamp as discu ssed in Section 2.8) was also routed to the input of the bit driver. The Phantom V7.3 was configured to trigger on the rising edge of the pulse, and recorded for a total time of about 4.357 s (34,861 frames). The camera was configured to record with 10% pre trigger. Video files, each with disk space of about 8 GB, were downloaded after each trigger. The save time was less than that of the Yokogawa DL750 DSOs, and thus was not a limiting factor in the time duration between attempts to trigger lightning. The Phantom V7.3 was powered by 24 V DC obtained from two 55 Ah Optima marine batteries connected in series. The batteries were charged with the outputs of two 30 W solar cells. During summer 2010, an additional three Phantom V310 high speed cameras we re loaned to the ICLRT by NASA. The cameras were intended to be tested on natural and triggered lightning at the ICLRT, and were to be eventually installed atop the lightning protection towers surrounding launch pad 39B at Kennedy Space Center. The Phant om V310 cameras were

PAGE 90

90 configured to record wide angle images of the site with one camera located in the Office Trailer and the other in Launch Control. Unlike the existing Photron and Phantom cameras at the ICLRT, the Phantom V310 cameras were equipped wit h full color 8 bit sensors. The camera sensors were 1280 x 800 pixels (horizontal x vertical) with a pixel size of 20 m square. The Phantom V310 in the Office Trailer was configured with a pixel region of 1280 x 600 (horizontal x vertical), a 24 85 mm N ikkor lens set to a focal length of 24 mm, and a fixed aperture of f/16. The camera recorded at a frame rate of 4.3 kfps or a framing interval of 232.57 s (actual exposure time of 230.61 s). The camera acquired a total record length of 846.7 ms (3641 f rames) with 40% pre trigger and was triggered on the output of the master trigger box, typically on the first return stroke. The Phantom V310 in Launch Control was configured with a larger pixel region of 1280 x 800 (horizontal x vertical), a 24 85 mm Nik kor lens set to a focal length of 24 mm, and a fixed aperture of f/16. The camera recorded at a frame rate of 3.2 kfps or a framing interval of 312.5 s (actual exposure time of 310.55 s). The camera acquired a total record length of 855.31 ms (2737 fr ames) with 30% pre trigger and was triggered on the output of the master trigger box, typically on the first return stroke. Both Phantom V310 cameras were typically configured with two memory partitions of the above stated lengths. Camera configuration a nd file saving for all Phantom high speed cameras is handled through the Phantom Camera Control software package. Prior to summer 2011, a new Cordin high speed camera was installed in the Optical Building for imaging the bottom 100 m of the leader phase and the attachment region of triggered lightning discharges. The Cordin camera was configured with a 50 mm lens set to an aperture of f/2.8, though the internal optics of the camera limit the system aperture to f/5.6. The Cordin camera has 64 individual CMOS sensors with dimensions of 1000 x 1000 pixels. The

PAGE 91

91 pixel size is 7.4 m square. The images are digitized with 10 bit grayscale amplitude resolution. The image is projected onto the sensors by a rotating mirror assembly that is spun either by an el ectrically driven or helium driven motor. The maximum possible frame rate from the electrically driven system is about 800 kfps and the maximum possible frame rate from the helium driven system is about 4 Mfps. During Summer 2011, only the electrically d riven system was used. The camera was always triggered on the first return stroke following the ICC of a triggered lightning discharge and recorded the first leader, generally a dart leader. In future experiments, the camera will be triggered to record d art stepped leaders or "chaotic" dart leaders preceding first or subsequent triggered lightning return strokes. External circuitry is currently being designed and tested for this purpose. A block diagram of the full ICLRT trigger system is given in Figu re 2.13. The system triggers for all DSOs (Section 2.4), the GPS time card (Section 2.8), and the high speed cameras (Section 2.9) are annotated. The blocks are color coded according to the key at the bottom of Figure 2.13. 2.10 ICLRT Still Cameras and HD Video Cameras In addition to the Photron, Phantom, and Cordin high speed cameras, the ICLRT is also equipped with a network of digital SLR (DSLR) still cameras and HD video cameras for imaging both natural and triggered lightning. SLR 35 mm film camer as have traditionally been used at the ICLRT to shoot time exposures of the triggered lightning channel. Details of channel geometry can be seen in still photographs that are poorly resolved in various types of video records. Starting in 2009, the 35 mm film cameras were phased out in favor of DSLR cameras. In 2011, a fully automated network of eight Nikon D5000 DSLR cameras was installed in the field at four locations (Blast Wall observation station, Southwest Optical observation station, IS2, and IS4). The D5000 has a 12.3 megapixel CMOS sensor. Each location housed a site

PAGE 92

92 coverage camera and a triggered lightning camera. Site coverage cameras were mounted with horizontal orientation on Manfrotto ball heads and were outfitted with Sigma 10 20 mm lens es set to 10 mm, providing a 109 degree field of view. The location of each camera station and the corresponding field of view of the site coverage camera are shown in Figure 2 14 A and an image of the camera installation is shown in Figure 2 14 B Each site coverage camera was outfitted with a B&W 6 stop neutral density filter and a B&W slim mount circular polarizing filter. The circular polarizer effectively reduces the light input by about 2 stops. Together, the neutral density and circular polarizer reduce the light reaching the camera sensor by a factor of about 64. The site coverage cameras were operated in manual focus mode and manual exposure mode with the aperture of each camera set to f/18 and the shutter set to bulb mode. For optimal long ex posure noise performance, the camera sensitivities were set to ISO 200, which is base sensitivity on the D5000. The remote shutter port of each site coverage camera was connected to an output port on a 2011 PIC controller (Section 2.2.2.3). The function of the output port is to provide a contact closure, which in turn triggers the camera's shutter mechanism. The contact closure is controlled by a custom Labview virtual instrument written by the author. The Labview program sets the exposure length in sec onds for each site coverage camera by sending the appropriate hexadecimal sequence in the last byte of the 5 byte command (Section 2.2.2). The workgroup (the first three bytes of the 5 byte command) are set to "CMM" for the site coverage cameras. Using t he different workgroup, the 2011 PIC controller can interpret the last two bytes of the command sequence as an exposure time and not a typical command for switching on calibration signals, attenuator networks, etc. After the exposure lengths are set on al l of the site coverage cameras, a typical power command (hexadecimal '80') is sent to initiate the exposures. The cameras take continuous exposures at the defined exposure length until the PIC controllers

PAGE 93

93 receive an "off" command (hexadecimal '00'). The exposure lengths can be changed at any time to compensate for changing light conditions. During typical storm conditions, exposure lengths are from 5 10 s. As seen in Figure 2.14 A the four site coverage cameras have overlapping fields of view, and often record the same natural lightning discharge from multiple angles, providing a good three dimensional view of the tortuous channel geometry. The triggered lightning channel is within the field of view of all site coverage cameras. Time exposures of natur al lightning discharges during daytime hours usually show the main channel sections in good detail, but often do not resolve branches, which are typically significantly less bright. This is a consequence of the neutral density filters, though without the filters, long exposure times are impossible when there is significant sunlight. Filters are manually removed when nighttime storms are present. The second camera at each camera station functioned as a dedicated triggered lightning camera. These cameras were mounted on Manfrotto ball heads with vertical orientation. The cameras farthest from the field launcher (Blast Wall and Southwest Optical) were outfitted with Nikkor 18 105 mm lenses set to focal lengths of 18 mm. The additional two cameras in IS2 and IS4 were outfitted with Nikkor 55 200 mm lenses set to focal lengths of 90 mm and 86 mm, respectively. The triggered lightning cameras also operated in manual focus and manual exposure modes, with the aperture set to f/22, the shutter set to bulb mode and the sensitivity set to ISO 200. The triggered lightning cameras were mounted with the same filters as the site coverage cameras. The remote shutter port on each camera was connected to the second output port on the 2011 PIC controller. The exposur e length for each camera was set by the Labview program as described for the site coverage cameras, though the workgroup was replaced with "CMS" such that the exposure lengths for the two sets of cameras could be controlled independently. The triggered li ghtning cameras were configured to fire at the same time by

PAGE 94

94 sending a hexadecimal 'FF' in the last two bytes of the command sequence. The fire command was sent at the same time as the trigger pulse for Scope 22, Scope 25, Scope 30, and the Phantom V7.3. A fifth triggered lightning camera, a Nikon D80 belonging to the author, was operated from Launch Control. The camera was outfitted with a Nikkor 70 300 mm lens set to a focal length of either 70 mm for the tower launching facility or 100 mm for the field launching facility. The camera was configured in manual focus and manual exposure mode with an aperture of f/32, an exposure length of 6 seconds, and a sensitivity of ISO 100. A 4 stop neutral density filter and a circular polarizer were mounted to the lens. The shutter was manually triggered when the rocket exited the launch tube. The network of eight D5000 cameras were powered with Nikon EH 5A AC adapters mated to Nikon EP 5 battery adapters. The cameras in IS2 and IS4 were connected directly to s hore power via APC 750 VA battery backup units while the cameras at the Blast Wall and Southwest Optical stations were powered by 500 W inverters that were connected to 55 Ah Optima marine batteries. The batteries were each charged with 12 W solar cells. The Nikon D80 in Launch Control was powered by the internal camera battery. All cameras saved their image files directly to 16 GB SDHC cards. Normal card capacity was about 4,000 images. The memory cards were collected after each storm so that the imag e files could be copied to hard disks. High definition (HD) video recorders are also used extensively at the ICLRT. In 2009, a network of four Sony HDR HC5 cameras was used to image the triggered lightning channel from four locations (IS2, Office Traile r, Launch Control, Runway). The cameras record at 30 frames/s in true 1080i resolution directly to mini DV tapes. The camera has a 2.1 megapixel CMOS sensor. The record time of each tape is approximately one hour. After a storm, the

PAGE 95

95 videos correspondin g to the time window around each rocket launch were saved into .AVI files using the free VirtualDub software package. In 2010, the Sony HDR HC5 recorders were replaced in favor of Canon HF S20 video cameras. The HF S20 records in 1920 x 1080 resolution a t 30 frames/s. The HF S20 has a 8.59 megapixel CMOS sensor. The primary advantage of the HF S20 is that it can record to both an internal 32 GB flash memory and also has two external slots for up to 64 GB of additional memory space. A 32 GB class 6 SDHC card was purchased for each camera and the cameras were set to record to the external SDHC card for each data transfer. In FXP recording mode (one step down from best quality), the camera can record over 4 hours of continuous HD video on a 32 GB memory c ard. The two Canon HF S20 cameras were placed in Launch Control and the Office Trailer, both focused on the triggered lightning channel. An additional two Canon HF S21 video recorders were purchased prior to summer 2011. The HF S21 is the successor of t he HF 20 and offers more internal memory (64 GB vs. 32 GB). The two HF S21 cameras were also configured to record to 32 GB class 6 external SDHC cards. Both cameras were located in the Office Trailer and were used for general site coverage. One camera w as mounted to view due south and the other to view east. The raw videos recorded by the Canon HD cameras are in the AVCHD format with the proprietary Canon .MTS file extension. In order to play the videos on standard media software, they are converted to HD .MOV or HD .AVI files using the free Emicsoft MTS converter. 2.11 ICLRT Rockets and Rocket Launching System An image of a rocket used to trigger lightning at the ICLRT is shown in Figure 2.15. The rockets are 1 m long fiberglass tubes with diameter of 2". The rocket includes a removable nose cone that is ejected when the parachute deploys following the full burn of the rocket motor. Four fiberglass fins are mounted to the sides of the rocket near the base of the tube to help the rocket track straig htly under the influence of high winds aloft. An additional set of metal "wings" are

PAGE 96

96 mounted to the mid section of the rocket body to stabilize the rocket while it exits the launch tube. The wings are bent such that the rocket fits snuggly inside the lau nch tube. An H series rocket motor is inserted into the bottom of the fiberglass tube and secured with a metal ring clamp. A spool of 32 AWG kevlar coated copper wire is mounted to the bottom of the rocket body with heavy gauge coated wire. The wire spo ol has total length of 700 m and total resistance of about 350 The wire wraps on the spools are glued in a precise manner such that only one wrap at a time peels off the spool as the rockets ascends. The rockets reach a peak velocity of about 150 m/s. During the data collection period for this dissertation, two different rocket launching system were utilized. In 2009, the tower launching facility was used exclusively. From June 5, 2010 through July 11, 2010, rocket launching operations were conduct ed from the field launching facility. Rocket launching operations were moved back to the Tower Launcher from July 11, 2010 through May 14, 2011. From June 23, 2011 to present, the field launcher has been used exclusively. 2.11.1 Tower Launching Facilit y An image of the tower launching facility is shown at left in Figure 2 16 A The tower is located about 50 m south of Launch Control. The tower deck is 11 m above local ground level. An aluminum rocket launcher with 12 individual launch tubes is mounted to the tower deck. The launcher is supported by tubular fiberglass legs for electrical insulation from the supporting wooden tower. The top of the launch tubes are approximately 14 m above local ground level. The launcher control box in the field is lo cated on a vertical support beam between the launch tubes. The control box houses a Motorola HC6811 micro controller and a bank of 12 pneumatic switches. The micro controller communicates with the rocket launcher control unit (Figure 2

PAGE 97

97 16 B ), located imme diately next to the NLDN computer in Launch Control, over a 200 m fiber link. The air supply to the bank of pneumatic switches is a compressor located in the NASA trailer immediately to the east of Launch Control. The compressor charges a large storage tank to approximately 75 psi. The storage tank is connected to the bank of pneumatic switches with 1/4" pneumatic tubing. The launcher control box in Launch Control (Figure 2 16 C ) has the ability to select a tube, arm or disarm the tube, and send a comma nd to fire the tube. When the command is sent to fire a tube, the appropriate pneumatic switch in the master launcher control box opens its valve and allows air pressure to flow to the individual control box mounted to the particular tube (Figure 2 16 D ). The individual tube control boxes house a second pneumatic switch, a 9 V battery, electrical ports where the rocket squib wire is attached, and moveable pins to configure the launcher in test mode or firing mode. When the switch in the particular tube co ntrol box receives air pressure from the master control box, the switch closes, putting 9 V across the rocket squib wire. The resulting current flow detonates the small charge at the end of the squib wire, igniting the rocket motor. The end of the spool of kevlar coated copper wire attached to the rocket is connected to the aluminum launcher with a ring terminal and stainless steel bolt. The launcher itself is grounded through a long section of braided wire that passes through an opening in the tower dec k below the launcher to the ground below, where it is connected to a single 25 m copper clad steel ground rod. Any current flowing on the ascending triggering wire, and all current from subsequent return strokes, is directly measured before passing to gro und using a T&M Research 1 m current viewing resistor (CVR). Subsequent return strokes typically attached directly to the top of the rocket tubes. The current measurements are located in the large box mounted to the underside of the launcher and will be discussed in detail in Section 2.12.

PAGE 98

98 2.11.2 Field (Ground) Launching Facility (2010) Rocket launching operations were moved from the tower launching facility to a select area on the ground prior to summer 2010. The move was prompted by a grant from NAS A to construct a 1/20th scale model of the newly installed lightning protection system at Pad 39B at Kennedy Space Center. The field (ground) launching facility is located about 140 m to the west northwest of the tower launching facility. The field launc hing facility was chosen to be centrally located within the measurement network and to provide good optical viewing angles from multiple buildings on the site. An image of the 2010 version of the field launching facility is shown in Figure 2 17. The fiel d launcher was built with six aluminum rocket tubes and tubular fiberglass leg supports. A current measurement box housing a T&M Research R 5600 8 1.25 m CVR was mounted to the underside of the launcher. The launcher was grounded through the current me asurement box to a set of three 12 m copper clad steel ground rods connected in parallel. The model of the Pad 39B lightning protection system consisted of three main support towers (in this case, telephone poles) spaced in a triangular arrangement a few tens of meters apart. The support towers were connected by a catenary wire system at a height of 30 ft above local ground level. Two down conductors were connected to the catenary wire system at each support pole and an addition three down conductors wer e connected to the catenary wires between the support poles. Each down conductor was grounded to a 40 ft copper clad steel ground rod. The current was measured at the base of each down conductor with a Pearson 110A current transformer. The launcher was placed about 3 m northwest of the western most support pole. A wooden platform was constructed on the top of the western support pole above the catenary wire system to hold a large current measurement box (Figure 2 17). A 3 m ground rod was secured to th e very top of the support pole with ceramic insulators and was then connected

PAGE 99

99 to the current measurement box. The top of the ground rod was about 14 m above local ground level. When a rocket was launched from the launcher on the ground, the current on th e wire during the wire ascent and typically the full current during the subsequent UPL/ICC process flowed to ground through the rocket launcher (these currents did not typically flow through the catenary wire system), and were measured in the current measu rement box mounted underneath the launcher. Subsequent return strokes normally attached to the intercepting rod mounted on top of the western support pole. Current from the return strokes was measured with a T&M Research 1 m CVR in the current measureme nt box atop the western support pole. The return stroke current then flowed into the catenary wire system and subsequently distributed to ground through the down conductor network. The launcher functionality for the field launching system was essentiall y identical to that of the Tower Launcher, though the hardware differed somewhat. A close up view of the field launcher is shown in Figure 2 18 A The field launcher control box, shown in Figure 2 18 B is located immediately on the northwest side of the l auncher. The box contains a 2006 edition PIC controller ( Section 2.2.2.2), a launcher control board, and a bank of eight pneumatic switches (six of which are used). The launcher control board is connected to the output header of the 2006 PIC via a ribbon cable. The logic signals on the 2006 PIC that control the switching of the relays attached to each attenuator network are used as control signals for each of the six launch tubes on the field launcher. When the control signal is received, the logic sign als are fed to opto isolators which in turn provide the selected pneumatic switch with 12 V power. Control signals for the field launcher control box are sent directly from a custom Labview program that runs on HAL in Launch Control. For safety purposes, the 62.5 m control fiber for the 2006 PIC in the launcher control box is connected to an independent, two port optical fan out board in Launch

PAGE 100

100 Control to isolate it from the full network control system (i.e. to make any accidental cross talk impossible). The air supply to the master launcher control box is the compressor in the NASA trailer. Pneumatic tubing was run from the NASA trailer to the field launcher in protective 1" PVC pipe. Like the tower launching configuration, each pneumatic switch in th e master launcher control box is connected with a short section of pneumatic tubing directly to a control box on each individual launch tube (Figure 2 18 C ). The individual tube control boxes house a second pneumatic switch, a 9 V battery, electrical ports where the rocket squib wire is attached, and moveable pins to configure the launcher in test mode or firing mode. Again, the rocket motors are ignited when the pneumatic switch closure transfers 9 V to the squib wire. A block diagram, schematic view of the rocket launching process from the field launching system is shown in Figure 2 19. The diagram is also accurate for the Tower Launcher, except that the control system utilizes a different micro controller and fiber optic link. 2.11.3 Field Launching Facility (2011) Prior to summer 2011, the model experiment of the Pad 39B catenary wire system was disassembled. A smaller, rectangular intercepting wire ring was installed in its place (Figure 2 20). The wire ring is at an altitude of about 5.4 m above local ground level and is constructed from heavy braided wire. Ceramic insulators are used to offset the conducting ring from the wooden support poles. Four down conductors tie the corners of the wire intercepting ring to the aluminum launcher. Current on the ascending triggering wire and current from the subsequent UPL/ICC process typically flowed directly to the launcher while subsequent return strokes typically attached to the intercepting wire ring. The small current measurement box that was mounted to the underside of the field launcher was removed and replaced with the larger measurement box that had been located atop the western tower during the 2010 experiment. The larger box was mounted on a wooden stand under the field launcher that was insula ted from

PAGE 101

101 ground by large ceramic insulators. A heavy section of braided wire connected the new current measurement box to the launcher frame. The output of the 1 m CVR mounted in the current measurement box was tied directly to three 12 m copper clad st eel ground rods connected in parallel. In 2011, the air supply to the master launcher control box was changed from the compressor in the NASA trailer to a gas powered compressor in a small metal building about 25 m southwest of the launcher. Otherwise, t he launcher control system was identical to that of summer 2010. 2.12 ICLRT Channel Base Current Measurements In this section, the channel base current measurements used at the ICLRT from 2009 2011 will be discussed in detail including the physical cons truction of the measurement boxes, the current measurement devices, and the accompanying amplification electronics. 2.12.1 2009 Channel Base Current Measurements The current measurement box for the Tower Launcher was bolted directly to the aluminum supp ort structure on the underside of the launcher. The stainless steel box had length and width of 0.91 m and depth of 0.3 m. Dielectric grease was applied to the junction between the dissimilar metals. A T&M Research R 7000 10 1 m CVR was mounted directl y to the bottom side of the metal box with the flange secured to the box with stainless steel bolts. The lug of the CVR was connected to the braided wire ground lead that terminated at the 25 m ground rod 11 m below. The CVR is a non inductive, large vol ume cylindrical resistor capable of dissipating many kilo joules of energy without damage. The device has a flat frequency response from DC 8 MHz and provides a voltage output on a BNC terminal proportional to the input current in accordance with Ohm's Law. Triggered lightning current flows from the launcher to the outside

PAGE 102

102 of the current measurement box, enters the CVR through the flange, produces a proportional measureable voltage, and then exits the CVR though the lug. In Figure 2 21 A a picture of the inside of the tower current measurement box is shown, and in Figure 2 21B a schematic view of the internal configuration of the box is given. There were three channel base current measurements in 2009, named "II High", "II Low", and "II Very Low", co rresponding to three different sensitivity ranges. The measured current amplitudes during typical triggered lightning discharges may span five orders of magnitude, hence the need for three different measurements to expand the possible dynamic range. In 2 009, the transducer factors (the numbers used to convert raw digitizer volts to meaningful physics units) for the three current measurements in order of increasing sensitivity were about 63 kA/V, 6.3 kA/V, and 21 A/V. The least sensitive current measureme nt (63 kA/V) is configured to record the return stroke peak current. The mid level measurement (6.3 kA/V) is designed to resolve the peak amplitudes of weaker return strokes, large ICC pulses, and M components following return strokes. The most sensitive measurement (21 A/V) is designed to measure precursor current pulses on the ascending triggering wire and the initial upward positive leader currents preceding subsequent return strokes. The PIC controller, fiber optic transmitter, and any amplification electronics for each current measurement were enclosed in smaller metal boxes inside the large measurement box. The smaller metal boxes were secured to a sheet of Plexiglass, which was bolted to the interior of the large outer box with standoffs, effectiv ely electrically isolating the inner measurement boxes from current flow on the exterior box. With possibly many tens of kilo amperes flowing on the outer box, it is very important to avoid any possible ground loops inside the measurement box. A 2006 PIC controller was placed in the II High measurement box. The PIC controller attenuated the incoming voltage signal from the CVR by 36 dB before

PAGE 103

103 passing the signal to the Opticomm transmitter. A 2001 PIC controller was placed in the II Low measurement box. It received control commands over plastic fiber from the 2006 PIC controller in the adjacent II High measurement box. The 2001 PIC controller attenuated the voltage output of the CVR by 16 dB before passing the signal to the Opticomm transmitter. Fina lly, a second 2006 PIC controller was placed in the II Very Low measurement box. The input signal from the CVR was first passed through an 820 current limiting resistor and then fed across back to back 4.7 V zener diodes to clamp the voltage at the inpu t to the following amplifier (input signals could be as high as 50 60 V for very large peak current return strokes, potentially damaging the amplifier). The non inverting amplifier circuit was designed by Dr. Christopher Biagi with gain of 46.5. Fifty oh m feed through resistors were placed on the inputs of the Opticomm transmitters on both the II High and II Low measurements. The output of the amplifier on the II VL measurement was injected directly into the 68 k input impedance of the Opticomm transmit ter. Fiber optic bundles connections are not shown in the measurement box schematic of Figure 2 21. An armored, 6 strand fiber bundle passes through a pipe nipple in the bottom of the outer measurement box. Internal fiber optic connections (62.5 m glas s fiber and 200 m plastic fiber) pass through the interior measurement boxes in small holes with rubber grommets to protect the fibers. 2.12.2 2010 Channel Base Current Measurements The rocket launching configuration during summer 2010 was discussed in Section 2.11.2. In Figure 2 22, schematics are shown for the channel base current measurement boxes during the Pad 39B catenary wire experiment. The steel current measurement bolted directly to the underside support of the launcher (Figure 2 18) and had width and length of 0.61 m and depth of 0.3 m. The box housed two measurements named "ICC" and "ICC Very Low", with respective

PAGE 104

104 transducer factors of 16 kA/V (ICC) and 400 A/V (ICC Very Low). Each measurement and its associated electronics were mounted in separate interior metal boxes, both of which were secured to a Plexiglass support that was offset from the outer metal box with standoffs (Figure 2 22A) The ICC measurement was configured with a 2001 PIC controller and was set to attenuate incoming sign als from the CVR by 20 dB. The ICC Very Low measurement passed through an identical current limiting and clamping circuit as that described in Section 2.12.1, and then passed through a non inverting amplifier circuit with gain of 2. The ICC Very Low mea surement was configured with a 2001 PIC controller with no attenuation. The output signals of both PIC controllers were routed to Opticomm transmitters, each with 50 feed through resistors at the inputs. The ICC and ICC Very Low current measurements we re powered with 28 Ah batteries that rested on the bottom of the outer measurement box next to the CVR. A pair of 62.5 m glass fiber jumpers were fed through a pipe nipple on the underside of the outer measurement box and were connected to each of the Op ticomm transmitters. The opposite ends of the fiber jumpers were fed into a junction box located next to the base of the western support tower where they were mated with ST ST connectors to individual fibers of a 6 strand armored bundle for transmission t o Launch Control. 200 m plastic fibers that transmitted control signals to the 2001 PIC controllers were also fed through the pipe nipple on the outer measurement box. The opposite ends of the plastic fibers were connected to a plastic fiber fan out boa rd located in a second junction box immediately next to the 62.5 m fiber junction box described above. The plastic fiber fan out board received its control signals from the 2006 PIC controller in TERA 24, located about 12 m to the south. I n Figure 2 2 2 B a schematic is shown for current measurement box mounted atop the western tower during the Pad 39B catenary wire experiment. As stated in Section 2.11.2, the

PAGE 105

105 measurements in this box were designed to record return stroke currents that attached to the 3 m intercepting rod above the western tower. The stainless steel box was 0.91 m in length, 0.61 m in width, and had depth of 0.3 m. Two interior metal boxes were fastened to a large sheet of Plexiglass that was offset from the bottom of the outer measur ement box with rubber standoffs and Styrofoam. A T&M Research R 7000 10 1 m CVR was bolted to the forward right side of the outer measurement box. The left interior measurement box housed the associated electronics for two current measurements, named "R S High" and "RS Low", having respective transducer factors of 63 kA/V and 6.3 kA/V, while the right interior measurement box housed electronics for a more sensitive current measurement named "RS Very Low" with transducer factor of 139 A/V. The RS High and RS Low measurements were both configured with 2001 PIC controllers with attenuations of the CVR output set to 36 dB and 16 dB, respectively. The output signals of the PIC controllers were routed to separate Opticomm transmitters with 50 feed through resistors at the inputs. The RS Very Low measurement passed through an identical current limiting and clamping circuit as the ICC Very Low measurement described above, and then passed through an amplifier with gain equal to 7.2. The RS Very Low measureme nt was also configured with a 2001 PIC controller set to 0 dB atteunation. The output of the PIC controller passed to an Opticomm transmitter with a 50 feed through resistor at the input. The associated electronics for all three current measurements we re powered by 33 Ah batteries. A set of three 62.5 m fiber jumpers were routed from the junction box on the ground, up the western support pole through a 2" PVC tube, into the current measurement box through a pipe nipple on the rear edge of the box, and then to the inner measurement boxes where they connected to the Opticomm transmitters. Plastic fiber pairs for the control of each 2001 PIC followed the same path and were connected to the plastic fiber fan out board on the ground.

PAGE 106

106 After the conclusion of the Pad 39B catenary wire experiment on July 11, 2010, rocket launching operations moved back to the Tower Launcher. The measurement configuration of the II High and II Low measurements was identical to that described in Section 2.12.1. The existing non inverting amplifier on the II Very Low measurement was replaced with the non inverting amplifier from the ICC Very Low measurement (gain of 7.2). The 2006 PIC controller in the II Very Low measurement was also set to attenuate the incoming signal from the CVR by 6 dB, giving an effective transducer factor of 278 A/V. 2.12.3 2011 Channel Base Current Measurements Through May 14, 2011, the channel base current measurements on the Tower Launcher were identical to those described in the bottom paragraph of Section 2.12.1. Rocket launching operations were moved back to the field launching facility after May 14, 2011. As described in Section 2.11.3, the large current measurement box that had been mounted on the platform atop the western support tower dur ing the 2010 experiment was mounted on a support structure directly underneath the field launcher during summer 2011. The lug of the 1 m CVR was tied directly to the three 12 m ground rods connected in parallel. A picture of the inside of the 2011 curre nt measurement box is shown in Figure 2 23A and a schema tic view of the box is shown in Figure 2 23B The configuration of the II High and II Low measurements was identical to that of the RS High and RS Low measurements described above in Section 2.12.2. The discrete clamping circuit and amplifier boards on the II Very Low measurement were replaced with a new, self contained circuit board that provided both functionalities. The non inverting amplifier had gain of 5, providing a transducer factor of 200 A /V. The 2001 PIC controller on the II Very Low measurement was configured with no attenuation. A fourth current measurement named "II XL" was also added to the right interior measurement box. An amplifier board, including the

PAGE 107

107 current limiting and clampi ng capability, was designed with gain of 2,500, providing a transducer factor of 400 mA/V. This extremely sensitive current measurement was designed to measure the low level current flowing into the base of the triggering wire as the wire ascends, the cur rent serving to keep the wire at ground potential as the electric field aloft induces charge on the ascending wire. The II XL measurement was also configured with a 2001 PIC controller with no attenuation applied. The PIC outputs of both the II VL and I I XL measurements were fed to Opticomm transmitters with 50 feed through resistors at the inputs. All four current measurements were powered with 33 Ah batteries. Four 62.5 m fiber jumpers passed through a large pipe nipple on the rear of the outer me asurement box and were connected to each of the four Opticomm transmitters. The opposite ends of the 62.5 m fiber jumpers connected to the junction box described in Section 2.11.2 for transmission to Launch Control. Plastic fiber pairs for each of the f our 2001 PIC controllers also passed through the pipe nipple on the rear of the measurement box. The plastic fibers were all connected to the plastic fiber fan out board discussed in Section 2.11.2. In addition to the four ICLRT current measurements, t wo HBM digitizers were connected to the output of the CVR. The HBM digitizers were configured to record two sensitivity ranges, the "II High" measurement recording signals ranging from +/ 50 kA and the "II Low" measurement recording signals ranging from +/ 250 A. The outputs of the HBM digitizers were connected to single mode fiber pairs that transmitted the signals to the HBM Gen16t Transient Recorder in the Office Trailer. The details of the HBM digitization system were discussed in Section 2.5. The HBM digitizers were powered by a pair of 28 Ah batteries. The two HBM digitizers rested on top of the right interior measurement box (Figure 2 2 3A ) and were isolated by pieces of Styrofoam.

PAGE 108

108 2.13 ICLRT dE/dt Measurements Electric field derivative (dE/ dt) measurements are primarily used for studying the impulsive electric field changes radiated by lightning leaders (propagation), and the sub microsecond processes that occur during the interaction of lightning leaders and ground based objects (attachment ). The network of dE/dt sensors at the ICLRT form a time of arrival (TOA) network used for locating sources in three dimensions (Chapter 4). At the ICLRT, the normal component of electric field and its derivative are measured using flat plate antennas fa bricated from 0.16 cm aluminum. A schematic and a corresponding image of a flat plate dE/dt antenna are shown in Figure 2 24 A and Figure 2 24B, respectively The antenna housing measures 0.6 m square by 5 cm in depth. A section is cut from the top plate of the housing for the antenna sensing plate. The circular antenna sensing plate measures 0.48 m in diameter, has area of 0.155 m 2 and is separated from the antenna housing by an annular space of about 1.25 cm. The sensing plate is supported by six r ubber standoffs that elevate the plate to a level flush with the top plate of the antenna housing. The signal from the sensing plate is routed to a BNC feed through mounted to the antenna housing with a short length of RG 223U coaxial cable. The cable is stripped so that a short length of the center conductor is exposed. A ring terminal is soldered to the center conductor and then connected to the center of the antenna sensing plate with a small stainless steel bolt. The housing of the antenna assembly is secured to a 1.2 x 2.4 m piece of galvanized screening which serves as a ground mesh. The antenna is placed on the ground surface and pieces of galvanized hardware cloth are secured to the existing ground mesh on three sides to expand the ground plane. A 3 m copper clad steel ground rod is driven along the front edge of the antenna. A short length of 12 guage solid core copper wire connects the ground rod to the antenna housing. A hole is dug in the ground on the front side of the antenna to house th e associated electronics for the antenna. The measurement holes typically have

PAGE 109

109 dimensions of about 1.2 x 0.76 x 0.61 m (L x W x H) and are supported on all sides by sections of pressure treated lumber. The wood helps to prevent the holes from caving in w ith heavy rain and also helps to reduce the amount of vermin (spiders, snakes, etc.). The holes are covered by 1.2 x 1.8 m sections of TUFF R insulation. The TUFF R is highly reflective and serves both to provide a protective covering for the measurement hole and to regulate the temperature. The long side of the galvanized screen is stretched over the TUFF R and is held in plate by piece of lumber. A concerted effort is made to keep the ground mesh around each antenna as flat as possible so not to disto rt the electric field measured at ground. In addition, vegetation around each antenna is regularly cut flush with the ground level. A cross section schematic of the flat plate antenna installation and the accompanying measurement hole is shown in Figur e 2 25. A schematic of the measurement electronics boxes, a top down image of the measurement hole, and images of the inside of each electronics box are given in Figure 2 26. Three concrete blocks are placed in the bottom of each hole to elevate the elec tronics boxes in the event the holes take on large amounts of water. A wooden support platform is rested on top of the concrete blocks. Two shielded Hoffman enclosures are screwed to the wooden support stand. The boxes are bonded together mechanically a nd electrically by a 1.9 cm galvanized pipe nipple. The leftmost Hoffman box has a BNC feed through mounted to the rear wall. A 1.2 m section of RG223 U coaxial cable is connected between the antenna housing and the Hoffman box, transferring the output of the antenna to the inside of the electronics box. Through the shield of the BNC cable, the Hoffman box is also at the same ground potential as the antenna housing, the ground rod, and the surrounding ground mesh. The connecting BNC cable is enclosed i n tinned copper braided wire for additional electromagnetic shielding. Additional pipe nipples are mounted to each Hoffman box and serve as feed through

PAGE 110

110 ports for fiber optic cables. The leftmost Hoffman boxes houses a 2001 PIC controller, an Opticomm transmitter, a solar charging circuit, and a 28 Ah battery for powering the aforementioned electronics. The antenna cable first passes from the BNC feed through on the Hoffman box to the 2001 PIC controller. Depending on the measurement configuration, th e PIC controller is a short circuit from input to output, or alternately, is commanded to attenuate the incoming signal by 3 dB or 6 dB. The output of the PIC controller is fed to the input of the Opticomm transmitter. A single 62.5 m multi mode fibe r carried the analog signal from the Opticomm transmitter to Launch Control. Prior to 2011, the signal path terminated at the Opticomm transmitter, where a 50 feed through resistor was placed at the input. In 2011, the 50 feed through resistor was re moved and the signal was branched with a BNC F connector at the Opticomm input. A 0.3 m section of BNC cable passed from the branch point through the pipe nipple to the rightmost Hoffman box. Here, the signal was injected through a 50 feed through resi stor to the input of a HBM digitizer. The rightmost Hoffman box also housed a solar charger and a 28 Ah battery that supplied power to the HBM digitizer. A 2 strand 9/125 m single mode fiber pair was connected to the HBM digitizer and carried the digita l output to the Genesis 16t Transient Recorder in the Office Trailer. The solar chargers in each Hoffman box were connected to separate 11 W solar panels that were mounted to pressure treated lumber and placed adjacent to the ground mesh at each antenna The outputs of the solar panels were regulated to 14.5 V and then connected to the battery in each Hoffman box. When storm conditions were present, the output power of the PIC controller was used to control a relay in each solar charger that shorted th e solar panel to the Hoffman box, effectively preventing the solar panel and lead from acting as an antenna and introducing parasitic RF noise into the measurement system.

PAGE 111

111 A thorough derivation of the time and frequency domain responses of the flat plat e antennas used at the ICLRT is given in Jerauld [2007]. An abbreviated version of the flat plate antenna response will be given here specific to the dE/dt antenna configuration. A flat plate antenna oriented horizontally (parallel with the ground surfac e) senses the normal (vertical) component of the electric field consistent with the electric field boundary condition at a perfectly conducting surface. (2.1) The vector quantity is the electric dis placement vector at the surface of the conductor (the sensing plate), the quantity is the unit vector in the vertical direction, and the quantity is the surface charge density on the sensing plate. T he dot product of the electric displacement vector and the normal unit vector is simply the normal component of the electric displacement vector. (2.2) With the assumption that the sensing plate is perfectly conducting, we can take the tangential component of the electric field to be equal to zero inside the sensing plate and at the plate surface. If the dielectric medium above the plate is air, the vertical component of the electric displacement vector can be rewritten as the product of the permittivity of free space ( 0 = 8.854 x 10 14 F/m) and the normal component of the electric field. (2.3) The total charge on the plate can be obtained by multiplying the surface charge density by the total area of the plate. The assumption is made that the surface charge density is uniform across the plate area based on the frequency content and corresponding wavelengths of the

PAGE 112

112 signals of interest. The bandwidth constraints of the fiber optic data transmission system limit the measurement frequency range to about 20 MHz for measurements transmitted over the Opticomm fiber optic links and about 25 MHz for measurements digitized by the HBM digitizers. The total charge on the plate is given by the following expre ssion. (2.4) By taking the time derivative of both sides of Equation 2.4, we can obtain an expression for the time dependent current which flows off of the plate in response to the external electric field. This quantity is t hen directly proportional to the time derivative of the vertical component of the electric field. (2.5) The frequency domain response of the flat plate antenna can be found by taking the Fourier tra nsform of Equation 2.5. A time derivative in the time domain is equivalent to multiplication by the complex number j in the frequency domain. Hence, the frequency domain equivalent of Equation 2.5 is given by Equation 2.6 as a function of the angular fr equency. (2.6) Viewed from the standpoint of a Norton equivalent circuit, the frequency domain response of the flat plate antenna is that of a current source in parallel with the source and load impedances of the antenna. The measured capacitance of the flat plate antennas used at the ICLRT is about 80 pF. The resistance and inductance of the flat plate is negligible and is not factored into the antenna source impedance. The series resistance and inductance of the coaxial ca ble that connects the antenna sensing plate to the load impedance is negligible compared to the capacitance of the flat plate antenna at the frequencies of interest (DC 25 MHz), and hence is

PAGE 113

113 not accounted for in the expression for the antenna source impe dance. For a flat plate antenna configured to measured dE/dt, the load impedance of the antenna is simply a 50 resistor. For a flat plate antenna configured to measure the electric field, the load impedance is typically the input impedance of the integ ration electronics, which for passive integrators is a parallel RC circuit. A discussion of the frequency domain response of the electric field sensor will not be presented in this document ( Jerauld [2007]). For the dE/dt sensor, the total impedance see n by the current source is then the parallel combination of the antenna capacitance and the 50 load resistor. (2.7) The equivalent frequency domain voltage output across the load impedance is then given by Equation 2.8. (2.8) The term in brackets in Equation 2.8 represents a single pole low pass filter in the frequency domain with 3 dB point given by Equation 2.9. (2.9) For a load resistor of 50 and the measured 80 pF capacitance of the flat plate antenna of area 0.155 m 2 the 3 dB bandwidth of the antenna is given by Equation 2.10. (2.10)

PAGE 114

114 In the case of the dE/dt antenna, 0 represents a high frequency roll off. If the frequencies of interest are much less than the 3 dB bandwidth of the antenna ( 0 ), Equation 2.8 simplifies to Equation 2.11, which represents the ideal frequency domain response of the flat plate dE/dt antenna. (2.11) The time domain equivalent of Equation 2.11 at frequencies 0 is given by Equation 2.12. (2.12) Equation 2.12 represents the ideal time domain response of the flat plate dE/dt antenna. The output voltage of the antenna is directly proportional to the vertical component of the electric field derivative, th e plate area, and the load resistance. Again, the load resistance is simply the 50 feed through resistor that is placed in parallel with the 68 k input impedance of the Opticomm transmitter. In cases where the onboard attenuator networks are used on t he PIC controllers to decrease the signal size passed to the Opticomm input, Equation 2.12 is modified by multiplying the right side of the equation by the appropriate attenuation value (Equation 2.13). The attenuation value is always less than one. (2.13) From Equation 2.13, the characteristic sensitivity of the flat plate antennas used at the ICLRT is about 14.5 kV/m/ s per digitizer volt assuming no attenuation is applied in the PIC controller. During the years of data co llection for this dissertation, the dE/dt antennas have been configured with either 6 dB of attenuation (sensitivity of about 29 kV/m/ s per digitizer volt) or

PAGE 115

115 with no attenuation. dE/dt measurements that are recorded on the HBM digitization system are c onfigured to record to thresholds of +/ 8 kV/m/ s. Due to the high frequency response of the dE/dt antennas, the outputs of all dE/dt antennas are recorded on LeCroy 44 Xi oscilloscopes sampled at 250 MS/s or are digitized by the HBM digitizers at 100 MS/ s. 2.14 ICLRT Energetic Radiation Measurements The ICLRT measurement network is equipped with three different types of energetic radiation detectors, 1) Sodium Iodide (NaI) scintillators, 2) Plastic scintillators, and 3) Lanthanum Bromide (LaBr 3 ) scint illators. The energetic radiation measurements are collectively 3 detectors are primarily used to detect x rays and gamma rays from lightning leader and attachment processes in the energy range from about 30 keV to about 15 MeV. The plastic detectors are likewise used for detecting x rays and gamma rays from lightning, but also serve as muon detectors. Muons are essentially heavy electrons and are bi products of cosmic ray a ir showers lightning initiation in the cloud. The incidence of muons on the ground surface at multiple stations spread over some hundreds of meters can be u sed to infer the presence of an air shower, the arrival time and incidence direction of which can then be correlated with independent lightning electromagnetic measurements. The details of this experiment are outside the scope of this dissertation and wil l not be further discussed. The NaI and LaBr 3 from 0.3 cm aluminum. Images of the outside and inside of a TERA box in addition to a box schematic are shown in Figure 2 27. The box is about 0.5 m square and is 0.67 m in height. The

PAGE 116

116 boxes are painted white to help with internal temperature regulation and are numbered according to the station location. The lid of the box overlaps the bottom portion of the box by 15 cm and rests on a rubber g asket to form a light tight seal. The lid is secured to the bottom of the box with four latches which serve to compress the aforementioned gasket. The inside of the TERA box has support structures for two NaI or LaBr 3 detectors and their accompanying pho to multiplier tube (PMT) bases, an enclosure to house a large battery for powering the electronics, and brackets to support a so buses for the detectors, a 2006 or 2011 PIC controller, a po wer switch and fuse for the box, and the Opticomm fiber optic transmitter(s). All support structures are welded to the sides of the box and are painted black (as is the full interior of the TERA box). Each TERA box is equipped with three ST type feed thr ough ports for 62.5 m fibers and two plastic fiber feed through ports for 200 m plastic fibers. As discussed in Section 2.2.2.2, the 2006 PIC controller in the TERA box at each station serves as the master device that transceives control signals to all other PIC controllers at the respective station. One of the 62.5 m feed through ports is used to bring the control fiber (which connects to the optical fan out board in the rear of Launch Control) to the 2006 PIC controller. A short fiber jumper connect s the feed through port to the PIC controller. The remaining two 62.5 m feed through ports are used to connect the Opticomm transmitters to fibers that transfer the collected signals back to Launch Control. Typically, the actual armored fiber bundle fro m Launch Control is secured in a junction box. Ten meter jumpers are coupled to the individual fibers in the armored fiber bundle. These jumpers then fan out to the TERA box and to any other measurements at the station. Similarly, the 2006 PIC controlle r broadcast received control signals through the plastic fiber feed through ports and to any 2001 PIC controllers at the given station.

PAGE 117

117 The PMT bases have pre attached power cords with male DB9 connectors. Female DB9 connectors are mounted to the bottom of the FRU. Two pins of the DB9 connector are wired to the power and ground buses in the FRU to transfer power to the PMTs. Power is transferred to the power and ground buses on command from HAL. The 2006 PIC controller output power port first powers t he Opticomm transmitter(s). The power signals are branched at the Opticomm power input and then connect to the power and ground buses. Stations with only NaI detector(s) are powered with 12V, 28 Ah batteries. The batteries are charged with an 11 W solar panel which connects to the battery through a solar controller identical that described in Section 2.13. The switch power to short the output of the solar panel to the TERA box when storm conditions are present is taken from the power and ground buses in the FRU. 2.14.1 NaI Detectors The NaI detectors are 7.6 cm x 7.6 cm cylindrical detectors (3M3 series) manufactured by Saint Gobain. The detector is contained in a hermetically sealed package and is optically coupled to a PMT. The PMT collects the li ght output from the NaI crystal when the crystal is excited by ionizing radiation and then converts the light output to an electrical signal. The entire detector and PMT are enclosed in a mu metal magnetic light shield. The assembly is mounted to an Orte c model 296 high voltage PMT base through a pin assembly. The PMT base contains the divider chains that amplify the electrical output of the PMT. The high voltage supply can be adjusted by a small screw on the bottom of the PMT base. The high voltage su pply is typically set to about 980 V. The PMT base has two outputs. The first is the output of a pre amplifier enclosed within the base. This output is not used. The second output is the anode of the PMT base. The signal from the divider chain passes through a DC blocking capacitor and across a 1 a short circuit in this case) and then to the input of the Opticomm transmitter. For NaI

PAGE 118

118 measurements, there is n impedance of the Opticomm. The unshielded detectors are supported by a 0.32 cm thick lead tu be that covers only the PMT base. The led tube rests inside the welded aluminum support structure described above. The only attenuation of incoming photons to the unshielded detector is that due to the outer aluminum box. The box attenuates photons with energies less than about 30 keV. With the exception of the boxes that house the LaBr 3 detectors, each TERA box is equipped with an configuration. The shielded detectors are completely covered by a 0.32 cm thick lead tube with a top cap. The lead shield serves to attenuate incoming photons with energies below about 300 keV. The shielded NaI detectors are suited to studying higher energy photons from lightning p rocesses (i.e., terrestrial gamma ray flashes, etc.). The standard convention is that all shielded NaI measurements correspond to Channel 1 on the 2006 PIC controller and all unshielded measurements correspond to Channel 2. All NaI detectors are calibrat ed using a Cs 137 662 keV radio active source. The source is placed on the top of the TERA box and the amplitude of the corresponding output pulse is measured in Launch Control by using the histogram feature on a LeCroy oscilloscope. Typical calibration factors for NaI detectors at the ICLRT range from about 3 15 MeV per digitizer volt. A subset of eight of the unshielded NaI detectors are part of the TOA network (Chapter 4). These eight measurements are recorded on LeCroy Waverunner I oscilloscopes a t a sampling rate of 250 MS/s. These measurements are used to locate x ray sources from lightning leaders in three dimensions and to examine the timing relationship between the x rays recorded at each

PAGE 119

119 TOA station and the corresponding dE/dt measured at th e same station. The NaI measurements at TOA stations are located about 10 m from the corresponding dE/dt antenna. While NaI detectors exhibit excellent sensitivity and linear spectral response across the typical spectrum of emission from lightning leader s, the light decay time of the sensor is about 0.23 s and the overall pulse width from a single photon excitation is about 1 s. This elongated pulse width is problematic when many photons strike the detector in a short time duration, which is typical du ring stepped leaders, dart triggered lightning. The result is pulse pile up, which quickly saturates the dynamic range of the Opticomm fiber optic transmitters. When pile up ensues, the phot ons from individual leaders steps cannot be resolved. 2.14.2 LaBr 3 Detectors To combat the issue of pulse pile up due to nearby lightning leaders, two LaBr 3 detectors were installed at the ICLRT in 2009. The LaBr 3 detectors were originally housed in a single TERA box with one detector operated in an unshielded configuration and the other in a shielded configuration. The shielded detector was then removed and placed in a separate TERA box and operated in the unshielded configuration to provide another p ossible TOA measurement. The TERA boxes containing the LaBr 3 detectors (Figure 2 28) are powered with 12 V, 33 Ah batteries. The LaBr 3 detectors are 7.6 cm x 7.6 cm cylindrical detectors manufactured by Saint he NaI detectors, the scintillator is optically coupled to a PMT, and both the scintillator and PMT are encased in a mu metal magnetic light shield. The PMT is mounted to an identical Ortec model 296 PMT base. The LaBr 3 detectors are also calibrated with the Cs 137 radio active source. The typical calibration factor is about 5 MeV per digitizer volt. The LaBr 3 detector has several advantages over the NaI detector (better

PAGE 120

120 energy resolution and higher light output), but for measuring energetic radiation a ssociated with lightning leaders, the primary advantage is the fast light decay time (about 16 ns). The total pulse width for a single photon excitation is about 184 ns. This fast decay time does not alleviate pulse pile up, particularly for highly energ etic lightning leaders, but it does help substantially. The single photon responses of the NaI and LaBr 3 detectors to the Cs 137 source are shown in Figure 2 29. The outputs of both LaBr 3 detectors are recorded on a LeCroy 44 Xi DSO at a sampling rate of 250 MS/s. The output of the LaBr 3 detector installed at Station 25, immediately south of the Launch Control trailer, is transmitted to the DSO directly over a length of RG223 U coaxial cable enclosed in braided wire. As a result, the digitized signal fr om this detector has a much hi gher signal to noise ratio ( Figure 2 29) than measurements transmitted over the analog fiber optic links. The second LaBr 3 detector at Station 17 is transmitted over the normal Opticomm fiber optic link. 2.14.3 Plastic Dete ctors The first generation of plastic detectors was installed at the ICLRT in 2007 ( e.g., Howard 20 09 ). These detectors were vertical paddles that were installed in TERA boxes with modified taller lids. While plastic detectors are not well suited for d etermining spectral content of incident photons and particles, they do have excellent timing characteristics. The typical full pulse width for a received x ray pulse on a plastic detector is between 35 50 ns. Unfortunately, the original plastic paddles l acked both a sufficient surface area and the thickness to provide adequate efficiency for detecting x rays from lightning. A set of eight new plastic scintillators was installed at the ICLRT in 2009 at eight of the TOA stations. The scintillation materia l was manufactured by Saint Gobain (model BC 408, Pilot F) and measured 1 m 2 in area and 2 cm in thickness. The plastic scintillator is composed of organic scintillating molecules (fluorescent material) suspended in a solvent base material (in this case P olyvinyltoluene). The scintillating

PAGE 121

121 molecules emit fluorescent light upon interaction with detected energetic radiation. The scintillators at the ICLRT emit at peak wavelength of about 430 nm. Two Hamamatsu R550 PMTs were mated to opposite sides of each plastic scintillator with optical jelly. The PMTs were coupled to Ortec model 296 PMT bases (identical to those used with the NaI and LaBr 3 detectors). The high voltage setting on each PMT base was set to about 980 V. Annotated images of the field in stallation of the plastic scintillators are shown in Figure 2 30. Images of the inside of the measurement box are shown at top in Figure 2 31 A and Figure 2 31C and a corresponding schematic is shown in Figure 2 31 B The field assembly for the plastic sci ntillators is composed of an outer aluminum box with dimensions of 1.2 x 1.8 m welded from 0.48 cm aluminum, inside of which is a second aluminum box that actually houses the scintillator and associated electronics. The outer aluminum box is mated to the TERA box at each station with a 1.5 m section of galvanized tubing. The tubing slides over welded nipples connected to each box. Braided wire was placed over each nipple to ensure a good electrical connection between the two boxes, while still allowing t he boxes to be separated easily for trouble shooting purposes. Power for the PMT bases mounted to the plastic scintillator was transferred through the galvanized tube using a two wire harness enclosed in braided wire. The power wires were connected to th e buses in the FRU in each TERA box, and were soldered to a BNC AV connector on the opposite end. The inner aluminum box is equipped with a series of BNC feed through connectors that are used for both power and signal transmission. Inside the inner box, t he plastic scintillator material was supported by a wooden structure crafted from pressure treated lumber. The PMT bases were secured to the wooden structure with padded plastic brackets. The box has a lid that is secured with about 30 individual stainle ss steel hex head bolts. The lid compresses a rubber gasket that serves to make the box light tight.

PAGE 122

122 In the original measurement configuration (starting June 1, 2009), the outputs of both PMT bases were combined with a BNC T connector which was coupled directly to one of the BNC feed through connectors on the wall of the inner box. The signal was carried through the galvanized connecting tube to the TERA box with a section of RG 223 U coaxial cable. The cable was connected to the Channel 1 input of the PIC controller (which was a short circuit in this case), and then from the Channel 1 output of the PIC controller to the input of the Opticomm transmitter. For the plastic scintillators, a 50 feed through resistor was placed in parallel with impedance of the Opticomm transmitter. It was quickly discovered that the signal amplitudes were insufficient without external amplification electronics. An inverter and pre fabricated T exas Instruments amplifier (model THS4011) with gain of 2 were installed at the output of each PMT base on July 17, 2009 by the team from FIT. The inverter was added to accommodate a muon coincidence trigger circuit that was installed in Launch Control. The coincidence circuit required a positive going pulse, and the characteristic output of the PMT base is negative polarity. The outputs of the two amplifiers were tied together with a BNC T connector. The signal connectivity to the Opticomm in the TERA box remained the same. The configuration of the plastic scintillators remained unchanged until August, 2010 when the scintillators were wrapped in Tyvec, a barrier material developed by Dupont. Small holes were cut in the Tyvec where the two PMTs could still mate directly to the plastic scintillator. The Tyvec effectively provided an additional gain of about 2 by reducing the number of photons that scattered out of the plastic scintillator before being collected by the two PMTs. Prior to Summer 2011 the amplification electronics on the plastic scintillators were completely overhauled. The performance had been generally unsatisfactory and the general design topology was not well conceived (tying the outputs of two separate voltage sources

PAGE 123

123 together i s not generally good practice). The inverters and Texas Instr uments amplifier boards were removed. Rob Olsen III designed a trans impedance amplifier with trans impedance gain impedance amplifiers were mounted in small Pomona electronics boxes. The outputs of each trans impedance amplifier drove equal mounted in a small Pomona electronics box, with unity gain. The output of the summing amplifier drove the coaxial cable that connected to the Opticomm transmitter in the TERA box. A terminal strip with power and ground buses was installed within the inner aluminum box. Power wires for the three amplifier circuits were routed from the terminal strip to each Pomona box where t hey were soldered directly to feed through power connectors. Due to the inherent lack of linearity with incident photon energy for plastic scintillators, the absolute output amplitudes of the detectors are un calibrated. From a lightning standpoint, the plastic scintillators are used strictly for TOA measurements of x rays from leader and attachment processes. As stated previously, the plastic scintillators are also very effective in detecting muons. The outputs of the eight plastic scintillators are d igitized on a set of two LeCroy 44 Xi oscilloscopes that are part of the TOA network, and also on an additional two LeCroy 44 Xi oscilloscopes that trigger on the output of a muon coincidence circuit designed by Rob Olsen III and Dr. Douglas Jordan. All p lastic scintillator outputs are digitized at 250 MS/s. 2.15 Lightning Mapping Array (LMA) Prior to Summer 2011, a 7 station Lightning Mapping Array (LMA) was installed on and around the ICLRT. The LMA was provided via the UF DARPA NIMBUS grant funds by Dr. Rich Blakeslee of the NASA Marshall Space Flight Center and Dr. Hugh Christian of the University of Alabama Huntsville. The LMA system was installed by John Pilkey, a graduate student at the ICLRT, with assistance from Jeff Bailey, an employee of NASA and the

PAGE 124

124 University of Alabama Huntsville. Dr. Douglas Jordan provided invaluable assistance in finding suitable LMA station locations. Finally, Dr. Paul Krehbiel and Dr. William Rison of the New Mexico Institute of Mining and Technology, the inventors of the system, provided continuous assistance in understanding the intricacies of the LMA hardware and data processing. The LMA stations were arrayed around the Field (Ground) Launching facility at the ICLRT at distances ranging from about 460 m to about 9 .6 km. The LMA station locations in absolute Latitude and Longitude are given in columns 2 4 of Table 2 2. The LMA station locations relative to the local ICLRT coordinate origin (Section 4.2) are given in columns 5 7. Finally, the straight line distanc e from each LMA station to the field launching facility and the azimuth of each station with respect to the field launching facility are given in columns 8 9. A plan view of the LMA station locations is given in Figure 2 32. The general purpose of the LMA is to map in three dimensions the electrical breakdown that occurs at the tips of propagating lightning leaders (both of negative and positive polarity discharges). This is accomplished by measuring the VHF radiation from the same impulsive event at m any stations, and then using the measured arrival times of the common signal at each station to calculate the spatial position and emission time of the radiation source. The LMA at the ICLRT operates in the Channel 4 band (66 72 MHz). The locations of LM A stations were chosen based on several factors, 1) the magnitude of the local noise within the Channel 4 frequency band, 2) the availability of consistent AC power, and 3) a good combination of accessibility and security. During 2011, one LMA station was located on the far northeast corner of the ICLRT (about 25 m to the northeast of the Blast Wall observation station), two stations were located on the Dupont property both north and south of the ICLRT, one station was located on a Camp Blanding communicat ions tower to the southeast

PAGE 125

125 of the ICLRT, one station was located at a military surplus warehouse in Starke, one station was located on private property just east of the Starke golf course, and one station was located on Florida Department of Transportatio n (FDOT) property to the northwest of the ICLRT. The stations at the golf course, communications tower, and warehouse were AC powered while the remaining stations ran on battery power. At stations with AC power, a battery charger was used to constantly c harge the battery, which was directly connected to the LMA box. All stations used 12 V Optima 55 Ah batteries. Each LMA station is equipped with a monopole VHF antenna (model Sirio GPA 40 70). The length of the monopole and the corresponding ground rad ials are adjusted according to the transformer (model North Hills 0102JB) before entering bandpass filter (Microwave Filter Company Model 3303, Channel 4) to filt er the received signal to the necessary frequency band. A 1 m length of RG 6 coaxial cable connects the output of the filter to a pre amplifier (Channel Master Titan 2 Model CM 7777) mounted on the antenna mast. The mast and ground radials are connected to a 3 m ground rod driven at the base of the mast. A photograph of the monopole antenna and associated electronics is shown in Figure 2 33 A A length of RG 6 coaxial cable connects the output of the pre amplifier filter to the LMA box (Figure 2 33 B) T he LMA box is usually located about 30 m from the antenna to avoid the antenna responding to local noise generated by the electronics within the LMA box. The same coaxial cable is also used to send 12 V DC power to the pre amplifier. The received signal passes through a logarithmic amplifier. The wide dynamic range of the logarithmic amplifier allows the LMA to record VHF sources with respective powers ranging from milliwatts to several hundred watts. After the received signal is amplified, it passes to an analog to digital converter located on the LMA board, a

PAGE 126

126 proprietary piece of hardware designed by the team at New Mexico Tech. The LMA obtains accurate timing across the network by using a 25 MHz oscillator that is locked to GPS time with an onboard G PS receiver (the corresponding GPS antenna is located in close proximity to the LMA box). At the ICLRT in 2011, the LMA board sampled the peak power of the received VHF signal in consecutive windows of 80 s length. At a sampling frequency of 25 MHz, the best possible time resolution is 40 ns. Hence, the time of the peak power within each 80 s window was recorded to 40 ns accuracy. This time and the corresponding received power were written to a 160 GB solid state hard disk. In one second time duratio n, there are a total of 12,500 individual 80 s windows. Once a minute, the LMA board adjusts the trigger threshold to yield an approximate average of 1000 triggers per second. A photograph of the inside of the LMA box is shown in Figure 2 33 C After a s torm, the hard disks from each LMA station were collected and the data were copied to RAID storage arrays at the ICLRT. The solutions were processed using software written for the Linux platform by the team at New Mexico Tech. The solution technique is a non linear least squares fitting algorithm (the Levenburg Marquardt algorithm) that is described in detail in Appendix A of Thomas et al. [2004]. The solution technique will be described in more detail in Section 4.8 as it pertains to obtaining three dim ensional source locations from dE/dt and x ray sources with the ICLRT TOA network. The solution algorithm produces lateral and altitude locations for sources detected on a minimum of five different stations. Typical LMA networks with 10 15 stations use a threshold of six stations for a valid solution, however, the ICLRT LMA network typically had only six operational stations during Summer 2011. The solution algorithm also returns a reduced chi squared value that is indicative of the goodness of fit of th e calculated solution to the measured arrival times. In displaying the data, the reduced

PAGE 127

127 chi squared value can be used as a thresholding metric to eliminate extraneous solutions that are likely not associated with the lightning discharge of interest. Ext raneous solutions can occur due to local noise sources and corona discharges in areas with high local electric fields. The processing algorithm also returns the VHF source power in dBW and the stations used in the solution. The source powers of the valid solutions can be used to infer the polarity of a single discharge, and also the general charge structure of a thundercloud by plotting the source powers of the LMA sources obtained for many discharges over a longer time window. Negative VHF sources typic ally radiate more strongly than positive VHF sources. The ICLRT LMA network is much smaller in total area than most LMA networks, and therefore is more efficient at resolving positive VHF sources during the ascent of the triggering wire and the subsequent UPL/ICC process. 2.16 SMART Radar From the end of June, 2011 through the second week of August, 2011, a C band dual polarimetric Shared Mobile Atmospheric Research and Teaching (SMART) radar [e.g., Biggerstaff et al ., 2005] was operated from the Keyston e Heights Airpark, located about 11.6 km to the south of the ICLRT. An image of the SMART radar is shown in Figure 2 34. The radar documented the hydrometeor and precipitation structure of the clouds over the study area. The radar was operated by Dr. Mi ke Biggerstaff and several graduate students from the University of Oklahoma. While storm conditions were present over the ICLRT, the radar operated in 0 60 elevation angle Range Height Indicator (RHI) scan mode over a five degree azimuthal sector center ed over the ICLRT collecting equivalent radar reflectivity factor (Z e ), Doppler velocity, spectrum width, differential radar reflectivity factor (Z DR ), differential phase ( DP ), and co polar correlation coefficient at zero lag ( HV ). These RHI sector volu me scans were collected every 90

PAGE 128

128 seconds. Data from the SMART radar were correlated with LMA data to examine the dependence of hydrometeor content on the geometrical progression of triggered lightning channels, and to infer the presence of high field regi ons in the cloud. 2.17 2009 Measurement Description A satellite view of the ICLRT is shown in Figure 2 35 A with all 2009 field measurements annotated in addition to the Launch Control trailer, Tower Launcher, and Office Trailer. General specifications f or each type of measurement are given in Table 2 3. There were a total of 75 measurements in 2009. A subset of the measurements shown in Figure 2 35 A comprised a small area, nine station TOA network utilized for determining the three dimensional spatial locations and emission times of both dE/dt (Section 2.13) and energetic radiation (Section 2.14) sources. A satellite view of the 2009 TOA network is provided in Figure 2 35 B The general photographic setup specifications for 2009 are given in Table 2 4 As discussed in Section 2.11, all rocket triggered lightning operations in 2009 were conducted from the Tower Launcher. There were two major additions to the ICLRT measurement network during Summer 2009, 1) a network of eight 1 m 2 plastic scintillators (Section 2.14.3) was installed at eight of the existing TOA stations in place of the shielded NaI detectors housed in the existing TERA boxes, and 2) two Lanthanum Bromide (LaBr 3 ) energetic radiation detectors were installed at the ninth TOA station. As d iscussed in Section 2.14.2, one of the LaBr 3 detectors was unshielded and the other was shielded with lead. The two LaBr 3 detectors were installed at Station 25, located immediately southwest of the Launch Control trailer ( Figure 2 35 B ). The outputs of b oth detectors were transmitted directly over double shielded RG 223 U coaxial cable to the oscilloscope inputs, providing better bandwidth and lower noise than signals transmitted over the normal Opticomm fiber optic links. A flat plate dE/dt antenna was installed about 10 m to the southwest of the LaBr 3 detectors at Station 25, comprising the ninth TOA station. As previously

PAGE 129

129 noted, both the plastic and LaBr 3 detectors have significantly faster time response than the NaI detectors that had previously been used by Howard et al. [2008] in the first attempt at employing the TOA technique to locate x ray sources in three dimensions. In 2009, high speed video images were acquired of triggered lightning discharges from the Office Trailer, a distance of 430 m fr om the Tower Launcher. The Phantom V7.3 and Photron SA1.1 high speed cameras were operated at frame rates from 5 10 kilo frames per second (kfps) and 50 300 kfps, respectively. The Phantom V7.3 was used primarily to image the ascent of the rocket, the s ustained upward positive leader, and subsequent explosion of the triggering wire. The Photron SA1.1 was used primarily to photograph leader processes within the bottom 130 m of the triggered lightning channel in addition to the attachment region. 2.18 2 010 Measurement Description A satellite view of the ICLRT is shown in Figure 2 36 A with all 2010 field measurements annotated in addition to the Launch Control trailer, Tower Launcher, Field Launcher, and Office Trailer. General specifications for each ty pe of measurement are given in Table 2 5. There were a total of 74 measurements in 2010. A satellite view of the 2010 T OA network is provided in Figure 2 36 B The general photographic setup specifications for 2010 are given in Table 2 6. From June 5, 2010 through July 11, 2010, all rocket triggered lightning operations were conducted from the F ield ( G round) L aunch er and after July 11, 2010, all rocket triggered lightning operations were conducted from the T ower L aunch er Prior to Summer 2010, a te nth TOA station (Station 17) was constructed about 40 m to the east of the Tower Launcher. The shielded LaBr 3 detector that had been installed at Station 25 was removed and placed in the TERA box at Station 17 as an unshielded detector. A flat plate dE/d t antenna was also installed about 10 m to the north of the TERA box at Station 17. As

PAGE 130

130 mentioned in Section 2.12.3, in August of 2010, the eight plastic scintillators were wrapped in Tyvec, effectively increasing the gain of the system by about a factor of two. In 2010, high speed video images were acquired of triggered lightning discharges from the Office Trailer, a distance of 300 m from the Field Launcher and 430 m from the Tower Launcher. The Phantom V7.3 and Photron SA1.1 high speed cameras wer e operated at frame rates of 8 kfps and 300 kfps, respectively. The Phantom V7.3 was used primarily to photograph the ascent of the rocket, the sustained upward positive leader, and subsequent explosion of the triggering wire. The Photron SA1.1 was used primarily to photograph leader processes within the bottom 130 m of the triggered lightning channel in addition to the attachment region. The Phantom V310 high speed cameras were used for site coverage. The cameras operated at frame rates of 3.2 kfps and 4.3 kfps respectively. 2.19 2011 Measurement Description A satellite view of the ICLRT is shown in Figure 2 37 A with all 2011 field measurements annotated in addition to the Launch Control trailer, Tower Launcher, Field Launcher, Office Trailer, and the new Optical Building. The Optical Building is located about 200 m to the northeast of the Field Launcher. General specifications for each type of measurement are given in Table 2 7. There were a total of 80 measurements in 2011. A satellite view of th e 2011 TOA network is provided in Figure 2 37 B There were numerous changes to the measurement network prior to Summer 2011. Station 17, a TOA station, was re located to a position about 20 m southeast of the Field Launcher. Station 12 was re located to the 2010 position of Station 17. Station 13 was re located to a position in line with and about equidistant to Station 7 and the Field Launcher. The electric field measurement network was completely revamped with a new inverted antenna design primarily for the purpose of using long decay time constant sensors with higher gain that otherwise would saturate the amplification electronics in the presence of rainfall.

PAGE 131

131 A picture of a 2011 inverted electric field antennas is shown in Figure 2 38. All electri c field measurements with the exception of the flat plate antenna at Station 2 were either re located or removed. A single inverted antenna was co located with a traditional flat plate antenna at Station 12, both with identical amplification electronics, which for the flat plate antenna provided a theoretical sensitivity of about 60 kV/m per digitizer volt. The sensitivity of the inverted antennas could not be calculated directly due to the combination of enhancement of the electric field from the elevate d plate and the shielding of the sensing plate by the grounded bowl above. Alternately, lightning return stroke field changes were measured on the co located flat plate antenna (with well characterized theoretical calibration, see Jerauld [200 7 ]) and inve rted antenna simultaneously to obtain a correction factor for the calibration of the inverted antenna. Results obtained during Summer 2011 indicate the sensitivity of the inverted antenna is about a factor of 3.27 less than the traditional flat plate assu ming identical amplification electronics. Two inverted antennas were designed specifically for recording precursor current pulses during the ascent of the grounded triggering wire (nominal sensitivity of about 300 V/m per digitizer volt and decay time con stant of about 10 ms) and were placed radially outward from the Field Launcher at Station 13 and Station 7. Two inverted antennas were designed for recording the shielding field change seen by the Campbell Scientific field mills during the ascent of the t riggering wire. These two antennas have nominal sensitivity of about 10 kV/m per digitizer volt and time constant of about 8 s and were placed radially outward from the Field Launcher at 60 m and 100 m respectively to the northwest (approximately co locat ed with two field mills). Finally, two inverted antennas were designed for measuring net charge transfers of positive and negative cloud to ground lightning discharges within 20 km of the ICLRT with nominal sensitivities of 10 kV/m and 3 kV/m per digitiz er volt, and time constants of 8 s and 6 s

PAGE 132

132 respectively. These two antennas were co located about 60 m north of the Field Launcher. The amplification electronics for all electric field measurements were designed and constructed by Robert Olsen III. As d escribed in Section 2.5, a redundant dE/dt network (the HBM dE/dt network) was constructed prior to Summer 2011. The outputs of the 10 existing flat plate dE/dt antennas were recorded on both the LeCroy DSOs in Launch Control and by the HBM Transient Reco rder in the Office Trailer. In addition to the dE/dt measurements, two channel base current measurements were also recorded on the HBM Transient Recorder. Setup specifications for the 10 dE/dt measurements and two channel base current measurements record ed on the HBM Transient Recorder are given in Table 2 8. There were many changes to the photographic setup at the ICLRT for Summer 2011. A network of eight Nikon D5000 still cameras were deployed on the periphery of the site in four different locations (Section 2.10) for imaging both natural and triggered lightning discharges. In addition, a new Cordin high speed camera (Section 2.9) was installed in the Optical Building for imaging the bottom 100 m of the leader phase and the attachment region of trig gered lightning discharges. The Phantom V7.3 high speed camera that had previously operated from the Office Trailer for photographing the ascent of the triggering wire and beginning of the sustained upward positive leader was relocated to a portable build ing immediately north of the Blast Wall on the far northeast corner of the site. The camera was located about 440 m from the Field Launcher. Finally, the Photron SA1.1 was used primarily to photograph leader processes within the bottom 130 m of the trig gered lightning channel in addition to the attachment region. The Photron was operated at frame rates of either 300 kfps or 450 kfps. Late in the summer, the Photron was angled upward to photograph the upward positive leader channel above the triggering wire. The

PAGE 133

133 camera imaged a region from about 215 m to 345 m above ground. The general photographic setup specifications for 2011 are given in Table 2 9. The 7 station LMA network (Section 2.15) recorded triggered and natural lightning events at the ICLRT beginning on June 23, 2011. The SMART radar (Section 2.16) captured radar scans during triggered and natural lightning events from June 23, 2011 to August 12, 2011.

PAGE 134

134 Figure 2 1. Perspective aerial view of the ICLRT. The manned operational facilities (Launch Control, Office Trailer, Optical Building), launching facilities (Tower Launcher, Field Launcher), and observation stations are labeled.

PAGE 135

135 Figure 2 2. An example of one of 25 individual ground based measurement stations at the ICLRT. Pictured is Station 5, located about 79 m from the Field Launcher and about 94 m from the Tower Launcher (pictured). Station 5 is equipped with flat plate electric field and electric field derivative (dE/dt) sensors, an unshielded NaI energetic ra diation detector, and a plastic energetic radiation detector. Lightning waveforms acquired from these sensors are transmitted to the central Launch Control trailer over 62.5 m multi mode fiber using Opticomm analog fiber optic links where they are digiti zed and stored. Photo courtesy of the author.

PAGE 136

136 Figure 2 3. PIC serial command and receive structures. A) The serial packet command structure sent by HAL to a given PIC controller in the field. B) The decomposition in bits of the hexadecimal serial c ommand (byte 5) in A. Each bit is coordinated with a different function of the PIC controller (power, calibration, attenuation). C) The serial packed receive structure returned to HAL by each PIC controller after a command packet is transmitted. The rec eived structure is parsed and displayed for the user to monitor the status of each measurement.

PAGE 137

137 Figure 2 4. Three generations of ICLRT PIC controllers. A) a 2001 PIC, B) a 2006 PIC, and C) a 2011 PIC. The fiber connectivity ports, battery power port s, output power ports, and channel I/O ports on each PIC controller are annotated. Photos courtesy of the author.

PAGE 138

138 Figure 2 5. Inside the rear door of the Launch Control trailer. The armored fiber bundles (far right) are stripped and the individua l fibers are terminated and coupled to Opticomm receiver cards. The optical fan out boards that broadcast control signals from HAL to the PIC controllers in the field are located in the middle rack. Photo courtesy of the author.

PAGE 139

139 Figure 2 6. Front ra ck space in Launch Control that supports the majority of the DSO network. The DSOs are annotated by scope number. Scopes 12, 13, 14, 17, 17, 20, 21, 28, and 29 (all LeCroy models) record the outputs of the dE/dt antennas, plastic energetic radiation dete ctors, and LaBr 3 energetic radiation detectors that comprise the TOA network, a portion of the MSE. Scopes 36 and 37 also record the outputs of the eight plastic energetic radiation detectors, but trigger independently on cosmic ray air showers. Note the above photograph depicts the 2012 installation of the DSO network. Photo courtesy of the author.

PAGE 140

140 Figure 2 7. The HBM digitization system. A) photograph of the HBM GEN16t transient recorder located in the Office Trailer, and B) photographs of the H BM 7600 Isolated Digitizer. Photos courtesy of the author.

PAGE 141

141 Figure 2 8. Screen capture of the NLDN display on the LTS2005 software during a storm on May 15, 2012. Cloud to ground and cloud lightning discharges are plotted according to their polarity and occurrence time. The location of the ICLRT is annotated.

PAGE 142

142 A B Figure 2 9. Campbell Scientific field mills. A) locations of the 8 Campbell Scientific field mills overlaid on an aerial photograph of the ICLRT. The naming of the field mills is consistent with Table 2 1. B) annotated photograph of the NE80 field mill. The Field Launcher is in the background. Photos courtesy of the author.

PAGE 143

143 Table 2 1. The absolute spatial positions of all Campbell Scientific field mills were measured to an accuracy of 1 cm to the center of the sensing plate using a Differential Global Positioning System (DGPS). Coordinate measurements were bas ed on the Florida North NAD83 data. ICLRT spatial coordinates are obtained by subtracting the absolute spatial coordinates from the ICLRT coordinate origin, which is located on the far southwest corner of the site. The distances from each field mill to b oth launching facilities are also given. Absolute Spatial Coordinates ICLRT Spatial C oordinates Field Mill Latitude Longitude Altitude (m) Easting (m) Northing (m) Altitude (m) Distance From Field Launcher (m) Distance From Tower Launcher (m) NE35 29 56 35.76074 82 01 58.81204 70.351 325.694 481.157 3.30 38.536 135.354 NE80 29 56 36.6 7610 82 01 57.62828 70.497 356.820 510.022 3.45 80.746 128.763 NE130 29 56 38.18070 82 01 56.60836 71.024 383.161 556.931 3.97 133.631 154.250 NW30 29 56 35.37842 82 02 00.85457 69.936 271.190 468.203 2.89 35.145 180.632 NW60 29 56 35.69977 82 02 01.802 95 69.955 245.550 477.545 2.90 62.226 207.889 NW100 29 56 35.80353 82 02 03.06405 69.723 211.671 480.008 2.67 94.699 241.071 LC 29.94272413 82.03111869 72.527 508.892 424.948 5.48 208.720 65.786 OT 29.94417846 82.03476245 70.302 153.731 578.507 3.25 195 .447 332.959

PAGE 144

144 Figure 2 10. The field installation of the Northeast Optical detector (NEO). The signals from the two optical detectors on the northeast and southwest corners of the ICLRT generate a trigger signal for the DSOs and high speed cameras when natural or triggered lightning return strokes of sufficient brightness occur within the field of view of both detectors. Photo courtesy of the author.

PAGE 145

145 Figure 2 11. ICLRT Master Trigger Box installed in the front r ack in Launch Control. Photo courtesy of the author.

PAGE 146

146 Figure 2 12. Photographs of ICLRT high speed cameras. A) Photron SA1.1 B) Phantom V7.3, and C) Cordin 550 high speed cameras. Specification for each camera are given at right. Timing is obtain ed for the Cordin framing camera by digitizing the framing output on a DSO with a timebase sychronized to IRIG B. A circuit is currently being tested so the Cordin camera will only trigger on dart stepped leaders. Presently the camera always triggers on the first return stroke. Photos courtesy of the author.

PAGE 147

147 Figure 2 13. A block diagram of the full ICLRT instrument triggering topology. The diagram corresponds to the 2011 measurement network. Blocks are color coded by instrument type according to th e key at the bottom of the figure.

PAGE 148

148 A B Figure 2 14. ICLRT DSLR still cameras. A) an aerial perspective view of the ICLRT showing the locations of the still cameras. The fields of view of the site coverage cameras are sho wn. B) Nikon D5000 DSLR still camera installation at the Blast Wall observation station. One camera operates in a site coverage mode (top camera ) and the one camera operates in a triggered lightning mode (bottom camera ). Both cameras are equipped with 6 stop neutral density filters and circular polarizing filters. The shutters of both cameras are controlled by the 2011 PIC controller. The cameras are AC powered from a large inverter. Photos courtesy of the author.

PAGE 149

149 Figure 2 15. An example of a fibe rglass rocket used at the ICLRT to trigger lightning. The rockets elevates the grounded kevlar coated copper wire at a peak velocity of about 150 m/s. The nose cone is ejected at the end of the rocket motor burn by a small parachute charge. Photo courte sy of the author.

PAGE 150

150 Figure 2 16. The Tower Launching Facility. A) a photograph of the Tower Launching facility. The Launch Control trailer is pictured in the background. The Tower Launcher consists of 12 aluminum rocket tubes on top of an 11 m tall wooden tower. The top of the rocket tubes are 14 m above local ground level. B) the launcher control box mounted between the rocket tubes. C) the rocket launcher box in Launch Control. D) bottom right, an individual rocket tube control box (note the tu be is not loaded and there is no squib wire present). Photos courtesy of the author. B C D A

PAGE 151

151 Figure 2 17. 1/20 th scale model experiment of the lightning protection system (LPS) over Pad 39B at Kennedy Space Center. Lightning return strokes were triggered to the intercepting rod atop the support pole closest to the ground based Field Launcher. Return stroke currents were measured in the large box atop the support pole and initial stage currents were measured in the large box mounted directly underneath th e launcher. Return stroke currents flowed through the catenary wire system to ground through a network of nine down conductors. Photo courtesy of the author.

PAGE 152

152 Figure 2 18. T he 201 0 Field Launch er A) the 2010 Field Launcher has 6 aluminum rocket tubes and is supported by tubular fiberglass legs. T he launcher control box is mounted to the right of the launch tubes. The initial stage current measurement box is pictured. B) the inside of the launcher control box. C) the inside of one of the tube control boxes. Photos courtesy of the author. B C A

PAGE 153

153 Figure 2 19. Block diagram of the rocket launching system for the Field Launcher (2010, 2011). The system for the Tower Launcher is very similar.

PAGE 154

154 Figure 2 20. The 2011 Field Launcher with intercep ting wire ring and down conductors. The large current measurement box was mounted beneath the launcher on supporting ceramic insulators. The pneumatic supply (air compressor) was moved to a small shed to the southwest of the launcher. Rocket tube extende rs were added to help the rocket track more straightly upon exiting the tube. Photo courtesy of the author.

PAGE 155

155 Figure 2 21. The Tower Launcher current measurements. A) photograph of the Tower Launcher current measurement box (2009 configuration). B) an electrical schematic of the current measurement box. Note that fiber optic connectivity is not shown in this diagram. Photo courtesy of the author. A B

PAGE 156

156 Figure 2 22. 2010 Initial Stage current measurements. A) electrical schematic of Initial Stage current measurement box during 2010 Pad 39B catenary wire experiment. The box was mounted to the underside of the Field Launcher. B) electrical schematic of the corresponding Return Stroke current measurement box. The box was mounted atop the westernmos t support pole on a wooden platform. A B

PAGE 157

157 Figure 2 23. 2011 current measurements. A) a photograph of the 2011 current measurement box mounted underneath the Field Launcher. B) an electrical schematic of the 2011 current measurement box. The box contai ned a total of six current measurements, four on the ICLRT digitization system and two on the HBM digitization system. Photo courtesy of the author. A B

PAGE 158

158 Figure 2 24. dE/dt flat plate antenna. A) a schematic of the dE/dt flat plate antenna installation B) a photograph of the dE/dt flat plate antenna (dE 7). Photo courtesy of the author. A B

PAGE 159

159 Figure 2 25. A vertical cross section of the dE/dt flat plate antenna installation.

PAGE 160

160 Figure 2 26. dE/dt measurement installation. A) a schematic of the dE/dt measurement electronics box installation. B) a corresponding top down photograph of the electronics boxes resting on the support structure (schematic shown in Figure 2 25). C) the inside of the ICLRT dE/dt measurement enclosure. D) the inside of t he HBM dE/dt measurement enclosure. Photos courtesy of the author. A B C D

PAGE 161

161 Figure 2 27. TERA box installation. A) photograph of the exterior of a TERA box (Station 12). B) an image of interior of a TERA box. C) a schematic drawing of the electronics and NaI detectors inside a TERA box. Photos courtesy of the author. A B C

PAGE 162

162 Figure 2 28. A photograph of the LaBr 3 detector TERA box (Station 17). Photo courtesy of the author.

PAGE 163

163 Figure 2 29. Plot of the single photon responses of the NaI (green trace) and LaBr 3 (blue trace) detectors to a Cs 137, 662 keV source. The vertical scale is in raw digitizer volts. The higher noise level on the NaI measurement is a result of the analog fiber optic link. The output of the LaBr 3 detector is trans mitted directly over double shielded coaxial cable.

PAGE 164

164 Figure 2 30. Plastic scintillator installation. A) photograph of the field installation of the plastic detectors. Signal and power cables pass through the galvanized pipe from the TERA bo x to outer box of the plastic detector. B) photograph of the inner box that holds the plastic detector. The signal and power feed through panel on the inner box is annotated at far left. Photos courtesy of the author. A B

PAGE 165

165 Figure 2 31. Interior of pla stic scintillator measurement box. A) photograph of the plastic scintillator during initial installation in 2009 (without Tyvec and with one PMT). B) photograph of the PMT mated to the plastic scintillator. C) schematic of the 2011 layout of the plastic detectors including the modified amplification electronics. Photos courtesy of the author. A C B

PAGE 166

166 Table 2 2. In Columns 2 4, the absolute spatial coordinates of all seven LMA stations (Columns 2 4) obtained from the GPS receiver at each station. In Columns 5 7, the coordinates of the LMA stations relative to the ICLRT coordinate or igin. ICLRT spatial coordinates are obtained by subtracting the absolute spatial coordinates from the ICLRT coordinate origin, which is located on the far southwest corner of the site. The distances from each LMA station to the Field Launcher are given i n Column 8 and the azimuth angle from the Field Launcher to each LMA station is given in Column 9. Absolute Spatial Coordinates ICLRT Spatial Coordinates LMA Station Latitude Longitude Altitude (m) Easting (m) Northing (m) Altitude (m) Distance From Field La uncher (m) Azimuth Angle From Field Launcher (degrees) Blast Wall 29.94474 82.02894 61.74 714.33 652.91 5.31 459.32 64 Golf 29.94023 82.06406 48.50 2664.53 80.15 18.55 2989.41 277 DupontS 29.89488 82.04172 50.94 399.96 4899.28 16.11 5396.03 188 DupontN 30.01243 82.03605 52.59 134.02 8139.53 14.46 7700.95 357 FDOT 29.96578 82.08430 37.50 4678.06 2870.08 29.55 5536.36 296 Warehouse 29.91754 82.12840 45.15 8822.21 2565.36 21.90 9609.66 252 Blanding 29.92448 82.01397 69.97 2208.08 1560.90 2.92 2771.53 134

PAGE 167

167 Figure 2 32. Plan view of the seven LMA station locations. The location of the Field Launcher is annotated.

PAGE 168

168 Figure 2 33. LMA statio n installation. A) photograph of the Blast Wall LMA antenna. B) the LMA electronics enclosure, located in the Blast Wall observation station. The LMA antenna input, GPS antenna input, and power input enter the box through a panel on the right side of th e box. C) the inside of the LMA electronics enclosure. Photos courtesy of the author. A B C

PAGE 169

169 Figure 2 34. A photograph of the C band dual polarimetric SMART Radar P hoto courtesy of Kyle Thiem, University of Oklahoma.

PAGE 170

170 Figure 2 35. 2009 ICL RT measurement network. A) aerial photograph of the ICLRT with all 2009 measurements annotated. B) aerial photograph of the 9 station 2009 TOA network measurements, a subset of the measurements shown in the top photograph. The measurements are labeled a ccording to the keys at the bottom of each photograph. A B

PAGE 171

171 Table 2 3. 2009 ICLRT measurement specifications. The measurement sensitivity ranges for each type of measurement in units (Column 2) per digitizer volt are given in Column 3. The DSO timebase and digitization set tings are given in Columns 5 7. The number identificati ons and models of the DSOs ( Figure 2 6) that digitize each type of measurements are given in Columns 8 9. Measurement Units Sensitivity (Units/V) # of Channels Sampling Frequency Record Length Memory Segments Oscilloscope Numbers Oscilloscope Model Electric Field kV/m .069 76 10 10 MS/s 2 s 1 22, 24, 25 Yokogawa DL750 Electric Field Derivative kV/m/s 30 9 250 MS/s 250 MS/s 5 ms 2 ms 10 2 18, 20, 21 14, 17 LeCroy 44 Xi LeCroy Waverunner I Magnetic Field W/ m 2 30 4 10 MS/s 2 s 1 22, 24, 25 Yokogawa DL750 Channel Base Current A 21 63000 3 10 MS/s 2 50 MS/s 2 s 5 ms 1 10 22, 24, 25, 30 26 Yokogawa DL750 LeCroy 44 Xi Optical 2 10 MS/s 2 s 1 23 Yokogawa DL750 X ray (NaI unshielded) MeV 2.6 10.5 24 10 MS/s 10 MS/s 250 MS/s 2 s 1.6 s 2 ms 1 1 2 23, 30 10, 19 12, 13, 17 Yokogawa DL750 Yokogawa DL 716 LeCroy Waverunner I X ray (NaI shielded) MeV 4.2 16.8 13 10 MS/s 10 MS/s 2 s 1.6 s 1 1 23 10, 19 Yokogawa DL750 Yokogawa DL716 X ray (Plastic) MeV 8 250 MS/s 250 MS/s 5 ms 2 ms 10 2 28, 29 14 LeCroy 44 Xi LeCroy Waverunner I X ray (LaBr 3 ) MeV 4.4 2 250 MS/s 5 ms 10 18 LeCroy 44 Xi

PAGE 172

172 Table 2 4. 2009 ICLRT photographic setup specifications. Camera Location Frame Rate or Exposure Length Focal Length Lens Aperture Filters Purpose P hotron SA1.1 Office Trailer 50 300 kfps 20 or 24 mm f/2.8 None Attachment Region Phantom V7.3 Office Trailer 8 kfps 20 or 24 mm f/4 f/11 None Full Lightning Channel Sony HDR HC5 IS2 30 fps Variable Variable None Midrange Lightning Channel Sony HDR HC5 Office Trailer 30 fps Variable Variable None Full Lightning Channel Sony HDR HC5 Launch Control 30 fps Variable Variable None Attachment Region Sony HDR HC5 Runway 30 fps Variable Variable None Midrange Lightning Channel Nikon D80 Launch Control 6 s 70 mm f/32 4 stop ND, CPOL Attachment Region

PAGE 173

173 Figure 2 36. 2010 ICLRT measurement network. A) aerial photograph of the ICLRT with all 2010 measurements annotated. B) aerial photograph of the 10 station 2010 TOA network measurements, a subset of the measurements shown in the top photograph. The measurements are labeled according to the keys at the bottom of each photograph. A B

PAGE 174

174 Table 2 5. 2010 ICLRT measurement specifications. The measurement sensitivity range s for each type of measurement in units (Column 2) per digitizer volt are given in Column 3. The DSO timebase and digitization settings are given in Columns 5 7. The number identificati ons and models of the DSOs ( Figure 2 6) that digitize each type of me asurements are given in Columns 8 9. Measurement Units Sensitivity (Units/V) # of Channels Sampling Freque ncy Record Length Memory Segments Oscilloscope Numbers Oscilloscope Model Electric Field kV/m .069 76 10 10 MS/s 2 s 1 22, 24, 25 Yokogawa DL750 Electric Field Derivative kV/m/s 30 10 250 MS/s 250 MS/s 5 ms 2 ms 10 2 18, 20, 21 14, 17 LeCroy 44 Xi LeC roy Waverunner I Magnetic Field W/ m 2 30 4 10 MS/s 2 s 1 22, 24, 25 Yokogawa DL750 Channel Base Current A 21 63000 3 10 MS/s 250 MS/s 2 s 5 ms 1 10 22, 24, 25, 30 26 Yokogawa DL750 LeCroy 44 Xi Optical 2 10 MS/s 2 s 1 23 Yokogawa DL750 X ray (Na I unshielded) MeV 2.6 10.5 24 10 MS/s 10 MS/s 250 MS/s 2 s 1.6 s 2 ms 1 1 2 23, 30 10, 19 12, 13, 17 Yokogawa DL750 Yokogawa DL716 LeCroy Waverunner I X ray (NaI shielded) MeV 4.2 16.8 13 10 MS/s 10 MS/s 2 s 1.6 s 1 1 23 10, 19 Yokogawa DL750 Yoko gawa DL716 X ray (Plastic) MeV 8 250 MS/s 250 MS/s 5 ms 2 ms 10 2 28, 29 14 LeCroy 44 Xi LeCroy Waverunner I X ray (LaBr 3 ) MeV 4.4 2 250 MS/s 5 ms 10 18 LeCroy 44 Xi

PAGE 175

175 Table 2 6. 2010 ICLRT photographic setup specifications. Camera Location Frame Rate or Exposure Length Focal Length Lens Aperture Filters Purpose Photron SA1.1 Office Trailer 300 kfps, 14 or 20 mm f/4 None Attachment Region Phantom V7.3 Office Trailer 8 kfps 20 mm f/5.6 f/8 None Full Lightning Channel Phantom V310 Office Trailer 3.2 kfps 24 mm f/16 None Site Coverage Phantom V310 Launch Control 4.3 kfps 24 mm f/16 None Site Coverage Canon HF S20 Launch Control 30 fps Variable Variable None Attachment Region Canon HF S20 Office Trailer 30 fps Variable Variable None Full Lightning Channel Nikon D80 Launch Control 6 s 70 mm f/32 4 stop ND, CPOL Attachment Region

PAGE 176

176 Figure 2 37. 2011 ICLRT measurement netwo rk. A) aerial photograph of the ICLRT with all 2011 measurements annotated. B) aerial photograph of the 10 station 2011 TOA network measurements, a subset of the measurements shown in the top photograph. The measurements are labeled according to the ke ys at the bottom of each photograph. A B

PAGE 177

177 Table 2 7. 2011 ICLRT measurement specifications. The measurement sensitivity ranges for each type of measurement in units (Column 2) per digitizer volt are given in Column 3. The DSO timebase and digitization settings ar e given in Columns 5 7. The number identificati ons and models of the DSOs ( Figure 2 6) that digitize each type of measurements are given in Columns 8 9. a,b Note that six electric field measurements and two sensitive channel base current measurements we re digitized continuously on the NI 6212 DAQ card that runs on HAL. Measurement Units Sensitivity (Units/V) # of Channels Sampling Frequency Record Length Memory Segments Oscilloscope Numbers Oscilloscope Model Electric Field kV/m 0.3 180 9 10 MS/s 50 KS/s 2 s Cont. 1 Cont. 22, 24, 25 Yokogawa DL750 NI 6212 DAQ a Electric Field Derivative kV/m/s 15 10 250 MS/s 250 MS/s 5 ms 2 ms 10 2 18, 20, 21 14, 17 LeCroy 44 Xi LeCroy Waverunner I Channel Base Current A 0.4 63000 4 10 MS/s 250 MS/s 50 KS/s 2 s 5 ms Cont. 1 10 Cont. 22, 24, 2 5, 30 26 Yokogawa DL750 LeCroy 44 Xi NI 6212 DAQ b Optical 2 10 MS/s 2 s 1 23 Yokogawa DL750 X ray (NaI unshielded) MeV 2.6 10.5 24 10 MS/s 10 MS/s 250 MS/s 2 s 1.6 s 2 ms 1 1 2 23, 30 10, 19 12, 13, 17 Yokogawa DL750 Yokogawa DL716 LeCroy Waver unner I X ray (NaI shielded) MeV 4.2 16.8 13 10 MS/s 10 MS/s 2 s 1.6 s 1 1 23 10, 19 Yokogawa DL750 Yokogawa DL716 X ray (Plastic) MeV 8 250 MS/s 250 MS/s 5 ms 2 ms 10 2 28, 29, 36, 37 14 LeCroy 44 Xi LeCroy Waverunner I X ray (LaBr 3 ) MeV 4.4 2 25 0 MS/s 5 ms 10 18 LeCroy 44 Xi

PAGE 178

178 Table 2 8. 2011 HBM digitization system measurement specifications. Note the electric field derivative measurements digitized on the HBM system were a factor of two more sensitive than those recorded on the ICLRT digitization system. Measurement Units Peak Amplitude Range # of Channels Sampling Frequency Record Length Memory Segments HBM Recorders Electric Field Derivative kV/m/s +/ 8 10 100 MS/s 20 ms Infinite C, D, E, F, G, H Ch annel Base Current A +/ 200 +/ 50000 1 1 100 MS/s 100 MS/s 20 ms 20 ms Infinite Infinite H C

PAGE 179

179 Figure 2 38. 2011 inverted electric field antenna (Station 7). Photo courtesy of the author.

PAGE 180

180 Table 2 9. 2011 ICLRT photographic setup specifications. Camera Location Frame Rate or Exposure Length Focal Lengt h Lens Aperture Filters Purpose Photron SA1.1 Office Trailer 300 kfps, 14 mm f/4 None Attachment Region Phantom V7.3 Office Trailer 8 kfps 20 mm f/5.6 f/8 None Full Lightning Channel Canon HF S20 Office Trailer 30 fps Variable Variable None Site Covera ge Canon HF S20 Office Trailer 30 fps Variable Variable None Site Coverage Canon HF S21 Office Trailer 30 fps Variable Variable None Full Lightning Channel Canon HF S21 Launch Control 30 fps Variable Variable None Attachment Region Nikon D5000 Blast Wa ll 5 10 s 10 mm f/18 6 stop ND, CPOL Site Coverage Nikon D5000 Blast Wall 6 8 s 18 mm f/22 6 stop ND, CPOL Full Lightning Channel Nikon D5000 IS2 5 10 s 10 mm f/18 6 stop ND, CPOL Site Coverage Nikon D5000 IS2 6 8 s 90 mm f/22 6 stop ND, CPOL Midrange L ightning Channel Nikon D5000 SWO 5 10 s 10 mm f/18 6 stop ND, CPOL Site Coverage Nikon D5000 SWO 6 8 s 18 mm f/22 6 stop ND, CPOL Full Lightning Channel Nikon D5000 IS4 5 10 s 10 mm f/18 6 stop ND, CPOL Site Coverage Nikon D5000 IS4 6 8 s 86 mm f/22 6 stop ND, CPOL Midrange Lightning Channel Nikon D80 Launch Control 6 s 100 mm f/32 4 stop ND, CPOL Attachment Region

PAGE 181

181 CHAPTER 3 TIME OF ARRIVAL MEASUREMENTS The majority of the data presented in this dissertation are related to time of arrival (TOA) measurements using, 1) wideband electric field derivative (dE/dt) waveforms obtained from flat plate antennas, 2) energetic radiation waveforms from both plastic and LaBr 3 scintillation detectors, and 3) narrowband (66 72 MHz) VHF radiation detected by the Lightning Mapping Array (LMA) sensors. These three measurement systems, which record spectral ranges from near DC to the x ray/gamma ray wavelengths, are all configured to study the radiation associ ated with propagating lightning leaders, and the case of the first two systems, the attachment of these lightning leaders to ground or ground based objects. The principle of any three dimensional TOA system is to measure a common radiated source, which is assumed to occur at a single point and time in space ( x, y, z, t) at numerous spatial locations ( x i y i z i ) and then, using the measured arrival times at each station ( t i ), assuming the signal propagated in a straight line and at the speed of light c, calculate the position and emission time of the source. The governing equation, often referred to as the "TOA Equation" is given by Equation 3.1. (3.1) Considering lightning leaders propagate in a three dimensional space, in theory, it is necessary to measure a radiated source at only four stations in order to reconstruct the spatial position of the source. In practice though, the physical geometry of the measuring network, the geometry of the network relative to the source, and the associated inherent errors in sensor location and the selected arrival times of the radiated signal make a four station network effective in only very idealized cases. A four station network provides only one possible combination of measured arri val times by which to reconstruct the three dimensional source location.

PAGE 182

182 Systems for locating lightning discharges via their radiated electromagnetic fields have been utilized since the early part of the 20th century. Early lightning locating systems us ed crossed loop magnetic field antennas to estimate the azimuth angle to the causative discharge. A thorough discussion of the history of magnetic direction finding (MDF) systems is given in Rakov and Uman [2003, Ch. 17.3]. The original TOA systems, such as the implementation described in Lewis [1960], primarily utilized long baseline (often greater than 100 km) networks of sensors that operated in the VLF/LF frequency range. Lewis [1960] used a network of four stations arranged in a triangular geometry around a central station. Lightning sferics were sensed by vertical antennas operating in the 4 45 KHz band. The data were relayed continuously to the central station over a microwave link. The stations were located in New England and were used to locat e sferics from lightning discharges in western Europe. The geometry of the sensor baseline made the network efficient at determining the azimuth to the causative discharge in the north south direction (perpendicular to the direction of propagation). Lewi s [1960] obtained azimuths for 150 sferics radiated from western Europe using this system. The azimuths were compared to ground strike estimates for the same causative lightning discharges obtained by the magnetic cross loop DF network operated by the Bri tish Meteorological Office (BMO). The BMO network reported fixes for the ground termination points of lightning discharges within about 30 nautical miles in the north south direction and about 18.5 nautical miles in the east west direction. The azimuth a ngles obtained by Lewis [1960] placed the causative discharges within an average absolute deviation of about 31 nautical miles from the positions reported by the BMO. Other long baseline TOA systems were developed in the later part of the 20th century th at provided actual two dimensional ground strike locations of individual lightning discharges. Lee

PAGE 183

183 [1986] operated a seven station TOA network operating in the 2 18 KHz frequency band with stations located in the United Kingdom, Gibraltar and Cyprus. The stations were separated by 250 3300 km. The system provided lightning flash locations at a rate not exceeding 400 per hour. The estimated system error was 2 20 km. In the late 1980s, the LPATS (lightning positioning and tracking system) network [e.g., Casper and Bent, 1992; Bent and Lyons, 1984] became the first commercially available TOA system. The system was developed by the Atmospheric Research Systems, Inc. (ARSI). LPATS also operated in the VLF/LF frequency band and consisted of four or more st ations separated by typical distances of 200 400 km. The system measured the differences in signal arrival times at the stations with time synchronization (of the order of 100 ns) provided by Loran C (and later GPS). Since 1987, LPATS systems have operat ed in 18 countries spanning five continents. In the early 1990s, a network of 59 of the LPATS sensors were combined with 47 IMPACT MDF sensors (manufactured by Lightning Location and Protection, Inc.) to form the 106 sensor National Lightning Detection Ne twork (NLDN). The combined benefits of the two types of sensors (MDF/TOA) provided a significant improvement in the location accuracy of the system (median accuracy of 500 m) and flash detection ratio (90% for events with peak currents above 5 kA) [e.g., Cummins et al. 1998]. The systems mentioned above operate in the VLF/LF frequency band and provide estimates for the ground strike locations of lightning discharges. They do not, however, provide any information about the physical channel structure (a nd underlying physics) of individual discharges. Systems capable of mapping the progression of lightning leader channels in three dimensions have traditionally operated in the VHF frequency range, which is typically associated with air breakdown processes The first three dimensional VHF lightning mapping system was developed in South Africa by D.E. Proctor in the early 1970s. Proctor [1971, 1981, 1983] and

PAGE 184

184 Proctor et al. [1988] used short base line (10 40 km) systems with five ground stations arranged o n two nearly perpendicular base lines. The fifth station was used to provide a redundant measurement. The systems operated at 250 MHz [e.g., Proctor 1971], 253 MHz [e.g., Proctor, 1981] or 355 MHz [e.g., Proctor, 1983; Proctor et al. 1988] with a 5 MHz bandwidth. Proctor's network had spatial resolution of about 100 m. Source coordinates were determined based on the hyperbolic approach. The equations were solved using a non linear iterative process to find the optimal source location from an initial guess, though the derivations were highly complex and only applicable to very idealized network geometries. Lateral source coordinates were determined with errors of the order of 25 m and altitude source coordinates were determined with errors of about 10 0 m for sources well above ground, but not more than several kilometers in altitude. The network was capable of recording a source location about every 70 s. A similar VHF lightning mapping system, the LDAR (Lightning Detection and Ranging), was built a t the Kennedy Space Center and is described in Lennon and Poehler [1982] and Maier et al. [1995]. The LDAR was comprised of seven stations that operated in the 56 75 MHz frequency band. Six of the stations were arranged in a concentric ring with diamete r of 16 km and the seventh station was located at the center of the measurement array. The geometry effectively formed two y shaped sensor networks, each including three outer sensors and the central sensor. Each station recorded 82 s segments of data w hen a pre defined trigger threshold was exceeded. The data were transmitted via a microwave link to the central station where they were processed as individual four station sets of arrival times for the two y shaped networks. The arrival time from each s tation was taken as the time of the peak signal recorded at the given station within the 82 s window. The differences in arrival times were used to calculate the three dimensional position of the source. Events were discarded if the solutions from both sets of four stations

PAGE 185

185 were not self consistent within a defined tolerance. The processing of the LDAR solutions was modified in the early 1990s to include all 20 of the possible four station combinations. The best solution was determined using a voting p rocess with weights attached to each four station combination. Despite the improved processing technique, Starr et al. [1998] estimated that the LDAR discarded about 60 % of detected events. Poehler [1977] used geometric dilution of precision (GDOP) formu lations and an assumed timing error of 20 ns to estimate the LDAR lateral coordinate errors to be about 7 11 m rms for sources below about 8 km in altitude. The LDAR system typically located anywhere from 10 50 sources per lightning flash. A thorough ana lysis of the performance characteristics of the LDAR system is given in Boccippio et al. [2001a, 2001b]. The relative success of the LDAR system prompted the development of a second VHF mapping system, the Lightning Mapping Array (LMA) [e.g., Rison et al ., 1999; Krehbiel et al., 2000; Thomas et al., 2004], in the mid 1990s. The LMA was conceived and developed by Dr. Paul Krehbiel, Dr. Bill Rison, and Dr. Ron Thomas at the New Mexico Institute of Mining and Technology. The methodology and operation of th e LMA system was described in Section 2.15. The general operation of the LDAR and the LMA are not in principle different. Both systems determine the time of the peak power of a received VHF impulse within a given time window and use the measured signal a rrival times at different stations to calculate the spatial position and emission time of the source. The main advantage of the LMA system is that the time base at each station is controlled by an onboard GPS receiver, the one pulse per second (1 PPS) out put of which is used to phase lock a 25 MHz clock that determines the endpoints of the data acquisition windows. For each data point above a defined threshold, the LMA simply stores the time and peak power of the pulse at the LMA station. In contrast, th e LDAR transmits the

PAGE 186

186 detected analog waveforms over a microwave link to a central station, introducing many sources of potential timing error. LMA systems were installed in Oklahoma and at Langmuir Lab in New Mexico in the late 1990s and later in northern Alabama. Typical LMA networks consist of 10 15 stations spread over a typical ground area of 60 km in diameter. Most LMA networks operate in the 60 66 MHz frequency band (Channel 3). The locations of sources recorded by six or more stations are initial ly determined using the hyperplane technique [e.g., Koshak and Solakiewicz 1996; Koshak et al., 2004], the output of which is used as an initial solution guess for a non linear least squares minimization technique for the over determined set of equations [e.g. Thomas et al., 2004]. Sources occurring over the LMA network at 6 12 km altitude are typically located with a horizontal uncertainty of 6 12 m rms and vertical uncertainty of 20 30 m rms. The rms timing error of the LMA network is typically 40 50 ns With 80 s data acquisition windows, the LMA often locates several thousand sources per lightning flash, a significant improvement over the LDAR system. A limitation of the LMA system is the poor altitude determination of low altitude sources. The alt itude error of any three dimensional TOA system is principally dependent on the ratio of the source height to the horizontal distance from the source to the closest sensor. Large area LMA networks are thus ideally suited for studying in cloud processes th at occur two kilometers or more above the plane of the network. A third type of TOA system was developed by Thomson [1994] and was installed at the Kennedy Space Center. This system differed from the LDAR and LMA networks described above in that it dete cted the broadband dE/dt signature radiated by propagating lightning leaders. The network consisted of four outlying stations covering a ground area of approximately 15 x 15 km and one central station. The outer stations had bandwidth from 800 Hz to 4 MH z and the central station had bandwidth from 800 Hz to 2 MHz. When the system was triggered, data frp,

PAGE 187

187 the outer stations were transmitted to the central station where they were digitized at 20 MS/s in 204.8 s segments. As many as 25 data segments could be recorded per lightning flash with as little as 40 s dead time between each subsequent data segment. The signal arrival times for common pulses at the five stations were determined to be the mean of the three waveform features for a given dE/dt pulse, 1) the rising portion half peak, 2) the peak, and 3) the falling portion half peak. The source locations and emission times were computed using a weighted hyperbola technique to solve the over determined set of equations. Thomson [1994] estimated that t he location error was less than 100 m for sources within the network. 3.1 ICLRT TOA Network The limitations of the previously described long baseline VHF TOA systems prompted the design and implementation of a small area dE/dt TOA system at the ICLRT sui table for locating radiated sources from lightning leaders with a high degree of accuracy and without any windowing timing constraints. The first dE/dt TOA system at the ICLRT capable of reproducing three dimensional "images" of lightning leaders was cons tructed by Dr. Joseph Howard and the author at the beginning of summer 2006. Previously, a network of four dE/dt antennas had been used at the ICLRT to calculate the approximate strike points of natural lightning discharges terminating on or near the site [e.g., Jerauld, 2007]. The original four antennas were located at Station 1, Station 4 Station 8, and Station 9 ( Figure 3 37). The measured signal arrival times at the four stations, obtained from measurement of the peak dE/dt associated with the retur n stroke, provided a set of four possible three station combinations. The antennas were assumed to be exactly co planar, removing the altitude dimension from the problem. Each three station combination ( i, j, k) yielded two time difference of arrival (T DOA) equations, given in Equation 3.2.

PAGE 188

188 (3.2) Each TDOA equation represents a hyperbola in two dimensions. The intersection of the two hyperbolas, which can be obtained numerically, gives the ground strike location ( x, y) of the source. For some cases, the two hyperbolas intersect at two locations, giving a non determinate solution. Then, the other three station combinations are used to eliminate the non physical location. The lateral coordinates of the strike point obtaine d from all available three station combinations were averaged to obtain the final strike point estimate. An additional four flat plate dE/dt antennas were installed at the beginning of summer 2006 to convert the existing two dimensional TOA system to a network capable of mapping lightning leader sources in three dimensions. The new antennas were located at Station 3, Station 5, Station 7, and Station 11 (Figure 2 37), forming two north/south baselines to complement the existing east/west baseline. Th e total enclosed network area was about 0.25 km 2 Unshielded NaI energetic radiation detectors (Section 2.14.1) were installed in TERA boxes at each TOA station within about 10 m of the flat plate dE/dt antenna. Howard et al. [2008, 2010] used the calcul ated TOA locations from the new 8 station network to initially establish the spatial and timing relationship between the dE/dt and x ray sources from stepped leader steps, and to analyze the sub microsecond structure of the lightning attachment process. In 2009 and 2010, the TOA network was substantially modified. Many of the additions and changes were described in Chapter 2. The network expanded to ten stations, each with a flat plate dE/dt antenna and either a plastic or LaBr 3 energetic radiation dete ctor. The additional stations (Station 17 and Station 25) were added in close proximity to the rocket launching

PAGE 189

1 89 facilities to provide better altitude resolution, primarily for locating radiated sources from triggered lightning dart stepped leaders. To c alculate the three dimensional locations and emission times of sources radiated by lightning leaders with a high degree of accuracy, two quantities must be determined with significant precision, 1) the three dimensional location of each sensor, and 2) the timing delay of each measurement system from the sensor to the digitizer input. The techniques by which these quantities were obtained for the 2009 2011 ICLRT TOA networks are discussed in the following sections. Similar discussions for the previous TOA network installations are found in the Ph.D dissertations of Jerauld [2007] and Howard [200 9 ]. 3.2 Determination of Sensor Locations In 2009, a professional team of surveyors from Pat Welch Surveying in Starke, FL was hired to determine the spatial loca tions of all sensors at the ICLRT. The lateral and altitude coordinates of all sensors were first determined using a Differential Global Positioning System (DGPS). For dE/dt measurements, the sensor location was taken to be the center of the antenna sens ing plate. For the eight plastic energetic radiation detectors, the sensor location was measured at the center of the plastic scintillator. For the NaI detectors included in the TOA network and the two LaBr 3 energetic radiation detectors, the sensor loca tion was measured at the center of the mu metal shield that encases the detector. After the DGPS measurements were conducted, the altitude coordinates of all measurements included in the TOA network were re measured using a closed level loop for greater a ccuracy. The spatial positions of the TOA sensors are measured to an accuracy of about 1 cm. The coordinates of the sensors were provided in both the state plane format and in latitude/longitude. Lateral state plane coordinates were based on the Florida North NAD83 data and altitude coordinates were based on the NAVD88 data. Considering that the state plane coordinates for the sensors are of the order of

PAGE 190

190 hundreds of thousands of meters, it was convenient for data processing purposes to establish a local coordinate system at the ICLRT. A single surveyed point was chosen on the far southwest corner of the ICLRT to be the coordinate origin. The location of the origin was chosen such that the coordinates of all sensors would be positive laterally and verti cally relative to the location of the origin point. The state plane coordinates of the origin point were subtracted from the state plane coordinates of all sensors to establish the local coordinate system. The dimensions of the full measurement network r elative to the origin were about 700 m to the east, 625 m to the north, and about 8 m in altitude (excluding the x ray measurements located on the 11 m deck that supports the Tower Launcher). A similar survey was conducted prior to summer 2011 by the same team of surveyors. The coordinates of newly added measurements and any existing measurements that had changed location since the 2009 survey were re measured. The state plane coordinates, latitude/longitude, and local ICLRT coordinates of all TOA sensor s from both the 2009 and 2011 site surveys are given in Table 3 1. The measurements are arranged by type and by increasing station number. Electric field derivative sensors are named according to the convention "dE X", plastic energetic radiation detecto rs are named with the convention "TX F", LaBr 3 energetic radiation detectors are named with the convention "LaBr X", and unshielded NaI radiation detectors are named with the convention "TX U". In all sensor naming conventions, "X" is the corresponding st ation number where the sensor is located. Refer to Sections 3.17 through 3.19 and Figures 3 35 through 3 37 for discussion on the measurement network from 2009 2011 and graphical representations of the position of the stations and corresponding sensors.

PAGE 191

191 3.3 Measurement of Time Delays When a signal arrival time is measured at a given station, the total delay between the actual emission time of the source and the measured arrival time on the digitized waveform is the sum of 1) the signal propagation time from the source location to the antenna and 2) the signal propagation time through the measurement electronics in the field, the fiber optic link, and the coaxial cable that connects the fiber optic receiver to the digitizer input. In order to accurately measure the signal propagation time, all delays introduced by electronics and cabling must be accounted for. In early two dimensional TOA measurements at the ICLRT, only the delay of the fiber optic cable was removed. The fiber delay was measured using an Optical Time Domain Reflectometer (OTDR). For three dimensional TOA measurements, more accurate timing was required, and thus, a system was developed to measure not only fiber optic cabling delays but all cabling and electronics delays between the sens or output and the digitizer input. The delay measurement system, which was designed by Rob Olsen III, is comprised of a single mode fiber transmitter/receiver pair and a 1 km 9/125 m single mode fiber. The first step in making the delay measurement is to accurately measure the nominal signal propagation delay through the fiber transmitter/receiver pair connected by the 1 km single mode fiber. A line diagram of the nominal time delay measurement process is given in Figure 3 1 A A pulse generator is use d to inject a pulse with width of about 1 s with repetition rate of several tens of kilohertz into a short section (about 1 m) of coaxial cable. This short section of cable (Cable 1 in Figure 3 1 A ) is connected to one input of a BNC F connector on Channe l 1 of a LeCroy 44 Xi DSO. A second short section of coaxial cable (Cable 2 in Figure 3 1 A ) is connected to the other input of the BNC F connector and then to the input of the single mode fiber transmitter. The 1 km fiber is connected between the fiber t ransmitter and receiver, both of which are packaged in individual

PAGE 192

192 shielded project boxes. The transmitter and receiver are powered by a 12 V battery. The output of the fiber receiver is connected to Channel 2 of the LeCroy 44 Xi DSO through two sections of coaxial cable coupled by a BNC barrel connector. Cable 3 is about 1 m in length and Cable 4 is mode on the rising edge of the incoming pulse on Channel 1. A vertical cursor is placed on the initial rising point of the incoming pulse measured on Channel 1, and a second vertical cursor is placed on the initial rising point of the incoming pulse on Channel 2, this pulse having propagated through Cable 2, the f iber transmitter, the 1 km fiber, the fiber receiver, Cable 3, and Cable 4. The time difference between the two cursors is the nominal delay of the system. The zoom feature of the LeCroy 44 Xi is used to insure that the selected point on the rising edge of each pulse is chosen as accurately as possible. After the nominal delay of the system is measured, the next step in the process is to unspool the 1 km fiber to the location of a given field measurement location. The other end of the 1 km fiber remain s in Launch Control. The single mode fiber receiver, a 12 V battery, and coaxial Cable 3 (Figure 3 1 B ) are also re located to the field measurement location. A line diagram of the field delay measurement configuration is given at bottom in Figure 3 1 B The sensor output in the field is disconnected and coupled directly to Cable 3 with a BNC barrel connector. For dE/dt antennas, the BNC cable is removed from the BNC feed through on the antenna housing (Section 2.13) and coupled to Cable 3. For NaI and L aBr 3 detectors, the output of the PMT base is disconnected and coupled to Cable 3. For the plastic detectors, the coaxial cable was disconnected from the BNC feed through on the outside of the inner box (prior to the cable entering the galvanized pipe con nected to the TERA box, see Section 2.14) and coupled to Cable 3. The PIC controller at the given measurement is then commanded to switch on its output

PAGE 193

193 power, providing 12 V to the Opticomm transmitter. In Launch Control, the coaxial cable is removed fro m the DSO channel input where the signal from the given sensor in the field terminates (after being transmitted over the Opticomm link). This cable is connected to Cable 4 (Figure 3 1 B ) with a BNC barrel connector. The opposite end of Cable 4 is connecte d to Channel 2 of the LeCroy 44 Xi. Similar to the nominal delay measurement, the LeCroy 44 Xi is configured to trigger on the rising edge of the incoming pulse from the pulse generator on Channel 1. The total delay of the system is then the time differe nce between the incoming pulse on Channel 1 and the delayed incoming pulse on Channel 2, this pulse having propagated through the single mode fiber link, the measurement electronics in the field, the Opticomm fiber link, and the length of coaxial cable con necting Channel 2 to the output of the Opticomm receiver. The actual measurement system delay can then be calculated by simply subtracting the previously measured nominal delay from the total delay. For TOA measurements, delays range from about 170 ns to about 3.5 s. The measured system delays for all dE/dt measurements are given in Table 3 2 and for all energetic radiation measurements (plastic, LaBr 3 and NaI) in Table 3 3. Prior to 2011, the delays of the PMTs attached to the plastic and LaBr 3 scin tillators were not accounted for in the delay measurements described above in Section 3.3. Some research revealed that the manufacturer stated delay for the Hamamatsu R550 PMTs coupled to the plastic scintillators is 70 ns. In order to verify this delay time and to insure that the delays of all the PMTs were the same, a system was devised by Rob Olsen III and the author to measure accurately the PMT delays. A line diagram of the system in shown in Figure 3 2. The objective was to replicate the optical s ignal output of the plastic scintillator with a repeatable waveshape. Considering the PMTs have peak sensitivity in the blue wavelengths, we elected to take the same

PAGE 194

194 multi mode fiber transceiver used on the 2006 and 2011 PIC controllers (model HFBR 1414 T ) and replace the standard red LED with a blue LED. The transceiver was mounted on a piece of perf board along with the necessary drive circuitry and input/output BNC ports. The HFBR 1414 T was coupled to a 10 m 62.5 m fiber jumper. The same pulse gene rator described in Section 3.3 was used to inject a 1 s pulse into a short length of coax that terminated at a BNC F connector at Channel 1 of the LeCroy 44 Xi. The HFBR 1414 T board was coupled directly to the other output of the BNC F connector. The l id of the plastic scintillator inner box was removed and the PMTs were de coupled from the plastic scintillator material. The output of one of the PMTs was disconnected from the summing amplifier. The end of the 10 m fiber jumper, which was terminated wi th a typical ST type connector, was mated directly to the face of the second PMT. The inner box lid was replaced and bolted down with light tension so that the fiber jumper, which rested on top of the rubber gasket, would not be damaged. A black sheet wa s placed over the entire measurement box to insure that no light entered the inner box. A length of coaxial cable was connected to the outside BNC feed through port on the inner box, replacing the cable that normally carries the signal through the galvani zed pipe to the TERA box. The opposite end of the coaxial cable was connected to Channel 2 of the LeCroy 44 Xi. The measured total delay between the Channel 1 and Channel 2 inputs was 156 ns. The signal propagated through the 10 m fiber jumper, the PMT, the PMT base, the trans impedance amplifier at the output of the PMT base, the summing amplifier, and the connecting coaxial cables. This delay measurement was repeated for all 16 of the PMTs connected to the 8 plastic scintillators. The nominal delay o f the LED circuit, the 10 m fiber jumper, and the external coaxial cables was later determined to be 61 ns. The delay of the internal coaxial cables that connect the output of the trans impedance amplifier to the summing amplifier and the output of

PAGE 195

195 the su mming amplifier to the BNC feed through on the inner box were measured to be 16 ns. Thus, the delay of the PMT and the PMT base was 79 ns. Fortunately, this measurement was highly consistent for all 16 PMTs. An additional 95 ns (the addition of the coax ial cable, PMT, and PMT/PMT base delays inside the inner box) was added to the delay measurements for all the plastic detectors. This number is slightly in error for the measurement configurations prior to February 2011 (when the amplification electronics were replaced on all plastic scintillators). The error is strictly due to differences in coaxial cable length inside the inner box and should be consistent and not more than about 4 ns. Ideally, a similar delay procedure would be conducted with the LaB r 3 detectors, however, the scintillator and PMT are coupled together inside a hermetically sealed mu metal shield, preventing direct access to the PMT. For the LaBr 3 detectors, the manufacturer stated delay for the PMT is 45 ns. This number was taken as valid and was added to the delay measurements for both LaBr 3 detectors. The additional delays discussed in this section are included in the stated measurements in Table 3 3. 3.4 DSO Time Base Synchronization A line diagram of the DSO channel configurati on in Launch Control for all TOA measurements in shown in Figure 3 3. As discussed in Chapter 3, Scopes 12, 13, 14, and 17 are LeCroy Waverunner I DSOs and Scopes 18, 20, 21, 28, and 29 are LeCroy 44 Xi DSOs. The TOA DSOs all trigger on the output of the ICLRT master trigger box (Section 2.7). The external trigger input of each DSO is connected to an individual output on the ICLRT master trigger box with a 68" (1.73 m) length of coaxial cable. Ideally, all nine of the TOA DSOs would trigger at exactly t he same time and have identical, sample aligned time bases. In practice though, the time bases of the nine DSOs do not align exactly. For TOA measurements with nanosecond level timing, synchronization of the DSO time bases is then very important to avoid systematic

PAGE 196

196 timing errors. DSO time base correlation is achieved through a complex series of physical cabling connections and post processing algorithms, the details of which are described in this section. The first step in synchronizing the time bases of the nine DSOs is to pick one reference time base. In this case, the reference DSO is Scope 21 and the reference channel is dE 7. As seen in Figure 3 3, the first channels of Scopes 12, 13, 20, and 21 are connected with 64" (1.63 m) coaxial cables to c hannels 1 4 of Scope 17. The signals are branched directly at the input channels of Scopes 12, 13, 20, and 21 with BNC T connectors. Similarly, the first channels of Scopes 28, 29, and 18 are connected to channels 1 3 of Scope 14, also with 64" coaxial c ables. Channel 4 of Scope 17 is then connected to channel 4 of Scope 14 with a 64" coaxial cable. Scopes 17 serves to synchronize the time bases of Scopes 12, 13, 20, and 21 (treating Scope 21 as the reference) and Scope 14 serves to synchronize the time bases of Scopes 28, 29, and 18. The Scope 14 time base can the be synchronized to the Scope 17 time base. The processing algorithm for actually performing the synchronization is described below in a step by step format. The processing code is written in Matlab. The waveforms from all nine DSOs are read into the Matlab workspace in individual time and data vectors. Time base synchronization is performed on the memory segment of interest, though due to the shallow memory depth of the LeCroy Waverunner I DSOs, synchronization can only be performed on the first two memory segments (even though the LeCroy 44 Xi DSOs record 10 memory segments). When TOA calculations are performed for memory segments on the LeCroy 44 Xi DSOs beyond the second segment, the s hift values returned for the second memory segment are used. 1) The first channel on Scope 21 (dE 7) is cross correlated with the duplicated channel on Scope 17. The cross correlation is performed using a 50 s section of the waveform

PAGE 197

197 recorded on each DSO, starting with a time window from t = 50 s to t = 0 s. The trigger point occurs at t = 0 s. The cross correlation operation returns the number of sample points to shift the time base on Scope 17 to match the time base of Scope 21. The cross correlation is then performed an additional four times with the 50 s window shifted backwards in time for each operation (i.e., the second operation uses times from t = 60 s to t = 10 s and so on). After the last operation is complete, the program checks to see that all five cross correlation operations returned the same shift value. If so, the shift for Scope 21 is stored in a variable as the reference shift value. 2) The first channels of Scopes 12, 13, and 20 are cross correlated with their duplicate channels on Scope 17 using the same five operation procedure described in Step 1. If the shift value for a particular scope is consistent, the actual shift for the scope time base is calculated as the difference between the shift for Scope 21 (the reference) and the returned shift value and saved to a variable. Effectively, the shift for Scope 21 is treated as zero and the time bases of the other DSOs are shifted accordingly. 3) The fourth channel of Scope 17 (dE 7) is cross correlated with the fourth channel of Scope 14 (dE 7) using the five operation procedure described in Step 1. In this step, the Scope 17 time base is considered to be the reference. The shift of Scope 14 with respect to Scope 17 is stored in a variable. 4) The first channels of Scopes 28, 29, and 18 are cross correlated with their duplicate channels on Scope 14, again using the five operation procedure described in Step 1. The returned shift values are then subtracted from the shift of Scope 14 with respect t o Scope 17 and stored to variables.

PAGE 198

198 5) The shift values for Scopes 28, 29, and 18 obtained in Step 4 are added to the shift of Scope 17 relative to Scope 21 (the reference time base) and stored to variables. This operation synchronizes the time bases of Scopes 28, 29, and 18 to that of Scope 21. Typical time base shifts vary from 1 3 sample points (4 12 ns) and rarely exceed 4 sample points (16 ns). 3.5 dE/dt Signal Arrival Time Selection (DSO dE/dt Network) The LeCroy waveforms from the ten dE/dt sensors, consisting of time and data vectors, are first read into Matlab. In order to simplify the waveform aligning process, an artificial time base is created for each data vector independent of the actual time vector (which contains time values to the precision of the sampling resolution). The time base is created by initially assigning the array index (integer values starting at 1) to each position of a vector with the same length as the data vector. Each data vector is then cross correlated with th e reference data vector, which is typically chosen to be dE 7 unless the measurement is unavailable. The time window of the cross correlation can be varied, but typically operates over a span of about 50 s prior to the trigger point (the return stroke). For dE/dt waveforms, this time window usually encompasses a portion of the descending leader with large, well resolved pulses (in the case of stepped leaders, dart correlation returns the optimal shi ft value, which may be either positive or negative depending on the location of the source and the time delay of the measurement link, to align the waveform from each station with that of the reference station. The time and data vectors for each of the te n measurements along with the shift values with respect to dE format allows the large number of contained vectors to be easily passed to other programs.

PAGE 199

199 The struct returned from the cross correlati on routine is passed to a plotting routine that displays the ten waveforms. Before the plot command is executed, the appropriate shift calculated in the cross correlation routine is added to the artificial time base for each particular measurement. As an example, the artificial time base array for a LeCroy 44 Xi waveform prior to shifting would have total length of 1.25 million sample points (array indices 1 to 1.25 x 10 6 with each array value equal to its index). If a shift was applied of 300 points, t he array indices would remain the same and the array values would then range from 299 to 1.2497 x 10 6 The shifting does not modify the original time vector in any way. After the shifting operations have been conducted, the ten waveforms are plotted in the same plot window with each subsequent waveform offset by a vertical quantity that can be specified within the plotting routine. The vertical shift can be used to zoom in or out on the waveforms depending on the signal to noise ratio of the pulse in qu estion. The waveforms are colored differently and are labeled according to station number on the vertical axis. A 64 s plot of the shifted (time aligned) waveforms for a triggered lightning dart stepped leader preceding the fourth return stroke of event UF 11 35 on August 18, 2011 is shown in Figure 3 4. Note that the entire length of the artificial time vector is not plotted in Figure 3 4. Typically, only the first 700,000 points are plotted (a portion of which are shown in Figure 3 4) in order to co nserve memory. For the dE/dt measurements, the portion of the recorded waveform well after the return stroke usually does not contain a great deal of useful information. Once the waveforms are plotted, the process of manually selecting the signal arriva l times for common pulses can begin. The time window is adjusted using the zoom tool within the Matlab plot window such that the fine structure of individual leader step pulses can be easily resolved. After the window is zoomed correctly, the user visual ly observes which stations have

PAGE 200

200 the common pulse sufficiently resolved for an arrival time to be accurately determined. A second program is called from the Matlab command line that prompts the user to select the signal arrival times. As input, this progr am takes a vector containing the station numbers that do not have the given pulse sufficiently resolved. The arrival time selection process uses the Matlab clicking on the plotted vector with the left mouse button. For each qualified station, the program takes in two data points one prior to the peak of the pulse and one after the peak of the pulse. The points are selected manually by the user with two mou se clicks. For each station, the program determines the data indices in the artificial time vector of the two data points, then finds the corresponding maximum amplitude value of the data vector between the two data indices. With the exception of dE 17 and dE 25, the vector index of the maximum amplitude value is returned in the position of a new array that corresponds to the station number. The indices of dE 17 and dE 25 are returned in positions 12 and 13 in the new array as opposed to positions 17 an d 25. In Figure 3 5, a single dart stepped leader pulse from the leader plotted in Figure 3 4 is shown with the user selected points marked with red circles and the software determined maximum amplitude points annotated. The length of the plot is about 1 .9 s. If two data points between the user selected points share the same amplitude, the point occurring first in time is selected by the software (reference the waveform for dE 4 in Figure 3 5). Data points that share the same amplitude near the peak of a dE/dt pulse are usually an artifact of the 8 bit digitization on the LeCroy DSOs. Note that in Figure 3 5, the peak of the dE/dt pulse on dE 17 is not resolved. From the user defined vector of unused stations, the data point selection program will aut omatically bypass dE 17 when assigning variables from user mouse clicks. The 13 element output vector will have a '0' in any position that does not correspond to a station number

PAGE 201

201 (positions 2, 6, and 10) and also a '0' in any position where the dE/dt puls e was unresolved (in this case, position 12). When the array of data indices are determined corresponding to the peak dE/dt amplitudes for a common pulse, the next step is to calculate the actual arrival times of the signal at each of the involved stati ons. The 13 element indices vector obtained in the prior step is passed to a different program. This program parses through the index vector, and for each non zero element, first adds the shift of the time base for the DSO where the measurement was digit ized (Section 2.4). For dE/dt measurements recorded on Scope 21 (the reference time base), there is no shift applied. The time bases for Scope 18 and Scope 20 are shifted accordingly. After the time base shift is applied, the program then finds the valu e of the actual time vector for each measurement (not the artificial time vector) at the given index. When the actual time is determined, the program then subtracts the appropriate measured cabling/electronics time delay (Section 3.3) for the measurement. When the arrival times for all involved stations are computed, a separate program is called that checks to see if the time differences between all possible sets of two stations are physically possible assuming straight line propagation at the speed of li ght. If not, an appropriate error message is displayed. For dE/dt measurements, this rarely, if ever, occurs, assuming the waveforms were shifted appropriately to line up the common pulses. The final output of the program is again a 13 element vector co ntaining the actual arrival times, accounting for all delays, of the common signal at each of the involved stations. The 13 element vector is automatically written to a Microsoft Excel worksheet using the Matlab Excel Link toolkit. 3.6 dE/dt Signal Arr ival Time Selection (HBM Network) The waveforms stored by the HBM Genesis recorder are merged into an archive file with the ".PNRF" extension. Each time the system is armed, a new archive file is created. Inside the

PAGE 202

202 archive file, the waveforms are stored according to the physical channel connectivity to the 4 channel receiver cards, each card having a corresponding number (1 7) and letter (B H) identifier. Each center triggered segment of data is 20 ms in length, or 2 million sample points. The ten dE/d t waveforms corresponding to the appropriate data segment are read into Matlab, though only the first 1.25 million data points are read to conserve memory. An artificial time base is created for each data vector. Each dE/dt waveform is similarly cross co rrelated with the dE 7 waveform to establish the necessary shift in order to align visually the leader step pulse structures. After the shift values are returned, the waveforms are plotted in an identical manner to the dE/dt waveforms described above in S ection 3.5. The arrival time selection process is also handled using the same "ginput" technique to pick data points on either side of the dE/dt peak of a given pulse. For each station, the index of the maximum dE/dt peak between the two selected data po ints is stored in a 13 element vector. In Figure 3 8, the HBM dE/dt waveforms and user selected data points are plotted for the same dart stepped leader pulse shown in Figure 3 5. In addition to demonstrating the arrival time selection process, Figure 3 8 also showcases the quality of the digitized data from the HBM digitization system versus that of the DSO digitization system. The peaks of the dE/dt pulses, which were generally rounded or squared off on the DSO digitization system due to the 8 bit ampl itude resolution, are very well resolved on the 14 bit HBM digitization system. After the software has determined the indices of the maximum dE/dt peaks for each available station, a program is then called that checks to see if the differences in arrival indices, considering the 10 ns sampling resolution of the system, for every two station combination are physically possible. The vector of indices are then written to an Excel spreadsheet. Unlike the DSO dE/dt system, once the indices of a given pulse ha ve been obtained and verified for each available station, all the necessary information is stored to call the

PAGE 203

203 solution algorithm. As discussed in Section 2.5, the HBM Genesis recorder automatically removes cabling and electronics delays each time the syst em is armed. 3.7 X ray Signal Arrival Time Selection The arrival time selection process for energetic radiation (primarily x ray) waveforms follows the same general series of steps as that described in Section 3.5 for the DSO dE/dt waveforms. The primar y difference for processing the x ray waveforms is that the cross correlation routine used to visually align the dE/dt waveforms does not work particularly reliably. As described in Section 2.14, the amplitude response of the plastic detectors is not line ar with photon energy, and thus, the structure of the x ray pulses recorded on different detectors often do not share a great deal of similarity. After some trial and error, it was found that the best way to visually align the x ray waveforms was to apply the shift values calculated for the dE/dt waveforms of the same leader event, implying that the source locations were similar. In Figure 3 6, the energetic radiation waveforms for the eight plastic detectors and two LaBr 3 detectors are plotted using the dE/dt shift values for the same triggered lightning dart stepped leader plotted in Figure 3 4. The time of the return stroke is annotated. Selecting the x ray arrival times associated with the formation of a leader step is a more subjective process than for the corresponding dE/dt pulses. Whereas most leader step pulses exhibit a relatively clear dE/dt peak that can be used as a distinguishing waveform feature for the arrival time, x ray waveforms simply deviate from the system noise level when the firs t x ray photon is detected. X ray arrival times are similarly selected using the Matlab "ginput" function, but instead of specifying a window where the arrival time occurs and asking the software to find the appropriate data point, the indices of the arri val times at each available station are stored to elements in a 13 element vector with a single mouse click on the user identified sample point where the initial deflection occurs. In Figure 3 7, an example x ray waveform is plotted for a

PAGE 204

204 single burst of photons associated with a dart stepped leader step (annotated in Figure 3 6). The user selected arrival times on the initial negative deflection of qualified waveforms (T4 F does not have a well resolved x ray pulse for this leader step) are annotated wit h red circles. The actual arrival times are computed using the exact same technique as described for the dE/dt waveforms, again accounting for DSO time base shifts and measured cabling/electronics delays. After the arrival times for all involved stations are computed, a similar program is called that checks to see if the time differences between all possible sets of two stations are physically possible. Occasionally, the arrival times of the x ray photons can be very difficult to select accurately, and t he time checking program prevents un realistic time values from being run through the solution algorithm. The measured arrival times contained in the 13 element vector are written to an Excel spread sheet. 3.8 Calculation of Source Locations and Emission Times After the arrival times at the involved stations are determined on either of the three measurement systems described above, the final step in the TOA process is to actually calculate the three dimensional location and emission time of the source. This is accomplished using a Matlab non linear least squares program based on the Levenburg Marquardt algorithm. The algorithm minimizes a chi squared goodness of fit value in an iterative manner by linearizing the TOA equations (Equation 3 1) for each st ation around successive trial solutions and subsequently solving the linearized equations to determine the next trial solution. The linear curvature matrix that is used to obtain the actual solution can be inverted to obtain the covariance matrix that des cribes the uncertainties of the three dimensional solution parameters and the source emission time. This procedure is described in Bevington [1969]. The chi squared goodness of fit value is defined according to Equation 3.3.

PAGE 205

205 (3.3) The quantity is the measured arrival time at the i th station, the quantity is the fitted arrival time at the i th station predicted by the iterative algorithm, and the quantity is the estimated Gaussian distributed, rms timing error of the network. The goal of the algorithm is to minimize the sum of squares differences between the measured arrival times and the predicted arrival times, normalized by the estimated system t iming error. The chi squared values are normalized with respect to the number of stations N, providing a reduced chi squared value, where is the number of degrees of freedom ( = N 4) necessary to obtain a three dimensional s olution. For the three ICLRT TOA networks described in Sections 3.5 3.7, the rms system timing error is assumed to be 10 ns. For automated networks such at the LMA where often many thousands of sources are collected per lightning flash, the rms timing er ror can be estimated by fitting the distribution of reduced chi squared values to the theoretical distributions for each degree of freedom, which are given by Bevington [1969], assuming Gaussian distributed errors. In the LMA processing code, the rms timi ng error is assumed to be 70 ns. The reduced chi squared values can be scaled to any rms timing error to fit the theoretical distributions by 2 of sources are ca lculated with the three ICLRT TOA networks to accurately fit the reduced chi squared vales to the theoretical distributions for each degree of freedom. Instead, the rms timing error was originally estimated in the Ph.D dissertation of Howard [2010], who u sed a Monte Carlo simulation to inject random timing errors to a known source location and then compared the covariance estimates for the source returned by the solution algorithm with those obtained

PAGE 206

206 from a lightning source radiated from the same spatial l ocation (i.e., same geometry relative to the measurement network). Howard [2010] found that the best match for the covariance estimates occurred when the injected timing error was between 6 10 ns. The duration of the non linear least squares algorithm is limited by two quantities, 1) the user defined tolerance parameter, and 2) the user defined number of iterations. The tolerance parameter sets the necessary precision at which the algorithm is determined to have converged, that is, the improvement in the normalized sum of squares goodness of fit between the measured and predicted arrival times between successive iterations. For the three ICLRT TOA networks, the tolerance is set to 0.0001. The maximum number of iterations is set to 40, though the solu tions typically converge with less than 20 iterations. From a processing time standpoint, limiting the iterations is much less critical with the ICLRT TOA networks than with networks like the LMA where the signal arrival time selection process is automate d (and more extraneous arrival times are selected). When the TOA solution algorithm is called from the Matlab command line, the program first calculates the number of possible station combinations for each degree of freedom using the built in "nchoosek" function, where n is the number of available stations with measure arrival times and k is the minimum number of stations to participate in the solution. For the three ICLRT TOA networks, at least five stations must be involved in any solution. For each c ombination of k stations, a k x 3 element array is then constructed with the spatial coordinates (x, y, z) of the given station. Next, an initial guess is defined for the location of the source. This guess is contained in a four element vector with the s patial location (x, y, z) and emission time (t) of the source. For the three ICLRT TOA networks, the spatial position guess is set to x = 200 m, y = 200 m, z = 300 m, and the source emission time guess is set to the average of the

PAGE 207

207 measured arrival times m inus 2 s. For large area networks like the LMA, the hyperplane approach [e.g. Koshak and Solakiewicz 1996; Koshak et al., 2004] is used to determine the initial guess for the non is calculated by squaring Equation 3 1 for each station and then subtracting the equations from each other to obtain N 1 difference equations. The quadratic terms in x, y, z, and t cancel in the process, leaving the unknown terms in the linear domain. T he difference equations define a plane in four dimensional hyperspace and can then be solved analytically for the four unknowns. The solution is the intersection of the four or more hyperplanes. This approach is similar to that used by Jerauld [2007] to calculate the two dimensional strike points of natural lightning flashes that terminated on or near the ICLRT. The resulting hyperplane solution is poorly determined in the altitude plane for netwo rks that are nearly planar (reference Appendix A of Thomas et al. [2004] for a thorough discussion of the altitude errors associated with the hyperplane solution), however, the initial guess drastically reduces the number of necessary iterations for the following least squares solution to converge. When many tho usands of source locations are being calculated, the savings in processing time are substantial. For the three small area ICLRT TOA networks, the four parameters do not have to be varied significantly in subsequent iterations for the solution to converge, and hence, computing a hyperplane solution prior to calling the non linear least squares program is not necessary. When the non linear least squares program converges for a particular set of stations, the program returns the vector of predicted arrival times and the covariance matrix that describes the uncertainties of the parameters. The trace of the covariance matrix provides the total spatial location uncertainty of the calculated solution. The estimated errors for the four parameters are obtained by taking the square root of the diagonal elements of the covariance matrix. Finally, the

PAGE 208

208 reduced chi squared value for the solution is calculated according to Equation 3 3. The solution, errors, reduced chi squared value, and the combination of stations are stored in a three dimensional array with the third dimension of the array corresponding to the number of stations used in the solution. The non linear least squares program is run for all possible combinations of five or more stations, and the result s described above are likewise stored for any solution that converges. After all possible station combinations have been computed, the program next has to determine the best solution. The metric for determining the best solution is identical to that us ed with the LMA processing code and is described in Appendix A of Thomas et al. [2004]. First, the reduced chi squared value for each convergent solution is compared to unity. If the value is greater than one, the returned reduced chi square value is use d. If the value is less than one, the returned reduced chi squared value is set equal to one. The best solution is then determined by the minimum product of the trace of the covariance matrix and the reduced chi squared value for a particular set of stat ions. Once the best solution has been determined, the four element solution chi squared value of the solution, and the ten element station combination vecto r (unused stations are zeroes) are all written to the same line of the Excel file where the signal arrival times were previously written. For DSO dE/dt waveforms, the emission times of the sources are calculated with respect to the trigger point of the Sc ope 21 time base. The absolute trigger time of Scope 21 is known to within a microsecond via the GPS time stamping software described in Section 2.8, and such, the source emission times can then be related to absolute time if necessary. For waveforms dig itized on the HBM Genesis recorder, the emission times of the sources are initially calculated with respect to the beginning of the data segment and are given in units of

PAGE 209

209 sample points. The emission times can be converted to absolute time by multiplying t he emission time by the sampling resolution of the system (10 ns) and then adding this number to the absolute time of the first point of the data segment. The absolute time of the first point of the data segment is known to within 10 ns. 3.9 Spatial and Timing Errors of the ICLRT TOA Networks Based on the 8 station TOA networks that operated from 2006 2008 at the ICLRT, Howard [2010] used a Monte Carlo analysis to predict that the network experiences timing errors of 6 10 ns for sources within the networ k boundaries, as noted above. The fitted data were from sources calculated from one triggered lightning dart stepped leader (event UF 07 07) and one natural lightning stepped leader (event MSE 06 04). The simulation predicted that sources were located wi th errors of only 2 3 m in the lateral directions and less than 10 m in the vertical direction. For event UF 07 07, the closest dE/dt sensor to the triggering facility was dE 5, located about 90 m from the Tower Launcher. From Thomas et al. [2004], the a ltitude uncertainty is given by Equation 3.4. (3.4) In Equation 3.4, d is the horizontal distance from the source to the closest measurements station, r is the radial distance from the source to the closest measurement station, and z is the source altitude. For a source at 50 m altitude in the lateral location of the Tower Launcher with the closest dE/dt sensor at 90 m, the expected altitude uncertainty is about 11.8 m. For the 2009 2011 TOA networks, the closest dE/dt sensor to the Tower Launcher was dE 25 (about 42 m) and the closest sensor the Field Launcher was dE 17 (about 27 m). For the 42 m distance, the expected altitude uncertainty is 6.5 m. For the 27 m distance, the expected altitude uncertainty falls to 5 m. Thus, we would expect the uncertainty in the predicted source altitudes to be about

PAGE 210

210 a factor two better than those reported by Howard [2010]. The TOA results calculated for the 2009 2011 data sets indicate that the location uncertainties are, in fact, somewhat better than the model predicted results. For centrally located sources (such as those associated with triggered lightning leaders) above about 50 m with respect to local ground level, the lateral coordinates are typically calculated with errors of less t han 1 m and the altitude coordinate is calculated with errors of the order of 1 m. For lower altitude sources, the lateral coordinate errors remain of the order of 1 m while the altitude error may increase to 2 3 m depending on the combination of station s used to calculate the solution. The improvement in spatial location uncertainty over the results reported in Howard [2010] is likely attributable to several factors, 1) the Field (Ground) Launcher, which was utilized for most of 2010 and 2011, is locate d closer to more of the 10 dE/dt sensors than is the Tower Launcher, 2) the spatial locations of the sensors are likely known with better precision, 3) for the DSO dE/dt network, the system sensitivity was doubled in 2011, resulting in better signal to noi se ratio and more well determined dE/dt peaks (and hence, more well determined arrival times), 4) for the HBM dE/dt network, the sensitivity is a factor of four greater than the data used by Howard [2010], there is essentially no uncertainty in cabling del ay, and the waveforms are digitized with 14 bit versus 8 bit amplitude resolution, all improvements that provide less uncertainty in the selected arrival times.

PAGE 211

211 Table 3 1. The spatial coordinates of all TOA sensors at the ICLRT from the 2009 and 2011 site surveys. With the exception of the absolute coordinates, all coordi nates are given in units of meters. State Plane Coordinates Absolute Coordinates ICLRT Local Coordinates Sensor Easting Nor thing Altitude Latitude Longitude Easting Northing Altitude Distance to Field Launcher Years Origin 106653.796 837811.858 67.050 29.93899065 82.03648403 0.000 0.000 0.00 542.601 2009 2011 dE 1 107153.348 837948.964 70.037 29.94346957 82.03495 231 137.106 499.551 2.99 171.717 2009 2011 dE 3 107222.974 838185.359 70.674 29.94405142 82.03248851 373.501 569.176 3.62 138.292 2009 2011 dE 4 107147.606 838492.623 71.720 29.94331163 82.02932329 680.764 493.809 4.67 381.387 2009 2011 dE 5 107133.248 838187.014 71.332 29.94324182 82.03249149 375.155 479.450 4.28 78.710 2009 2011 dE 7 106898.477 838204.961 70.510 29.94112083 82.03235826 393.102 244.680 3.46 225.555 2009 2011 dE 8 106939.945 838382.195 71.915 29.94146024 82.03051350 570.336 286.148 4.8 6 315.066 2009 2011 dE 9 106751.047 837946.795 72.786 29.93984149 82.03506485 134.937 97.251 5.74 391.060 2009 2011 dE 11 106914.515 837980.056 69.199 29.94130938 82.03468380 168.198 260.718 2.15 232.438 2009 2011 dE 17 107066.902 838293.208 72.997 29.9 4262269 82.03140659 481.349 413.105 5.95 183.494 2009 2010 dE 17 107087.991 838134.404 70.580 29.94284391 82.03304648 322.545 434.194 3.53 26.657 2011 dE 25 107108.296 838257.477 72.055 29.94300302 82.03176734 445.619 454.499 5.00 143.868 2009 2011 T1 F 107142.567 837952.830 70.354 29.94337157 82.03491468 140.972 488.769 3.30 165.215 2009 2011 T3 F 107228.838 838182.757 70.558 29.94410482 82.03251415 370.898 575.041 3.51 142.054 2009 2011 T4 F 107137.924 838503.140 72.168 29.94322224 82.029 21654 691.281 484.127 5.12 390.898 2009 2011 T5 F 107122.511 838193.843 71.473 29.94314365 82.03242317 381.985 468.713 4.42 82.139 2009 2011 T7 F 106908.963 838204.427 71.305 29.94121552 82.03236144 392.568 255.166 4.26 215.783 2009 2011 T8 F 106934.096 838390.320 71.799 29.94140589 82.03043067 578.461 280.299 4.75 325.051 2009 2011 T9 F 106758.656 837945.655 73.018 29.93991034 82.03507496 133.797 104.860 5.97 384.689 2009 2011 T11 F 106906.362 837985.286 69.373 29.94123483 82.03463147 173.428 252.565 2.32 236.274 2009 2011 LaBr 17 107057.363 838293.173 73.551 29.94253666 82.03140909 481.314 403.566 6.50 185.662 2009 2010 LaBr 17 107087.174 838122.910 70.866 29.94283878 82.0331657 311.051 433.377 3.82 19.842 2011 LaBr 25 107112.517 838268.042 72.768 29.94303903 82.03165698 456.184 458.720 5.72 154.585 2009 2011 T1 U 107142.954 837950.077 70.735 29.94337560 82.03494311 138.219 489.156 3.69 167.984 2009 2011 T3 U 107231.522 838182.805 70.714 29.94412902 82.03251305 370.947 577.725 3.66 144 .429 2009 2011 T4 U 107140.435 838504.093 72.457 29.94324471 82.02920611 692.234 486.638 5.41 392.068 2009 2011 T5 U 107122.509 838191.169 71.732 29.94314415 82.03245086 379.311 468.712 4.68 79.530 2009 2011 T7 U 106908.438 838201.405 71.494 29.94121138 82.03239285 389.546 254.641 4.44 215.008 2009 2011 T8 U 106932.842 838387.864 72.219 29.94139507 82.03045638 576.006 279.045 5.17 323.627 2009 2011 T9 U 106758.280 837942.835 73.283 29.93990750 82.03510425 130.977 104.484 6.23 386.266 2009 2011 T11 U 1 06907.847 837987.628 69.766 29.94124776 82.03460687 175.770 254.050 2.72 233.758 2009 2011

PAGE 212

212 Figure 3 1. Measurement of time delays. A) l ine diagram of nominal delay measurement configuration, and B) line diagram of the delay measurement configurations for all TOA measurements.

PAGE 213

213 Table 3 2. Measured time delays from each dE/dt sensor to the DSO input. Sensor DSO Injection Point Delay (ns) Year Comments dE 1 Scope 20 Antenna BNC 2572 2009 2011 dE 1 Scope 17 Antenna BNC 2578 2009 2011 dE 3 Scope 20 Antenna BNC 1182 2009 2011 dE 4 Scope 20 Antenna B NC 1502 2009 2011 dE 5 Scope 20 Antenna BNC 670 2009 2011 dE 7 Scope 21 Antenna BNC 1624 2009 2011 dE 7 Scope 17 Antenna BNC 1636 2009 2011 dE 7 Scope 14 Antenna BNC 1642 2009 2011 dE 8 Scope 21 Antenna BNC 1893 2009 2011 Events before 5/18/10 (O ld St. 8 fiber) dE 8 Scope 21 Antenna BNC 1896 2010 2011 Events after 5/18/10 (New St. 8 fiber) dE 9 Scope 21 Antenna BNC 3166 2009 Events before 6/28/09 (Old St. 9 fiber) dE 9 Scope 21 Antenna BNC 3067 2009 2010 Events after 6/28/09 (New St. 9 fiber) dE 9 Scope 21 Antenna BNC 3448 2010 2011 Events after 5/12/10 (New St. 9 fiber) dE 11 Scope 21 Antenna BNC 3014 2009 2011 dE 17 Scope 18 Antenna BNC 567 2010 Location next to Tower Launcher dE 17 Scope 18 Antenna BNC 1420 2011 Location next to Field La uncher dE 25 Scope 18 Antenna BNC 399 2009 2011

PAGE 214

214 Table 3 3. Measured time delays from each energetic radiation detector to the DSO input. Sensor DSO Injection Point Delay (ns) Year Comments T1 F Scope 28 PMT 2685 2009 2011 T1 F Scope 14 PMT 2693 2009 2011 T3 F Scope 28 PMT 1335 2009 2011 T4 F Scope 28 PMT 1606 2009 2011 T5 F Scope 28 PMT 776 2009 2011 T7 F Scope 29 PMT 1779 2009 2011 T7 F Scope 14 PMT 1788 2009 2011 T8 F Scope 29 PMT 1989 2009 2010 Events bef ore 5/18/10 (Old St. 8 fiber) T8 F Scope 29 PMT 2036 2011 Events after 5/18/10 (New St. 8 fiber) T9 F Scope 29 PMT 3141 2009 2010 Events after 6/28/09 T9 F Scope 29 PMT 3520 2010 2011 Events after 5/12/10 (New St. 9 fiber) T11 F Scope 29 PMT 3175 2011 LaBr 17 Scope 18 PMT 664 2010 2011 LaBr 17 Scope 18 PMT 1108 2011 LaBr 25 Scope 18 PMT 167 2009 2011 T1 U Scope 12 PMT Base 2558 2009 2011 T1 U Scope 17 PMT Base 2562 2009 2011 T3 U Scope 12 PMT Base 1227 2009 2011 T4 U Scope 12 PMT Ba se 1497 2009 2011 T5 U Scope 12 PMT Base 664 2009 2011 T7 U Scope 13 PMT Base 1666 2009 2011 T7 U Scope 17 PMT Base 1672 2009 2011 T8 U Scope 13 PMT Base 1878 2009 2010 Events before 5/18/10 (Old St. 8 fiber) T8 U Scope 13 PMT Base 1935 2010 2011 Events after 5/18/10 T9 U Scope 13 PMT Base 3140 2009 Events before 6/28/09 (Old St. 9 fiber) T9 U Scope 13 PMT Base 3032 2009 2010 Events after 6/28/09 (New St. 9 fiber) T9 U Scope 13 PMT Base 3414 2010 2011 Events after 5/12/10 (New St. 9 fiber) T11 U Scope 13 PMT Base 3062 2009 2011

PAGE 215

215 Figure 3 2. Line diagram of plastic scintillator PMT/PMT base measurement delay configuration.

PAGE 216

216 Figure 3 3. Line diagram of the DSO channel connectivity for all TOA measurements. Scope 17 and Scope 14 ser ve to synchronize the time bases of Scope 12, Scope 13, Scope 20, Scope 18, Scope 28, and Scope 29 to the time base of Scope 21, which is the reference time base.

PAGE 217

217 Figure 3 4. A plot of the 10 dE/dt waveforms from a triggered lightning dart stepped le ader on August 18, 2011. The waveforms have been shifted using the cross correlation routine described in Section 3 .5 to align the common leader step pulses. The waveform region enclosed with the dotted line is expanded in Figure 3 5.

PAGE 218

218 Figure 3 5. An expanded view of the 10 dE/dt waveforms shown in Figure 3 4. The user selected times on either side of the dE/dt peak for a single dart stepped leader pulse on each of the qualified channels are annotated with red circles. The software determined times of the dE/dt peaks are indicated with arrows.

PAGE 219

219 Figure 3 6. A plot of the 10 energetic radiation waveforms (plastic and LaBr 3 detectors) from the same triggered lightning dart stepped leader shown in Figure 3 4. The waveforms have been shifted to alig n the common pulses using the shift values obtained from cross correlation of the dE/dt waveforms plotted in Figure 3 4. The time of the return stroke is annotated. The pulse enclosed in the dotted lines is expanded in Figure 3 7.

PAGE 220

220 Figure 3 7. An ex panded view of the 10 energetic radiation waveforms shown in Figure 3 6. The user selected arrival times on the initial negative deflection from the system noise level on each of the qualified channels are annotated with red circles.

PAGE 221

221 Figure 3 8. A p lot of the 10 dE/dt waveforms recorded on the HBM Genesis system for the same dart stepped leader pulse shown in Figure 3 5. The user selected times on either side of the dE/dt peak for a single dart stepped leader pulse on each of the qualified channels are annotated with red circles. The software determined times of the dE/dt peaks are indicated with arrows.

PAGE 222

222 CHAPTER 4 DATA DOCUMENTATION, DATA, AND GENERAL DA TA STATISTICS This chapter will first provide a detailed discussion about the protocols used to save, catalogue, and document both lightning waveform and photographic data acquired at the ICLRT (Section 4.1). A synopsis of the natural and triggered lightning data collected at the ICLRT from 2009 2011 will follow (Section 4.2). Finally, general sta tistics are computed for various parameters of the lightning discharges recorded from 2009 2011 and compared with prior studies. Statistics for triggered lightning return strokes and return stroke currents are given in Section 4.3 and statistics for trigg ered lightning initial stage (IS) currents are given in Section 4.4. The intent of the statistical portion of this chapter is to provide an overview of the dataset (and how it may be unique) that can be referenced when reading the detailed discussions of particular events that follow in the subsequent chapters. 4.1 Data Cataloguing and Documentation The procedures for cataloguing, documenting, and safeguarding acquired lightning data are involved, yet very necessary processes. Each lightning event recor ded by the ICLRT network of DSOs, high speed cameras, and the LMA generates about 40 GB of data. At the conclusion of a day where lightning data have been collected, the most important task is to download all waveform and calibration files saved on the DS Os. Each DSO has an independent static IP address that allows for communication over a local area network with ICLRT computers. Traditionally, DSO data are downloaded on HAL, the NLDN computer, and on the author's personal computer. Downloaded data are then transferred over the local network to an 8 TB RAID storage array located in the Office Trailer. The RAID storage array contains folders for each year of data collected. Inside the year folder, sub folders are created for each storm date within the c alendar year. The DSO data, which includes the calibration files recorded when the

PAGE 223

223 network armed and disarmed (Section 2.6) and any waveform data files, are copied to the appropriate date folder with sub folders for each individual DSO. The DSO sub folde rs have the naming convention "ScopeXX", where "XX" is the DSO number (and also the last field in its IP address). A separate sub folder is created for the PNRF file(s) generated by the HBM Genesis system. The PNRF files are downloaded directly from the computer that controls the HBM system using a USB drive. Another sub folder is created with the title "Pictures&Videos". Data acquired and saved by the Photron SA1.1, Phantom V7.3, and Cordin 550 high speed cameras are copied into respective sub folders within the "Pictures&Videos" folder. Sub folders are also created to hold still images and raw HD movie files. Finally, the GPS time stamp file generated by the NLDN computer, the launch times file generated by HAL, and the NLDN data file queried from th e LTS2005 software are all copied to the appropriate date folder. When all data from a given day have been saved to the RAID array, the full dataset is copied to a second 8 TB RAID array located in the Office Trailer. The data from the backup RAID array are not globally accessible over the local network. The data from the backup RAID array in the Office Trailer are regularly copied to a 12 TB RAID array located in 311 Larsen Hall on the University of Florida campus. When all backups are complete, the fu ll dataset is stored on three independent RAID storage arrays in addition to the personal computers of UF graduate students. For each day during which lightning data are collected, there are three documentation files created, 1) the Equipment Table, 2) t he Data List, and 3) the Summary. These files were prepared by the author during 2009 2010 and by graduate student William Gamerota during 2011. The Equipment Table is a Microsoft Excel file that contains the full list of measurements grouped by type and ascending station number. The information listed for each measurement include the PIC hexadecimal address, the DSO channels where the measurement is recorded, the

PAGE 224

224 data fiber color, the gain/attenuation value (if applicable), and the reference voltage for the calibration square wave. For electric field and dE/dt measurements the area of the flat plate antenna is given, and for NaI and LaBr 3 measurements, the detector response (in millivolts) to the Cs 137 662 keV source is given. Two columns are left ope n for the pre storm and post storm calibration factors. These values are calculated by measuring the peak to peak voltages of the square waves generated by the PIC controller at each station. Measurements of the calibration square waves are handled in a custom waveform viewing program named "DF" written by Dr. Douglas Jordan. When the calibration waveform is opened for a given channel, simply pressing the letter "c" on the keyboard will generate a text box including the peak to peak voltage of the square wave to millivolt accuracy. The pre storm and post storm calibration amplitudes for each measurement are averaged and stored in a separate column within the Equipment Table. The most important value in the Equipment Table is the actual measurement calib ration factor used to convert the raw digitizer volts recorded on each DSO to meaningful physical units. For each type of measurement, this value takes into account the theoretical expression for the vo ltage output of the sensor (reference Section 2.13 fo r time domain response of the flat plate dE/dt antenna), any external gain or attenuation, and the calibration signal reference voltage (1 V peak to peak if a 50 resistor is placed across the input of the Opticomm transmitter, 2 V peak to peak if the sig nal is injected directly into the 68 k Opticomm input impedance). The raw digitizer volts recorded in each data file are multiplied by the calibration factor to obtain the proportional data file in meaningful physical units. Within the Equipment Table, measurements are color coded according to various criteria (i.e., configuration changed since last storm date, no calibration signal recorded, did not save appropriately, recorded but did not function

PAGE 225

225 appropriately, did not record). The key for the color coding is provided to the right of the primary table. The Data List file is also prepared in Microsoft Excel and is an equally useful document. The Data List contains a row for every DSO channel. Within each row, the name of the recorded measurement is l isted in addition to the sampling rate and the record length. For each natural lightning or triggered lightning attempt during the given day, a column is created to the right of the measurement name with the heading "MSE XX YY" or "UF XX YY", where "XX" i s the suffix of the calendar year and "YY" is the event number of the calendar year. Natural lightning events have the prefix "MSE" and triggered lightning events have the prefix "UF". This naming convention is carried throughout this document. Within each event column, the file names of the collected DSO waveform files are given for every channel. The Data List file is extremely helpful and saves the user a great deal of time when processing large number of waveforms from many different DSOs (i.e., co mputing TOA locations). The final documentation file is the Summary document. This document is prepared in Microsoft Word. The Summary document is an important reference tool for data processing purposes. The document first provides a general descript ion of the local atmospheric conditions during any natural or triggered lightning events and a list of the functioning experiments during the given storm day. Tables are given at the beginning of the document containing the calibration factors for the cha nnel base current measurements and the list of functioning cameras and their settings. A written description is given for each triggered lightning attempt including such parameters as the quasi static electric field at the time of the launch, the initiati on altitude of the sustained UPL, the overall peak current of the flash, the flash multiplicity, and the UPL/ICC charge transfer, duration, and average current. If the event contained any unusual characteristics,

PAGE 226

226 they are noted. For each event, tables ar e given including the quasi static electric field at ground on all eight Campbell Scientific field mills (Section 2.6), a list of the return strokes (if applicable) and their corresponding measured peak currents, GPS times, and LeCroy memory segments, and a list of the number of channels recorded for each type of measurement. A written description is also provided for the high speed video data of the rocket ascent, sustained UPL, and subsequent leader/return stroke sequences. If the event contains return strokes, the corresponding frame numbers in the Photron and Phantom high speed video files are given in a table. At the end of the written portion of the Summary document, a thorough list of known problems that may have occurred during the data collection period is provided. Typically, plots of the channel base current records for triggered lightning events are given at the end of the document. Additional plots may be included of other unique waveform characteristics (e.g., dE/dt and x ray waveforms of d art stepped or "chaotic" dart leaders, abnormally long inter stroke continuing currents, large M components). 4.2 Synopsis of Collected Lightning Data In Table 4 1, a list is provided of the natural lightning discharges recorded by the ICLRT instrumentati on from 2009 2011. These discharges, 12 in total, terminated within or very close to the network boundaries. Historically, the average number of natural lightning discharges terminating on ground within the ICLRT network boundaries is 5 8 per year. In 2 009 and 2010, a total of only four natural lightning discharges were recorded, a number well below the expected average. In 2011, there were eight natural lightning discharges recorded, including six during one spectacular convective storm on July 31, 201 1 (four of which terminated within the network boundaries). There were a total of three natural return strokes recorded that were preceded by dart NLDN reported p eak currents for the 12 first return strokes ranged from 10.6 99.4 kA with

PAGE 227

227 arithmetic mean (AM) of 33.1 kA, geometric mean (GM) of 28.2 kA, standard deviation (SD) of 24.3 kA, and standard deviation of the base ten logarithm (SD(log)) of 0.24. The high st andard deviation is due mostly to the 99.4 kA peak current of flash MSE 09 01. The peak current distribution of the natural return strokes is in good agreement with the 50% value of 30 kA reported by Berger et al. [1975] for 101 negative return strokes m easured at two telecommunications towers on Monte San Salvatore in Switzerland. The NLDN reported peak currents of the six subsequent return strokes ranged from 17.0 38.3 kA with AM of 25.1 kA, GM of 24.2 kA, and SD of 7.7 kA. The peak currents of the su bsequent strokes are about a factor of two greater than the 50% value of 12 kA reported by Berger et al [1975] for 135 negative subsequent strokes to the towers on Monte San Salvatore. This statistic is not surprising considering that all subsequent retur n strokes identified by the NLDN were preceded by either dart For 22 strokes initiated by dart stepped leaders and 15 strokes initiated by "chaotic" dart leaders, Rako v and Uman [1990b] found that the peak electric fields normalized to 100 km were 4.8 V/m and 4.3 V/m, respectively. In contrast, for 232 subsequent strokes not categorized by preceding leader type, Rakov et al. [1994] found that the peak electric field no rmalized to 100 km was 2.7 kV/m. The peak electric fields for dart stepped and "chaotic" dart leaders are in better agreement with the normalized value of 4.1 kV/m reported by Rakov et al. [1994] for subsequent strokes that created a new ground terminatio n point. From optical measurements, Guo and Krider [1982] reported that natural lightning return strokes preceded by dart stepped leaders have higher light intensity than return strokes preceded by dart leaders. This result, coupled with the work of Idon e and Orville [1985], who found that the return stroke light pulse peak was highly correlated with the measured channel base current in 39 triggered lightning subsequent return strokes,

PAGE 228

228 suggests that dart stepped leaders may typically be associated with hi gher peak current subsequent return strokes. All three recorded dart stepped leaders preceded the second return stroke in the given flash, a result in agreement with the observations of Schonland [1956], who analyzed six dart stepped leaders, all associat ed with the second strokes, and also in agreement with Rakov and Uman [1990b], who found that second strokes were initiated by dart stepped leaders more than five times as frequently as all higher order strokes combined. High speed video data were acquire d of four natural lightning flashes, including an event on June 18, 2010 that was photographed at a frame rate of 300 kfps, the fastest video of a natural lightning discharge recorded to date. The high speed video observations of this flash, particularly concerning the formation of stepped leader steps, are discussed in detail in Chapter 6. In Tables 4 2, 4 3, and 4 4, lists are provided of the triggered lightning attempts that resulted in, at minimum, a full initial stage (IS) process in 2009, 2010, an d 2011, respectively. For each triggered flash, statistics are given, when available, for the duration (ms), charge transfer (C), and average current amplitude (A) during the initial continuous current (ICC), the time of the initial current variation (ICV ) relative to upward positive leader (UPL) initiation (ms), the flash multiplicity and overall peak current (kA), and the Photron SA1.1 frame rate (kfps). In this document, a full IS process is defined as a wire launch with measured precursor current pul ses on the ascending triggering wire, followed by the initiation of a sustained UPL, the subsequent explosion of the triggering wire (ICV), and the full duration of ICC. In 2009, there were 22 events having at least a full IS process, 17 of which had at l east one subsequent return stroke. All events were triggered from the Tower Launcher. Sixteen of the events were photographed with the Photron SA1.1 high speed camera at frame rates from 108 300 kfps. A total of eight dart

PAGE 229

229 triggered lightning return strokes. In 2010, there were a total of 14 triggered lightning flashes with at least a full IS process, 10 of which had at least one subsequent return stroke. All events with su bsequent return stroke(s) were photographed at a frame rate of 300 kfps. Flashes UF 10 01 through UF 10 18 were triggered from the Field (Ground) Launcher and flashes UF 10 20 through UF 10 26 were triggered from the Tower Launcher. A total of one dart s tepped leader the "chaotic" dart leaders recorded during 2010 is given in Chapter 7. Finally, in 2011, there were 15 triggered lightning events with at least a full IS process, 12 of which had at least one subsequent return stroke. Flashes UF 11 01 through UF 11 08 were triggered from the Tower Launcher while flashes UF 11 11 through UF 11 35 were triggered from the Field (Ground) Launcher. Nine of the trigg ered lightning flashes with subsequent return strokes were photographed at either 300 kfps or 450 kfps. A total of 10 dart dart leaders were recorded during 2011. General statistics for all dart stepped leaders and order and subsequent return stroke peak current are given in Tables 4 5 and 4 6, respectively. 4.3 General Statistics for Triggered Lightning Return Strokes (2009 2011) In Figure 4 1, the distributions of return stroke peak currents are shown graphically for triggered lightning events during 2009, 2010, 2011, and 2009 2011. Statistics for each data set are overlaid on the respective plot including the sample size ( N), the arithmetic mean (AM), geometric mean (GM), standard deviation (SD), standard deviation of the base ten logarithm (SD(log)), minimum peak current (Min), and maximum peak current (Max). Statistics for peak currents greater than about 6 kA were measu red using data recorded on Scope 26, a LeCroy 44 Xi DSO with bandwidth of 20 MHz. Statistics for peak currents less than 6 kA were measured using data recorded on Scope 24, a Yokogawa DL750 with bandwidth of 3 MHz (2009), or with

PAGE 230

230 the HBM Genesis system wi th bandwidth of 25 MHz (2010, 2011). Large current pulses (amplitudes greater than 1.5 kA) that occurred during the IS processes of three triggered flashes (flash UF 10 23, flash 11 08, and flash UF 11 15) were classified as return strokes if their rise t imes were less than 40 s. Over the three year period, there were a total of 156 triggered lightning return strokes recorded with amplitudes ranging from 1.5 46.5 kA. These strokes had AM of 13.1 kA, GM of 10.9 kA, SD of 7.9 kA, and SD(log) of 0.27 kA. The 2009 dataset contained both more return strokes (61) and higher average peak current return strokes (AM of 15.6 kA and GM of 13.6 kA) than either the 2010 or 2011 datasets. The total return stroke peak current statistics for the 2009 2011 dataset are given at bottom in Table 4 7, along with similar statistics for triggered lightning studies conducted at the Kennedy Space Center in Florida [ Depasse 1994], Saint Privat d'Allier in France [ Depasse, 1994], the Kennedy Space Center and Fort McClellan in Al abama [ Fisher, 1993], and at Camp Blanding in Florida [ Crawford, 1998; Rakov et al., 1998; Uman et al., 2000; Schoene et al., 2003; Jerauld et al., 2005; Nag et al., 2011]. From Table 4 7, it is clear that the statistics for the peak currents of triggered lightning return strokes at Camp Blanding during 2009 2011 are in very good agreement with those of previous studies. The distribution of triggered lightning return stroke peak currents can also be viewed in relation to the preceding leader type. To the best of the author's knowledge, this comparison is not found elsewhere in the literature for triggered lightning return strokes. Of the 156 return strokes with measured channel base current recorded from 2009 2011, 121 were preceded by dart leaders, 18 w ere preceded by dart stepped leaders, and 17 were preceded by "chaotic" dart leaders. The leader type is determined using dE/dt waveforms measured within several hundred meters of the triggered lightning channel. A graphical representation of the distrib ution of return

PAGE 231

231 stroke peak currents versus preceding leader type is shown in Figure 4 2. The AM peak currents for strokes preceded by dart leaders, dart stepped leaders, and "chaotic" dart leaders were about 11.2 kA, 19.9 kA, and 19.7 kA, respectively. Similarly, the GM peak currents for the three leader types were 9.5 kA, 17.5 kA, and 17.7 kA. The AM and GM peak currents of strokes associated with dart stepped and "chaotic" dart leaders were about 75% and 85% higher than those associated dart leaders. This result is not entirely unexpected considering the higher electric field peak values and higher peak luminosities of dart stepped and "chaotic" dart leaders associated with natural subsequent strokes, but is important nonetheless. A histogram of t he occurrences of dart stepped leaders and "chaotic" leaders versus triggered lightning return stroke order is shown in Figure 4 3 for the 2009 2011 dataset. Six out of 19 (~32%) dart stepped leaders and 8 out of 17 (~47%) "chaotic" dart leaders were asso ciated with the first return stroke following the IS. For dart stepped leaders, this statistic is consistent with the findings of Rakov and Uman [1990b] (discussed above in Section 4.2) for dart stepped leaders associated with natural lightning second str okes. To the author's knowledge, comparable statistics are not available in the literature for "chaotic" dart leaders associated with natural lightning second strokes. A final statistical return stroke parameter measured for the 2009 2011 dataset was th e overall stroke multiplicity (the number of strokes in a given flash). The distributions of triggered lightning flash multiplicities for the three individual years of study and for the full dataset are given in Figure 4 4. For the 39 flashes with subseq uent return strokes and measured channel base current, stroke multiplicities were recorded ranging from 1 16 strokes. The triggered flashes had AM of 4.2 strokes, GM of 3.0 strokes, and SD of 3.4 strokes. Eleven flashes (about

PAGE 232

232 26% of the total number) ha d one return stroke following the IS. Eight flashes (about 21% of the total number) had seven or more return strokes following the IS. 4.4 General Statistics for Triggered Lightning UPL/ICC Currents (2009 2011) For the 2009 2011 datasets, specific valu es for the duration, charge transfer, and average current amplitude of the UPL/ICC process were given in Tables 4 2 through 4 4 for each triggered lightning event with available channel base current measurements. In Figure 4 5, the distributions of ICC ti me durations for each year (2009 2011) are plotted in addition to the distribution for the full dataset. The UPL/ICC duration is defined as the elapsed time between the initiation of the sustained UPL and the point at which the continuous current ceases t o flow. The UPL/ICC durations were measured using waveform data from a sensitive channel base current measurement (minimum resolvable amplitude of several hundred milliamperes) recorded on Yokogawa DL750 DSOs. ICC durations of 46 flashes were measured fr om 160 945 ms, with AM of 433 ms, GM of 387 ms, and SD of 197 ms. The variations of the UPL/ICC duration statistics within the three years of study were relatively minor. The GM UPL/ICC duration for the 2009 2011 dataset was about 39% greater than the re ported GM duration of 279 ms reported by Wang et al. [1999 c ] for 37 triggered flashes at Fort McClellan and Camp Blanding, and about 27% greater than the GM duration of 305 ms reported by Miki et al. [2005] for 45 triggered flashes at Camp Blanding. In F igure 4 6, a similar set of distributions are plotted for the UPL/ICC charge transfers of the 46 flashes. The UPL/ICC charge transfer is calculated by numerically integrating the channel base current waveform using the trapezoidal integration technique. Care is taken to remove any DC offset in the waveform data prior to performing the integration. For the full dataset, UPL/ICC charge transfers were measured ranging from 8 225 C, with AM of 67 C, GM of 50 C, and SD of 47 C. There were a total of nine fla shes with UPL/ICC charge transfers in

PAGE 233

233 excess of 100 C, including six during 2011. The variation of the mean UPL/ICC charge transfers over the three years of study is less than 20%. The GM UPL/ICC charge transfers measured during 2009 2011 were about 85% larger than the GM value of 27 C reported by Wang et al. 1999 c ] and about 64% larger than the GM value of 30.4 C reported by Miki et al. [2005]. In Figure 4 7, the distributions of UPL/ICC average current amplitudes are plotted individually for the three years of study and for the full dataset. The UPL/ICC average current amplitude is obtained by averaging the current values over the duration of the UPL/ICC process, taking into account any DC offset prior to calling the averaging routine. For the 46 eve nts, average current amplitudes were measured from 49 834 A, with AM of 145 A, GM of 130 A, and SD of 68 A. Interestingly, the two events with abnormally large UPL/ICC average current durations (834 A for flash UF 09 30 and 328 A for flash UF 11 26) both exhibited current polarity reversals during the UPL/ICC process. Yoshida et al. 2012] used a broadband interferometer to map the IS of flash UF 09 30. Similarly, Hill et al. [2012] present LMA data for the IS of flash UF 11 26. The correlated LMA and ch annel base current observations for flash UF 11 26 are also presented in detail in Chapter 8 of this document. Similar to the statistics for UPL/ICC charge transfer, there was less than 20% difference between the mean values of UPL/ICC average current amp litude between the three years of study. The GM UPL/ICC average current amplitude of 130 A was about 35% larger than the GM value of 96 A reported by Wang et al. [1999 c ] and about 31% larger than the GM value of 99.6 A reported by Miki et al. [2005]. I n Figure 4 8, distributions are plotted of the time duration between the initiation of the sustained UPL and the ICV for 37 triggered lightning flashes. Recall that the ICV is associated with the vaporization of the copper triggering wire. In nine flashes the ICV could not be clearly

PAGE 234

234 determined. Time durations for this statistic were measured ranging from 2.1 69.4 ms with AM of 13.1 ms, GM of 7.5 ms, and SD of 17.1 ms. The GM duration for our 37 cases was in good agreement with the GM duration of 8.6 ms reported by Wang et al. [1999 c ] for 22 triggered flashes at Fort McClellan and Camp Blanding. Olsen et al. [2006] classify ICVs in two categories, 1) Type I events exhibit a pronounced decrease in channel base current towards zero over a period of some h undreds of microseconds, followed by a flattening of the current at the zero level (within the amplitude resolution of the measurement system) that persists for typically 1 4 ms, and 2) Type II events exhibit a similar decrease in the channel base current, again lasting some hundreds of microseconds, but the current does not flatten at the zero level and the overall current reduction period is generally less than 500 s. Type I events often exhibit return stroke like current pulses at ground during the ze ro current period. These pulses have rise times less than 1 s and have typical amplitudes from 40 250 A [e.g., Olsen et al. 2006]. For both Type I and Type II events, the current is re established between the UPL and ground following the current reduct ion period by a larger pulse with typical amplitude of 1 kA. Olsen et al. [2006] refer to the return stroke like pulses in Type I events as attempted reconnection pulses (ARP) and the larger pulse that re establishes current between the UPL and ground as the reconnection pulse (RP). If the metric described in Olsen et al. [2006] is used to classify ICVs measured from 2009 2011, there were a total of eight Type I events and 31 Type II events. The eight Type I events had zero current periods with durations ranging from 290 s to 1.523 ms with AM of 806 s, GM of 706 s, and SD of 413 s. The zero current durations for Type I events in this study are somewhat shorter than those reported by Olsen et al. [2006]. Six of the eight (75%) Type I events exhibited at least one ARP during the zero current interval. The nine total ARPs had AM of 116 A, GM of 113 A, and SD of 33 A. In addition, four ICVs that would otherwise be

PAGE 235

235 classified as Type II events exhibited smaller ARP like pulses during the current reducti on period. There were a total of four of these pulses recorded ranging in amplitude from 9 16 A. For Type I events, the RP amplitudes ranged from about 250 A to 2.75 kA with AM of 820 A, GM of 600 A, and SD of 830 A. It is also interesting to relate th e occurrences of Type I and Type II ICVs to other parameters measured during the IS process. The AM (GM) time durations between the initiation of the sustained UPL and the ICV for Type I and Type II events were 22 (19) ms and 11 (6) ms, respectively. Sim ilarly, the AM (GM) UPL/ICC duration for flashes containing Type I and Type II ICVs were 547 (462) ms and 409 (373) ms, respectively. Further, the AM (GM) UPL/ICC charge transfer for flashes containing Type I and Type II ICVs were 83 (56) C and 64 (49) C respectively. To summarize, with calculated GM values used for reference, flashes containing Type I ICVs exhibited longer time durations between the initiation of the sustained UPL and the ICV by a factor of 3, had 24% longer UPL/ICC durations, and 14% larger UPL/ICC charge transfers. The action integrals (the time integral of the square of the channel base current waveform) between the initiation of the UPL and the time of the ICV were calculated for the 37 flashes where the ICV was clearly defined. The distribution s of calculated action integrals for the 2009 2011 dataset are shown in Figure 4 9 The action integral (or specific energy) is the energy that would dissipated in a 1 resistor if the lightning current were to flow through it. According to Rakov and Uman [2003], the value of the action integral determines the heating of electrically conducting materials and the explosion of non conducting materials. Considering the triggering wire spools are relatively uniform in their resistance per u nit length, one would expect the action integral between the initiation of the UPL and the ICV to be constrained to a reasonably narrow

PAGE 236

236 range of values. Indeed, 65% of the calculated action integrals for the 2009 2011 dataset fall between 120 150 A 2 s with a GM for the full dataset of 119 A 2 s. About 95 % of the action integrals have values between 60 150 A 2 s. The two outliers (flash UF 09 12 with action integral of 305 A 2 s and flash UF 10 13 with action integral of 29 A 2 s) had no unusual characteristics suggesting that the wire spools themselves like ly had different characteristics Wang et al. [1999c] calculated the GM action integral between the UPL initiation and the ICV for 22 events at Fort McClellan and Camp Blanding to be 110 A 2 s, a value in good agreement with the present study. Finally, in Figure 4 1 0 the distribution s of average current amplitudes for the time period between the UPL initiation and the ICV are shown for the 2009 2011 dataset. The average current amplitudes range from 29 A to 127 A with a GM value for the full dataset of 67 A. The variability in the average current amplitudes between the UPL initiation and the ICV suggests that that action integral is primarily responsible for the determining when the triggering wire exploded and not the average current amplitude. In Table 4 8, measured statistics from this study in addition to those of Wang et al. [1999 c ] and Miki et al. [2005] are provided for channel base currents during the IS of triggered lightning discharges.

PAGE 237

237 Tabl e 4 1. A list of natural lightning discharges that terminated on or very near the ICLRT from 2009 2011. Date Flash Return Stroke Leader Type Time (UT) NLDN Latitude NLDN Longitude NLDN Peak Current (kA) High Speed Video Frame Rate (kfps) 063009 MSE 09 01 1 Stepped 11:21:29.496232 29.9472 82.0211 99.4 061810 a 1 Stepped 19:38:29.853946 10.6 300 063010 MSE 10 01 1 Stepped 17:16:35.715782 29.9408 82.0368 21.5 4.3 092710 MSE 10 02 1 Stepped 18:05:06.796647 29.9443 82.0317 27.7 4.3 2 Dart Ste pped 18:05:06.865118 29.9441 82.0322 38.3 4.3 070711 MSE 11 01 1 Stepped 19:37:34.737534 29.9389 82.0297 30.4 2 Dart Stepped 19:37:34.794163 29.9396 82.0336 22.1 3 "Chaotic" Dart 19:37:34.837564 29.9401 82.0348 20.6 4 "Chaotic" Dart 19:37: 34.881848 29.9396 82.0348 22.8 073111 MSE 11 02 1 Stepped 20:23:40.166011 29.9442 82.0120 25.3 073111 MSE 11 03 1 Stepped 20:27:48.682425 29.9368 82.0248 28.5 073111 MSE 11 04 1 Stepped 20:34:57.121671 29.9396 82.0343 37.6 2 Dart Stepped 20:3 4:57.157557 29.9376 82.0336 17.0 073111 b MSE 11 05 1 Stepped 20:44:15.489141 2 "Chaotic" Dart 20:44:15.922976 29.9393 82.0270 29.7 073111 MSE 11 06 1 Stepped 20:58:22.833734 29.9417 82.0274 20.8 073111 MSE 11 07 1 Stepped 21:02:00.11693 6 29.9412 82.0327 29.8 081811 b MSE 11 08 1 Stepped 20:24:50.095451 a The ground strike point of the natural lightning event on 061811 was about 1 km to the northwest of the ICLRT. b Return strokes not recorded by NLDN.

PAGE 238

238 Table 4 2. 2009 triggered lightning event characteristics. Only events including, at minimum, a full IS process are tabulated. For some events, there is no clear ICV in conjunction with the triggering wire explosion. Date Shot Time (UT) UPL/ICC Duration (ms) UPL/ICC Cha rge Transfer (C) UPL/ICC Average Current (A) Time of ICV Relative to UPL Initiation (ms) Current Zero During ICV Return Strokes Overall Peak Current (kA) Photron SA1.1 Frame Rate (kfps) 032809 UF 09 06 01:55:24 575 94 164 6.9 No 6 17.3 120 052609 UF 09 1 2 20:31:496924 422 71 169 2.9 No 1 34.9 108 060409 UF 09 15 20:20:20 164 8 49 17.7 Yes 0 0.4 060409 UF 09 17 20:37:13.416051 576 74 128 4.6 No 5 46.5 135 061809 UF 09 20 16:34:03.119539 418 46 110 No 4 20.1 180 061809 UF 09 21 16:44:42 556 71 128 1 1.1 No 0 2.0 061809 UF 09 22 16:58:10.953253 433 67 162 10.9 No 8 15.5 180 062909 UF 09 25 21:09:17.612277 412 51 124 No 5 32.9 240 062909 UF 09 26 21:18:36.306798 522 82 157 3.1 No 5 28.7 240 062909 UF 09 27 17:31:23.005941 188 16 83 7.3 No 6 28.4 240 063009 UF 09 29 13:49:17.096270 351 66 171 2.2 No 5 16.7 300 063009 UF 09 30 a 14:01:04.391149 234 93 834 2.9 No 1 31.6 300 063009 UF 09 31 14:12:24.486065 356 51 144 2.2 No 5 14.8 300 070709 UF 09 32 15:12:57.261135 719 193 272 2.7 No 1 19.2 180 070909 UF 09 34 17:15:40.445418 204 15 77 No 3 17.7 180 071409 UF 09 35 21:07:57 668 96 144 No 0 0.4 071409 UF 9 37 b 21:21:21 0 071409 UF 09 38 21:25:18.526361 537 86 160 8.6 No 1 29.8 180 071809 UF 09 40 200 26 137 5.3 No 0 1.4 081809 UF 09 41 c 16:22:25 0 081809 UF 09 42 16:24:41.585945 280 62 228 4.8 No 0 14.5 300 081809 UF 09 43 d 16:30:12.252595 5 25.3 300 a Flash UF 09 30 exhibited a bi polar IS process. The charge transfer and average current giv en is only for the negative charge transport to ground. b Scopes 22/25 had finished recording at the time of the IS process. c Scopes 22/25 were not armed and did not record the IS process. d Scopes 22/24/25 were not finished saving from the previo us shot at the time of the rocket launch.

PAGE 239

239 Table 4 3. 2010 triggered lightning event characteristics. Only events including, at minimum, a full IS process are tabulated. For some events, there is no clear ICV in conjunction with the triggering wire exp losion. Date Shot Time (UT) UPL/ICC Duration (ms) UPL/ICC Charge Transfer (C) UPL/ICC Average Current (A) Time of ICV Relative to UPL Initiation (ms) Current Zero During ICV Return Strokes Overall Peak Current (kA) Photron SA1.1 Frame Rate (kfps) 060510 U F 10 01 18:12:26.015687 190 9 49 No 1 11.6 300 061710 UF 10 06 23:59:21.392145 742 102 137 13.7 Yes 8 14.9 300 061810 UF 10 09 20:08:19.883969 246 44 174 No 3 24.6 300 062110 UF 10 13 20:09:37.357362 684 68 98 26.2 Yes 1 43.1 300 063010 UF 10 14 a 1 7:26:23 5.8 No 0 063010 UF 10 15 17:32:05.060857 265 38 145 4.9 No 3 12.7 300 071110 UF 10 16 22:08:29 401 75 186 2.1 No 0 3.7 071110 UF 10 17 22:41:14.813597 332 45 135 2.6 No 7 22.4 300 071110 UF 10 18 22:55:05 404 69 169 2.6 No 0 4.1 071510 UF 10 20 17:28:25.235689 652 87 133 49.1 Yes 4 17.1 300 071510 UF 10 21 17:35:06.232322 402 42 107 14.0 Yes 3 22.2 300 073110 UF 10 23 20:25:28.639877 719 107 149 5.7 No 16 23.8 300 081310 UF 10 24 19:44:35.013453 548 100 182 42.6 No 9 28.3 300 092710 UF 10 26 19:44:29 212 18 87 9.5 No 0 2.3 a Scopes 22/25 had finished recording at the time of the IS process.

PAGE 240

240 Table 4 4. 2011 triggered lightning event characteristics. Only events including, at minimum, a full IS process are ta bulated. For some events, there is no clear ICV in conjunction with the triggering wire explosion. Date Shot Time (UT) UPL/ICC Duration (ms) UPL/ICC Charge Transfer (C) UPL/ICC Average Current (A) Time of ICV Relative to UPL Initiation (ms) Current Zero D uring ICV Return Strokes Overall Peak Current (kA) Photron SA1.1 Frame Rate (kfps) 012511 UF 11 01 21:43:36 187 12 64 4.1 No 0 042011 UF 11 04 20:33:18 171 8.9 52 66.7 No 0 051411 UF 11 08 a 17:23:35.154788 10 8.3 450 062311 UF 11 11 19: 06:27.834612 345 22 63 No 5 13.5 300 070711 UF 11 15 19:02:19.528147 161 18 114 8.0 Yes 11 20.5 300 070711 UF 11 18 19:26:27.471664 211 11 51 5.8 No 1 8.2 300 071011 UF 11 20 19:01:28.051825 398 33 83 No 2 7.7 300 080511 UF 11 24 19:33:19.5435736 4 25 46 107 4.7 No 1 32.8 300 080511 UF 11 25 19:43:31.4899735 404 28 70 No 1 12.1 300 080511 UF 11 26 b 19:49:55 433 120 328 5.2 No 0 081211 UF 11 28 23:39:21.288556 694 136 196 No 3 19.5 300 081811 UF 11 32 20:37:30.841497 945 225 236 18.7 Yes 2 19.8 300 081811 UF 11 33 c 20:45:13.497381 630 110 219 28.4 Yes 1 7.6 081811 UF 11 34 c 20:51:42.136853 567 120 211 3.9 No 2 13.4 081811 UF 11 35 c 20:58:11.665782 726 128 176 3.9 No 7 27.4 a The II Low and II Very Low channel base current measureme nts malfunctioned and no data were recorded b Flash UF 09 30 exhibited a bi polar IS process. The charge transfer and average current given is only for the negative charge transport to ground. c The Photron SA1.1 had filled its three memory partitio ns and did not record flashes UF 11 33 through 11 35.

PAGE 241

241 Table 4 5. A list of dart stepped leaders preceding triggered lightning return strokes between 2009 2011. The following return stroke peak current is given for each leader event. Date Shot Stroke Peak Current (kA) 052609 UF 09 12 1 34.9 060409 UF 09 17 1 24.2 060409 UF 09 17 2 22.8 060409 UF 09 17 4 46.5 062909 UF 09 25 5 32.9 062909 UF 09 26 4 24.6 063009 UF 09 29 5 20.0 070709 UF 09 32 1 19.2 061710 UF 10 06 a 3 051411 UF 11 08 10 8.3 070711 UF 11 15 10 10.2 080511 UF 11 25 1 12.1 081811 UF 11 32 1 14.5 081811 UF 11 32 2 19.8 081811 UF 11 34 1 13.4 081811 UF 11 34 2 12.1 081811 UF 11 35 4 27.4 081811 UF 11 35 6 7.5 081811 UF 11 35 7 8.4 a The final five return strokes of flash UF 10 06 did not attach to the intercepting rod over the western tower on the Pad 39B catenary wire experiment. No current data are available for these returns strokes.

PAGE 242

24 2 Table 4 6. A list of "chaotic" dart leaders preceding triggere d lightning return strokes between 2009 2011. The following return stroke peak current is given for each leader event. Date Shot Stroke Peak Current (kA) 063009 UF 09 30 1 31.6 062110 UF 10 13 1 43.1 071510 UF 10 20 1 17.1 071510 UF 10 21 1 22.2 08 1310 UF 10 24 3 28.3 062311 UF 11 11 1 13.3 062311 UF 11 11 2 11.2 062311 UF 11 11 3 10.5 062311 UF 11 11 4 10.6 062311 UF 11 11 5 13.5 070711 UF 11 15 7 20.5 070711 UF 11 15 11 10.5 070711 UF 11 18 1 8.2 080511 UF 11 24 1 32.8 081211 UF 11 28 1 19.5 081811 UF 11 35 2 18.4 081811 UF 11 35 5 23.5

PAGE 243

243 Figure 4 1. Distributions of triggered lightning return stroke peak currents at Camp Blanding in A) 2 009, B) 2010, C) 2011, and D) 2009 2011. Computed statistics for each dataset are overlaid on the respective plots. A B C D

PAGE 244

244 Table 4 7. Triggered lightning return stroke peak current statistics for studies at the Kennedy Space Center in Florida, Saint Privat d'Allier in France, Fort McClellan in Alabama, and Camp Blanding in Florida from 1985 2011. The statistics for the present study (2009 2011) are given in the bottom row of the table. Location Years Number of Strokes Min (kA) Max (kA) AM (kA) GM (kA) SD (kA) Log(SD) Kennedy Space Center, Florida 1985, 1987 1991 305 2.5 60.0 14.3 9.0 Saint Privat d'Allier, France 1986, 1990 1991 54 4.5 49.9 11.0 5.6 Kennedy Space Center, Florida & Fort McClellan, Alabama 1990, 1991 45 < 2.0 38.0 12 0.28 Camp Blanding, Florida 1993 37 5.3 44.4 15.1 13.3 0.23 Camp Blanding, Florida 1997 11 5.3 22.6 12.8 11.7 5.6 0.20 Camp Blanding, Florida 1998 24 5.9 33.2 14.8 13.5 7.0 0.19 Camp Blanding, Florida 1999 2000 64 5.0 36.8 16.2 14.5 7.6 0.21 Camp Blanding, Florida 2001 2003 122 2.9 43.0 15.0 13.0 7.4 0.20 Camp Blanding, Florida 2004 2009 12 2 2.9 45.0 14.0 12.0 7.9 0.23 Camp Blanding, Florida 2009 2011 156 1.5 46.5 13.1 10.9 7.9 0.27

PAGE 245

245 Figure 4 2. Distribution of triggered lightning return stroke peak currents versus preceding leader type. From left, A ) dart leaders, B) dart s tepped leaders, and C ) "chaotic" dart leaders. A C B

PAGE 246

246 Figure 4 3. Histogram of the occurrences of dart stepped leaders (blue) and "chaotic" dart leaders (red) versus triggered lightning subsequent return stroke order (2009 2011 dataset).

PAGE 247

247 Figure 4 4. Distributions of triggered lightning flash multiplicities at Camp Blanding in A) 2009, B) 2010, C) 2011, and D) 2009 2011. A B C D

PAGE 248

248 Figure 4 5. Distributions of triggered lightning UPL/ICC time durations for 46 flashes trigger ed at Camp Blanding in A) 2009, B) 2010, C) 2011, and D) 2009 2011. A B C D

PAGE 249

249 Figure 4 6. Distributions of triggered lightning UPL/ICC charge transfers for 46 flashes triggered at Camp Blanding in A) 2009, B) 2010, C) 2011, and D) 2009 2011. A B C D

PAGE 250

250 Figure 4 7. Distributions of triggered lightning UPL/ICC average current amplitudes for 46 flashes triggered at Camp Blanding in A) 2009, B) 2010, C) 2011, and D) 2009 2011. A B C D

PAGE 251

251 Figure 4 8. Distributions of the time duration between the i nitiation of the sustained UPL and the ICV for 37 triggered lightning flashes at Camp Blanding in A) 2009, B) 2010, C) 2011, and D) 2009 2011. A B C D

PAGE 252

252 Figure 4 9 Distribution of calculated action integrals between the initiation of the UPL and the ICV for 37 triggered flashes at Camp Blanding in A) 2009, B) 2010, C) 2011, and D) 2009 2011. A B C D

PAGE 253

253 Figure 4 1 0 Distribution of average current amplitudes b etween the initiation of the UPL and the ICV for 37 triggered flashes at Camp Blanding in A) 2009, B) 2010, C) 2011, and D) 2009 2011. A B C D

PAGE 254

254 Table 4 8. Measured channel base current statistics during the initial stage (IS) of triggered lightning discharges at Fort McClellan in Alabama, and Camp Blanding in Florida. The statistics for the present study (2009 2011) are given in the bottom row of the table. Location Years Events UPL/ICC Duration (ms) UPL/ICC Charge Transfer (ms) UPL/ICC Average Current (A) ICV (ms) Action Integral from UPL to ICV (A 2 s) Fort McClellan, Alabama Camp Blanding, FL 1994 1996 1997 37 a 279 27 96 8.6 (22 events) 110 (22 events) Camp Blanding, FL 1996 1997 1999 2000 45 b 305 30.4 99.6 Camp Blanding, Florida 2009 2011 46 387 50 130 7.5 ( 37 events) 119 (37 events) a Events discussed in Wang et al. [1999 c ] b Event s discussed in Miki et al. [2005]

PAGE 255

255 CHAPTER 5 HIGH SPEED VIDEO OBSERVAT IONS OF A NATURAL LI GHTNING STEPPED LEADER In this chapter, high speed video observations are presented for one branch of a stepped leader that preceded a natural lightning first return stroke. The focus of this work is to 1) develop a more thorough understanding of the propagation mechanisms (i.e. the leader step formation process) of lightning leaders through virgin air given the present knowledge obtained from long laboratory spark ex periments, and 2) to compare the measured properties with those of dart stepped leaders preceding triggered lightning return strokes. Due to the inherent repeatable nature of both laboratory spark experiments and triggered lightning, the optically observe d properties of leader propagation for both scales of discharges are better quantified than are the similar properties of natural stepped leaders. The results presented in this chapter are also discussed in Hill et al. [2011]. 5.1 Literature Review Pro perties of stepped leaders preceding negative natural lightning first strokes have been studied optically via streak photography [e.g., Schonland et al., 1935; Berger, 1967; Orville and Idone, 1982], through the use of photodiode array photographic system s [e.g., Chen et al., 1999; Krider, 1974; Lu et al. 2008], and through electric field measurements [e.g., Kit a gawa, 1957; Krider and Radda, 1975; Krider et al, 1977; Thomson, 1980; Beasley et al., 1982; Cooray and Lundquist, 1985]. Using a Boys continuou s moving film camera with time resolution of approximately 600 ns, Schonland et al. [1935] measured individual step lengths from 10 to 200 m, interstep intervals ranging from 40 to 100 s, and average two dimensional stepped leader speeds of 3.8 x 10 5 m/s. Berger [1967] used a similar streak camera to photograph four downward negative stepped leaders to the towers atop Mount San Salvatore in Switzerland. They also photographed 14 negative stepped leaders that terminated on ground. The step lengths

PAGE 256

256 for ne gative stepped leaders to the towers were 8 to 10 m with interstep intervals from 40 to 52 s, and the step lengths for negative stepped leaders to ground were 3 to 17 m with interstep intervals from 29 to 47 s. Stepped leader speeds ranged from 0.9 to 4 .4 x 10 5 m/s. Orville and Idone [1982] imaged three stepped leaders using a streak camera, but only the overall propagation speeds for the three leaders could be determined due to the weak optical intensity of the steps. Their measured leader speeds ra nged from 5.9 to 15 x 10 5 m/s. Chen et al. [1999] recorded two stepped leaders using the ALPS photodiode array imaging system in both Australia and China. For the event in Australia (time resolution of 500 ns), individual step lengths were measured betwe en 7.9 and 20 m, interstep intervals were from 5 to 50 s, and leader speeds were from 4.9 to 11 x 10 5 m/s. For the event in China (time resolution of 100 ns), step lengths were reported to be 8.5 m, interstep intervals were from 18 to 21 s, and leader speeds ranged from 4.9 to 5.8 x 10 5 m/s. Using a photoelectric detector in Arizona, Krider [1974] measured interstep intervals from 17 to 32 s for stepped leader pulses within 70 s of the return stroke. Lu et al. [2008] also used the ALPS optical imagi ng system to record a downward branched stepped leader in Florida at an estimated distance of about 1.3 km. They measured 60 values for interpulse interval for the main channel ranging from 0.2 to 15.7 s with geometric mean of 3.3 s and an average two d imensional leader propagation speed of 1.5 x 10 6 m/s. Through measurements of the electric field produced by descending negative stepped leaders obtained within a few hundred microseconds of the return stroke, interstep intervals have been estimated to be from 5 to 25 s [e.g., Kitigawa, 1957; Krider and Radda, 1975; Krider et al., 1977; Thomson, 1980; Beasley et al, 1982; Cooray and Lundquist, 1985]. Summary statistics for interstep interval, step length, and stepped leader propagation speed from previo us optical studies of stepped leaders are given in Table 5 1.

PAGE 257

257 While the overall characteristics of negative stepped leaders have been well studied, as noted above, the actual formation process of individual leader steps remains poorly documented and poo rly understood. From the photographic and photoelectric work of Schonland et al. [1935] and Orville and Idone [1982], the step formation appears to occur within about a microsecond, and hence is unresolved. The formation mechanism of natural negative lea der steps has however been inferred from meter length laboratory spark experiments, although the time and size scales of the two phenomena differ. Long laboratory spark development has been described by Gorin et al. [1976], Les Renardieres Group [1978], O rtega et al. [1994], Reess et al. [1995], Bazelyan and Raizer [1998], and Gallimberti et al. [2002]. Long laboratory spark discharges are initiated by applying megavolts of potential difference between electrodes separated by up to about 17 meters. Typi cally, the voltage magnitude across the gap and the rate of voltage rise are adjusted to control the discharge properties. The voltage across the gap and current through the gap are measured and the spark gap is often imaged using image converter cameras in either frame or streak mode. A general sequential description of the negative leader step formation process in long laboratory sparks follows: 1) when the negative voltage impulse is applied to the high voltage electrode, a burst of branched filamentar y corona streamers moves into the gap, heating the air in the immediate proximity to the electrode and establishing the initial leader channel at the electrode, 2) a luminous, apparently isolated space stem forms ahead of the primary negative leader tip wi th positive streamers propagating back from it towards the negative high voltage electrode (and initial leader channel) and negative streamers propagating downward from the space stem into the gap, 3) a new section of leader channel is formed when a suffic ient number of positive backwards propagating streamers reach the negative primary leader channel, then instantly transferring the potential of the high voltage electrode to the end of the

PAGE 258

258 secondary channel established by the space stem, and 4) a burst of negative streamers is emitted from the end of the secondary channel, initiating a current pulse that propagates up the existing channel segment towards the high voltage electrode, illuminating the full channel. A new space stem forms and the sequence note d above repeats during the formation of each new step. A drawing of the process, modified from Gorin et al. [1976], is given by Biagi et al. [2010]. While lightning stepped leaders should share some characteristics with long laboratory spark leaders, t he larger available gap, different charge source, and higher potential voltage at the tip of a lightning leader likely generates larger leader step currents with higher charge densities and greater propagation speeds [e.g., Rakov and Uman, 2003]. The fact that the type of leader channel development shown in laboratory sparks also occurs in lightning was first shown by Biagi et al. [2009] who obtained an image of a 2 m long space stem located roughly 4 m below the tip of a dart stepped leader that preceded the eighth return stroke of a triggered lightning discharge (flash UF 08 18 triggered on September 17, 2008). The space stem was photographed using a Photron SA1.1 high speed camera operated at a frame rate of 50 kfps, or 20 s per frame. Two frames of t he descending dart stepped leader preceding the return stroke are shown in Figure 5 1 (adapted from Biagi et al. [2009]). Each frame is shown in both a true and pixel inverted form to better illustrate the low level luminosity features of the leader. Bia gi et al. [2010] show additional instances of space stems and/or space leaders in seven high speed video frames of a dart stepped leader that preceded the fifth return stroke of a triggered lightning discharge (flash UF 09 25 triggered on June 29, 2009). The frames were recorded with the same camera, but at a frame rate of 240 kfps, or 4.17 s per frame. A sequence of 10 consecutive frames (~42 s) of the triggered lightning dart stepped leader are shown in Figure 5 2. In each of the seven frames where s pace stems and/or space leaders are visible, the luminous segments

PAGE 259

259 were 1 to 4 m in length and separated from (and were below) the primary leader channel by 1 to 10 m. In three frames, there were two clearly separated segments of luminosity vertically bel ow the primary leader channel (inset, Figure 5 2). In some instances, the luminous intensity of the separated channel segments was comparable to that of the primary leader channel, and in all cases, the luminous intensity of the separated channel segments was greater than that of the surrounding corona streamers. In this study, the shortest time exposure (3.33 s) video observations to date of a natural lightning stepped leader are analyzed. From the video data, accompanying statistics are measured for interstep interval, step length, and two dimensional leader propagation speed. In addition, the first observations of space stems/leaders associated with stepped leaders are presented. A schematic view of the formation of a natural negative leader step is provided from examination of 82 individual steps imaged during this event. The measured parameters from the high speed video analysis are compared to observations obtained in association with dart stepped leaders in triggered lightning discharges. Fin ally, the high speed video data confirm the observations of Wang et al. [1999 a ] and the suggestion of Chen et al. [1999] that following the progression of the leader channel due to a new step, a luminosity wave propagates from the leader tip back up the ex isting channel. 5.2 Experimental Setup and Data Processing Techniques The high speed video images acquired of the negative stepped leader presented here were recorded at the ICLRT on June 18, 2010. The flash occurred at 19:38:29.853946 (UT). The Nationa l Lightning Detection Network's (NLDN) most probable ground strike location of the 10.6 kA negative return stroke associated with the photographed stepped leader was about 1 km to the northwest of the ICLRT. High definition video records indicate that the stepped leader branches imaged with the high speed video traversed a path over and slightly to the southeast of

PAGE 260

260 the ICLRT. The stepped leader entered the field of view of the Photron SA1.1 high speed camera that was configured with a 20 mm Nikon lens set to an aperture of f/4 for imaging rocket triggered lightning discharges at a distance of 300 m. The Photron SA1.1 recorded the stepped leader at a frame rate of 300 kfps, or 3.33 s per frame. The resolution was 320 x 32 pixels (vertical x horizontal) w ith 12 bit grayscale amplitude resolution. As discussed in Section 2.7, the Photron SA1.1 is triggered to record on the output of the ICLRT master trigger box, which generates a trigger pulse anytime a natural lightning terminates within the site boundari es or a triggered lightning current is measured in excess of about 6 kA. In this case, neither of the trigger criteria were met. The Photron fortuitously triggered in conjunction with a "glitch" in the generator power that supplies the Office Trailer whe n storm conditions are presented. It is undetermined (yet probable) that the glitch in generator power was associated with an induced effect from the photographed lightning discharge. The remainder of the ICLRT data acquisition network did not record dat a for the flash. The best estimate of the distance from the camera to the stepped leader is 1 km, providing a spatial resolution of 1 m per pixel, with a possible range from as far away as 2 km to as close as 700 m. The best estimate and distance limits are based on the following: 1) comparison of the luminous features of streamers and space stems/leaders observed in the present high speed video with video recorded at the known distance of 430 m of dart stepped leaders in triggered lightning with the sam e camera at essentially the same frame rate [e.g., Biagi et al., 2010], 2) the fact that the stepped leader speed computed from the present data at a dista nce between 700 m and 2 km ( Section 5.4 and Table 5 2) is consistent with speeds found in the literat ure, 3) knowledge of the ground strike point of the lightning in question from the NLDN, and 4) the spatially smaller and less luminous phenomena evident in the video would have been difficult to resolve at distances

PAGE 261

261 too much beyond 1 km in the presence of the light precipitation that was falling at the end of the storm when the video was acquired. All stepped leader size and velocity measurements scale linearly with the assumed distance to the stepped leader. For all spatial measurements presented, the stepped leader channel is assumed to propagate only within the plane perpendicular to the field of view of the camera. Since each pixel represents 1 m at a range of 1 km, the approximate error in all length measurements is 1 m at 1 km. In addition, all s patial measurements are straight line lengths and are presented to tenths of a meter accuracy because of propagation along directions other than the horizontal or vertical of the pixel array. The video data were analyzed and processed in a darkened room using a Dell UltraSharp U2410 LCD monitor. The monitor was configured to display 1920 x 1200 pixels at 94 dots per inch (DPI) with a typical contrast ratio of 1000:1. Video images were manipulated (inverted and contrast enhanced) using Matlab and Adobe P hotoshop CS3. 5.3 Results A total of 225 frames (about 750 s) of downward stepped leader were recorded. After the ground attachment of the primary leader channel out of the field of view of the camera (one of many branches until it connected to ground ), the branches being observed were illuminated by the return stroke. There were 45 frames (about 150 s) captured during the return stroke illumination and subsequent decay. The stepped leader channel entered the field of the view of the camera at an alt itude of 358 m. The leader propagated downward within the field of view, splitting into two distinct branches about 417 s later at an altitude of 268 m. The two stepped leader branches continued to propagate downward and to the left in the frame, ultima tely exiting the horizontal field of view of the camera at altitudes of 217 m and 202 m respectively, and 227 s and 287 s respectively after branching from the original leader channel. Shown in Figure 5

PAGE 262

262 3 is a full frame image of the flash obtained by t ime integrating the luminosity from four consecutive frames during the return stroke illumination of the leader channel. All stepped leader branches are labeled with the notation "P X" indicating a primary branch and "S X" indicating a secondary branch, w here "X" is an integer. Differentiation between primary and secondary branches was to some degree subjective. However, primary branches were characterized by more extensive propagation within the field of view (and hence a greater number of measured lead er steps) while secondary branches were channels with smaller extent originating from the three primary leader channels. Three primary branches and five secondary branches have been analyzed. In Table 5 2, measured statistics are presented for each indiv idual branch including the number of individual leader steps, the average interstep interval, the average step length, and the average two dimensional leader propagation speed, all assuming a range of 1 km. Located below each value in Table 5 2 for averag e step length and average two dimensional leader speed, uncertainties are provided in the presented statistics in the form "(+ A / B)", where "A" represents the range correction in measurement for a maximum range of 2 km and "B" represents the correction for a minimum range of 700 m. For the three primary branches, there were a total of 64 individual leader steps analyzed. The average interstep intervals for the three primary branches ranged from 13.7 s to 15.1 s, the average step lengths ranged from 4.8 m to 5.2 m, and the average two dimensional leader speeds ranged from 4.4 to 4.6 x 10 5 m/s. For the five secondary branches, there were a total of 18 individual leader steps analyzed. The average interstep intervals for the five secondary branches r anged from 12.2 s to 40.0 s, the average step lengths ranged from 5.4 m to 7.1 m, and the average two dimensional leader speeds ranged from 2.7 to 6.2 x 10 5 m/s. Average two dimensional leader speeds could only be calculated for two of the five secondar y branches (S 1 and S 4). Leader speeds were

PAGE 263

263 calculated by dividing the distance between the ending points in space of two consecutive leader steps in a respective branch by the number of frame integration times of 3.33 s apiece between the steps. The s peeds obtained between each set of two consecutive steps were summed and divided by the total number of steps in the respective branch. It appears that there is not much difference between the step lengths and average leader speeds in the primary and seco ndary channels, but that the interstep times in the secondary channels are longer. Before presenting video frames of the stepping mechanism, an artist sketch of the author's view of the leader step formation process is provided in Figure 5 4. This sequ ential process, decomposed into five stages, was derived from analyzing the high speed video images with frame integrations ending randomly at different times within the leader step formation process. Here, a 50 m segment of leader channel is drawn. In t he first stage, a negative stepped leader channel is shown having decayed in luminosity following a prior leader step. In the second stage, the formation of a space stem/leader is shown with vertical extent of several meters located several meters below t he tip of the leader channel above. It is likely that the space stem/leader actually forms within the negative streamer zone of the leader channel above as illustrated in Biagi et al. [2010]. However, due to the increased distance between the channel and camera, estimated at roughly twice that of Biagi et al. [2010], and the ambient atmospheric conditions, there was insufficient spatial and luminosity resolution in the present high speed video data to well spatially resolve the streamer zones, though it i s inferred that they exist in each case by the observed low level "glow" surrounding both the leader tip and the space stem/leader. In the third stage, the space stem/leader begins to reconnect to the leader channel above with low level luminosity, and in the fourth stage, the space stem/leader fully connects to the above leader channel, brightly illuminating the space stem/leader and the connection region.

PAGE 264

264 Finally, in the fifth stage, after the new bright segment of stepped leader channel has been formed a luminosity wave propagates back up the existing channel for some tens of meters, illuminating the channel above with intensity comparable to that of the leader tip. It is likely that all five stages shown in Figure 5 4 occur within about a microsecond The first examples of space stems/leaders shown in association with natural stepped leader steps are now presented. In two cases, shown in Figure 5 5 and Figure 5 6, five consecutive 3.3 s frames (about 16.7 s total) are plotted which have been in verted and contrast enhanced to show better the more faint luminous characteristics of the step formation process. Time progresses from left to right and the frames span 70 m in altitude and 32 m horizontally. Note that Figure 5 5 and Figure 5 6 show con secutive leader steps. The final frame shown in Figure 5 5 is the first frame shown in Figure 5 6. In both examples the top of the frames are located about 16 m below the split of primary branch P 1 into primary branches P 2 and P 3 (as shown in Figure 5 3). The first example of the leader step formation process is shown in Figure 5 5. A new leader step is forming in Frame A in branch P 2, brightly illuminating the leader tip. An upward propagating luminosity wave in branch P 3 moves out of the frame from a new leader step in the previous frame (not shown). In Frame B, the upward propagating luminosity wave from the new leader step in Frame A moves up branch P 2 and out of the frame and the luminosity in branch P 2 begins to decay. In Frame C, two d istinct space stems/leaders are imaged, one below branch P 3 and one to the right of branch P 2, both annotated by arrows in the figure. Interestingly, the space stem/leader below branch P 3 is already partially connected to the leader channel above with low level luminosity, likely a result of the frame integration ending during the third stage of the leader step formation process outlined in Figure 5 4. The space stem/leader shown to the

PAGE 265

265 right of branch P 2 appears to be completely separated from the ex isting leader channel. In this case, the space stem/leader below branch P 3 is significantly brighter than the leader channel above. However, this could be a result of the existing partial connection. The space stem/leader to the right of branch P 2 is quite faint in comparison to the leader channel which is still decaying in luminosity from the new leader step two frames prior. In Frame D, both space stems/leaders fully connect to their respective leader channels, at left forming a new section of branc h P 3 and at right forming a new step in branch S 5. Also in Frame D, upward propagating luminosity waves have moved up branch P 3 and from branch S 5 into branch P 2. It is interesting to note that the upward propagating luminosity wave originating from the new step in branch S 5 appears to only move into the existing channel section above the junction point of branches S 5 and P 2 and does not appear to influence the new channel section of branch P 2 formed in Frame A. A second example of the leader s tep formation process in shown in Figure 5 6. In Frame A of Figure 5 6, three distinct leader channels have already decayed in luminosity from prior leader steps, P 3 on the left, P 2 in the middle, and S 5 on the right. In Frame B, all three leader chan nels decay further in luminosity where P 3 and P 2 are only faintly visible. In Frame C, two distinct space stems/leaders form below leader channels P 3 and P 2 and are annotated in the figure with arrows. The luminosity levels of the space stems/leaders are comparable to or somewhat brighter than the leader channels above. In Frame D, both leader channels P 3 and P 2 connect with the space stems/leaders formed in Frame C, brightly illuminating the connection regions. Also in Frame D, the propagation of upward luminosity waves over several tens of meters in both channels P 3 and P 2 is evident. In Frame E, the upward propagating luminosity

PAGE 266

266 waves move out of the frame and the luminosity of the new steps in channels P 3 and P 2 begin to decay. A total of 16 instances of space stems/leaders have been analyzed from the high speed video data. In Table 5 3, measured statistics are provided for all imaged space stems/leaders including length and separation from the leader channel above assuming a range of 1 km. The separation is measured as the distance between the top of the space stem and the lowest point in space of the prior leader step along the direction of the space stem formation. The average space stem/leader length was 3.9 m and the average separ ation from the leader channel above was 2.1 m. For all statistics presented in Table 5 3, range corrections (shown to the right of each primary value) are provided in the form "(+ A / B)", where "A" represents the range correction in measurement for a maximum range of 2 km and "B" represents the correction for a minimum range of 700 m. Figure 5 5 includes space stems/leader 12 from Table 5 3 and Figure 5 6 includes space stems/leaders 13 and 14 from Table 5 3. Though a detailed discussion of the rem aining 13 instances of photographed space stems/leaders will not be given, the processed images for the remaining events are shown in Figures 5 7 though 5 13. The number of the space stem/leader in each figure is given relative to Table 5 3. Of the 82 leader steps identified, the lengths of the upward propagating luminosity waves following the step formation were successfully measured in 28 instances. The measurements were restricted by the field of view and dynamic range of the camera and by the relat ive difference between the background luminosity preceding the formation of each leader step and the luminosity of the upward propagating luminosity wave. Measurements were obtained by subtracting the luminosity one frame prior to a new step from the fram e containing the new step

PAGE 267

267 and the subsequent three frames. In seven cases, upward propagating luminosity was measured in only one frame, for 20 cases, in two consecutive frames, and for one case, in three consecutive frames. The total lengths of the upwa rd propagating waves ranged from 14 to 85 m with an average value of 43 m. The velocities calculated from these length values ranged from about 4.0 x 10 6 to 1.3 x 10 7 m/s with an average value of about 7.5 x 10 6 m/s. The velocities of the upward propagati ng waves were calculated by dividing the total length traveled by the elapsed time of either one, two, or three frame integrations of 3.33 s apiece. The stated velocity measurements are clearly the minimum possible velocities considering the impossibilit y of determining what fraction of the 3.33 s frame integration time imaged the actual motion of the luminosity wave. 5.4 Discussion of Results Clearly, the space stem/leader plays an integral role in determining the propagation characteristics of the neg ative stepped leader. The measured statistics for space stem/leader length and separation from the previous leader channel tip given in Table 5 3 are in relatively good agreement with those obtained by Biagi et al. [2010] for a dart stepped leader precedi ng a rocket triggered lightning return stroke at a range of 430 m. Biagi et al. [2010] reported space stem/leader lengths ranging from 1 to 4 m and separations from the above leader channel ranging from 1 to 10 m ( Figure 5 2). Comparing the measured st atistics for the natural stepped leader in this study with previously obtained statistics for interstep interval and step length using continuous moving film (streak) cameras, it was found that both the interstep intervals and step lengths from this study are, in general, shorter in duration and length. The measured statistics for interstep interval and step length are in better agreement with those measured for stepped leaders using photodiode arrays, and also in better agreement with those measured optic ally and through electric field waveforms for dart stepped leaders in both natural and triggered lightning discharges [e.g.,

PAGE 268

268 Schonland, 1956; Krider, 1977; Orville and Idone, 1982, 1984, Davis, 1999]. It is likely that differences in measurement technique s are at least in part responsible for the shorter step lengths measured in comparison to past studies of negative stepped leaders. From examination of previously obtained streak photographs, the author believes that step length could have been overestima ted by measuring not only the length of the newly formed section of leader channel, but the superposition of the newly formed section of the leader channel and the luminosity wave traveling back up the existing leader channel after the step formation, the upward propagating luminosity wave often exhibiting luminosity comparable to that of the leader tip for some tens of meters up the existing channel. Finally, the measured stepped leader propagation speeds in this study, assuming a range of 1 km, are in go od agreement with those of Schonland et al. [1935] and Schonland [1956], on the upper boundary of those measured by Berger [1967], and on the lower boundary of those measured by Chen et al. [1999]. Including the maximum and minimum range corrections as li sted in Table 5 2, the values for stepped leader speed remain within or very close to the statistical ranges calculated from past studies. The good agreement between the spatial measurements of space stems/leaders compared to those measured in associati on with dart stepped leaders in triggered lightning and of the stepped leader propagation speeds in comparison with previous optical studies of stepped leaders suggest that the estimate of 1 km range is reasonable. It should also be noted that prior optic al studies of leader propagation utilized such techniques as thunder ranging, cloud base height estimation, and approximations for return stroke speed to estimate the distance between a given leader channel and an optical measurement. The inherent errors in these lightning location techniques likely contribute to the relatively wide range of values for step length and two dimensional stepped leader propagation speed given in Table 5 1.

PAGE 269

269 Wang et al. [1999 a ] used the ALPS photodiode system to measure the total lengths of the upward propagating waves following the step formation, in this case the bottom 400 m of a dart stepped leader preceding a rocket triggered lightning return stroke, to be from several tens of meters to more than 200 m and the velocitie s of the upward propagating waves to be from 1.9 x 10 7 to 1.0 x 10 8 m/s with an average value of about 6.7 x 10 7 m/s. The disparity between the measured average velocity (a minimum possible value) of about 10 7 m/s in this study and that of Wang et al. [1 999] is likely mostly due to the necessary assumption that, for the video data acquired, the observed propagation occurred for the full 3.33 s frames.

PAGE 270

270 Table 5 1. Previous optically obtained stepped leader statistics Interstep Interval (s ) Step Length (m) Leader Speed (x 10 5 m/s) Schonland et al. [1935] 10 200 40 100 3.8 Schonland [1956] 37 124 10 200 0.8 8 Berger [1967] 29 47 3 17 0.9 4.4 Orville and Idone [1982] 5.9 15 Chen et al. [1999] (Aus) 5 50 7.9 20 4 .9 11 Chen et al. [1999] (China) 18 21 8.5 4.9 5.8 Krider [1974] 17 32 Lu et al. [2008] 0.2 15.7 15

PAGE 271

271 Figure 5 1. Two frames of a dart stepped leader preceding the eighth return stroke of flash UF 08 18 on September 17, 2008. Each frame is shown in both true value and pixel inverted forms to better show the low luminosity features of the leader. A space stem with length of about 2 m is evident in the frame immediately prior to the return stroke. The space stem is sepa rated from the above leader tip by about 4 m. Adapted from Biagi et al. [2009].

PAGE 272

272 Figure 5 2. Ten consecutive frames (~42 s) of a dart stepped leader preceding the fifth return stroke of a triggered lightning discharge (flash UF 09 25 on June 29, 2009). Space stems and/or space leaders are visible in seven of the frames. Adapted from Biagi et al. [2010].

PAGE 273

273 Figure 5 3. A full frame time integrated image of the flash and identification of all branching structure. The notation "P X" refers to primary branches and the notation "S X" refers to secondary branches, where "X" is an integer. There were three primary branches and five secondary branches analyzed. Photo courtesy of the author.

PAGE 274

274 Table 5 2. Measured stepp ed leader statistics assuming a range of 1 km. Given in parentheses are range corrections for a maximum range of 2 km and a minimum range of 700 m. Branches a Number of Steps Avg. Interstep Interval (s) Avg. Step Length (m) Avg. Leader Speed (x 10 5 m/ s) P 1 28 13.7 4.8 (+ 4.8 / 1.4) 4.4 (+4.4 / 1.3) P 2 21 14.9 5.2 (+5.2 / 1.6) 4.6 (+4.6 / 1.4) P 3 15 15.1 5.1 (+5.1 / 1.5) 4.5 (+4.5 / 1.4) S 1 5 21.3 6.0 (+6.0 / 1.8) 2.7 (+2.7 / .8) S 2 6 23.9 7.1 (+7.1 / 2.1) S 3 2 40.0 6.0 (+6.0 / 1 .8) S 4 3 12.2 5.8 (+5.8 / 1.7) 6.2 (+6.2 / 1.9) S 5 2 26.7 5.4 (+5.4 / 1.6) a "P X" represents a primary branch and "S X" represents a secondary branch

PAGE 275

275 Figure 5 4. A five step sequential artist sketch of the step forma tion process inferred from the 3.3 s frame data. Time progresses from left to right and a total of 50 m of leader channel are drawn.

PAGE 276

276 Figure 5 5. First example of space stem/leaders of a natural negative stepped leader step. Five c onsecutive 3.33 s frames are shown (about 16.7 s total). The frames span 70 m in altitude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos courtesy of the author.

PAGE 277

277 Figure 5 6. Second example of space s tem/leaders of a natural negative stepped leader step. Five consecutive 3.33 s frames are shown (about 16.7 s total). The frames span 70 m in altitude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos courtes y of the author.

PAGE 278

278 Table 5 3. Measured space stem/leader statistics assuming a range of 1 km. Given in parentheses are range corrections for a maximum range of 2 km and a minimum range of 700 m. a The ending point in space of the leader step and the top of the subsequent space stem/leader occurred in the same pixel but after the leader step luminosity became undetectable. Space Stem/Leader Number Length (m) Distance from Prev ious Leader Channel Tip (m) 1 3.0 (+3.0 / 0.9) 3.2 (+3.2 / 0.9) 2 5.0 (+5.0 / 1.5) 1.0 (+1.0 / 0.3) 3 3.2 (+3.2 / 1.0) 2.2 (+2.2 / 0.7) 4 3.0 (+3.0 / 0.9) 3.0 (+3.0 / 0.9) 5 4.2 (+4.2 / 1.2) 0.0 a (+0.0 / 0.0) 6 4.2 (+4.2 / 1.2) 2.0 ( +2.0 / 0.6) 7 3.2 (+3.2 / 1.0) 3.2 (+3.2 / 0.9) 8 2.4 (+2.4 / 0.7) 3.2 (+3.2 / 0.9) 9 3.2 (+3.2 / 1.0) 3.6 (+3.6 / 1.1) 10 3.2 (+3.2 / 1.0) 2.2 (+2.2 / 0.7) 11 5.0 (+5.0 / 1.5) 2.0 (+2.0 / 0.6) 12 5.1 (+5.1 / 1.5) 1.4 (+1.4 / 0.4) 13 4.6 (+4.6 / 1.4) 1.4 (+1.4 / 0.4) 14 5.5 (+5.5 / 1.6) 0.0 a (+0.0 / 0.0) 15 4.6 ( +4.6 / 1.4) 1.4 (+1.4 / 0.4) 16 3.0 (+3.0 / 0.9) 3.0 (+3.0 / 0.9) Averages 3.9 (+3.9 / 1.2) 2.1 (+2.1 / 0.6) Biagi et al. [2010] 1 4 1 10

PAGE 279

279 Figure 5 7. Images of space stems. A) s pace stem #1, and B) space stem #2. Five consecutive 3.33 s frames are shown (about 16.7 s total). The frames span 70 m in altit ude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos courtesy of the author. A B

PAGE 280

280 Figure 5 8. Images of space stems. A) space stem #3, and B) space stem #4. Five consecutive 3.33 s frames are shown (abo ut 16.7 s total). The frames span 70 m in altitude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos courtesy of the author. A B

PAGE 281

281 Figure 5 9. Images of space stems. A) space stem #5, and B) space ste m #6. Five consecutive 3.33 s frames are shown (about 16.7 s total). The frames span 70 m in altitude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos courtesy of the author. A B

PAGE 282

282 Figure 5 10. Images of space stems. A) space stem #7, and B) space stem #8. Five consecutive 3.33 s frames are shown (about 16.7 s total). The frames span 70 m in altitude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos cour tesy of the author. A B

PAGE 283

283 Figure 5 11. Images of space stems. A) space stem #9, and B) space stem #10. Five consecutive 3.33 s frames are shown (about 16.7 s total). The frames span 70 m in altitude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos courtesy of the author. A B

PAGE 284

284 Figure 5 12. Images of space stems. A) space stem #11, and B) space stem #15. Five consecutive 3.33 s frames are shown (about 16.7 s total). The frames span 70 m i n altitude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos courtesy of the author. A B

PAGE 285

285 Figure 5 13. Space stem #16. Five consecutive 3.33 s frames are shown (about 16.7 s total). The frames span 70 m in altitude are 32 m horizontally, assuming a range of 1 km. Time progresses from left to right. Photos courtesy of the author.

PAGE 286

286 CHAPTER 6 "CHAOTIC" DART LEADE RS IN TRIGGERED AND NATURAL LIGHTNING This chapter presents the first published observations of "chaotic" dart leaders preceding triggered lightning return strokes, and the first close range (< 1 km) observations of "chaotic" dart leaders preceding natural lightning subsequent strokes. The data presented include electric field deriv ative (dE/dt) and energetic radiation waveforms in addition to high speed photographs. The chapter begins with an in depth literature review of "chaotic" leader processes associated with natural lightning, followed by a brief description of the experiment al setup. The "chaotic" dart leader process is then compared to the more well understood dart and dart stepped leader processes that occur in association with triggered and natural lightning return strokes, and to previous accounts of "chaotic" dart leade rs (or similar phenomena) preceding natural lightning subsequent return strokes. Detailed results and discussions are presented for two of the four "chaotic" dart leaders associated with triggered lightning return strokes recorded during summer 2010. Fi nally, waveforms are presented and discussed for two "chaotic" dart leaders that occurred in the same natural lightning flash during summer 2011. Some of the results for triggered lighting "chaotic" dart leaders have been published in Hill et al. [2012]. 6.1 Literature Review The electromagnetic emissions of descending negatively charged dart leaders and dart stepped leaders preceding subsequent return strokes in natural and triggered lightning have been well studied [e.g., Schonland et al., 1935; Krider et al., 1977; Orville and Idone, 1982; Guo and Krider, 1985; Jordan et al., 1992; Wang et al., 1999]. References in the literature to the "chaotic" leader process are relatively few [e.g., Weidman, 1982; Bailey et al., 1988; Willett et al., 1990; Rakov an d Uman, 1990; Davis, 1999; Gomes et al., 2004; Makela et al., 2007; Lan et al., 2011] Further, the term "chaotic" leader is not well defined but has, in general, has been used to refer to

PAGE 287

287 a relatively continuous sequence of irregular electric field puls es occurring within the roughly 2 ms preceding a subsequent return stroke in natural lightning. These pulses are different from the more well documented dart stepped leader pulses. In his Ph.D dissertation, Weidman [1982, p. 77] was the first to use the term "chaotic" to differentiate the newly identified leader process from the dart or dart stepped leader process. Weidman [1982] in his Figure 3.17b and 3.17c shows two leader electric field derivative (dE/dt) waveforms demonstrating an irregular sequence of pulses in the final 22 s prior to the return stroke that are clearly different from the dart stepped leader record in his Figure 3.17a. The causative discharges occurred mostly over salt water at distances of 10 to 60 km from the measuring station lo cated on Anna Maria Island near the southern entrance to Tampa Bay, Florida. Bailey et al. [1988] reported on four events termed "chaotic subsequent strokes" with similar rapidly varying irregular pulses immediately prior to the onset of the return stroke in the dE/dt waveforms. The discharges occurred over salt water at distances from 15 to 45 km from the measuring station located at the Kennedy Space Center in Florida. They also found that "chaotic subsequent strokes" produced larger positive and negat ive peak dE/dt (return stroke) signatures than those preceded by normal dart leaders or dart stepped leaders. Willett et al. [1990] recorded 15 events classified as "chaotic" leaders from three storms at ranges less than 35 km using the equipment describe d in Bailey et al. [1988]. In their Figure 1, they give a "chaotic" leader dE/dt waveform showing an irregular pulse sequence similar to those reported by Weidman [1982] and by Bailey et al. [1988]. The pulse sequence extends backwards in time to at leas t 60 s before the return stroke. Rakov and Uman [1990] analyzed 15 subsequent return strokes with preceding "chaotic" leader signatures in 76 close, over land, negative cloud to ground lightning discharges containing a total of 270 subsequent return stro kes. The data were recorded near Tampa, Florida. Similar to the results of Bailey et

PAGE 288

288 al. [1988], they reported that the "chaotic" leaders are more likely to precede the larger peak electric field return strokes in a given flash. They suggest that the ir regular pulse signature of the "chaotic" leader could be associated with an unknown stepping process of a dart leader that had previously moved continuously downward along the pre conditioned channel, and also propose that the "chaotic" leader process migh t be related to the electromagnetic interaction between the simultaneously propagating downward moving, negatively charged leader channel and the upward moving, positively charged connecting leader channel. Perhaps the most thorough analysis of the "chao tic" leader process is found in the Ph.D dissertation of Davis [1999]. A total of 12 "chaotic" leaders were recorded in 16 negative cloud to ground lightning discharges at the Kennedy Space Center, Florida. Time of arrival (TOA) techniques were employed to locate dE/dt sources in three dimensions from a five station network of flat plate antennas covering an area of 15 km x 15 km (this network was previously discussed in Chapter 4). The average leader speeds for two "chaotic" leader events were calculate d to be 3.2 x 10 7 m/s and 5.0 x 10 7 m/s, speeds higher than those typically associated with dart stepped leaders and in better agreement with the speeds of dart leaders. For two "chaotic" leader events, the median interpulse intervals were determined to b e 0.9 s and 1.45 s respectively for a threshold 4.4% above the background noise level. A total of eight "chaotic" leader waveforms are shown, with durations varying from a few tens of microseconds to about 180 s. Of particular interest is the bottom d E/dt trace in Davis [1999, p.111]. What appears as a typical dart stepped leader for the first 870 s of the record (accounting for a 30 s dead time between each recording) transitions into a "chaotic" leader in at least the final 180 s before the retur n stroke. Gomes et al. [2004] recorded Chaotic Pulse Trains (CPT) in 95 of the 169 lightning discharges they recorded in Sweden, Denmark, and Sri Lanka. The authors chose to use the term "Chaotic Pulse Train" as

PAGE 289

289 opposed to "chaotic" leader or "chaotic su bsequent stroke" because in several cases they observed irregular sequences of pulse activity independent of subsequent return strokes. The causative storms were 10 to 150 km distant. It was reported that "Chaotic Pulse Trains" immediately preceding subs equent return strokes had pulse widths generally from 2 to 4 s (measured as short as 400 ns), interpulse intervals from 2 to 20 s, and mode value duration of 400 to 500 s. Makela et al. [2007] also performed an analysis of "chaotic" leaders in Sri Lan ka using both a wideband (upper frequency limit of about 12 MHz) flat plate electric field antenna and a vertical antenna tuned to a resonant frequency of 10.44 MHz with bandwidth of 2.02 MHz. They analyzed 34 lightning flashes containing 74 subsequent re turn strokes, over 30% of which appeared to be preceded by "chaotic" leaders. The discharges were estimated to occur at distances from 1 to 7 km from the measuring station. They reported that the "chaotic" component during the leader phase had average du ration of about 300 s but rarely exceeded 500 s. Using the narrowband HF data, they also found that the HF intensity of individual peaks of the "chaotic" component present during the leader phase preceding a subsequent return stroke can approach the int ensity level recorded from the actual return stroke, in stark contrast to the typically weak emission from both dart and dart stepped leaders. Lan et al. [2011] analyzed 210 return strokes in 74 negative cloud to ground flashes in China. They reported th at 48% of the subsequent return strokes in the study were preceded by a chaotic component (referred to as CPT) with average duration of 472 s. The widths of individual CPT pulses were found to range from 0.5 to 8 s. From two station interferometer meas urements, the propagation velocities of the downward leaders associated with CPT were found to be about 2.0 x 10 7 m/s. Here, two classes of previously un published electric field derivative and energetic radiation waveforms are presented, 1) the first chaotic" dart leader waveforms associated with

PAGE 290

290 triggered lightning return strokes, 2) the first close range (< 1 km) "chaotic" dart leader waveforms associated with natural lightning subsequent strokes. Recall that triggered lightning return strokes are s imilar to natural lightning subsequent strokes. In the previous studies of "chaotic" leaders preceding natural subsequent return strokes, no detailed analysis has been performed of the characteristics of the sub microsecond irregular field variations. Of four observed triggered lightning "chaotic" dart leaders, the sub microsecond field variations of two "chaotic" dart leaders within 6 s of the triggered lightning return stroke and within about 200 m of ground are examined in detail, establishing new ter minology to describe the slower background field changes of the order of hundreds of nanoseconds and the faster field variations of the order of tens of nanoseconds superimposed on the slower background field changes. A similar analysis is performed for t wo "chaotic" dart leaders that preceded the third and fourth return strokes of a natural lightning flash. Measured statistics are provided for the pulse width of the slower background field changes, and pulse width and amplitude for the faster superimpose d field variations. Finally, the term "chaotic" dart leader is adopted instead of previously used terms because the overall characteristics of the leader process closely resemble those of a typical dart leader with superimposed irregular sequences of puls es in the electric field derivative signature observed in at least the bottom 200 m of the leader channel. Some of the new data presented here are also found in Hill et al. [2012]. Measurements from ground based energetic radiation detectors have shown that roughly 80% of dart and dart stepped leaders preceding rocket triggered lightning return strokes emit detectable energetic radiation (primarily x rays and gamma rays) when they are within several hundred meters of ground, that is, within a few hundre d microseconds of the subsequent return stroke initiation [e.g., Dwyer et al. 2003, 2004, 2011; Saleh et al., 2009]. Howard et al. [2008]

PAGE 291

291 used TOA technique to show that the high frequency electric field sources emitted from a descending negatively charg ed dart stepped leader preceding a rocket triggered lightning return stroke were both in close spatial proximity (less than 50 m) and temporal relation (0.1 to 1.3 s preceding) to the corresponding x ray sources. In this chapter, the first measurements o f energetic radiation specifically associated with "chaotic" dart leader processes are presented and compared to the emissions radiated by typical dart and dart stepped leaders preceding rocket triggered lightning return strokes. Finally, high speed vid eo images are provided of two triggered lightning "chaotic" dart leaders within 140 m of the ground and compared to the optical characteristics of those recorded for dart and dart stepped leaders. 6.2 Experimental Description and Data Processing Techniqu es The electric field, energetic radiation (x rays and gamma rays), and high speed photographic measurements described in this chapter were obtained during the summers of 2010 and 2011 at the ICLRT. As discussed in Section 2.11, lightning was artificiall y initiated (triggered) by launching a 1 m fiberglass rocket carrying a 700 m spool of 32 AWG kevlar coated copper wire toward a region of high negative charge concentration produced by an overhead or nearby thunderstorm. The end of the wire spool was con nected to the grounded rocket launcher. A detailed description of the triggered lightning process was given in Section 1.5. Triggered lightning return strokes discussed in this chater were triggered to the Tower Lau ncher ( Section 2.11.1) and to t he Field (Ground) Launcher ( Section 2.11.2). Refer to Figure 2 1 for the spatial locations of both launching facilities. In both launching configurations, current was measured just below the strike object, the rocket launcher in the Tower Launcher configuration and the intercepting rod in the Field Launcher configuration, with a non inductive T&M Research R 7000 10 current viewing resistor (CVR) having a bandwidth from DC to 8

PAGE 292

292 MHz. The implementation of the current measuring systems during summer 2010 was discu ssed in detail in Section 2.12.2, and shown graphically in Figures 2 21 and 2 22. Electric field derivative (dE/dt) waveforms were recorded at ten distances from the lightning channel base ranging from 37 m to 443 m with flat plate antennas of area 0.155 m 2 The 3 dB bandwidth of the antenna response is about 40 MHz (determined by the antenna capacitance and the load impedance), but the received signals are bandwidth limited to 20 MHz by the fiber optic data transmission system and the anti aliasing fil ter at the digital storage oscilloscope (DSO) input. A thorough description of the dE/dt antennas used at the ICLRT was given in Section 2.13. The ten dE/dt antennas comprise a small area (around 0.25 km 2 ) time of arrival (TOA) network, similar to that d escribed in Howard et al. [2008, 2010], capable of locating high frequency electric field sources in three dimensions with high spatial and temporal resolution when the sources are within about 500 m of the ground. The dE/dt TOA network was described in d etail in Chapter 3. Sources are located spatially and temporally by using the signal of the peak dE/dt with 4 ns accuracy, the sampling resolution of the system as input to a non linear least squares Marquardt algorithm similar to that described in Thomas et al. [2004] and Koshak et al [2004]. Actual signal arrival times are calculated by removing the artificial propagation delays introduced by BNC and fiber o ptic cable runs from each sensor to the DSO inputs. The propagation delays are measured with an accuracy greater than the sampling resolution of the system. The removal of artificial time delays, the signal arrival time selection process, and the TOA sol ution algorithm were described in detail in Sections 3.3, 3.5, and 3.8. E nergetic radiation measurements were co located within 10 m of each dE/dt sensor. Two lanthanum bromide (LaBr 3 ) and eight 1 m 2 plastic scintillation devices were employed. The L aBr 3 detectors were each mounted to a photomultiplier tube (PMT). Two PMTs were

PAGE 293

293 mated to opposite sides of each plastic scintillator. The LaBr 3 detectors were calibrated using a 662 keV Cs 137 gamma ray source. A plot of the LaBr 3 detector single photon response to the calibration source in given in Figure 6 1. The typical amplitude response of the sensor to the calibration source is 152 mV and the typical full pulse width is 184 ns. A thorough discussion of the design, implementation, and operation of the LaBr 3 and plastic energetic radiation detectors at the ICLRT was given in Section 2.14. As noted in Section 3.2, the spatial positions of all dE/dt sensors and energetic radiation detectors were measured to an accuracy of 1 cm to the center of the se nsor using a Differential Global Positioning System (DGPS) for the lateral coordinate measurements and a closed level loop for the altitude coordinate measurements. State plane coordinate measurements were based on the Florida North NAD83 data. The coord inates were converted to a local coordinate system with origin at the far sou thwest corner of the ICLRT ( Section 3.2). In this chapter, all altitude measurements are referenced to the local coordinate system origin. The spatial locations of all dE/dt and energetic radiation measurements included in the TOA network were given in Table 3 1 and are shown graphically in Figure 2 36 (2010 configuration). Video images of the triggered lightning discharges were acquired from a distance of 430 m (Tower Launcher) using a Photr on SA1.1 high speed camera ( Section 2.9) operated at a frame rate of 300 kfps. A 20 mm Nikon lens was mounted to the camera and was set to an aperture of F/4. This configuration provided a spatial resolution of 0.43 m/pixel and a total vert ical field of view of 138 m. The camera was configured to record a pixel region measuring 320 x 32 (vertical x horizontal) with 12 bit grayscale amplitude resolution. During summer 2010, with the exception of one LaBr 3 detector, all dE/dt and energetic radiation measurements were transmitted to the Launch Control trailer via Opticomm MMV

PAGE 294

294 120C FM fiber optic links with nomi nal bandwidth of DC 30 MHz ( Section 2.3). The output from the LaBr 3 detector at Station 25 was transmitted directly to the DSO input over approximately 20 m of double shield coaxial cable to achieve higher bandwidth and signal to noise ratio. The coaxial cable was enclosed in braided wire for additional electromagnetic shielding. All signals were digitized at a sampling rate of 250 M S/s by LeCroy Waverunner 44 Xi DSOs with 8 bit amplitude resolution ( Section 2.4.3) The Waverunner 44 Xi DSOs were operated in segmented memory mode and were capable of recording 10 segments, each of 5 ms length. The segments for all DSOs were configur ed to r ecord with 50% pre trigger ( Section 2.4.3). As noted in Section 2.7, the data acquisition system and high speed camera were triggered to begin recording when the current measured at the lightning channel base exceeded a threshold of about 6 kA. During summer 2011, the outputs of the 10 dE/dt sensors were recorded in parallel with the ICLRT DSO network by the HBM Genesis digitization system described in Section 2.5. The output of each dE/dt sensor was digitized in the field at 100 MS/s and the di gital data were transmitted to the HBM GEN16t transient recorder in the Office Trailer over single mode fiber. Waveform data surrounding each return stroke were recorded in 20 ms segments with 50% pre trigger. 6.3 Background and Overall Characteristics of "Chaotic" Dart Leaders To describe the "chaotic" dart leader process appropriately, it was necessary to develop criteria to differentiate "chaotic" dart leaders from the more common dart leaders and dart stepped leaders that also precede triggered ligh tning return strokes. In Figure 6 2, dE/dt and associated energetic radiation waveforms corresponding to the three classifications of leader types are shown on a 14 s time scale. The vertical scales of all three dE/dt measurements have been truncated to better show the lower amplitude field changes occurring during the leader

PAGE 295

295 phases. The peak dE/dt measured in association with each subsequent return stroke is given for reference. The three dE/dt waveforms were measured 177 m from the triggered lightnin g channel base (Station 7) and the three corresponding energetic radiation waveforms were recorded with a LaBr 3 detector 45 m from the lightning channel base (Station 25). A typical dart leader dE/dt waveform is shown at the top Figure 6 2 of Panel A. T he dart leader waveform is characterized by subtle, relative slow (of the order of several hundred nanoseconds) variations in the electric field within the roughly 5 s preceding the subsequent return stroke. The energetic radiation measured in associatio n with the same dart leader is shown at the bottom of Panel A. A short burst (about 2 s in duration) of relatively weak energetic radiation was recorded immediately prior to the onset of the return stroke. There were no additional deviations from the sy stem noise with the exception of several weak signals detected about 4 s prior to the return stroke. At the top of Panel B in Figure 6 2, a dE/dt waveform is shown of a typical dart stepped leader preceding a triggered lightning return stroke. The dart stepped leader is characterized by pronounced pulses corresponding to individual leader steps occurring at intervals of 1 to 2 s. The energetic radiation waveform measured from the same dart stepped leader is shown at the bottom of Panel B. Here, discre te bursts of energetic radiation were measured in association with most leader step pulses presented at the top of Panel B. Note that the pulses of energetic radiation are not exactly time aligned with the above dE/dt leader step pulses because the dE/dt measurement co located with the LaBr 3 sensor saturated during this event. The output of a different antenna is plotted, which introduces propagation path length differences. Finally, at the top of Panel C in Figure 6 2, a dE/dt waveform is shown from a chaotic" dart leader. The "chaotic" dart leader demonstrates the same subtle electric field variations within the 5 s preceding the return stroke as does the typical dart leader shown at the

PAGE 296

296 top of Panel A. In addition, however, superimposed on the slow er electric field variations is a relatively continuous series of very narrow (of the order of tens of nanoseconds) pulses, frequently with amplitudes comparable to the pulses recorded in association with dart stepped leader dE/dt steps shown at the top of Panel B. At the bottom of Panel C, the energetic radiation measured from the "chaotic" dart leader is shown. A relatively continuous burst of energetic radiation was recorded beginning at about t = 8.0 s and continuing to the return stroke. The "pile up" characteristic shown is a result of energetic photons arriving at the detector at shorter intervals than the single photon response pulse width (shown in Figure 6 1 to be about 184 ns) and is not necessarily indicative of the detection of more energet ic photons. From the dE/dt and energetic radiation waveforms presented in Figure 6 2, it is clear that the electromagnetic emission of the "chaotic" dart leader is distinctly different than that of typical dart and dart stepped leaders. "Chaotic" dart leaders exhibit large, high frequency irregular variations in the electric field immediately prior to the return stroke that are not observed from typical dart leaders. The characteristics of these pulses are also generally dissimilar from dart stepp ed leader pulses and occur nearly continuously as opposed to discretely. In some cases, the distinction between "chaotic" dart leaders and dart leaders is less obvious. Some dart leaders exhibit very pronounced slower background field changes within 10 s of the return stroke similar to "chaotic" dart leaders, but do not exhibit a significant series of superimposed high frequency pulses. As suggested by Bailey and Willett [1989] and Lan et al. [2011], there may be a continuum of characteristics between t he observed electric field emissions of "chaotic" dart leaders and ordinary dart leaders, in part influenced by the system noise level, system sensitivity and frequency response, and measurement location. A more clearly defined metric than the characteris tics of the dE/dt used previously for differentiating between the

PAGE 297

297 "chaotic" dart leader and dart leader processes is apparently the intense emission of energetic radiation by the "choatic" dart leader. As shown in Figure 6 2, the energetic radiation emitt ed from the "chaotic" dart leader is more continuous and longer in duration than the typical emission from dart leaders and is also dissimilar from the discrete bursts of energetic radiation emitted in association with dart stepped leader steps. 6.4 Data and Analysis A total of four "chaotic" dart leaders were observed between June and August of 2010 in association with triggered lightning return strokes. General information for each "chaotic" dart leader is shown in Table 6 1 (these events are also tab ulated in Table 4 3). Shot designations are given in the form "UF 10 XX" where "XX" denotes the shot number of the calendar year. The first event recorded was triggered to the Field (Ground) Launcher while the remaining three events were triggered to the Tower Launcher. The first three "chaotic" dart leaders preceded the first return stroke following the init ial stage process ( Section 1.5). The final event preceded the third return stroke in a flash that had nine return strokes. The peak return stroke currents measured at the channel base for these events ranged from 17.1 to 43.1 kA, all larger than typical triggered lightning return stroke peak currents of 10 to 15 kA [e.g., Rakov and Uman, 2003]. The dE/dt and energetic radiation emissions of two "c haotic" dart leaders that occurred in triggered lightning flashes UF 10 13 and UF 10 24 are analyzed in detail in Sections 6.4.1 and 6.4.2, respectively. In Section 6.4.3, high speed video images of two "chaotic" dart leaders triggered to the Tower Launch er are presented and analyzed. A discussion of the observations and results presented for "chaotic" dart leaders associated with triggered lightning return strokes is given in Section 6.4.4. In Section 6.4.5, dE/dt and energetic radiation waveforms are p resented for two "chaotic" dart leaders that preceded the third and fourth return strokes of natural flash MSE 11 01 on July 7, 2011. Finally, in Section 6.4.6, the observations and results

PAGE 298

298 presented for the triggered lighting "chaotic" dart leaders are c ompared with those of the two natural lightning "chaotic" dart leaders. 6.4.1 Triggered Lightning Flash UF 10 13 Flash UF 10 13 exhibited one return stroke and was triggered to the Field (Ground) Launcher at 20:09:37.357362 (UT) on June 21, 2010. The m easured peak return stroke current was 43.1 kA. An 18 s plot of the dE/dt waveform measured at Station 17, located 183 m from the lightning channel base, in shown in Figure 6 3 A The pronounced "chaotic" emission is recorded beginning at about time t = 7.0 s in the plot, about 6.0 s prior to the onset of the return stroke. There were a total of seven slowly varying field changes that are termed "bursts". The seven bursts ranged in width from 144 ns to 384 ns with an average of 261 ns. Burst width m easurement was to some extent subjective. Superimposed on each burst was a series of narrow pulses. Shown in Figure 6 3 B is a 2 s window of the waveform in Figure 6 3A bounded by the vertical dotted lines. The amplitude scale in Figure 6 3B has been re duced to better show the fine structure of the waveform. There are three bursts annotated in Figure 6 3 B by vertical dotted lines, the third, fourth, and fifth bursts measured during the event. Burst 3 had total width, as shown by the vertical lines, of 292 ns with a total of eight superimposed pulses, Burst 4 had total width of 264 ns with seven superimposed pulses, and Burst 5 had total width of 240 ns with six superimposed pulses. For each of the three bursts, there are four individual pulses annotate d that are numbered sequentially with increasing time. Shown in Figure 6 3C is a second dE/dt waveform measured 143 m from the lightning channel base at Station 25, corresponding to the time region shown in Figure 6 3 B. The purpose of this plot is to sho w the correlation of the burst and pulse waveform features across multiple channels. A similar aligning process is used for all ten dE/dt channels when TOA calculations are performed (Section 3.5).

PAGE 299

299 There were a total of 73 pulses analyzed for shape and amplitude. Measurement criteria for burst width, pulse width, and pulse amplitude are illustrated in Figure 6 4. Pulses ranged in full width from 12 to 68 ns and in amplitude above the burst level from 0.5 to 3.6 kV/m/s (at a distance of 183 m). The average full pulse width was 37 ns with standard deviation of 11 ns and the average pulse amplitude was 1.6 kV/m/s with standard deviation of 0.5 kV/m/s. The full width of the pulses were measured as opposed to the full width at half maximum because the sampling frequency was typically insufficient to establish an accurate half maximum amplitude. Pulse amplitude was measured from the pulse peak to the lesser amplitude of the two pulse endpoints. It is estimated that both the true pulse width and amplit ude could differ from the somewhat subjectively measured values, width by 30 to 50% and amplitude by as much as a factor of two. In addition, it is possible the pulses are in fact narrower than observed, but are bandwidth limited by the antenna and fiber optic link. The three technique described in Section 3.8 are given in Table 6 2. Also prov ided in Table 6 2 are the given only for those cases in which successive pulses in the record were located. Considering the typically straight and vertical geome try of a triggered lightning channel at low altitude, it is not surprising that the lateral (x, y) source location estimates for the located pulses in all three bursts fall within a plane of dimensions roughly 10 m square. The estimated altitude source lo cations, however, tend to vary fairly significantly among the pulses within a given burst. For Burst 3, the estimated altitude source locations for Pulses 1 3 decrease linearly. In Burst 4, the estimated altitude source locations for each locatable pulse appear to "bounce" up and down for

PAGE 300

300 each subsequent pulse, differing by as much as 33 m between adjacent pulses. Interestingly, the calculated velocities between Pulse 1 and Pulse 2 and between Pulse 2 and Pulse 3 both exceed the speed of light. The most likely explanation for this fact is that high frequency emission is being radiated from multiple (perhaps many) points within a few tens of meters of the leader tip nearly simultaneously. The four consecutive pulses located within Burst 5 also have altit ude source location estimates that bounce up and down for each subsequent pulse, varying by up to 15 m between adjacent pulses. Once again, the calculated velocities between all three sets of subsequent pulses approach or exceed the speed of light. The re were a total of 73 pulses located within the 6 s prior to the return stroke, ranging from a maximum altitude of about 211 m to a minimum altitude of about 32 m (or from 207 m to 28 m above local ground level at the location of launcher). The top of th e intercepting rod at the Field (Ground) Launcher was located at an altitude of about 18 m with respect to the local coordinate system origin, or about 14 m above local ground level. In Figure 6 5 three dimensional source locations for all located pulses are plotted. The data points are color coded in one microsecond intervals based on the emission time relative to the return stroke. Two additional data points in a small black box at the bottom of the channel correspond to the source locations for the tw o fast transition peaks associated with the onset of the return stroke (two large pulses annotated in Figure 6 3A, peaking at approximately t = 1.0 s and t = 0.5 s respectively), both at altitudes of about 37 m, and likely associated with the junction point(s) of the downward and upward connecting leaders from the launch facility [e.g., Jerauld et al., 2007; Howard et al., 2010]. The spatial locations of the ten dE/dt sensors are also plotted in Figure 6 5 for reference.

PAGE 301

301 The vertical velocity of the "chaotic" dart leader was estimated by fitting a line to a scatter plot of the altitude source locations versus emission time for the 73 pulses located prior to the return stroke. The plot and corresponding line of best fit are shown in Figure 6 6. The e stimated vertical velocity was 3.3 x 10 7 m/s with a correlation coefficient of 0.95. In Figure 6 7 A a dE/dt waveform is plotted that was measured 183 m (Station 17) from the lightning channel base. In addition, three channel base current waveforms are shown at different sensitivities in Figure 6 7B, C, and D The channel base current waveforms were obtained from three different measurements with sensitivities to record peak currents to about 150 A for the Very Low Current, about 6.7 kA for the Low Curr ent, and about 55 kA for the High Current. Cabling and fiber optic time delays have been removed from all waveforms. In addition, the propagation delay from the spatial location of the first fast transition pulse to the dE/dt sensor has also been removed from the waveform in Figure 6 7A Finally, the time delay to the current viewing resistor of the downward moving return stroke current wave initiated at the spatial location of the fast transition pulse has been removed from the all current waveforms, as suming a current propagation velocity of 1.1 x 10 8 m/s. In Figure 6 7 B the channel base current begins to increase from the system noise level (current is assumed to be zero at this point) at about t = 7.2 s. This point is annotated by a dotted vertic al line. The current increases relatively slowly over the next 4 s to a level of about 28 A, after which the current increases more rapidly over the next 1.4 s. At t = 1.8 s, the onset of the return stroke occurs in the dE/dt record plotted above in Figure 6 7 A. The relatively slow and smooth current rise over the time span of about 5.4 s likely indicates the initiation of an upward positive leader [e.g., Lalande et al., 1998] The current at this time is annotated with a dotted vertical line and i s at a level of about 1.4 kA, apparently the peak value of the upward leader current. With the

PAGE 302

302 assumption that the two fast transition pulses, which were located at an altitude of about 37 m, originate with the connections of the upward and downward leade rs, the upward positive leader would have needed to propagate at an average velocity of about 3.7 x 10 6 m/s to traverse the 20 m distance between the top of the intercepting rod and the spatial location of the fast transition pulses. The "chaotic" dart leader of UF 10 13 was an intense emitter of energetic radiation. In Figure 6 8, an 18 s record is plotted of the energetic radiation measured by a LaBr 3 detector located 154 m (Station 25) from the lightning channel base. A relatively continuous burst of energetic radiation pulses begins at about t = 12.5 s in the plot, about 11 s prior to the return stroke. For the next 5 s time period, ending at about t = 7.5 s, the emission is often recorded with enough time between incident photons to allow t he detector output to decay back near the zero level prior to the arrival of the next photon(s). Many of the pulses recorded during this time period have widths of the order of 200 ns, suggesting, from Figure 6 1, that the detector recorded single photons A fit of the detector response to the apparently single photon events reveals that most of the detected photons have energies between several hundred keV to about 2 MeV. The energies for five single photon events were calculated and are shown in Figure 6 8 beneath the vertical arrows. With the assumption that the downward leader propagated with relatively constant velocity at low altitude, the velocity estimate calculated from the dE/dt TOA locations in Figure 6 5 can be used to estimate the height of the leader at the beginning of the continuous burst of energetic radiation at t = 12.5 s. This calculation yields an altitude of 395 m, about 184 m higher in altitude than the highest located dE/dt pulse. After about time t = 7.5 s in the plot, the d etected photons arrive with interpulse intervals too short to allow the detector to decay prior to the arrival of the next photon. This pronounced "pile up" effect extends through the time of

PAGE 303

303 the return stroke, which is labeled and annotated in Figure 6 8 by a black dotted line. The time of the return stroke was determined by aligning the waveform from the co located dE/dt sensor with the LaBr 3 detector waveform, accounting for signal transit delays over coaxial and fiber optic cables. To understand why the single, higher energy photons were detected preceding the pile up of photons, the measured energies of which are undetermined, Monte Carlo simulations were performed by Dr. Joseph Dwyer of the Florida Institute of Technology [e.g., Hill et al. 2012] st arting with an initially spatially localized region of energetic electrons at several heights above ground. The results of the simulation show that for distant sources that occur a relatively long time before the return stroke, primarily the single, highe r energy photons arrive at the detectors without appreciable Compton scattering, while for closer sources occurring just before the return stroke, many photons from the assumed source arrive at the detector closely spaced in time. Interestingly, pulses of energetic radiation were detected for over a microsecond after the time of the return stroke. It is generally thought that energetic radiation is only associated with the leader processes and most records do indeed show the radiation terminating at the s tart of the return stroke. 6.4.2 Triggered Lightning Flash UF 10 24 Flash UF 10 24 with nine return strokes was triggered to the Tower Launcher on August 13, 2010 at 19:44:35.013453 (UT). The first two strokes of this flash had peak currents of 11.2 kA and 10.2 kA respectively, and both were preceded by typical dart leaders. The third stroke had a peak current of 28.3 kA and was preceded by a "chaotic" dart leader. The fourth, sixth, and eighth strokes of the event were preceded by dart stepped lea ders, and the fifth, seventh, and ninth return strokes were preceded by normal dart leaders. Of the four "chaotic" dart leaders recorded during summer 2010, this was the only one that did not precede the first return stroke.

PAGE 304

304 Like the other three events, this "chaotic" dart leader preceded the stroke with the largest peak current of the flash. In Figure 6 9 A a 12 s dE/dt record is shown of the "chaotic" dart leader measured 179 m (Station 8) from the lightning channel base. The pronounced "chaotic" emission is recorded beginning at about t = 6.0 s and continues through the time of the return stroke at about t = 0.3 s. Similar to the event discussed in Section 6.4.1, the "chaotic" emission during this event can be generally characterized as relat ively continuous series of narrow (tens of nanoseconds) pulses superimposed on slower (hundreds of nanoseconds) background field changes, which were termed "bursts". There were a total of seven bursts recorded during this event. Four bursts are annotated in Figure 6 9 B, the third, fourth, fifth, and sixth bursts measured during this event. The temporal extent of each burst is marked by vertical dotted lines. Burst 3 had total width of 252 ns with ten superimposed pulses; Burst 4 had total width of 416 n s with 13 superimposed pulses; Burst 5 had total width of 456 ns with 15 superimposed pulses, and Burst 6 had total width of 224 with seven superimposed pulses. For each burst, individual pulses are further annotated with integers increasing in time. Sho wn in Figure 6 9C C is a second dE/dt waveform measured 177 m (Station 7) from the lightning channel base corresponding to the time region shown in Figure 6 9 B. Again, the purpose of this plot is to show the correlation of the burst and pulse waveform feat ures for this event across multiple channels. There were a total of 45 individual pulses analyzed for waveshape and amplitude. These pulses ranged in width from 24 to 48 ns and in amplitude from 0.7 to 4.2 kV/m/s (at a distance of 179 m). The average pulse width was about 32 ns with standard deviation of about 6 ns and the average pulse amplitude was about 2.0 kV/m/s with standard deviation of 0.7 kV/m/s.

PAGE 305

305 The three dimensional source locations, associated location uncertainties, emission times, and calculated velocities between subsequent source locations for these pulses are given in Table 6 3. Similar to event UF 10 13, the lateral source location estimates for all of the located pulses during the four bursts annotated in Figure 6 9 A occur with in a spatial region of about 10 m square. In Burst 3, three of the final four pulses of the burst were located. The altitude coordinates of the three source locations covered a spatial region of about 6 m and tended to bounce up and down with increasing time over a time span of about 75 ns. There were a total of 13 pulses superimposed on Burst 4, all of which were located. The first pulse of the burst occurred at an altitude of 110 m, followed over the next 191 ns by seven subsequent pulses all with alt itude coordinates varying from 88 to 96 m. There was no apparent spatial pattern to the source locations of these seven pulses. The ninth pulse of the burst was located at an altitude of 76 m followed by the tenth pulse at 103 m. Interestingly, the emis sion time of the tenth pulse was calculated to be a fraction of a nanosecond prior to the ninth pulse, but occurred about 27 m higher in altitude. The final three pulses of the burst fluctuated in altitude from about 75 to 82 m over a time span of about 6 5 ns and also tended to bounce up and down with increasing time. There were a total of eight located pulses in Burst 5. The first three located pulses were consecutive at the beginning of the burst and decreased in altitude from 89 to 78 m over a time s pan of about 82 ns. The next six pulses in the record increasing in time were not locatable. The final five pulses in the record were located. Pulses 4 through 6 decrease in altitude from 71 to 54 m over a time period of about 94 ns. Pulse 7 was emitte d only 10 ns later in time than Pulse 6, but at an altitude of almost 72 m, a distance of about 18 m above Pulse 6. The final pulse in the burst was located with similar altitude to Pulse 7. There were four pulses successfully located within Burst 6. Th e first two pronounced pulses in the burst were not locatable. Pulse 1 was

PAGE 306

306 located at an altitude of about 50 m. Pulse 2 and Pulse 3 were separated in emission time by about 45 ns, but occurred within a meter of each other in altitude. Pulse 4 was the f inal pulse in the burst and was located at an altitude of 34 m, separated in altitude from Pulse 3 by over 23 m and in emission time by 60 ns. Similar to the results presented for event UF 10 13 in Section 6.4.1, in the four bursts discussed in detail abo ve, there were six cases in which the calculated velocities between subsequent source locations exceeded the speed of light and an additional four cases where the calculated velocities between subsequent source locations approached the speed of light Ther e were a total of 45 pulses located during this "chaotic" dart leader within the four microseconds prior to the return stroke. In Figure 6 10 the three dimensional source locations are plotted for all 45 located pulses. The located pulses ranged in altit ude from a maximum of about 185 m to a minimum of about 34 m referenced to the local coordinate system origin (or from 181 m to about 31 above local ground level at the location of the launcher). The top of the rocket tubes on the Tower Launcher are at an altitude of about 20.6 m referenced to the origin of the local coordinate system, or about 14 m above local ground level, so the lowest source occurred about 13.4 m above the top of the rocket tubes. The data points are color coded based on the emission time relative to the return stroke in 500 ns intervals. One additional data point is annotated by a small black box near the bottom of the channel corresponding to the source location of the fast transition peak (the large pulse annotated on Figure 6 9A peaking at approximately t = 0.2 s), at an altitude of about 46 m. The vertical velocity of this "chaotic" dart leader was estimated by fitting a line to a scatter plot of the altitude source locations versus emission time for the 45 pulses located pri or to

PAGE 307

307 the return stroke, as shown in Figure 6 11. The estimated vertical velocity was 4.3 x 10 7 m/s with a correlation coefficient of 0.94. In Figure 6 12 A a dE/dt waveform is plotted that was measured 179 m (Station 8) from the lightning channel base in addition to three channel base current waveforms in Figure 6 12 B, C, and D. The channel base current waveforms were obtained with three different sensitivities, to currents of about 290 A for the Very Low Current, about 6.3 kA for the Low Current, and about 51 kA for the High Current. Cable and free space propagation delays have been removed. In Figure 6 12B, the channel base current increases from the system noise level at about t = 6.1 s in the plot, annotated by the dotted vertical line. Similar to the case presented for event UF 10 13, the current increases slowly over the next 4 s to a level of about 38 A, followed by a more rapid increase over the next approximately 1.2 s to the time of the return stroke onset, again marked by a dotted verti cal line. The relatively slow and smooth increase in channel base current during the 5.16 s prior to the return stroke likely indicates the initiation of an upward positive leader, a hypothesis supported by the correlated high speed video data. The curr ent at the time of the return stroke onset is about 760 A, apparently the peak value of the upward leader current. An upward leader with length of 11.5 m was imaged in a single 3.33 s frame integration immediately prior to the return stroke. The downwar d leader was also imaged in the same frame with the bottom of the streamer zone located about 28 m above the upward leader tip. The high speed video data for this event will be shown and discussed in detail in Section 6.4.3. The single fast transition p ulse for this event was located at an altitude of about 46 m (with respect to the local coordinate origin) and slightly south of the launcher, a total distance of about 30 m from the top of the rocket tubes on the Tower Launcher. Given a duration of 5.16 s, the upward positive leader would have needed to propagate at an average velocity of about 5.8 x

PAGE 308

308 10 6 m/s in order to traverse the distance between the rocket tubes and the spatial location of the fast transition pulse. The "chaotic" dart leader of UF 10 24 was, like all of the "chaotic" dart leaders, an intense emitter of energetic radiation. Shown in Figure 6 13 is a 12 s record of the energetic radiation measured a distance of 45 m from the lightning channel base with a LaBr 3 detector. This is th e same detector used in the analysis of event UF 10 13 discussed in Section 6.4.1. The beginning of the energetic radiation burst occurs at about t = 8.0 s in the plot. Over the next approximately 2.6 s, a number of pulses are recorded with widths ind icating the sensor was excited by single photons. For three of these apparently single photon events, a fit of the detector response to the pulses of energetic radiation provides energies ranging from 329 to 649 keV. The three energetic radiation pulses analyzed are annotated in Figure 6 13 with vertical arrows. From the leader velocity estimate calculated from the dE/dt TOA locations in Figure 6 10, the extrapolated height of the leader at the beginning of the continuous burst of energetic radiation is at an altitude of about 350 m, about 165 m higher in altitude than the highest located dE/dt pulse. After about t = 5.4 s in the plot, the interpulse intervals from photons incident on the detector are insufficient to allow the sensor to decay. A prono unced pile up effect ensues for the next 6 s. Similar to event UF 10 13, the energetic radiation appears to continue for about a microsecond after the return stroke, which is labeled and marked with a dotted vertical line in Figure 6 13. The time of the return stroke is determined using the same method discussed in Section 6.4.1. 6.4.3 High Speed Video Images of Triggered Lightning "Chaotic" Dart Leaders High speed video images were obtained for three of the four "chaotic" dart leaders recorded during summer 2010, all except event UF 10 13 on June 21, 2010. The three events were triggered to the Tower Launcher, a distance of 430 m from the camera location (Office

PAGE 309

309 Trailer). The video frame rate was 300 kfps, an exposure time of 3.33 s. In Figure 6 1 4A, three consecutive frames are plotted of the "chaotic" dart leader that preceded the first return stroke, peak current of 17.1 kA, of event UF 10 20, triggered on July 15, 2010. In Figure 6 14B, eight consecutive frames are plotted of the dart leader p receding the fourth return stroke of the same event. The fourth return stroke had peak current of 11.7 kA and occurred 122 ms after the first return stroke. The purpose of this figure is to illustrate the differences in optical characteristics between a "chaotic" dart leader and a dart leader measured during the same triggered lightning discharge. For both leaders shown, the images have been inverted, equally contrast enhanced, and the pixel gain (gamma) has been increased by a factor of two to better sh ow the lower luminosity features of the leaders. The altitude scales in Figure 6 14 are referenced to the local coordinate system origin. There are three notable differences: 1) from the images shown, the velocity of the "chaotic" dart leader is estimate d to be about 2.0 x 10 7 m/s while the dart leader velocity is estimated to be about 6.6 x 10 6 m/s, a factor of three difference, 2) the apparent width of the channel and the apparent surrounding corona region of the "chaotic" dart leader are significantly larger than that of the dart leader, and 3) the length of the streamer zone at the tip of the "chaotic" dart leader is longer than that of the dart leader. The beginning of the streamer zone are defined spatially as the pixel beneath the brightly illumina ted leader tip (which typically saturates the sensor), where the pixel intensity falls below 75% of the saturation point. Using this criterion, the streamer zones in Frames 2, 3, and 4 for the "chaotic" dart leader in Figure 6 14 A are all 8 to 9 m in leng th, and the streamer zones in Frames 2 7 for the typical dart leader in Figure 6 14 B are all 2 to 5 m in length. In Frame 4 of Figure 6 14 A, it is unclear whether the lower luminosity section of channel connecting the brightly illuminated leader tip to t he top of the rocket tubes is an elongated streamer zone, an upward leader, or a combination of

PAGE 310

310 both. The frame integration ended immediately prior to the onset of the return stroke, which is shown fully saturating the sensor in Frame 5. In Figure 6 15 A, three consecutive frames are plotted of the "chaotic" dart leader preceding the third return stroke of event UF 10 24. The peak current of the return stroke was 28.3 kA. The electric field derivative and energetic radiation emissions of this event wer e presented in Section 6.4.2. As a result of the channel tilt to the north, only one frame of the leader phase was captured (Frame 2). The images have been inverted, contrast enhanced, and the pixel gain (gamma) has been increased by a factor of two to b etter show the lower luminosity features of the leaders. The altitude scales in Figure 6 15 are referenced to the local coordinate system origin. The downward leader is imaged in Frame 2, in addition to an upward connecting leader. The tips of both lead ers and the top of the rocket tubes are annotated Figure 6 15 A with arrows. In Figure 6 15 B, the subtraction of Frame 1 from Frame 2 is shown with a compressed altitude scale to better depict the downward and upward leaders. The streamer zone of the down ward leader was about 25 m in length and the upward connecting leader measured about 11.5 m in length. By increasing the pixel gain further, so as to see the details of the luminosity, it has been determined that the faint luminosity between the bottom of the downward leader streamer zone and the upward leader tip is clearly the structure of rocket exhaust smoke illuminated by the converging leader channels, not actual leader channel luminosity. At the end of the frame integration period, the bottom of th e downward leader streamer zone and the upward leader tip were separated by about 28 m. 6.4.4 Discussion of Results for Triggered Lightning "Chaotic" Dart Leaders The "chaotic" dart leader process exhibits characteristics that differ from those of a typ ical dart leader or a typical dart stepped leader. The four "chaotic" dart leaders recorded during summer 2010 all preceded return strokes with higher than average peak current, a finding

PAGE 311

311 in agreement with the results of Bailey et al. [1988] and Rakov and Uman [1990] in which return stroke peak electric field was used as a proxy for peak current. The features of the measured dE/dt waveforms of the order of tens of nanoseconds have not been previously documented. Further distinguishing "chaotic" dart lead ers from other types of leaders, all four "chaotic" dart leaders followed periods of unusually long continuing current and unusually large charge transfer. The "chaotic" dart leader recorded during event UF 10 13 on June 21, 2010 followed the Initial Cont inuous Current (ICC) resulting from the sustained upward positive leader that originated from the ascending triggering wire connecting to the cloud charge above. In this case, the ICC had duration of 684 ms and transferred about 68 C of negative charge to ground. The channel base current dropped to a level at or near zero for 100 ms between the end of the ICC and the first return stroke. The "chaotic" dart leaders recorded during the two triggered lightning discharges on July 15, 2010 also immediately fo llowed the ICC period after each respective wire launch. The ICC for event UF 10 20 had duration of 652 ms, transferring about 87 C of negative charge to ground. The channel base current dropped to a level at or near zero for 69 ms between the end of the ICC and the first return stroke. The ICC for event UF 10 21 had duration of 402 ms, transferring 42 C of negative charge to ground. The channel base current dropped to a level at or near zero for 95 ms following the ICC. Both the duration of the ICC pe riods and charge transfers for these three events were larger than the means reported by Wang et al. [1999 c ] (GM duration of 279 ms and GM charge transfer of 27 C) and by Miki et al. [2005] (GM duration of 305 ms and GM charge transfer of 30.4 C) for trigg ered lightning discharges at the ICLRT. Interestingly, the channel base current dropped to a level at or near zero during the wire explosion (producing the ICV, see Sections 1.5 and 4.4) for all three of these events. This is not always the case [e.g. O lsen et al., 2006]. Finally, the "chaotic" dart leader recorded during event

PAGE 312

312 UF 10 24 on August 13, 2010 occurred after an inter stroke continuing current of 135 ms duration following the second return stroke of the flash. This inter stroke continuing cu rrent transferred about 15 C of negative charge to ground. The channel base current dropped to a level at or near zero for 45 ms at the end of the inter stroke continuing current and before the initiation of the third return stroke. The dE/dt signature s of the four "chaotic" dart leaders recorded during summer 2010 shared some similar characteristics. The slower, but sub microsecond, background field changes termed "bursts" in Sections 6.4.1 and 6.4.2 were evident in all four "chaotic" dart leader dE/d t records. However, the relatively continuous trains of pulses superimposed on the bursts were only clearly evident and easily locatable via TOA techniques for events UF 10 13 and UF 10 24, which were also the two events preceding the stronger peak curren t return strokes (43.1 kA and 28.3 kA respectively). The peaks of each burst and only a small number of pulses were locatable via TOA techniques for events UF 10 20 and UF 10 21, probably due to low system sensitivity, although the energetic radiation emi ssion of both events was characteristic of a "chaotic" dart leader. The fact that the estimated source locations of the pulses tend to bounce up and down in altitude over a distance of often 5 to 30 m as the leader descends, coupled with the straight line velocity between successive pulses often exceeding the speed of light, suggests that the leader is emitting pulses of high frequency electric field radiation from multiple altitudes along the channel nearly simultaneously. This view is perhaps supported by the longer streamer zones (measured up to 25 m in length) of the "chaotic" dart leaders than those of typical dart leaders and dart stepped leaders, the latter typically displaying a more fan shaped characteristic and often including such luminous featu res as space stems/leaders [e.g., Biagi et al., 2010].

PAGE 313

313 The trains of narrow dE/dt pulses recorded during the four "chaotic" dart leaders are generally dissimilar in both wave shape and interpulse interval from dart stepped leader pulses. However, the e nergetic radiation (x rays and gamma rays) recorded from all four "chaotic" dart leaders might be expected from a very rapid stepping process. For the nearly continuous emission of energetic radiation to be supported, the electric field intensity at or ne ar the tip of the "chaotic" dart leader must remain nearly continuously at a level similar to that which occurs every microsecond or so in the bottom hundreds of meters of the propagation of a dart stepped leader, where energetic radiation appears to be on ly emitted in close time correlation with the formation of a new leader step. If dE/dt sources are being emitted from multiple altitudes near the tip of the "chaotic" dart leader, it is reasonable that the source region for the energetic radiation detecte d in association with the descending leader is often some tens of meters in vertical extent at any instance in time, possibly resulting in the pronounced pile up of energetic photons within five microseconds of the subsequent return stroke that was observe d for all four "chaotic" dart leader events. For both events UF 10 13 and UF 10 24, there appear to have been upward positive leaders initiated about 5 s prior to the return stroke From correlated channel base current records and TOA locations of t he fast transition pulses, which are assumed to correspond to the connection(s) of the upward and downward leaders, the upward leaders traversed distances of about 20 m at an average velocity of 3.7 x 10 6 m/s and about 30 m at an average velocity of 5.8 x 10 6 m/s for events UF 10 13 and UF 10 24, respectively. The findings of Weidman [1982], Willett et al. [1990], Davis [1999], Gomes et al. [2004], Makela et al. [2007], and Lan et al. [2011] suggest that the "chaotic" component of descending leaders prece ding subsequent return strokes may have duration from several tens of

PAGE 314

314 microseconds up to a half a millisecond. For the four events discussed here, the "chaotic" component of the leader phase was only detected during the final 10 to 12 s prior to the ret urn stroke fast transition pulses, and was only sufficiently above the system noise level for suitable analysis for the final 6 s prior to the return stroke. The short duration of the observed "chaotic" component compared to prior studies is likely a res ult of the relatively low system sensitivity. The dE/dt measurement sensitivities are configured to resolve both leader step and return stroke field changes from close distance (within 500 m) and low altitude. The fact that significant pulses of energeti c radiation were detected up to about 13 s prior to the return stroke suggests that the leaders likely had "chaotic" components from higher altitude. Energetic radiation measurements from higher altitude sources are restricted by a combination of atmosph eric attenuation and system sensitivity. 6.4.5 Natural Lightning Flash MSE 11 01 Natural flash MSE 11 01 terminated on the southwest quadrant of the ICLRT at 19:37:34.794163 (UT) on July 7, 2011. From calculated TOA locations of the final dart stepped l eader steps prior to the second return stroke, the ground strike point was about 50 m to the north northwest of the d E/dt antenna at Station 11 ( Table 3 1). The flash had a total of four return strokes, all of which were reported by the NLDN ( Table 4 1). The second return stroke was preceded by a dart stepped leader and the third and fourth return strokes were preceded by "chaotic" dart leaders, the first known "chaotic" dart leaders recorded at the ICLRT in association with natural lightning subsequent strokes. The NLDN reported peak current estimates for the third and fourth strokes were 20.6 kA and 22.8 kA, respectively. The third and fourth strokes occurred 100 ms and 144 ms following the first return stroke. A 44 s plot of the dE/dt measured ab out 213 m (Station 9) from the lightning channel base prior to the third return stroke is shown in Figure 6 16 A The waveform was recorded by

PAGE 315

315 t he HBM digitization system ( Section 2.5). The initial rise of the return stroke dE/dt waveform occurs at t = 0. The two fast transition peaks are labeled. Note that the second fast transition peak is saturated (the peak amplitude of the second fast transition peak was about 13.2 kV/m/s). A 4 s window of the waveform plotted in Figure 6 16A, bounded by the vert ical dotted lines is expanded in Figure 6 16 B. The "chaotic" emission was resolved on the HBM dE/dt waveform measured 213 m from the lightning channel base for about 100 s prior to the third return stroke (the final 40 s of which are plotted in Figure 6 16A). Within 10 s of the third return stroke, the dE/dt record is characterized by repeated slowly varying field changes (which were termed "bursts" in Sections 6.4.1 and 6.4.2 for "chaotic" dart leaders associated with triggered lightning return stroke s) on which were superimposed higher frequency pulses. In Figure 6 16B, the temporal extent of five bursts are annotated with dotted vertical lines. These five bursts ranged in full width from 672 813 ns, and contained as many as 10 superimposed pulses ( Burst 5). The pulses have typical widths from about 40 70 ns. In Figure 6 17B, a 44 s plot of the energetic radiation measured 206 m (Station 17) from the lightning channel base prior to the third return stroke is shown (Figure 6 17A, shows the same dE/ dt waveform plotted in Figure 6 16A). The energetic radiation waveform was measured using a LaBr 3 detector. Single photons associated with the descending "chaotic" dart leader were recorded up to 45 s prior to the return stroke, and the x ray emission w as recorded nearly continuously for about 18 s prior to the return stroke. It is worth noting that the closest plastic detector to the flash termination point (T11 F at Station 11) recorded nearly continuous x ray emission for about 37 s prior to the re turn stroke. In Figure 6 17B, the energies of four single photon events have been calculated by fitting the pulses with the single photon response of the LaBr 3 detector to the Cs 137 662 keV calibration source (Figure 6 1). The calculated single photon e nergies ranged from about 736 keV to 1.76

PAGE 316

316 MeV. From Figure 6 17B, it is clear that many additional single x ray photons were recorded with typical energies from about 200 600 keV. The single photon energies for x ray pulses within about 3 s of the retur n stroke cannot be calculated accurately due to the pulse pile up characteristic discussed previously. A similar analysis was performed for the "chaotic" dart leader preceding the fourth return stroke of natural flash MSE 11 01. In Figure 6 18A, a 36 s plot is shown of the HBM dE/dt waveform measured about 213 m (Station 9) from the lightning channel base. The dE/dt waveform for the "chaotic" dart leader preceding the fourth return exhibited a "burst" structure within about 10 s of the return stroke ve ry similar to that shown in Figure 6 16A, for the "chaotic" dart leader that initiated the third return stroke. A 2.5 s window of the dE/dt waveform bounded by the dotted vertical lines in Figure 6 18A, is expanded in Figure 6 18 B. Five individual burst s are annotated in Figure 6 18 B with full widths ranging from 280 512 ns. The bursts are somewhat narrower than those observed preceding the third return stroke. Each burst exhibits superimposed pulses, though the pulses are generally both smaller in amp litude and slightly narrower (typically 30 50 ns) than those observed for the "chaotic" dart leader preceding the third return stroke. The "chaotic" emission was observed in the dE/dt waveform measured 213 m from the lightning channel base for about 56 s prior to the return stroke, about a factor of two shorter duration than the "chaotic" emission prior to the third return stroke. A 36 s record of the energetic radiation measured in association with the "chaotic" dart leader preceding the fourth stroke is shown in Figure 6 19 B The waveform was recorded using the same LaBr 3 detector located about 206 m from the lightning channel base. Single x ray photons were recorded up to about 33 s prior to the return stroke, and the measured emission became nearl y continuous about 17 s prior to the return stroke. The continuous emission was recorded

PAGE 317

317 on the plastic detector at Station 11 (50 m from the lightning channel base) for about 20 s prior to the return stroke. Single photon energies were calculated for five x ray photons using the method described previously for the "chaotic" dart leader preceding the third return stroke. The x ray photons had energies ranging from 391 keV to 1.26 MeV. Similar to the previous event, from Figure 6 19B, it is clear that many x ray photons were recorded with energies from about 200 600 keV. 6.4.6 Discussion of Results for Natural Lightning "Chaotic" Dart Leaders The two "chaotic" dart leaders analyzed here associated with consecutive subsequent return strokes in a four stroke natural lightning discharge share many dE/dt characteristics with the two "chaotic" dart leaders analyzed in Sections 6.4.1 and 6.4.2, both of which occurred preceding triggered lightning return strokes. Within 10 s of the return stroke, "bursts" with widths of the order of hundreds of nanoseconds were observed for "chaotic" dart leaders associated with both triggered and natural lightning return strokes, as were the superimposed "pulses" with widths of the order of tens of nanoseconds. The burst widths for the "chaotic" dart leader preceding the third return stroke of natural flash MSE 11 01 (672 813 ns) were wider than those recorded for triggered flash UF 10 13 (184 384 ns), triggered flash UF 10 24 (200 456 ns), and the fourth stroke of flash M SE 11 01 (280 512 ns). The pulse widths of the natural and triggered "chaotic" dart leaders were generally similar. Though TOA locations have not yet been computed for the pulses observed in either of the two natural "chaotic" dart leaders, the similarit ies in the dE/dt waveform characteristics compared to the triggered "chaotic" dart leaders suggests that the radiation mechanisms (and hence the observed tendency of successive pulses to "bounce" up and down in altitude discussed in Section 6.4.4) are like ly quite similar. The durations of the "chaotic" emission observed for the two natural "chaotic" dart leaders in flash MSE 11 01 were considerably longer than those observed for the triggered events, and in

PAGE 318

318 much better agreement with statistics reported b y Weidman [1982], Willett et al. [1990], Davis [1999], Gomes et al. [2004], Makela et al. [2007], and Lan et al. [2011] for natural "chaotic" leader events. This result is not surprising considering the system sensitivity of the HBM dE/dt network is a fac tor of four greater than the ICLRT DSO dE/dt network, and as previously stated, the degree of the "chaotic" component observed in association with descending leaders, is, to some extent, a function of the measuring system gain and its noise level. The e nergetic radiation waveforms measured in association with the two natural "chaotic" dart leaders shared manycharacteristics with those recorded for the triggered lightning events. Single x ray photons were recorded up to 45 s prior to the return stroke f or the natural "chaotic" dart leader preceding the third stroke of flash MSE 11 01, about a factor of three longer time interval than the observed x ray photons recorded in association with the "chaotic" dart leader preceding the first return stroke of tri ggered flash UF 10 13. The pulse pile up characteristic observed within about 8 s of the return strokes for ev ents UF 10 13 and UF 10 24 ( Figure 6 8 and Figure 6 13) was evident for the natural "chaotic" dart leaders, but the durations were shorter (abou t 3 s). The energies of the x ray photons measured for the natural "chaotic" dart leaders (maximum of about 1.76 MeV) were within the range measured for the triggered lightning events. A more complete measure of the energy of single photons could be obt ained for the natural "chaotic" dart leader events considering more single photons were recorded prior to the detector experiencing pulse pile up. Finally, x ray photons were recorded for 1 2 s following the initiation of the return stroke for both natur al "chaotic" dart leaders, a result in agreement with the observation for the two triggered lightning "chaotic" dart leaders of events UF 10 13 and UF 10 24.

PAGE 319

319 Figure 6 1. LaBr 3 /PMT detector response to a CS 137 662 keV source.

PAGE 320

320 Figure 6 2. Electric field derivative (dE/dt) and LaBr 3 energetic radiation records of A ) Dart Leader, B ) Dart Stepped Leader, and C ) "Chaotic" Dart Leader preceding rocket triggered lightning return strokes.

PAGE 321

321 Table 6 1. General information on the four "chaotic" d art leaders recorded between June and August 2010. Date Shot Number GPS Time (UT) Stroke Order Peak Current (kA) Launching Configuration 062110 UF 10 13 20:09:37.357362 1 43.1 Field Launcher 071510 UF 10 20 17:28:25.235689 1 17.1 Tower Launcher 071510 U F 10 21 17:35:06.232322 1 22.2 Tower Launcher 081310 UF 10 24 19:44:35.214196 3 28.3 Tower Launcher

PAGE 322

322 Figure 6 3. "Chaotic" dart leader dE/dt waveforms. A ) an 18 s dE/dt record of the "chaotic" dart leader on June 21, 2010, measure d 183 m f rom the lightning channel base. B ) a 2 s section of the waveform in Figure 6 3A with three bursts annotated along with sequentially numbered TOA located pulses in each respective burst C ) a 2 s dE/dt waveform measured 143 m from the lightning channel base corresponding to the same time window as Figure 6 3B

PAGE 323

323 Figure 6 4. Measurement criteria for burst width, pulse width, and pulse amplitude.

PAGE 324

324 Table 6 2. Calculated three dimensional TOA locations, emission times, velocities between succes sive located pulses, and associated errors for individual pulses occurring during three bursts for event UF 10 13 on June 21, 2010, derived from measurement of the peak dE/dt with 4 ns accuracy. Pulse # X (m) Y (m) Z (m) T (s) (10 8 m/s) Burst #3 1 301.3 447.3 111.0 4.057 0.25 0.63 1.35 2 301.2 447.9 96.3 3.895 1.87 0.16 0.23 0.49 3 307.0 456.4 87.5 3.823 2.82 0.01 0.02 0.06 4 308.3 447.3 88.7 3.790 1.84 2.19 5.24 Burst #4 1 299.2 454.7 95.0 3.391 0.04 0.0 9 0.20 2 298.5 447.5 61.7 3.289 7.64 0.01 0.03 0.11 3 298.5 453.6 80.5 3.263 1.53 0.01 0.01 0.02 4 302.0 449.1 69.6 3.182 0.02 0.04 0.19 Burst #5 1 302.9 450.1 65.6 3.022 15.40 0.70 0.71 2.91 2 301.6 453.0 79.4 3.013 2.95 0.02 0.02 0.09 3 300.2 451.6 63.8 2.960 4.28 0.03 0.04 0.30 4 299.7 451.2 74.2 2.936 0.04 0.07 0.11

PAGE 325

325 Figure 6 5. TOA source locations for 75 pulses located during the "chaotic" dart leader on June 21, 2010. Points are color coded base d on calculated emission time relative to the return stroke. The locations of the 10 dE/dt sensors are shown.

PAGE 326

326 Figure 6 6. Best fit estimate for vertical velocity of the "chaotic" dart leader on June 21, 2010 using TOA source locations.

PAGE 327

327 Fi gure 6 7. "Chaotic" dart leader dE/dt and channel base current waveforms. A ) a dE/dt waveform measured 183 m from the lightning channel base, B ) RS Very Low current waveform, C ) RS Low current waveform and D ) RS High current waveform The time of the i nitial current rise from zero and the time of the return stroke onset with corresponding current amplitude are labeled with dotted vertical lines. The duration of the upward positive leader is measured to be 5.4 s.

PAGE 328

328 Figure 6 8. LaBr 3 energetic radia tion record for the "chaotic" dart leader on June 21, 2010 measured a distance of 154 m from the lightning channel base. The time of the return stroke and the arrival time of the first located dE/dt pulse are marked by dotted vertical lines. Five detecte d single photon events are labeled with their respective energies.

PAGE 329

329 Figure 6 9. "Chaotic" dart leader dE/dt waveforms. A ) a 12 s dE/dt record of the "chaotic" dart leader on August 13, 2010, measured 179 m from the lightning channel base. B ) a 1.75 s section of the waveform in Figure 6 9A with four bursts annotated along with sequentially numbered TOA located pulses in each respective burst. C ) a 1.75 s dE/dt waveform measured 177 m from the lightning channel base corresponding to the same time win dow as Figure 6 9 B.

PAGE 330

330 Table 6 3. Calculated three dimensional TOA locations, emission times, velocities between successive located pulses, and associated errors for individual pulses occurring during four bursts for event UF 10 24 on August 13, 2010, der ived from measurement of the peak dE/dt with 4 ns accuracy. Pulse # X (m) Y (m) Z (m) T (s) (10 8 m/s) Burst #3 1 445.3 419.5 106.1 2.733 1.31 0.01 0.02 0.06 2 445.7 421.7 100.6 2.687 0.01 0.01 0.02 3 449.9 422.9 103.6 2.658 3.91 0.43 0.35 1.28 Burst #4 1 453.3 424.9 110.4 2.638 3.38 0.84 0.30 1.17 2 440 .9 424.6 86.6 2.559 5.17 0.02 0.02 0.08 3 445.4 417.7 96.9 2.533 0.91 0.12 0.13 0.38 4 444.1 417.6 92.7 2.486 1.30 0.10 0.11 0.32 5 446.9 419.2 93.0 2.461 1.08 0.40 0.55 1.76 6 445.4 419.2 88.9 2.421 0.28 0.18 0.31 0.92 7 444.6 419.6 88.9 2 .391 1.71 0.18 0.25 0.81 8 446.1 420.6 92.4 2.368 3.45 0.02 0.04 0.11 9 443.7 419.6 76.6 2.321 1845 0.09 0.10 0.93 10 445.7 418.4 103.8 2.322 0.47 0.86 2.34 11 442.5 418.8 75.6 2.265 2.60 0.12 0.13 1.25 12 444.5 419.3 82.5 2.237 1.43 0.05 0.07 0.23 13 444.4 416.7 77.8 2.199 2.67 0.02 0.03 0.13 Burst #5 1 443.4 419.8 89.1 2.155 0.17 0.18 1.55 2 443.7 419.6 79.0 2.104 0.42 0.64 0.69 2.94 3 443.7 418.3 78.6 2.073 0.37 0.49 1.85 4 442.3 419.3 71.1 1.865 1.06 0.47 0.28 2.55 5 443.1 416.7 66.4 1.814 2.82 1.49 1.83 8.53 6 443.1 417.7 54.3 1.771 19.0 0.03 0.03 0.28 7 444.4 417.5 71.9 1.762 0.71 0.37 0.42 1.44 8 444.5 419.2 70.1 1.727 0.28 0.31 1.20 Burst #6 1 444.4 416.5 49.8 1.602 2.80 0.06 0.10 0.44 2 443.3 416. 5 56.4 1.578 0.17 0.20 1.07 3 446.4 418.9 57.4 1.533 1.10 1.19 5.40 4 444.0 415.0 34.6 1.473 0.03 0.04 0.31

PAGE 331

331 Figure 6 10. TOA source location estimates for 45 pulses located during the "chaotic" dart leader on August 13, 2010. Points are color coded based on calculated emission time relative to the return stroke. The locations of the 10 dE/dt sensors are shown.

PAGE 332

332 Figure 6 11. Best fit estimate for vertical velocity of the "chaotic" dart leader on August 13, 2010 using TOA source loca tions.

PAGE 333

333 Figure 6 12. "Chaotic" dart leader dE/dt and channel base current waveforms. A) a dE/dt waveform measured 179 m from the lightning channel base, B) II Very Low current waveform, C) II Low current waveform, and D) II High current wavefo rm. The time of the initial current rise from zero and the time of the return stroke onset with corresponding current amplitude are labeled with dotted vertical lines. The duration of the upward positive leader is measured to be 5.16 s.

PAGE 334

334 Figure 6 13 LaBr 3 energetic radiation record for the "chaotic" dart leader on August 13, 2010 measured a distance of 45 m from the lightning channel base. The time of the return stroke and the arrival time of the first located dE/dt pulse are marked by dotted vert ical lines. Three detected single photon events are labeled with their respective energies.

PAGE 335

335 Figure 6 14. Comparison of high speed video images recorded during A) a "chaotic" dart leader and B) a typical dart leader during the same triggered lightning discharge, event UF 10 20 on July 15, 2010. Each frame corresponds to a 3.33 s exposure time. Altitudes are given with respect to the local coordinate system origin. Photos courtesy of the author.

PAGE 336

336 Figure 6 15. High speed video images of event UF 10 24 on August 13, 2010. A ) three consecutive 3.33 s frames. The downward leader enters the field of view in Frame 2. An upward connecting leader is also seen in Frame 2. The attachment process and return stroke occur during Frame 3. B ) the subtrac tion of Frame 1 from Frame 2 is shown with a compressed altitude scale. The lengths of the downward leader's streamer zone and upward leader are annotated. Altitudes are given with respect to the local coordinate system origin. Photos courtesy of the a uthor.

PAGE 337

337 Figure 6 16. Natural "chaotic" dart leader dE/dt waveforms. A ) a 44 s dE/dt waveform of the "chaotic" dart leader preceding the third return stroke of flash MSE 11 01 on July 7, 2011. B ) a 4 s window of the waveform in Panel A with five bur sts annotated.

PAGE 338

338 Figure 6 17. Natural "chaotic" dart leader dE/dt and x ray waveforms. A ) a 44 s dE/dt waveform of the "chaotic" dart leader preceding the third return stroke of flash MSE 11 01 on July 7, 2011 B ) a 44 s waveform of the associated en ergetic radiation (x rays) measured by a LaBr 3 detector. Single photon energies are annotated for four x ray pulses.

PAGE 339

339 Figure 6 18. Natural "chaotic" dart leader dE/dt waveforms. A ) a 36 s dE/dt waveform of the "chaotic" dart leader preceding the fo urth return stroke of flash MSE 11 01 on July 7, 2011 B ) a 2.5 s window of the waveform in Panel A with five bursts annotated.

PAGE 340

340 Figure 6 19. Natural "chaotic" dart leader dE/dt and x ray waveforms. A ) a 36 s dE/dt waveform of the "chaotic" dart le ader preceding the fourth return stroke of flash MSE 11 01 on July 7, 2011 B ) a 36 s waveform of the associated energetic radiation measured by a LaBr 3 detector. Single photon energies are annotated for five x ray pulses.

PAGE 341

341 CHAPTER 7 LIGHTNING MAPPING ARRAY OBSERVATIONS O F THE INITIAL STAGE OF TRIGGERED LIGHTNING DISCHARGES The geometrical and electrical characteristics of the initial stage (IS) processes of nine triggered lightning flashes that occurred during summer 2011 are analyzed in this chapter u sing a combination of data from a local Lightning Mapping Array (LMA), channel base current measurements, and when available, dual polarization radar images from the C band SMART radar. The chapter begins with a review of the triggered lightning discharg e with emphasis on the processes, that, together, compose the IS. A brief description of the experimental setup follows. Detailed analyses are presented for the initial stages of four triggered lightning discharges, 1) flash UF 11 24, 2) flash UF 11 25, 3) flash UF 11 26, and 4) flash UF 11 32. The first three analyzed flashes were triggered on August 5, 2011 and the final analyzed flash was triggered on August 18, 2011. Plots of the LMA source locations and measured geometrical and electrical statisti cs are given for all nine triggered lightning flashes including the initial height of the IS branches, the current at ground when IS branches are initiated, and the three dimensional propagation speed and overall length of the initial UPL channel and IS br anches. Some of the data analyzed in this chapter were presented in Hill et al. [2012]. The LMA network was installed prior to summer 2011 by University of Florida graduate student John Pilkey and University of Alabama Hunstville researcher Jeff Bailey. Dr. Douglas Jordan of the University of Florida provided invaluable assistance in locating LMA sites and processing the LMA source locations. The LMA network was maintained during summer 2011 by John Pilkey. 7.1 Background and Experimental Setup The p rocess of triggering lightning using the rocket and wire technique begins with the launching of a small rocket trailing a grounded triggering wire typically in the presence of negative cloud charge over the triggering site. In Florida, rockets are launche d when the quasi

PAGE 342

342 static electri c field measured at ground ( Section 2.6) surpasses a typical threshold of about 5 kV/m (atmospheric electricity sign convention electric field vector pointing upward towards the negative charge overhead). As the rocket asc ends, electrical breakdown called precursor current pulses (or just "precursors") occurs at the wire tip [e.g ., Lalande et al ., 1998; Willett et al ., 1999 ; Biagi et al. 2009, 2012]. The precursors are essentially small sparks at or near the top of the wi re that fail to evolve into an upward propagating leader channel. When the wire top reaches a typical height of 200 400 m, an upward propagating, positively charged leader (UPL) is launched from the wire top and subsequently propagates towards the negativ e cloud charge. The triggering wire typically explodes when the UPL reaches an altitude of about 2 km, often 10 ms or so after its initiation. The current measured at the triggered lightning channel base rises to a typical amplitude of about 100 A as th e UPL ascends, often dropping sharply to a level at or near zero during the triggering wire explosion, and then resumes to a background level of about 100 A that persists for typically several hundred milliseconds. This long duration current, which can ha ve super imposed current pulses with amplitudes of the order of a kilo ampere, is referred to as the initial continuous current (ICC). The precursor current pulses, the UPL, and the ICC together comprise the initial stage (IS) of the triggered lightning d ischarge. The transition between the UPL and ICC is not well defined. Often, after the cessation of the IS, dart leader/return stroke sequences follow the path of the IS channel between the negative cloud charge and ground. These dart leader/return stro ke sequences are very similar to subsequent strokes in natural lightning. Example waveforms of the IS process were provided in Section 1.5. Measured statistics for various parameters of the ICC and subsequent return strokes were presented in Chapter 4 fo r triggered lightning discharges at the ICLRT between 2009 2011. Such triggered lightning discharges lower negative charge during the IS as well as during the

PAGE 343

343 period after the IS when dart leader/return stroke sequences may occur. The negative charge low ered to ground likely originates in the primary negative charge region of the cloud located at an altitude above the 0 C level, the 0 C level being at 4 5 km in north central Florida [e.g., Harris et al., 2000; Hansen et al., 2010], with a lower and uppe r bound for the primary negative charge region of roughly 6 km and 8 km corresponding to temperatures of 10C and 20C, respectively [e.g., Krehbiel 1986; Gremillion et al ., 1999]. The same primary negative charge region is also the source for natural negative lightning to ground. In Florida at sea level, the IS and the other processes that later illuminate the IS's path generally appear on photographs and to the human eye as a single channel below the cloud base [e.g., Biagi et al ., 2009]. In contra st, photographs of triggered lightning in New Mexico at Langmuir Laboratory at about 3000 m altitude [e.g ., Idone et al ., 1984; Winn et al., 2012] and St. Privat d'Allier in France at about 1000 m altitude [e.g ., Fieux et al ., 1978; Hubert and Mouget 1981 ] typically show an extensively branched UPL. These visual differences have long been a subject of discussion among researchers. Cloud bases in New Mexico in summer are typically 4 4.5 km above sea level, but only 1 1.5 km above the level of Langmuir L aboratory, while typical cloud bases in Florida in summer are about 1 km (+/ 500 m) above sea level. The atmospheric pressure, and likely the temperature and relative humidity, are all generally higher in Florida near sea level than at the higher altitud e triggering sites noted above; and pressure, temperature, and humidity may well influence the height and degree of IS branching. In this chapter, the channel shape and branch characteristics of the IS processes in nine lightning flashes that were trigge red during summer 2011 at the ICLRT are analyzed. The IS channels are mapped in three dimensions via the TOA locations of positive polarity VHF sources associated with the leader tip breakdown obtained from a local seven station Lightning Mapping

PAGE 344

344 Array (L MA) [e.g., Rison et al ., 1999; Krehbiel et al ., 2000 ; Thomas et al ., 2004 ]. As discussed in Section 2.15, each LMA station records the time of the peak VHF power (66 72 MHz) received in consecutive 80 s windows. The time windows are synchronized at the different geographic locations via integrated GPS receivers. The station locations of the seven LMA stations (both absolute and relative to the ICLRT coordinate origin) were given in Table 2 2 and were shown graphically in Figure 2 32. The three dimensio nal source locations and emission times are calculated using the non linear least squares optimization technique described in Section 3.8. Unless otherwise stated, for all LMA data shown in this chapter, source locations are plotted only to the time of th e end of the IS period, this time being determined from the correlated channel base current measurements. All LMA plots including lateral coordinates have positive axis labels corresponding to increasing distance in the northerly and easterly directions. Finally, the axes of all LMA plots are formatted with equal aspect ratios so the true geometry and scale of the discharges in either two or three dimensions are accurately depicted. In Table 7 1, the working condition of the seven station LMA network is given for each day during summer 2011 where triggered lightning data were collected (i.e., rockets were launched that resulted in at least a full IS process). For the events discussed in this chapter on August 5, 2011 and August 18, 2011, six LMA station s were operational. The Blast Wall LMA station was not operational due to a direct natural lightning strike (MSE 11 06) that struck the Blast Wall structure 5 m away from the LMA electronics enclosure, causing signal cabling damage. LMA data used in the analyses presented in this chapter are thus for 5 or 6 station solutions with reduced chi squared values less than 4, when corrected for the 30 ns actual timing errors of the network.

PAGE 345

345 For all triggered flashes discussed here, the current during the IS was measured at the T&M Research R 7000 10 CVR ( Section 2.12.3) and digitized at 10 M S/s on Yokogawa DL750 DSOs ( Section 2.4.5). For the three triggered flashes on August 5, 2011 (UF 11 24, UF 11 25, and UF 11 26) ve rtical scan RHI images were obtained by the C band dua l polarimetric SMART radar ( Section 2.16), which was located 11.6 km to the south of the ICLRT at the Keystone Airpark. For these three events, radar images were used to determine the hydrometeor conte nt in the vicinity of the triggering site and to determine the altitude of the 0 C level. Radar images were also obtained for two additional triggered flashes. These data are not analyzed as part of this dissertation, but will be in the future. Finally data were available from the S band National Weather Service (NWS) Weather Surveillance Radar 1988 Doppler (WSR 88D) [e.g., Crum and Alberty, 1993] located near Jacksonville, FL, 68 km to the north northeast of the ICLRT, to document the larger horizont al scale structure of the cloud systems. The center of the 0.5 elevation radar beam from the WSR 88D was at an altitude of about 900 m at the range of the ICLRT from the Jacksonville radar site. 7.2 The Events of August 5, 2011 At about 17:20 (UT), on A ugust 5, 2011, convection initiated at 35 40 km to the southeast of ICLRT, well inland of the sea breeze circulation that was located near the Florida east coast about 40 km farther away. Within one hour the cloud system grew to form a north south oriente d quasi linear convective band that was roughly 35 km in length. In the deepest part of the cluster, radar echo tops from the S band WSR 88D extended to ~14 km in altitude, consistent with the Jacksonville sounding taken at 12:00 (UT) on August 5. The co nvective band propagated westward at a speed of 5.5 m/s, arriving just east of ICLRT at 19:00 (UT). Shortly afterward, the central part of the system started to dissipate while new convective cells continued to form to the north northeast and south southw est of the ICLRT. The new cells also exhibited

PAGE 346

346 echo tops of about 14 km. Three lightning flashes, designated UF 11 24, UF 11 25, and UF 11 26, were triggered between 19:33:19 (UT) and 19:49:58 (UT). Correlated observations of the LMA source locations, m easured channel base currents, and SMART radar RHI images of the IS processes of UF 11 24, UF 11 25, and UF 11 26 will be discussed in detail in the following sections. Still photographs of the three triggered lightning discharges are shown in Figure 7 1. The photographs were taken from IS2 (view looking due south towards the SMART radar) and are each five second time exposures. 7.2.1 Flash UF 11 24 Flash UF 11 24 was triggered at 19:33:19 (UT) on August 5, 2011 with quasi static electric field at gr ound of about 5.5 kV/m. The flash contained a full IS process and one subsequent return stroke with peak current of 32.8 kA, the largest peak current recorded during summer 2011. The ICC had duration of 425 ms, transferring 46 C of negative charge to gr ound. A three dimensional view of the LMA source locations during the IS period of flash UF 11 24 is shown in Figure 7 2. Four projection views of the LMA source locations for flash UF 11 24 are shown in Figure 7 3. In both Figure 7 2 and Figure 7 3, t he source locations are identically color coded in 100 ms bins according to the respective keys at right. The 0 C level is annotated in all plots including LMA source altitude coordinates. The LMA located a significant number of precursor current pulses (evident at bottom left in Figure 7 3 in the altitude versus time projection) during the ascent of the triggering wire within 400 ms of the initiation of the sustained UPL. The UPL initiated at about 0.95 s in Figure 7 3 and is clearly differentiated fro m the preceding precursor pulses by the abrupt change in propagation speed. The triggering wire exploded about 4.6 ms after the initiation of the sustained UPL when the UPL was at an altitude of about 1.3 km. The UPL propagated generally vertically to an altitude of about 4.7 km with no detected upward branching (Figure 7 2), then turned 90 degrees to the north and propagated

PAGE 347

347 generally horizontally for about 6 km. Here, the UPL channel branched with one channel propagating to the northeast and the other to the west. Not accounting for channel tortuosity, the northeast IS channel progressed for about 5 km, branching once more, at an average altitude of about 4.7 km. The western IS channel also branched once more over a propagation distance of about 4 km as it ascended to an altitude of about 6.7 km (Figure 7 3 A ), after which the LMA sources became more diffuse as the IS channel exhibited significant horizontal branching to the north and west (Figure 7 3 D ). The two primary regions of positive electrical br eakdown from the northeast and western IS branches were eventually separated by as much as 14 km laterally and by about 2 km in altitude. The initial UPL channel prior to channel branching had a total length, including channel tortuosity, of about 12.6 km the longest observed un branched IS channel during the 2011 experiment by more than a factor of three. In Figure 7 4, a three dimensional view of the IS is shown on a spatial scale designed to emphasize the IS branching geometry. Only sources correspo nding to clearly defined IS channels are plotted. Sources corresponding to precursor current pulses and sources not obviously associated with the extension of IS channels plotted in Figures 7 2 and 7 3 are omitted. The initial UPL and the four IS branche s are labeled in Figure 7 4 by increasing initiation time. The sources are color coded in 50 ms time windows according to the key in Figure 7 4. Estimates for the lengths and average propagation speeds of the initial UPL and all IS branches were calcula ted by averaging the emission times and three dimensional spatial locations of the LMA points over time periods of 1 4 ms. The raw data indicate that VHF sources are often emitted from spatial locations well behind the probable tip of the propagating IS c hannel. As a result, the absolute time progression of the LMA source points cannot be used as a proxy for accurately calculating the length and average speed of the propagating channel.

PAGE 348

348 The averaged points are overlaid with the raw data points to insure the channel geometry is not altered significantly and if necessary, the averaging time windows are adjusted accordingly. It is estimated that the errors in the calculations for overall branch length and average speed are less than 25%. As stated previo usly, the initial UPL of flash UF 11 24 (shown in pink, gray, and red in Figure 7 4) traversed a distance of about 12.6 km with average speed of about 8.4 x 10 4 m/s. The two western IS channels (Branch 1 and Branch 3 in Figure 7 4) propagated for total di stances of 5.2 km and 6.2 km at average speeds of 1.7 x 10 4 m/s and 1.9 x 10 4 m/s, respectively. The northeast IS channels (Branch 2 and Branch 4 in Figure 7 3) propagated for total distances of 6.5 km and 7.2 km at speeds of 2.1 x 10 4 m/s and 3.3 x 10 4 m /s, respectively. In Figure 7 5, the altitude projection of the LMA source locations is overlaid on a 910 ms record of the channel base current. The channel base current is plotted only to the end of the IS period. In this case, the mid level sensitiv ity current measurement (II Low) was plotted because the most sensitive measurement (II Very Low) saturated on the 2.4 kA ICC pulse that occurs at about 490 ms in Figure 7 5. The initiation times of the four IS branches are annotated on the channel base c urrent waveform with red diamonds. The branches initiated at current amplitudes ranging from 90 125 A and all occurred following the initiation current variation (ICV) associated with the explosion of the triggering wire. The ICV occurred 4.7 ms after th e initiation of the sustained UPL. The IS branch initiation current amplitudes were determined from the II Very Low current waveform (not shown) with increased vertical resolution. SMART radar RHI scans collected at the time of flash UF 11 24 are shown in Figure 7 6. An equivalent radar reflectivity factor (dBZ) image is plotted in Figure 7 6A and differential radar reflectivity factor image (Z DR ) is plotted in Figure 7 6A The northing projection (Figure 7 3 B ) of

PAGE 349

349 the LMA sources are overlaid on each radar image. LMA sources located within the radar sweep plane are colored black and sources to the west of the sweep plane are colored in dark green. The 0 C level is clearly visible at the tops of the high reflectivity rain shafts to the north northeas t and south southwest of the ICLRT. From the dBZ image of Figure 7 6 A the rocket was launched with light precipitation at ground level (also see Figure 7 1 A ). The IS propagated upward between the rain shafts with little deviation from the vertical until it reached an altitude around the 0 C level. The IS then turned and followed the general contour of the 0 C level for about 6 km prior to exhibiting the branching structure described previously. The northeast IS branch propagated for several kilometer s along the tops of the rain shafts to the north northeast of the ICLRT while the western IS branch propagated towards a separate cell outside the plane of the RHI scan. The distribution of LMA source altitude coordinates for the IS of flash UF 11 24 is shown in Figure 7 7. The distribution includes 909 sources and the histogram is smoothed with bin width of 30 points. The large peak at altitudes less than 1 km is due to the precursor current pulses and the initial UPL points. The peak between 4 5 km is due to the horizontally propagating UPL following the 90 degree northerly turn when the discharge reached the altitude of the 0 C level ( Figure 7 3 B ). The peak between 5 6 km is due to the breakdown region initiated by the northeast IS branches, and t he higher altitude peak between 6 7 km is due to the breakdown region initiated by the western IS branches ( Figure 7 3 A ). 7.2.2 Flash UF 11 25 Flash UF 11 25 was triggered at 19:43:30 (UT) on August 5, 2011. The flash occurred about ten minutes followi ng UF 11 24. The quasi static electric field measured at ground at the time of the rocket launch was 4.6 kV/m. Flash UF 11 25 contained a full IS process and one subsequent return stroke with peak current of 12.1 kA. The ICC had duration of 404 ms and

PAGE 350

350 transferred 28 C of negative charge to ground. A three dimensional plot of the LMA source locations obtained during flash UF 11 25 is shown in Figure 7 8 and four projections views of the LMA sources are shown in Figure 7 9. The 0 C level is annotated i n all plots including LMA source altitude coordinates. The LMA sources span four seconds and are color coded according to the color key at right in 400 ms bins. Flash UF 11 25 exhibited a large number of precursor current pulses during the wire ascent ( Figure 7 9 C ). The LMA recorded source locations for more than 50 precursor current pulses in a time of about 3 s. The HBM dE/dt digitization system (Section 2.5) also recorded waveforms for 33 of the precursor pulses located by the LMA. T he source locat ions of common events with the two TOA systems are used to evaluate the source location accuracy of the LMA system for low altitude sources (treating the TOA locations obtained from the HBM dE/dt network as ground truth). The sustained UPL initiated at ab out 19:43:34.377 (UT), or about 3.6 s in the altitude versus time projection of Figure 7 9 C and is again distinguished from the preceding precursor pulses by the abrupt change in propagation speed. In Figure 7 10, the three dimensional geometrical struc ture of the IS is shown on reduced spatial scale to better show the upward IS branching. The LMA sources in Figure 7 10 are plotted with the same criteria specified for the three dimensional plot of the IS branching of Flash UF 11 24 given in Figure 7 4. Flash UF 11 25 was the first event triggered at the ICLRT where more extensive upward branching of the IS was observed via the LMA source locations. The initial UPL ascended to an altitude of about 750 m at an average speed of 7.9 x 10 4 m/s. Branch 1 (Fi gure 7 10) initiated at 750 m altitude and propagated upward and in a southeasterly direction for about 1.3 km at an average speed of 7.9 x 10 4 m/s, eventually stopping at an altitude of about 2 km. Branch 2 initiated nearly simultaneously with Branch 1 a t an altitude of about 750 m and

PAGE 351

351 propagated in a northerly and upward direction for about 3.3 km at a speed of 7.7 x 10 4 m/s. Branch 3 split from Branch 2 about 4 ms after the initiation of initiation of Branch 2. Branch 3 traveled at an average speed of 7.0 x 10 4 m/s in an upward and slightly southwesterly direction for about 1.7 km before apparently halting at an altitude of about 2.3 km. Branch 4 similarly split from Branch 2 at an altitude of about 2.3 km (25 ms after the initiation of Branch 2) and subsequent propagated for about 700 m to the northwest at an average speed of 3.5 x 10 4 m/s before stopping an altitude of 2.5 km. Branch 2 finally split into Branch 5 and Branch 6 at an altitude of about 2.6 km. Branch 5 traveled to the east northeast for about 900 m at a speed of 2.8 x 10 4 m/s and Branch 6 traveled to the north northeast for about 1.9 km at a speed of 4.3 x 10 4 m/s. Branch 5 and Branch 6 initiated extensive horizontally oriented branching that eventually led to the widespread area of positive breakdown evident in the easting, northing, and plan projections of Figure 7 9 located between 3 6 km altitude. The altitude projection of the LMA source locations (Figure 7 9 C ) is overlaid on a 590 ms waveform of the measured channel base curr ent in Figure 7 11. The current waveform plotted is from the II Very Low measurement (unlike the ICC of flash UF 11 24, there were no large ICC pulses that saturated the vertical scale of the measurement). The initiation times of the six IS branches desc ribed above are annotated on the current waveform with red diamonds. Branches 1 3 initiated within 7 ms of the beginning of the sustained UPL. There was no clearly defined ICV in the channel base current waveform, but given the average time duration betw een the UPL and the ICV of 7.5 ms (measured for 37 triggered flash at the ICLRT between 2009 2011, see Table 4 8), it is likely at least Branch 1 and Branch 2 occurred prior to the wire explosion. Branch 1 and Branch 2 both initiated within 3 ms of the be ginning of the UPL. The six IS branches initiated with channel base current amplitudes ranging from 12 60 A.

PAGE 352

352 SMART radar RHI scans at the time of flash UF 11 25 are shown in Figure 7 12 (dBZ image in Figure 7 12A and Z DR image in Figure 7 12B ). The no rthing projection of the LMA source locations (Figure 7 9 B ) is overlaid on both radar images with the same source coloring convention described for flash UF 11 24 (Figure 7 6). In the dBZ image of Figure 7 12 A the reflectivity in both the rain shafts to the north northeast and south southwest of the ICLRT decreased from the time of flash UF 11 24. In addition, the reflectivity immediately over the launching facility had ceased and there were clear sky conditions to the east ( Figure 7 1 B ). The initial UP L and Branches 1 3 propagated upward through relatively clear air. Branch 1 and Branch 3 both halted slightly above 2 km altitude in an area of light precipitation. Branch 2, Branch 5, and Branch 6 propagated in northerly directions generally within the plane of radar RHI scan, producing the high density area of LMA sources between 3 4 km altitude that extends from about 14 19 km in Figure 7 12 A (about 2.4 7.4 km from the launching facility). The IS branches appear to have propagated towards and into an area of higher reflectivity centered at about 18 km from the radar ( 5 .4 km from the launching facility) and about 3 km in altitude. The LMA sources from the IS then extended upward to around the 0 C level at about 4.7 km, where the discharge propagated in a northerly direction for about 10 km along the top of the high reflectivity region in a similar manner to flash UF 11 24. Extensive IS branching also occurred to the west out of the plane of the radar RHI scan (LMA sources shown in dark green in Figur e 7 12 A ). The western IS branches occurre d at altitudes from 3 7 km ( Figure 7 9 A ). The higher altitude sources between 5 7 km propagated into the same region where the western IS branches of flash UF 11 24 produced significant positive polarity electrica l breakdown. Interestingly, the lower altitude western LMA sources between 3 4 km appear to have been associated with a natural negative cloud to ground

PAGE 353

353 discharge that occurred during the IS process of flash UF 11 25. The natural flash likely terminated 5 7 km to the northwest of the launching facility ( Figure 7 9 A ). In Figure 7 13, the distribution of LMA source altitudes during the IS of flash UF 11 25 is shown. The histogram includes a total 1036 sources plotted with bin width of 30 sources. The f eatures of the histogram of flash UF 11 25 are somewhat more complicated than that of flash UF 11 24 due to more extensive electrical breakdown that occurred at different altitudes. The large collection of sources below 2 km are from the initial UPL, Bran ch 1, Branch 3, and the initial portion of Branch 2. The peaks around 3 km altitude are mostly due to the initial IS branches (Branch 2, Branch 5, and Branch 6) that traveled to the north within the plane of the SMART radar RHI scan. There is also some c ontribution to the histogram peaks around 3 km from the western IS branches that occurred later in the IS process than initiated the natural cloud to ground discharge 5 7 km northwest of the launching facility. The dominant histogram peaks between 5 6 km are from the IS branches that propagated along and above the 0 C level following the initial progression in the 3 km altitude range. As discussed above a large number of precursor current pulses were measured on the ascending triggering wire for flas h UF 11 25, in part due to the relatively long duration between the rocket launch and the initiation of the sustained UPL (about 4.4 s). The 33 precursor pulses located by both the LMA and the HBM dE/dt TOA network are used to evaluate the accuracy of the LMA system for low altitude sources, treating the dE/dt TOA source locations as ground truth. For flash UF 11 25, the closest LMA station to the launching facility was the Blanding site, located 2771 m east southeast of the launcher ( Table 2 2 and Figure 2 32). In Figure 7 14, the easting (longitudinal) coordinates of the dE/dt (blue circles) and LMA (red squares) source locations for the 33 common pulses are plotted as a function of increasing emission time. In

PAGE 354

354 addition, the differences between the eas ting coordinates of the LMA sources and the dE/dt sources are plotted (black diamonds). In general, the LMA overestimated the easting coordinate of the precursor pulses by less than 25 m, with a gradually decreasing error with increasing height of the tri ggering wire. The easting source locations of the two TOA systems agreed to within 5 m for seven pulses. A similar plot is shown in Figure 7 15 for the northing (latitudinal) coordinates of the dE/dt and LMA sources. The LMA tended to overestimate the n orthing coordinate of the precursor sources by less than 20 m when the triggering wire was less than 350 m in altitude, though for 11 pulses, the locations of the two system agreed to within 5 m. From 350 m to 450 m in altitude, the LMA tended to underes timate the precursor source location by less than 15 m, but similarly for 7 pulses, the locations of the two systems agreed to within 5 m. The accuracies of the LMA easting and northing source coordinates were not strong functions of altitude, as expecte d [e.g., Thomas et al., 2004]. In Figure 7 16, a plot is shown of the LMA and dE/dt source altitudes for the 33 precursor pulses versus increasing emission time. The differences between the LMA and dE/dt source altitudes are also plotted. For source alt itudes (determined by dE/dt TOA locations) from 160 m to 360 m, the LMA overestimated the altitude location of the source by an average of about 81 m. For source altitudes between 375 m and 455 m, the LMA overestimated the altitude location of the source by an average of only 34 m. 7.2.3 Flash UF 11 26 Triggered lightning discharges in Florida normally transport only negative charge to ground through the IS process and any subsequent leader/return stroke sequences. Flashes UF 11 24 and UF 11 25 discus sed above are examples of normal polarity triggered lightning discharges. Occasionally, triggered lightning current measured at the channel base, which, as noted above, usually exhibits a polarity consistent with the lowering of negative cloud charge, und ergoes a polarity reversal in its usual current and lowers positive charge. Positive charge

PAGE 355

355 regions are reported to be present in typical thunderstorms both below and above the primary negative charge region, that is, above about 8 km and below about 6 km [e.g., Marshall and Rust, 1991; Bringi et al., 1997]. Yoshida et al. [2012] described a flash triggered at the ICLRT in 2009 (flash UF 09 30) in which the IS first lowered negative charge, then positive charge. Yoshida et al. [2012] used a two dimension al VHF interferometer to infer the activity of a series of in cloud leaders that were not initially connected to the IS channel but that subsequently led to the connection of an "upper level positive charge region" with the IS channel to ground. Jerauld e t al. [2004] presented data for a two stroke triggered flash in Florida whose first stroke lowered the expected negative charge following a negative IS while the second stroke unexpectedly lowered positive charge. The location of the charge source for the positive stroke could not be determined. Here, data are presented for flash UF 11 26, a triggered discharge that first lowered negative charge to ground via the initial portion of the IS, and then induced a more or less natural appearing bi level intra cloud discharge via an upward propagating negatively charged leader between the inferred main negative cloud charge region and the primary positive charge region above it. Typical natural intracloud lightning flashes have previously been studied with LMA systems [e.g., Thomas et al., 2001; Behnke et al., 2005] and with two dimensional interferometric systems [e.g., Shao and Krehbiel, 1996]. For triggered flash UF 11 26, some of the charge from the positive charge region of the cloud flowed to ground throu gh the path of the previously negative charge lowering IS, causing a current polarity reversal of about 57 ms duration to be observed at ground. The coordinated data presented in this section: LMA source locations, SMART radar data, and electrical current at the channel base, provide convincing evidence that the process of triggering lightning with a grounded wire can directly induce a

PAGE 356

356 typical normal polarity, bi level intracloud discharge between the primary negative and upper positive cloud charge region s, as well providing information on the physics of both the triggering process and the resultant lightning discharges. Flash UF 11 26 was triggered at 19:49:58 (UT) on August 5, 2011 with a quasi static electric field at ground of 5.6 kV/m. The disch arge was triggered about 7 minutes after flash UF 11 25. At the time of the rocket launch, the low level reflectivity observed over the ICLRT by the Jacksonville WSR 88D had decreased to about 30 dBZ. From the still image shown in Figure 7 1 C the condit ions at ground level had continued to clear following flash UF 11 25. UF 11 26 was composed of an IS process followed by an intracloud discharge. There were no return strokes to ground in the flash. A three dimensional plot of the LMA source locations f or flash UF 11 26 is shown in Figure 7 17. In Figure 7 18 four projections of the LMA source locations obtained during UF 11 26 are shown. The 0 C level is annotated in all plots of Figure 7 18 that include LMA source altitude coordinates. Figures 7 1 7 and 7 18 both show a 1 s time period with points color coded according to the keys at right in 100 ms bins. The LMA detected only four precursor current pulses during the wire ascent, though many precursors were measured on the ascending triggering wire A plot of the LMA source locations showing the branching structure of the IS process is given in Figure 7 19. The points are plotted with the same criteria specified for flashes UF 11 24 and UF 11 25 (Figures 7 4 and 7 10). In Figure 7 19 (and in Figu re 7 18 A ), the UPL is seen to propagate upward for about 2.5 ms before diverging to the west at an altitude of about 580 m (Branch 1). About 2.6 ms after the UPL turned to the west, an IS branch (Branch 2) initiated at an altitude of about 710 m and propa gated towards the east. The triggering wire exploded when the IS branches reached about 750 m in altitude, about 5.2 ms following the initiation of the sustained UPL. The two IS channels then propagated generally

PAGE 357

357 upward, eventually becoming separated lat erally by about 2 km, at an altitude of about 4.5 km, before turning horizontal. Branch 1 traversed a total distance of about 2.8 km at an average speed of 1.8 x 10 5 m/s and Branch 2 propagated for about 4.1 km at an average speed of 4.7 x 10 4 m/s. In Figure 7 20, a 600 ms waveform of the measured channel base current and the altitude projection of the LMA source locations (Figure 7 18 C ) are plotted. Recall that a positive current corresponds to negative charge transported to ground and vice versa. T he current waveform shows the final 120 ms of the rocket ascent and the full 433 ms duration of the ICC process. About 114 ms after the initiation of the UPL, the channel base current, which had exhibited a polarity indicative of negative charge transport to ground, decreased sharply towards zero and subsequently changed polarity in a time span of about 2 ms. The time of the current polarity reversal is annotated in Figure 7 20 with a vertical line and the baseline current level is shown with a dotted hor izontal line. The current waveform was plotted using the II Low measurement considering the ICC background current level was above the saturation point of the II Very Low measurement and the ICC contained many super imposed pulses with amplitudes greater than 1 kA. About 7 ms prior to the change in current polarity, an upward negative leader, characteristic of typical naturally occurring bi level intracloud discharges, initiated at an altitude of about 5.6 km. The negative leader propagated from an altit ude of 5.6 km to about 9.3 km in a time of 11 ms (LMA sources shown in red color in Figure 7 20 and confirmed as negative polarity by their relatively high source power). The upward negative leader source powers recorded by the LMA ranged from 4 16 dBW, generally 1 to 2 orders of magnitude stronger than the preceding positive VHF sources recorded during the IS. The upward negative leader initiated about 3.5 km to the southwest of the launching facility and traversed a total distance of about 4.5 km with an

PAGE 358

358 average speed of about 4.1 x 10 5 m/s. Without accounting for channel tortuosity, the leader propagated at an angle from the vertical of about 18 degrees at an azimuth of about 109 degrees (southeast). The leader initiated widespread negative breakdow n within the inferred upper positive charge region for about 270 ms following its propagation. The negative breakdown covered a lateral area of about 20 x 23 km and ranged in altitude from about 8 km to about 10 km. Positive breakdown at an altitude near 5 km, eventually covering a lateral area of about 18 x 25 km, occurred simultaneously with the higher altitude negative breakdown,. Individual intracloud channels were poorly resolved at both altitude ranges, particularly within 5 km of the triggering si te, likely a result of the presence of many simultaneous channels and the 80 s acquisition window per LMA source location (e.g., compare the branch detail of the intracloud channels in Figure 7 3 for flash UF 11 24 with the lack of branch detail in the ca se of flash UF 11 26). A 35 ms segment of the channel base current waveform plotted in Figure 7 20 is shown in Figure 7 21 (the time scale of Figure 7 21 corresponds to 110 145 ms in Figure 7 20) with the altitude projection of the LMA source locations overlaid. The waveform shows the final 6.5 ms of the rocket ascent and the first 28.5 ms of the ICC (prior to the current polarity reversal). The times of the two IS branches are annotated on the current waveform with red diamonds. The IS branches initi ated with channel base current amplitudes of 9 A and 43 A, respectively. These current values were measured using the II Very Low current waveform (not shown). Flash UF 11 26 exhibited a clear ICV (Type II event from Section 4 3), which, as stated previo usly, occurred about 5.2 ms following the initiation of the sustained UPL. In Figure 7 22, a 75 ms plot of the current polarity reversal is shown. The time scale of Figure 7 22 corresponds to 225 300 ms in Figure 7 20. About 4.8 ms prior to the initia l drop in

PAGE 359

359 magnitude of the channel base current between 230 231 ms in Figure 7 22, a bipolar dE/dt pulse (~115 V/m/s peak to peak) was recorded by all 10 dE/dt sensors on the HBM dE/dt network. The initial polarity of the pulse is indicative of the upwar d movement of negative charge. The pulse corresponds in time to the first LMA source location of the upward negative leader. A 1 s window surrounding the pulse is shown in the inset of Figure 7 22 and the arrow shows where in time the pulse occurred rel ative to the channel base current waveform. The TOA location of the pulse from the HBM dE/dt network was within 290 m of the LMA source location (about 280 m laterally and about 20 m in altitude). Given that the source was well outside the boundary of th e dE/dt TOA network, the lateral location errors of the dE/dt TOA system are expected to be of the order of 100 m and the altitude location error of the order of 200 m. It is also worth noting that the initial LMA source location of the upward negative le ader occurred coincident with a small rise in the measured channel base current (immediately following the location of the arrow in Figure 7 22. No additional HBM dE/dt pulses clearly corresponding to the ascending LMA sources of the upward negative leade r were resolved. About 2.3 ms following the current polarity reversal, the current magnitude fell sharply to a level of about 3.2 kA. The current waveform during the full 56.8 ms duration of the polarity reversal is characterized by a continuous series of wide (~ 0.5 3 ms) pulses with amplitudes from several hundred amperes to several kilo amperes. The pulses are generally similar to negative ICC pulses [e.g., Wang et al., 1999 c ; Miki et al., 2005] but, in this case, appear to be superimposed on an oppo site polarity background current that gradually tends back towards zero. The implication is that both positive and negative charge sources were simultaneously available to the channel to ground. About 19 C of positive charge were transferred to ground du ring the current polarity reversal while the negative portion of the ICC transferred 120 C of charge to ground. The progression of the

PAGE 360

360 intracloud negative breakdown, as indicated by the LMA source locations, does not appear to be significantly altered whe n, after 56.8 ms of reverse current flow, the channel base current reverted back to a polarity indicative of negative charge transport to ground. In Figure 7 23 A a 300 ms record of the electric field is plotted corresponding to the time region from 100 400 ms of Figure 7 20. The channel base current is plotted in Figure 7 23 B for reference. The electric field was measure d with an inverted antenna ( Section 2.19) at Station 12 with sensitivity of about 197 kV/m/V. The antenna has a decay time constant o f 8 s. The more sensitive electric field antennas all saturated the dynamic range of the fiber optic transmitter s at the beginning of the UPL and did not recover from saturation quickly enough to view the current polarity reversal. In order to better res olve lower amplitude field changes contaminated by the system noise, the electric field waveform was first filtered with a 1000 point moving average filter. The most interesting characteristic of the electric field waveform is the pronounced hump that ris es from the system noise level about 5 ms prior to the current polarity reversal. The peak of the electric field hump occurred about 7.2 ms later, near the time of the next to last LMA source corresponding to the upward negative leader at an altitude of a bout 8.3 km. The electric field change could have started earlier in the waveform, closer to the time of the initiation of the upward negative leader, and be simply unresolved due to lack of system sensitivity. Yoshida et al. [2012] show a very similar e lectric field waveform feature in their Figure 8F immediately prior to the current polarity reversal of flash UF 09 30. SMART radar RHI scans at the time of flash UF 11 26 are shown in Figure 7 24 (dBZ image in Figure 7 24A and Z DR image in Figure 7 24B ). The northing projection of the LMA source locations (Figure 7 18 B ) during flash UF 11 26 is overlaid on each radar image. The precipitation at the ground surface was observed to have ended prior to the rocket launch that

PAGE 361

361 triggered flash UF 11 26 (also see Figure 7 1 C ). An analysis of the RHI scans taken with the SMART radar performed by Dr. Michael Biggerstaff and Mr. Pat Hyland of the University of Oklahoma shows that the rocket and subsequent UPL rose through a tilted descending reflectively packet (DRP) that was 2 km wide, on average, in the plane of the RHI and extended at least 1 km in the east west direction, as determined by the five degree azimuthal RHI sector scan centered on the ICLRT. At the time of the rocket launch, the north south width of the DRP base was 600 m over the ICLRT and was found at an altitude of 500 m. A few hundred meters farther east, however, the DRP had already reached the surface. The DRP formation may have been associated with remnants of the convective cell in the ce ntral part of the storm system but appeared as localized coalescence of raindrops just below the 0 C level. The vertical extent of the DRP increased as larger drops fell faster than smaller drops, which led to an appreciable size sorting signature in the differential radar reflectivity (Z DR ). Indeed, the lower portion of DRP had areas where Z DR values exceeded 2 dB while the upper portion had Z DR values closer to 1 dB. Given the low reflectivity and high Z DR values in the base of the DRP, it is likely t hat this region had low concentrations (less than one drop per cubic meter of air) of medium to large sized raindrops. When the IS channel turned horizontal near 4.5 km altitude, it propagated for some kilometers across the top of the DRP, and along the r adar determined 0 C level ( Figure 7 18). Thus, the path of the initial UPL and the subsequent channel development were closely associated with precipitation boundaries. The majority of the LMA sources during the current polarity reversal at ground w ere located in a layer between 8 10 km in altitude and were associated with precipitation near the cores of dissipating convective cells to the north and southwest of the ICLRT. SMART radar reflectivity indicates that the anvil of the cloud system was inh omogeneous, with regions of

PAGE 362

362 reflectivity approaching 35 dBZ in the 8 10 km layer (Figure 7 24 A ). The LMA sources during the charge reversal suggest the negative breakdown occurred near the top of these enhanced reflectivity regions. One of the enhanced r eflectivity regions, observed between 5 8 km range and 8 9.5 km in altitude at 19:50 (UT), had Z DR values that were negative on average. The surrounding regions, with lower radar reflectivity, had Z DR values closer to 0. Negative Z DR aloft is often assoc iated with ice crystals that become vertically aligned by strong vertical electric fields [e.g., Hendry and McCormick 1976]. It is noteworthy that projection of LMA sources onto the plane of the RHI showed that the majority of the initial few LMA sources during the charge reversal fell within this region of mean negative Z DR While the narrow section sampled by the radar during this time did not include the dissipating cell to the southwest of the ICLRT where the initial few LMA sources occurred, adjacen t RHIs suggest this region did extend westward. Aspects of the cloud charge structure during flash UF 11 26 can be inferred from the altitude distribution of LMA source locations. Figure 7 25 shows a histogram (bin width of 30 source points) of the LMA s ource altitudes for 667 sources. The bimodal distribution of source altitudes together with the corresponding source powers measured by the LMA indicates a region of predominantly positive electrical breakdown spanning an altitude from about 4 5.5 km and a region of predominantly negative breakdown spanning in altitude from about 7.5 10.5 km. The electric fields are highest on the outer boundaries of the charge region so the exact relation between the LMA breakdown location and the charge sources before t he intracloud discharge is not completely determinate. The unknown is how far and how thoroughly the discharge penetrates the charge region. The vertical structure of the electrical breakdown suggests that the upward negative leader, which propagated fro m 5.6 to 9.3 km in altitude, connected the expected

PAGE 363

363 top of the negative charge region to the lower part of the upper positive charge region. In the process, a path was developed for some positive charge to flow through the IS channel to ground. 7.3 Flas h UF 11 32 (August 18, 2011) Flash UF 11 32 was triggered at 20:37:29 (UT) on August 18, 2011 when the quasi static electric field at ground was about 6.4 kV/m. The flash was the first of four triggered lightning events on August 18, 2011 (UF 11 32 thro ugh UF 11 35). Flash UF 11 32 had a full initial stage process followed by two subsequent return strokes with peak currents of 14.5 kA and 19.8 kA, respectively. The IS process of flash UF 11 32 had several unique features, 1) the total duration of the I CC period was 945 ms, the longest ICC observed between 2009 2011 by more than 200 ms; 2) the ICC charge transfer was 225 C, also the largest observed from 2009 2011 by a margin of about 32 C; 3) the channel base current dropped to zero during the wire expl osion (Type I ICV) for a time of 950 s, the second longest zero current interval during the ICV period recorded from 2009 2011; and 4) two attempted reconnection pulses (ARPs) were recorded during the ICV period with amplitudes of 102 A and 119 A followed by a successful reconnection pulse (RP) with amplitude of 511 A. In Figure 7 26, a three dimensional plot is shown of the LMA source locations during the IS of flash UF 11 32, and in Figure 7 27, four projections views of the LMA source locations are sh own. In both Figure 7 26 and 7 27, the sources span 2 s and are color coded according to the key at right in 200 ms bins. During the triggering wire ascent and before its destruction, the LMA recorded source locations for 12 precursor pulses. These sour ces are annotated in the altitude versus time projection of Figure 7 27 C and occurred over a time period of about 650 ms prior to the initiation of the sustained UPL. The UPL is annotated in Figure 7 27 C primarily in orange color and, like the previous f lashes, is distinguishable from the preceding precursor pulses by the abrupt change in propagation speed.

PAGE 364

364 The triggering wire vaporized when the IS was at an altitude of about 2 km, about 19 ms after the initiation of the sustained UPL. The IS of flash UF 11 32, like that of flash UF 11 25 discussed in Section 7.2.2, exhibited more extensive low altitude upward branching than did most flashes recorded by the LMA during summer 2011. In Figure 7 28, a three dimensional view of the IS of flash UF 11 32 is shown on a spatial scale designed to emphasize the upward branching geometry. LMA sources for the IS branches are plotted with the same criteria used for the previously analyzed flashes. There are a total of six clearly defined IS branches labeled in Fi gure 7 28 by increasing initiation time. The sources are color coded in 20 ms time windows according to the key at right. The initial UPL (shown in bright green in Figure 7 28) propagated upward and generally southward for about 630 m at an average spee d of about 9.5 x 10 4 m/s before branching into two distinct IS branches at altitudes of about 880 m (Branch 1) and 870 m (Branch 2) respectively. The two initial IS branches are also clearly visible in Figure 7 27 at top right, Branch 1 on the right and B ranch 2 on the left. Branch 1 propagated generally upward for a distance of about 3.6 km at an average speed of about 9.4 x 10 4 m/s before splitting into Branch 3 and Branch 4. Likewise, Branch 2 moved generally upward and southward for about 14.7 km at an average speed of about 1.2 x 10 5 m/s, eventually splitting into Branch 5 and Branch 6. Branch 3 split from Branch 1 at an altitude of about 3.2 km and propagated in an upward and easterly direction for a distance of about 2.1 km at an average speed of about 5.6 x 10 4 m/s. Branch 4 initiated at an altitude of about 3.1 km and propagated in a northeasterly direction for about 500 m at an average speed of about 3.5 x 10 4 m/s. Finally, Branch 5 split from Branch 2 at an altitude of about 4.9 km and propag ated in a southerly direction for 810 m at an average speed of 2.8 x 10 4

PAGE 365

365 m/s, and Branch 6 initiated at an altitude of 4.8 km and propagated for 770 m in a southwesterly direction at an average speed of 2.0 x 10 4 m/s. In Figure 7 29, 200 ms of the LMA sou rce altitudes are plotted with the measured channel base current waveform (II Low). LMA sources corresponding to three precursor current pulses are shown between 10 40 ms in the plot. The sustained UPL is initiated at about 67 ms. The times of the six IS branches are annotated on the current waveform by red diamonds. These branches were initiated with corresponding channel base current amplitudes ranging from 11 A (Branch 1) to 156 A (Branch 5). LMA sources enclosed in the black box at about 86 ms cor respond to the time of the explosion of the triggering wire and subsequent reconnection processes discussed above. A final observation from Figure 7 29 is that Branch 1, while having a slightly slower average propagation speed than Branch 2, propagated wi th higher speed over its first roughly 8 ms. This speed difference is evident in the LMA altitude sources in Figure 7 29 as the upward splitting of the black data points immediately following in time the red diamond indicating the initiation of Branch 2. The distribution of LMA altitude sources for the IS of flash UF 11 32 is shown in Figure 7 30. The histogram is plotted for 974 sources with bin width of 30 sources. The collection of histogram peaks below 4 km are due to the lower altitude upward IS branches described above. The dominant histogram peak at about 4 km is due to the dense area of electrical breakdown that developed to the north and west of the launching facility following the progression of the clearly defined IS branches. This area of electrical breakdown is clearly visible in Figure 7 27 primarily in the dark pink and black colored LMA points. The higher altitude histogram peaks in Figure 7 30 between 4.5 7 km are a result of the more diffuse electrical breakdown that propagated to

PAGE 366

366 t he northwest, southwest, and southeast from the launching facility. These LMA sources are shown in Figure 7 27 primarily in dark blue color. Interestingly, the rocket launch of flash UF 11 32 occurred during the final stage of a classical bi level intr acloud discharge located southeast of the ICLRT. Four projection views of the LMA source locations during the intracloud discharge and the full extent of flash UF 11 32 (including the subsequent return strokes) are shown in Figure 7 31. The plots span 5 s in time and the points are color code according to the key at right in 500 ms bins. Note the coloring of the LMA points associated with flash UF 11 32 is not the same as for Figures 7 26 and 7 27 due to different binning. The intracloud flash initiated about 25 km east and 10 km south of the launching facility. The LMA points associated with the initiation of the intracloud flash are shown in bright green. Note that the source altitudes and ranges shown for the intracloud flash are likely both overest im ated by several kilometers (reference Thomas et al. [2004] for a thorough discussion on the LMA altitude and range uncertainties for sources well outside the LMA network boundaries). The intracloud flash developed for about a time period of about 650 m s, through the time of the rocket launch just prior to 1 s in Figure 7 31 C An animation of the LMA sources shows that the intracloud discharge halted abruptly as the rocket reached altitude. The LMA sources during the latter stage of the IS period (show n in dark blue in Figure 7 27) that propagated to the southeast from the launch facility continued to travel to the southeast after the cessation of the IS current at ground. As shown in the plan view of Figure 7 31 D the southeastern horizontal branches of flash UF 11 32 eventually propagated towards and possible into the region where the intracloud discharge had terminated 3 s earlier, at an altitude between 5 6 km.

PAGE 367

367 7.4 LMA Observations of Additional 2011 Triggered Flashes In addition to the four tr iggered lightning flashes discussed in detail in the previous sections, LMA data were obtained for an additional five flashes, each containing a full IS process and subsequent return strokes. The LMA data for these remaining flashes will not be discussed in detail in this chapter, though the data are shown for completeness. Statistical channel base current parameters for these flashes can be found in Table 4 4. In order of increasing shot number, both a three dimensional plot of the LMA source locations and a plot of the four projection views of the LMA sources are given for each flash in accordance with the following list: 1) Flash UF 11 11 (062311), Figures 7 32 and 7 33 2) Flash UF 11 28 (081211), Figures 7 34 and 7 35 3) Flash UF 11 33 (081811), Figur es 7 36 and 7 37 4) Flash UF 11 34 (081811), Figures 7 38 and 7 39 5) Flash UF 11 35 (081811), Figures 7 40 and 7 41 7.5 Discussion of 2011 Triggered Lightning LMA Observations Parameters of the measured channel base current during the full duration of th e UPL/ICC, geometrical characteristics of the initial UPL prior to channel branching, and geometrical characteristics of all IS channel branches are given in Table 7 2 for the nine triggered lightning flashes presented in the preceding sections. Some of t he measured channel base current parameters in Table 7 2 were listed previously in Table 4 4, but are given here again for reference. The LMA source altitude distributions for all nine triggered lightning flashes during summer 2011 are shown in Figure 7 4 2. The distributions are only plotted for LMA sources extending to 8 km in altitude. The panels of Figure 7 42 corresponding to flashes UF 11 24, UF 11 25, and UF 11 32 are identical to those shown in Figures 7 7. 7 13, and 7 30. The

PAGE 368

368 panel in Figure 7 4 2 corresponding to flash UF 11 26 does not show the higher altitude sources associated with the negative breakdown region of the triggered bi level intracloud discharge (Section 7.3). The LMA source altitude distributions for the nine triggered flashes sh own in Figure 7 42 are combined and plotted in Figure 7 43. The peak of the cumulative LMA source altitude distribution for the nine events occurs between about 4 5.5 km. Branching of the IS was observed via the LMA in eight of the nine triggered light ning events (the exception being flash UF 11 11 on June 23, 2011), occurring from altitudes as low as 580 m (flash UF 11 26), but more typically from altitudes of about 700 m to 5 km. Due to the low cloud bases, 1 km (+/ 500 m) during typical Florida thu nderstorms, and the fact that the array of cameras at the ICLRT typically views only about 450 m above the launching facility, IS branches are rarely viewed by eye or optically imaged. On the basis of work done to date, there is no definitive answer to wh y there is an apparent difference in branching characteristics between IS's at the ICLRT and those occurring at higher altitude triggering sites (e.g., Langmuir Lab in New Mexico and St. Privat d'Allier in France). For the nine triggered flashes in sum mer 2011, the time of each IS branch obtained from the LMA source locations was correlated with the measured channel base current to determine the current amplitude when up ward IS branching occurred ( Table 7 2) and, more importantly, if the initiation of t he IS branch caused a notable change in the current at ground. Unexpectedly, no significant change in the measured current at ground was observed concurrent with the initiation of IS branches. There is likely +/ 1 4 ms of error in the relation of the LM A branching times with the channel base current due to the averaging method discussed in Section 7.2.1, though accounting for the errors does not appear to significantly alter the above observation.

PAGE 369

369 For two triggered lightning flashes at the ICLRT in 2009, Yoshida et al. [2010] used a two station interferometer to map two IS's from altitudes of 1.1 to 2.4 km and 1.5 to 3.7 km, with three dimensional velocities of 2.2 x 10 6 m/s and 3.3 x 10 6 m/s, respectively. The average three dimensional speed calcul ated for the initial UPL channels before any branching occurs (Table 7 2) for the nine triggered flashes in summer 2011 was about 8.7 x 10 4 m/s, more than an order of magnitude slower than those calculated by Yoshida et al. [2010], and in much better agree ment with the two dimensional UPL speeds of 5.6 x 10 4 m/s given by Biagi et al. [2009] for the first 100 m of propagation of a UPL at the ICLRT, and of 1.0 x 10 5 m/s given by Jiang et al. [2011] for an UPL in China propagating between 130 730 m above ground Yoshida et al. [2010] reported that the highest source altitudes of the two recorded IS's were below the typical 0 C level in Florida storms of 4 5 km. Yoshida et al. [2010] suggest that the cloud charge structure was atypical of a Florida convecti ve thunderstorm and that there may have been a negative charge layer between 2 4 km, similar to that reported by Stolzenburg et al. [2002] for mesoscale convective systems. The distributions of LMA altitude sources shown in Figure 7 42 for the nine trigge red flashes in summer 2011, with typical peaks between 3 6 km, suggest that the two events discussed in Yoshida et al. [2010] may, in fact, have been typical for triggered lightning in Florida thunderstorms. In contrast to the results of Yoshida et al. [2 010], who reported no sources following the clearly defined IS channel, the LMA located sources following the end of the clearly defined IS channels generally propagated horizontally and outward from the ending point of the IS channels at altitudes near th e 0 C level of 4 5 km. In the cases of flashes UF 11 24 and UF 11 25, the horizontally propagating IS channels clearly followed the contour of the 0 C level for many kilometers along the tops of high reflectivity rain shafts to the northeast of the ICLR T ( reference SMART radar images in Figures 7 6 and 7

PAGE 370

37 0 12). The fact that the IS channels turn horizontal in the 3 6 km range indicates a preferred path for propagation either because of high electric fields there or perhaps because of the characteristics o f the particular hydrometeors present in that range, near the 0 C level, allow for a lowered breakdown electric field. Current polarity reversals during the ICC periods of triggered lightning flashes (e.g., flash UF 11 26) at the ICLRT have been rare ly observed, having occurred in only 2 out of a total of 51 (~ 4%) events from 2008 2011. The first case occurred in 2009 and is documented in Yoshida et al. [2012]. The current polarity reversal measured at ground in Yoshida et al. [2012] was shorter in duration (39 ms vs. 56.8 ms for flash UF 11 26) but transferred more positive charge (29 C vs. 19 C for UF 11 26) and exhibited a higher magnitude peak current during the reversal period ( 5.5 kA vs. 3.2 kA for UF 11 26). Yoshida et al. [2012] used a br oadband interferometer to infer the presence of a connecting leader between the main negative and upper positive charge regions that initiated 7.6 ms prior to the observed current polarity reversal, but they do not provide geometrical characteristics of th e leader. Unlike UF 11 26, which had no subsequent return strokes, the event discussed in Yoshida et al. [2012] had one subsequent return stroke with peak current of 29.6 kA following a 140 ms zero current interval at the cessation of the IS. The time d uration (11 ms) and average 3 D speed (4.1 x 10 5 m/s) of the upward negative leader recorded by the LMA for flash UF 11 26 are in good agreement with statistics reported by Shao et al. [1996], who used a narrowband interferometer to determine upward negati ve leader durations of 10 20 ms and propagation speeds of 1.5 to 3 x 10 5 m/s for bi level intracloud flashes in central Florida, and with statistics reported by Behnke et al. [2005], who used LMA sources to calculate the median initial upward negative lead er speed to be 1.6 x 10 5 m/s for 24 intracloud flashes in New Mexico and Kansas. The upward negative leader in this Florida study appeared to have

PAGE 371

371 initiated at an altitude of about 5.6 km, several kilometers lower than those reported by Shao et al. [1996] and Behnke et al. [2005] for naturally occurring bi level intracloud discharges, both studies reporting initial altitudes of about 7 8 km. Although one could argue that the horizontally propagating, positively charged IS channels that propagate along t he contour of the 0 C level would deposit positive charge or neutralize negative charge in the lower part of the main part of the negative charge region, thereby reducing the electric field in the cloud and the probability that a negatively charged leade r (such as the one observed during the ICC of flash UF 11 26) would initiate, in the case of flash UF 11 26, the propagating IS channels and their charge deposition appear to have provided suitable conditions at an altitude of 5.6 km for the initiation of an upward negative leader that started a more or less natural bi level intracloud discharge. Finally, Yoshida et al. [2010], based on two triggered lightning events with abnormally large ICC pulses, suggest that positive VHF sources, which have traditio nally been thought to have insufficient power to be routinely recorded by VHF imaging systems [e.g., Shao et al., 1996], can be recorded "if the current is sufficiently high (> 1 kA) and/or is impulsive". The LMA at the ICLRT resolved positive VHF sources from precursor current pulses during the triggering wire ascent with current amplitudes as small as 10 A, with the closest LMA station at about 2.7 km.

PAGE 372

372 Table 7 1. A list of working LMA stations on each day during summer 2011 where triggered lightning data were collected (flashes with at least a full IS process). Checkmarks denote proper operation. a Flash MSE 11 06 on July 31, 2011 struck the Blast Wall on the northeast corner of the ICLRT, damaging the LMA signal cabling. The Bla st Wall LMA station was operational for the first five onsite natural lightning discharges on July 31, 2011. Date Blast Wall Golf Dupont_S Dupont_N FDOT Warehouse Blanding Comments 062311 Bad phase on Dupont_N 062611 Bad phase on Dup ont_N & Warehouse 070711 Bad phase on Dupont_N & Warehouse 071011 Bad phase on Dupont_N & Warehouse 073111 Blast Wall stopped working a 080511 Blast Wall not operational 081211 Blast wall not operational 081311 Blast Wall not operational 081811 Blast Wall not operational

PAGE 373

373 Figure 7 1. Still photographs of A) flashes UF 11 24, B) UF 11 25, and C) UF 11 26 on August 5, 2011. The photographs were taken from IS2 and the view is looking due south. Photos courtesy of the author. A B C

PAGE 374

374 Figure 7 2. Three dimensional plot of the LMA source locations for flash UF 11 24 on August 5, 2011. Sources span 1 s beginning at 19:33:19.400 (UT) and are color coded accor ding to the key at right, each color corresponding to a 100 ms time window. The location of the launching facility is annotated.

PAGE 375

375 Figure 7 3. Four projection views of the LMA source locations for flash UF 11 24 on August 5, 2011 A ) an easting versus altitude p lot (view looking due north), B ) a northing versus altitude plot (view looking due west), C ) an altitude versus time plot, and D ) an easting versus northing plot (plan view, launching facility annotated by large red circle). The sources span 1 s in time beginning at 19:33:19.400 (UT) and are color coded according to the key at bottom right, each color corresponding to a 100 ms time window. The altitude of the 0 C level is annotated.

PAGE 376

376 Figure 7 4. Three dimensional view of the LMA source lo cations beginning at 19:33:19.900 (UT) associated with the initial UPL and subsequent IS branches of flash UF 11 24 on August 5, 2011. Sources correlated in time with precursor current pulses and diffuse sources not obviously corresponding to extensions o f previous IS branches have been removed. Branches are labeled in order of increasing initiation time. The sources span 500 ms and are color coded in time according to the key at right in 50 ms time windows. The launching facility is labeled.

PAGE 377

377 Figure 7 5. LMA source altitude locations (black points) for flash UF 11 24 on August 5, 2011 overlaid on a 910 ms window of the measured channel base current (II Low measurement). The final 485 ms of the wire ascent and the full 425 ms duration of the UPL/ICC a re shown. Sources are annotated corresponding to precursor current pulses. Red diamonds superimposed on the current waveform indicate the times of the four IS branches shown in Figure 7 4.

PAGE 378

378 Figure 7 6. RHI scans taken by the SMART radar at the tim e of flash UF 11 24 on August 5, 2011 A) dBZ image with northing projection of LMA source locations overlaid B) Z DR image with northing projection of LMA source locations overlaid. LMA sources within the plane of the RHI scan are colored black and sou rces outside the plane are colored dark green. Images are courtesy of Dr. Michael Biggerstaff and Mr. Pat Hyland of the University of Oklahoma. A B

PAGE 379

379 Figure 7 7. Histogram (bin size equal to 30 sources) of the altitude distribution of LMA sources of flash UF 11 24 on August 5, 2011.

PAGE 380

380 Figure 7 8. Three dimensional plot of the LMA source locations for flash UF 11 25 on August 5, 2011. Sources span 4 s beginning at 19:43:30.777 (UT) and are color coded according to the key at right, each color correspo nding to a 400 ms time window. The location of the launching facility is annotated.

PAGE 381

381 Figure 7 9. Four projection views of the LMA source locations for flash UF 11 25 on August 5, 2011 A) an easting versus altitude plot (view looking due nort h), B) a northing versus altitude plot (view looking due west), C) an altitude versus time plot, and D) an easting versus northing plot (plan view, launching facility annotated by large red circle). The sources span 4 s in time beginning at 19:43:30.377 (UT) and are color coded according to the key at bottom right, each color corresponding to a 400 ms time window. The altitude of the 0 C level is annotated.

PAGE 382

382 Figure 7 10. Three dimensional view of the LMA source locations beginning at 19:43:34.377 (UT) associated with the initial UPL and subsequent IS branches of flash UF 11 25 on August 5, 2011. Sources correlated in time with precursor current pulses and diffuse sources not obviously corresponding to extensions of previous IS branches have been r emoved. Branches are labeled in order of increasing initiation time. LMA sources span 120 ms and are color coded in time according to the key at right in 12 ms time windows. The launching facility is labeled.

PAGE 383

383 Figure 7 11. LMA source altitude locatio ns (black points) for flash UF 11 25 on August 5, 2011 overlaid on a 590 ms window of the measured channel base current (II Very Low measurement). The final 186 ms of the wire ascent and the full 404 ms duration of the UPL/ICC are shown. Sources are anno tated corresponding to precursor current pulses. Red diamonds superimposed on the current waveform indicate the times of the six IS branches shown in Figure 7 10.

PAGE 384

384 Figure 7 12. RHI scans taken by the SMART radar at the time of flash UF 11 25 on Aug ust 5, 2011 A) dBZ image with northing projection o f LMA source locations overlaid. B) Z DR image with northing projection of LMA source locations overlaid. LMA sources within the plane of the RHI scan are colored black and sources outside the plane are colored dark green. Images are courtesy of Dr. Michael Biggerstaff and Mr. Pat Hyland of the University of Oklahoma. A B

PAGE 385

385 Figure 7 13. Histogram (bin size equal to 30 sources) of the altitude distribution of LMA sources of flash UF 11 25 on August 5, 20 11.

PAGE 386

386 Figure 7 14. LMA (red squares) and HBM dE/dt TOA (blue circles) easting coordinates (longitudinal) for 33 commonly located precursor current pulses during the wire ascent of flash UF 11 25 on August 5, 2011. The plot spans 4 s and time increases f rom left to right. The black diamonds represent the difference between the LMA and dE/dt easting coordinates.

PAGE 387

387 Figure 7 15. LMA (red squares) and HBM dE/dt TOA (blue circles) northing coordinates (latitudinal) for 33 commonly located precursor curre nt pulses during the wire ascent of flash UF 11 25 on August 5, 2011. The plot spans 4 s and time increases from left to right. The black diamonds represent the difference between the LMA and dE/dt northing coordinates.

PAGE 388

388 Figure 7 16. LMA (red squar es) and HBM dE/dt TOA (blue circles) altitude coordinates for 33 commonly located precursor current pulses during the wire ascent of flash UF 11 25 on August 5, 2011. The plot spans 4 s and time increases from left to right. The black diamonds represent the difference between the LMA and dE/dt altitude coordinates.

PAGE 389

389 Figure 7 17. Three dimensional plot of the LMA source locations for flash UF 11 26 on August 5, 2011. Sources span 1 s beginning at 19:49:58 (UT) and are color coded according to the ke y at right, each color corresponding to a 100 ms time window. The location of the launching facility is annotated.

PAGE 390

390 Figure 7 18. Four projection views of the LMA source locations for flash UF 11 26 on August 5, 2011 A) an easting versus altit ude plot (view looking due north), B) a northing versus altitude plot (view looking due west), C) an altitude versus time plot, and D) an easting versus northing plot (plan view, launching facility annotated by large red circle). The sources span 1 s in time beginning at 19:49:58 (UT) and are color coded according to the key at bottom right, each color corresponding to a 100 ms time window. The altitude of the 0 C level is annotated.

PAGE 391

391 Figure 7 19. Three dimensional view of the LMA source locations beginning at 19:49:58.507 (UT) associated with the initial UPL and subsequent IS branches of flash UF 11 26 on August 5, 2011. Sources correlated in time with precursor current pulses and diffuse sources not obviously corresponding to extensions of previo us IS branches have been removed. Branches are labeled in order of increasing initiation time. The LMA sources span 100 ms and are color coded in time according to the key at right in 10 ms time windows. The launching facility is labeled.

PAGE 392

392 Figure 7 20. LMA source altitude locations (black points) of flash UF 11 26 overlaid on a 600 ms waveform of the measured channel base current (II Low measurement). The final 120 ms of the wire ascent and full 433 ms of the UPL/ICC are shown. The time of the curren t polarity reversal measured at ground is annotated by a dotted vertical line. LMA sources corresponding to the upward negative leader between the inferred main negative charge region and the lower portion of the upper positive charge region are shown in red color. The baseline current level is marked with a dotted horizontal line for reference.

PAGE 393

393 Figure 7 21. A 35 ms window of the channel base current waveform (II Low measurement) shown in Figure 7 20 from 110 145 ms. Red diamonds superimposed on th e current waveform indicate the times of the two IS branches of flash UF 11 26 shown in Figure 7 19. The ICV (Type II) is annotated.

PAGE 394

394 Figure 7 22. A 75 ms window of the channel base current waveform (II Low measurement) shown in Figure 7 20 from 225 300 ms during the current polarity reversal of flash UF 11 26. The inset shows a 1 s window around a dE/dt pulse corresponding to the initial LMA source of the upward negative leader. The arrow indicates the time of the dE/dt pulse relative to the curre nt waveform. The current baseline level is annotated with a dotted horizontal line for reference.

PAGE 395

395 Figure 7 23. Electric field and channel base current waveforms of flash UF 11 25. A ) a 300 ms waveform of the electric field measured by the inverted e lectric field antenna at Station 12, and B ) a 300 ms waveform the channel base current (II Low measurement), including the polarity reversal. The time scale corresponds to times from 100 400 ms in Figure 7 20. The time of the current polarity reversal is annotated with a dotted vertical lines. Note the prominent hump in the electric field that begins about 5 ms prior to the current polarity reversal.

PAGE 396

396 Figure 7 24. RHI scans taken by the SMART radar at the time of f lash UF 11 26 on August 5, 2011. A) dBZ image with northing projection of LMA source locations overlaid B) Z DR image with northing projection of LMA source locations overlaid. LMA sources within the plane of the RHI scan are colored black and sources outside the plane are colored dark green. Images are courtesy of Dr. Michael Biggerstaff and Mr. Pat Hyland of the University of Oklahoma. A B

PAGE 397

397 Figure 7 25. Histogram (bin size equal to 30 sources) of the altitude distribution of LMA sources of flash UF 11 26 on August 5, 2011.

PAGE 398

398 Figure 7 26. Three dimensional plot of the LMA source locations for flash UF 11 32 on August 18, 2011. Sources span 2 s beginning at 20:37:28.800 (UT) and are color coded according to the key at right, each color corresponding to a 200 ms time window. The loc ation of the launching facility is annotated.

PAGE 399

399 Figure 7 27. Four projection views of the LMA source locations for flash UF 11 32 on August 18, 2011 A) an easting versus altitude plot (view looking due north), B) a northing versus altitude plo t (view looking due west), C) an altitude versus time plot, and D) an easting versus northing plot (plan view, launching facility annotated by large red circle). The sources span 2 s in time beginning at 20:37:28.800 (UT) and are color coded according to the key at bottom right, each color corresponding to a 200 ms time window.

PAGE 400

400 Figure 7 28. Three dimensional view of the LMA source locations beginning at 20:37:29.870 (UT) associated with the initial UPL and subsequent IS branches of flash UF 11 32 o n August 18, 2011. Sources correlated in time with precursor current pulses and diffuse sources not obviously corresponding to extensions of previous IS branches have been removed. Branches are labeled in order of increasing initiation time. The LMA sour ces span 200 ms and are color coded in time according to the key at right in 20 ms time windows. The launching facility is labeled.

PAGE 401

401 Figure 7 29. LMA source altitude locations (black points) for flash UF 11 32 on August 18, 2011 overlaid on a 200 ms w indow of the measured channel base current. The final 67 ms of the wire ascent and initial 133 ms of the UPL/ICC (total duration of 945 ms) are shown. Sources are annotated corresponding to precursor current pulses and to the reconnection processes follo wing the explosion of the triggering wire. Red diamonds superimposed on the current waveform indicate the times of the six IS branches shown in Figure 7 28.

PAGE 402

402 Figure 7 30. Histogram (bin size equal to 30 sources) of the altitude distribution of LMA so urces of flash UF 11 32 on August 18, 2011.

PAGE 403

403 Figure 7 31. Four projection views of the LMA source locations for flash UF 11 32 on August 18, 2011 A) an easting versus altitude plot (view looking due north), B) a northing versus altitude plot (view loo king due west), C) an altitude versus time plot, and D) an easting versus northing plot (plan view, launching facility annotated by large red circle). The sources span 5 s in time beginning at 20:37:27 (UT) and are color coded according to the key at bott om right, each color corresponding to a 500 ms time window. The time region shown includes the intracloud discharge that occurred simultaneously with the rocket launch in addition to the full duration of flash UF 11 32 (including subsequent return strokes ).

PAGE 404

404 Figure 7 32. Three dimensional plot of the LMA source locations for flash UF 11 11 on June 23, 2011. Sources span 2 s beginning at 19:06:25.795 (UT) and are color coded according to the key at right, each color corresponding to a 200 ms time wind ow. The location of the launching facility is annotated.

PAGE 405

405 Figure 7 33. Four projection views of the LMA source locations for flash UF 11 11 on June 23, 2011 A) an easting versus altitude plot (view looking due north), B) a northing versus altitude pl ot (view looking due west), C) an altitude versus time plot, and D) an easting versus northing plot (plan view, launching facility annotated by large red circle). The sources span 2 s in time beginning at 19:06:25.795 (UT) and are color coded according to the key at bottom right, each color corresponding to a 200 ms time window.

PAGE 406

406 Figure 7 34. Three dimensional plot of the LMA source locations for flash UF 11 28 on August 12, 2011. Sources span 2 s beginning at 23:39:19.204 (UT) and are color coded ac cording to the key at right, each color corresponding to a 200 ms time window. The location of the launching facility is annotated.

PAGE 407

407 Figure 7 35. Four projection views of the LMA source locations for flash UF 11 28 on August 12, 2011 A) an ea sting versus altitude plot (view looking due north), B) a northing versus altitude plot (view looking due west), C) an altitude versus time plot, and D) an easting versus northing plot (plan view, launching facility annotated by large red circle). The sou rces span 2 s in time beginning at 19:06:25.795 (UT) and are color coded according to the key at bottom right, each color corresponding to a 200 ms time window.

PAGE 408

408 Figure 7 36. Three dimensional plot of the LMA source locations for flash UF 11 33 on Augus t 18, 2011. Sources span 2 s beginning at 20:45:11.507 (UT) and are color coded according to the key at right, each color corresponding to a 200 ms time window. The location of the launching facility is annotated.

PAGE 409

409 Figure 7 37. Four projection views of the LMA source locations for fl ash UF 11 33 on August 18, 2011. A) an easting versus altitude plot (view looking due north), B) a northing versus altitude plot (view looking due west), C) an altitude versus time plot, and D) an easting versus no rthing plot (plan view, launching facility annotated by large red circle). The sources span 2 s in time beginning at 20:45:11.507 (UT) and are color coded according to the key at bottom right, each color corresponding to a 200 ms time window. LMA sources associated with the intracloud discharge that occurred 10 km to the southeast of the ICLRT during the wire ascent are annotated.

PAGE 410

410 Figure 7 38. Three dimensional plot of the LMA source locations for flash UF 11 34 on August 18, 2011. Sources span 1 s beginning at 20:51:41.070 (UT) and are color coded according to the key at right, each color corresponding to a 100 ms time window. The location of the launching facility is annotated.

PAGE 411

411 Figure 7 39. Four projection views of the LMA source loca tions for flash UF 11 34 on August 18, 2011 A) an easting versus altitude plot (view looking due north), B) a northing versus altitude plot (view looking due west), C) an altitude versus time plot, and D) an easting versus northing plot (plan view, launc hing facility annotated by large red circle). The sources span 1 s in time beginning at 20:51:41.070 (UT) and are color coded according to the key at bottom right, each color corresponding to a 100 ms time window.

PAGE 412

412 Figure 7 40. Three dimensional plot of the LMA source locations for flash UF 11 35 on August 18, 2011. Sources span 1 s beginning at 20:58.10.637 (UT) and are color coded according to the key at right, each color corresponding to a 100 ms time window. The location of the launching facilit y is annotated.

PAGE 413

413 Figure 7 41. Four projection views of the LMA source locations for fl ash UF 11 35 on August 18, 2011. A) an easting versus altitude plot (view looking due north), B) a northing versus altitude plot (view looking due west), C) an altitude versus time plot, and D) an easting versus northing plot (plan view, launching facility annotated by large red circle). The sources span 1 s in time beginning at 20:58:10.637 (UT) and are color coded according to the key at bottom right, each color corresponding to a 100 ms time window.

PAGE 414

414 Table 7 2. At left, measured channel base current parameters for the duration of the UPL/ICC. At middle, three dimensional UPL length from the first detected LMA source of the UPL to the first branch loc ation. At right, three dimensional IS channel branching statistics. a Negative charge transport to ground only, although this discharge exhibited a current polarity reversal of 57 ms during the I CC. b Prior to branching, the initial UPL of flash UF 11 24 p ropagated vertically for about 5 km, then turned horizontal and propagated eastward for over 6 km. Total UPL/ICC Current Parameters Initial UPL Length & Avg. Speed Prior to Channel Branching IS Branch Geometry & Correlated Current Date Shot Charge Transfer (C) Duration (ms) Avg. C urrent (A) Length (km) Avg. Speed (10 4 m/s) # of IS Branches Initiation Altitudes (km) Lengths (km) Avg. Speeds (10 4 m/s) Initiation Currents (A) 062311 UF 11 11 22 345 63 3.7 13.5 0 080511 UF 11 24 46 425 107 12.6 b 8.4 4 5.2 5.4 5.2 7.2 1.7 3.3 90 125 080511 UF 11 25 28 404 70 0.15 7.9 6 0.75 2.7 1.2 2.4 2.8 7.9 12 60 080511 UF 11 26 120 a 433 328 a 0.07 3.8 2 0.58 0.71 2.8 4.1 4.7 18 9 43 081211 UF 11 28 136 694 196 2.5 9.3 3 2.1 2.4 0.81 3.1 4.0 9.3 29 42 0 81811 UF 11 32 225 945 236 0.63 9.5 6 0.87 4.9 0.51 14.7 2.0 11.9 11 156 081811 UF 11 33 110 630 219 0.81 7.5 6 1.1 5.0 0.82 11.3 4.0 7.5 24 183 081811 UF 11 34 120 567 211 5.0 11.8 3 4.4 4.7 5.0 13.7 2.0 14.8 193 203 081811 UF 1 1 35 128 726 176 3.7 6.6 2 4.0 4.3 2.1 3.1 3.5 5.9 104 169

PAGE 415

415 Figure 7 42. Histograms (bin size equal to 30 sources) of the altitude distributions of LMA sources for nine triggered lightning flas hes during summer 2011. A) UF 11 11, B) UF 11 24, C) UF 11 25, D) UF 11 26, E) UF 11 28, F) UF 11 32, G) UF 11 33, H) UF 11 34, and I) UF 11 35. Source altitudes below 8 km are plotted. The histograms are derived from sources extending only from the time o f the rocket launch through the end of the IS period as determined by the current terminating at ground. The peaks of the histograms are indicative of the altitude region where the IS transitioned from vertical to more horizontal propagation, often along the contour of the 0 C level.

PAGE 416

416 Figure 7 43. Cumulative LMA source altitude distribution for the nine triggered lightning flashes.

PAGE 417

417 CHAPTER 8 PROPAGATION CHARACTE RISTICS AND ATTACHME NT PROCESSES OF DART STEPPED LEADERS IN T RIGGERED AND NATURAL LIGHN TING VIA DE/DT AND X RAY TOA MEASUREMENTS In this chapter, the following data and analyses are presented, 1) dE/dt and energetic radiation (x ray) waveforms of four dart stepped leaders associated with triggered lightning return strokes recorded at the ICL RT during summer 2011, 2) dE/dt and energetic radiation (x ray) waveforms associated with a dart stepped leader preceding the second return stroke of a natural lightning flash that terminated within the ICLRT measurement network during summer 2011, 3) dE/d t TOA measurements with emphasis on the time period within 20 s of the return stroke and the attachment process to the intercepting wire ring (four triggered lightning dart stepped leader events) or to ground (one natural lightning dart stepped leader eve nt), 4) x ray TOA measurements of two triggered lightning dart stepped leaders and one natural lightning dart stepped leader with spatial and temporal comparison to the corresponding dE/dt TOA measurements of the causative leader stepping processes, 5) ana lysis of the correlated channel base current and dE/dt TOA measurements for four dart stepped leader events associated with triggered lightning return strokes and the implications on the mechanisms of the attachment process, 6) timing comparison of the mea sured dE/dt and the numerical derivative of the channel base current (dI/dt) and the implications on the mechanisms of the attachment process. In Section 8.1, a discussion and literature review are presented of the characteristics of dart stepped leaders in both triggered and natural lightning. A similar discussion and literature review of the lightning attachment process follows. The data and analyses listed above are presented for each triggered lightning dart stepped leader event in chronological orde r in Sections 8.2.1 through 8.2.4. Similar data and analysis are presented in Section 8.2.5 for the

PAGE 418

418 natural lightning dart stepped leader. The results of the analyses are discussed in Section 8.3 and compared with prior studies. 8.1 Background and Lite rature Review Dart stepped leaders occur preceding both natural subsequent return strokes and triggered lightning return strokes. Compared to stepped leaders preceding natural first strokes, dart stepped leaders propagate about an order of magnitude fast er, and usually have shorter step lengths and interstep intervals. Dart stepped leaders propagate along the same general path as a previously conditioned channel. In the case of natural negative cloud to ground lightning, this path to ground consists of the remnants of the channel following the stepped leader preceding the first stroke, and at times, one or more subsequent leader/return strokes sequences. In triggered lightning discharges, dart stepped leaders follow the general path to ground of the con ditioned channel established by the IS process. In natural lightning, dart stepped leaders occur prior to the second return stroke over five times more frequently than all higher order strokes combined [e.g., Rakov and Uman, 2003]. Analogous to this obse rvation for natural second strokes, in Chapter 5, it was shown that triggered lightning dart stepped leaders recorded at the ICLRT between 2009 2011 occurred preceding the first return stroke following the IS process in 6 out of 19 cases (32%). Dart stepp ed leaders may initiate in the cloud as dart leaders, and upon encountering a more poorly conditioned channel, transition to a stepped manner of propagation. The salient characteristics of dart stepped leaders satisfy the ideal set of criteria for exami ning both the leader step formation process and the attachment process to ground. Unlike stepped leaders, dart stepped leaders typically propagate in a single channel to ground without branching. As a result, the dE/dt waveforms recorded for dart stepped leaders are, in comparison to stepped leaders, relatively easy to interpret due to the lack of superposition of pulses from simultaneously propagating, multiply branching leader channels. Likewise, the energetic

PAGE 419

419 radiation (x ray) bursts associated with d iscrete dart stepped leader steps are also easy to identify. Individual dart stepped leader steps are typically clearly defined, providing a reliable measure of interstep interval. In addition, dart stepped leader step waveforms observed at multiple dE/d t stations are cross correlated and aligned with very minimal ambiguity for processing using the TOA technique described in Chapter 3. Low altitude TOA measurements of dart stepped leaders are highly accurate for triggered lightning events where the leade r descends centrally within the network of sensors. For the measurements discussed in this chapter, a dE/dt antenna was placed within 27 m of the triggered lightning channel. Though the dart stepped leader step formation process is likely not a direct pr oxy for the formation of leader steps in virgin air, the mechanisms have been shown to share significant similarities with both long negative laboratory sparks [e.g., Biagi et al., 2010] and stepped leader steps [e.g., Hill et al., 2011]. In the case of triggered lightning dart stepped leaders, the current measured at ground in response to the descending leader provides significant insight to the sequence of events that occur during the attachment process, particularly when coupled with TOA measurements of dE/dt pulses likely radiated from the leader tip. Channel base currents are also measured preceding typical dart leaders in triggered lightning, though dart leaders usually do not exhibit a pulse structure that can be analyzed using the TOA technique. Characteristics of natural and triggered lightning dart stepped leaders have traditionally been measured using optical techniques (streak photography, photo diode arrays, and more recently, high speed digital photography). Schonland et al. [1956] used a streak camera to calculate the average speeds of six natural dart stepped leaders to range from 0.5 to 1.7 x 10 6 m/s with typical step lengths of about 10 m and typical interstep intervals of about 10 s. Orville and Idone [1982] also used streak photog raphy to record four natural dart stepped leaders. They

PAGE 420

420 reported average overall propagation speeds from 2.1 to 4.6 x 10 6 m/s, typical step lengths from 10 to 20 m, and typical interstep intervals of the order of 4 to 10 s. From electric field records, Krider et al. [1977] found that dart stepped leaders have interstep intervals of 6.5 s and 7.8 s within 200 s of the return stroke in Florida and Arizona, respectively. Davis [1999] used dE/dt TOA measurements to estimate the speeds of seven natural da rt stepped leaders to be 3.5 x 10 6 m/s within 1 km of the ground. Davis [1999] also reported average dart stepped leader interstep intervals to be 4.1 s near ground. The statistics for dart stepped leader propagation speed, step length, and interstep intervals for dart stepped leaders preceding triggered lightning return strokes are in good agreement with those statistics reported above for natural dart stepped leaders. Wang et al. [1999] used the ALPS photodiode system to measure the speed of a dart stepped leader preceding a triggered lightning return stroke to increase from 2 x 10 6 m/s to 8 x 10 6 m/s as the leader descended from 200 m to 40 m in altitude. Biagi et al. [2010] used video frames recorded with a Photron SA1.1 high speed camera to measu re the speed of a dart stepped leader preceding a triggered lightning flash at the ICLRT (flash UF 09 25 on June 29, 2009) to be between 2.7 x 10 6 m/s and 3.1 x 10 6 m/s. Images of the dart stepped leader photographed by Biagi et al. [2010] were shown in F igure 5 2. Howard et al. [2010] estimated the speed of a dart stepped leader preceding a triggered lightning return stroke at the ICLRT (flash UF 07 07 on July 31, 2007) to be 4.8 x 10 6 m/s using dE/dt TOA source locations. Finally, Idone and Orville [19 84] used streak photographic techniques to estimate the steps lengths and interstep intervals of two dart stepped leaders in triggered lightning discharges in New Mexico to be from 5 to 10 m and 2 to 8 s, respectively.

PAGE 421

421 The electric and magnetic field w aveforms and their derivatives of natural first and subsequent strokes have been studied at distances of typically several tens of kilometers with the signal propagation path primarily over seawater to minimize propagation effects (attenuation of higher fr equency signal components) of the radiation field from signal propagation over poorly conducting ground. Such measurements have been described by Weidman and Krider [1978], Cooray and Lundquist [1982], and Murray et al. [2005]. Similar electric and magne tic field measurements of natural first strokes were reported by Jerauld et al. [2008] for 18 natural lightning discharges that terminated within or near the ICLRT measurement network. The measurements described in Jerauld et al. [2008] were recorded with in 1 km of the discharges and the signal propagation paths were entirely over land. The above measurements, relatively independent of range, demonstrated a common sequence of waveform characteristics within 10 s of the return stroke. This sequence of pr ocesses has been generally characterized as the Weidman and Krider [1978], who measured electric field waveforms over some tens of kilometers of seawater, noted that the electric field demonstrated a gradual rise prior to the return stroke with typical durations of 1 8 s (mean of about 4 s) followed by a fast rise to peak with duration of 200 ns or less. The gradual rise (the slow front) typically reached a value 40 50% of the electric field peak due to the following return stroke. Weidman and Krider [1978] reported similar electric field measurements of the slow front and fast transition sequence for subsequent strokes initiated by dart and dart stepped leaders. For strokes initiated by dart leaders, they found that the slow front had shorter duration (typically 600 900 ns) and rose to an average value of about 20% of the return stroke electric field peak. Strokes preceded by dart stepped leaders were found to exhibit typical slow front durations of about 2.1 s (be tween slow front durations observed for stepped leaders and dart leaders) with

PAGE 422

422 peak amplitude ratios comparable to first strokes. The fast transition amplitudes for first and subsequent strokes were found to be similar. Similar slow front and fast transi tion sequences have also been observed in the measured current waveforms of direct lightning strikes to tall towers [e.g., Berger et al ., 1975; Eriksson, 1978; Visacro et al ., 2004]. In this study, the attachment process is considered to begin following the final dart stepped leader step, and includes all interactions between the streamer zones of downward and upward propagating leaders in addition to the major subsequent connections that result in significant current flow. Efforts to model the return s troke and later portion of the attachment process (several microseconds prior to the return stroke) have traditionally used transmission line (TL) models that relate a distribution of current elements along a typically vertical conducting channel (the ligh tning channel) to the radiated electric and magnetic fields observed at an observation point on the ground surface. The original single wave TL model was proposed by Uman and McLain [1969]. The single wave TL model predicts the radiated electromagnetic f ields by injecting a propagating current wave into the bottom of a channel with the assumption that the current wave propagates upward without attenuation or distortion. The electromagnetic fields are calculated assuming propagation over perfectly conduct ing ground. The source of the current waveform can be a measured channel base current or a numerically generated waveform representative of a lightning current waveform. Rakov and Dulzon [1987] later introduced an adaptation of single wave TL model with a linearly decaying current as a function of altitude. Nucci et al. [1988] introduced a second adaptation of the single wave TL model with an exponentially decreasing current as a function of altitude. The single wave TL model can be easily modified to e xpress the predicted electric and magnetic fields in terms of the line charge density of each longitudinal current element (instead of the current itself) by the continuity

PAGE 423

423 equation [e.g., Thottappillil et al., 1997]. A unique feature of the single wave T L model is that it predicts that the distant radiation field components of the electric and magnetic fields [e.g., Uman et al., 1975], assuming propagation over perfectly conducting ground (no attenuation of high frequency signal components), have the same shape as the current waveform with appropriate amplitude scaling factors. The physical mechanism that radiates the observed slow front process in natural first and subsequent strokes has long been a topic of discussion among researchers. An original h ypothesis proposed by Berger and Vogelsanger [1969] suggested that the slow front for lightning discharges to tall towers was a result of an upward connecting positive discharge initiated in response to the downward negative leader. Their hypothesis was d erived from the observation that natural positive lightning strikes to tall towers often demonstrated long duration, gradually rising current that was correlated in time with optical observations of long upward leaders. In testing the suggestion of Berger and Vogelsanger [1969], Weidman and Krider [1978] utilized a single wave (TL) model to show that a single upward connecting leader could not accurately reproduce the slow front amplitude in relation to the subsequent field peak. As input to the TL model, they used a current wave with both exponentially increasing amplitude and velocity as a function of altitude (both functions having the same time constant). The rationalization for the current wave characteristics was derived from the observation reporte d by Wagner [1960] that upward streamers in long laboratory sparks demonstrated increasing velocity with time. With reasonable upward leader lengths (constrained to 30 m) and final current amplitudes (constrained to 10 kA), Weidman and Krider [1978] calcu lated electric fields that were similar in wave shape to the measured quantities, but with peaks that were typically much smaller than the measured 5 7 V/m field peaks (normalized to 100 km). As a physically

PAGE 424

424 plausible alternative, they hypothesize that th e slow front could be the result of multiple simultaneously propagating upward leaders. Additional attempts to model the slow front process have been conducted by Thottappillil and Uman [1993] and Cooray et al. [2004]. The attachment process of trigger ed lightning return strokes at the ICLRT have been previously studied by Jerauld et al. [2007] and Howard et al. [2010]. Jerauld et al. [2007], based on the unsuccessful results of previous researchers to model the slow front with a single wave TL model, extended the approach to utilize a two wave TL model in which the current wave injection point is elevated above ground. Jerauld et al. [2007] used a triggered lightning channel base current waveform as input to the model. The current waveform was record ed during the third return stroke of a triggered flash that demonstrated a slow front and fast transition sequence similar to those observed in association with natural first strokes. Electric and magnetic field waveforms were recorded for the event at di stances from the termination point of 15 m and 30 m. Jerauld et al. [2007] describe the slow front as being radiated by a pair of microsecond time scale current waves initiated from the elevated junction point that propagate simultaneously away from the j unction point as the downward propagating, negatively charged stepped leader and the upward propagating, positively charged leader meet. The bi directional current waves each have amplitude of the order of some tens of kilo amperes. Jerauld et al. [2007] describe the fast tr ansition as being radiated by the same bi directional current wave that establishes the full connection between the stepped leader and the upward connecting leader. The measured dE/dt and dB/dt waveforms at 15 m and 30 m were well rep roduced using a downward propagating wave speed of 1.55 x 10 8 m/s and an upward propagating wave speed of 0.95 x 10 8 m/s. From video records that showed multiple loops in the channel above the strike object, the junction point was determined to be at an a ltitude of 6.5 m. The results obtained by

PAGE 425

425 Jerauld et al. [2007] can also be extended to subsequent strokes, which as stated previously, exhibit slow front durations that are shorter than those radiated by first strokes (difference of about a factor of fiv e) and smaller slow front peak amplitude ratios that those radiated by first strokes (difference of about a factor of two). Considering the sources of the slow front are the charges on the descending negative and ascending positive leaders, and that typic al subsequent leaders have about a factor of five lower line charge density than do stepped leaders, it is not surprising that subsequent strokes demonstrate shorter duration slow fronts with smaller peak amplitude ratios. Subsequent strokes are associate d with shorter upward connecting leader than first strokes as a result of the lower charge density and faster propagation speed of the descending leader. Howard et al. [2010] extended the work of Jerauld et al. [2007] by analyzing the attachment process through the use of dE/dt TOA source locations. The locations were obtained from the original 8 station TOA network at the ICLRT that was constructed at the beginning of summer 2006 by Dr. Joseph Howard and the author. Howard et al. [2010] analyzed the le ader and post leader phases of three stepped leaders preceding natural first strokes and one dart stepped leader preceding a triggered lightning return stroke. For both stepped leaders and the triggered lightning dart stepped leader, Howard et al. [2010] identified pulse characteristics within several microseconds of the return stroke that were different than the preceding leader slow front period in the dE/dt waveforms. The TOA locations of the leader burst pulses typically descended rapidly (and in at least one case, traversed a large horizontal distance), often at speeds more than order of magnitude faster than the average propagation speed of the preceding leader. In all events analyzed, Howard et al. [2010] also found that the leader burst was a very strong

PAGE 426

426 emitter of energetic radiation (x rays). Pulses resembling the leader burst have previously been documented by Murray et al. [2005] and Jerauld et al [2007] immediately preceding the beginning of the slow front, though no physical explanation or mechanism is provided in either study. For distant electric field waveforms, Murray et al. [2005] found that about 57% (75 out of 131 events) exhibited a lea der burst in the interval extending from 4 9 s prior to the fast transition pulse associated with the return stroke. Murray et al. [2005] also identified many pulses in the interval from 1 4 s of the dominant dE/dt pulse associated with the return strok e that could be related to the leader burst process. Wang et al. [2001] also documented a low altitude (about 35 m above ground) pulse of light measured with the ALPS photodiode array that occurred less than 2 s prior to the onset of the slow front in a negative cloud to ground discharge that terminated about 2 km from the measuring station. The electric field pulse radiated by the optically observed feature was also recorded and had risetime of about 0.5 s. At the time the pulse of light was recorded, the upward connecting leader had already extended to an altitude of 88 m at an average speed of 1.7 x 10 6 m/s. The electric field pulse characteristics, the timing of the pulse relative to the initiation of the slow front, and the low altitude of the sou rce suggest that the observed pulse may have been associated with a leader burst process. Hence, the leader burst may emit a distinct optical signature and may also occur following the initiation of the upward connecting leader. Pulses super imposed on the slow front of the dE/dt waveforms were also analyzed by Howard et al. [2010] using the TOA technique. Slow front pulses have also been previously documented by Murray et al. [2005] and Jerauld at al. [2007]. Though no physical explanation is given fo r the slow front pulses, Jerauld et al. [2007] did note that the radiation fields of the slow front pulses were similar in wave shape yet smaller in amplitude compared to the

PAGE 427

427 subsequent fast transition pulses, and that the mechanisms may, in fact, be quite similar. Murray et al. [2005] also found that the distant radiation fields of slow front pulses in their Type B and Type C events demonstrated strong resemblance to the subsequent fast transition pulses associated with the return stroke. The analysis of Howard et al. [2010] did not refute these prior observations. Howard et al. [2010] found that the TOA locations of the slow front pulses occurred at very low altitude and in the same spatial proximity as the TOA location of the fast transition pulse, sugg esting that the slow front pulses may be involved in the interaction between the upward and downward leaders (the mechanism by which the slow front is radiated according to the modeling results of Jerauld et al. [2007]) prior to the full connection being e stablished. Howard et al. [2010] suggests that there may be no physical difference between the slow front and fast transition pulses, and that the defining characteristic is simply emission time. This view is perhaps supported by an observation reported by Howard et al. [2010] for a slow front pulse preceding a rocket triggered lightning return stroke that occurred when the channel base current had already risen to a level of 20 kA. Such a high current value is uncharacteristic of an upward connecting le ader current, and indicates that a connection between the descending dart stepped leader and upward connecting leader already existed. Howard et al. [2008] also reported the first TOA locations of energetic radiation (x rays) associated with both leader stepping processes and post leader processes. Howard et al. [2008] located seven individual x ray events, three from natural flash MSE 06 04 on June 2, 2006 and four from triggered flash UF 07 07 on July 31, 2007. The x rays were measured using a network of eight NaI scintillation detectors (un shielded with lead) that were co located with the eight flat plate dE/dt antennas. While the x ray TOA locations were typically poorly determined (large covariance estimates of the spatial parameters) in compariso n to the TOA locations of the dE/dt

PAGE 428

428 leader pulse associated with the causative step formation, Howard et al. [2008] determined that the x rays were emitted within 50 m of the dE/dt source location. In all but one case, the x ray location was below the loc ation of the related dE/dt pulse. Howard et al. [2008] also reported that the x ray emission time followed the emission time of the dE/dt source by times ranging from 0.1 to 1.3 s. By integrating the dE/dt waveform of the leader step pulse and aligning the pulse with the x ray measurement at the same station, Howard et al. [2008] found that the x rays were being emitted coincident with the electrostatic field change following the dominant step pulse. From the analysis of the step formation process prese nted in Chapter 5 of this dissertation and given by Biagi et al. [2010], the electrostatic field change following the dominant step pulse is likely associated with the movement of charge after the large transient current pulse that establishes the connecti on between the previous leader channel and the space stems/leaders that form within the streamer zone ahead of the previous leader tip. The close proximity of the dE/dt and x ray sources also provides evidence that the leader tip is likely responsible for producing the necessary conditions that allow for the propagation of runaway electrons that generate x ray photons through the bremmstrahlung mechanism, and that the runway electrons traverse relatively short path lengths. The spatial and temporal relati onship of the dE/dt and x ray sources for stepped leader steps preceding a natural first stroke and dart stepped leader steps preceding a triggered lightning stroke were also similar, indicating that the underlying physics of the two discharge processes ar e likely similar. In this chapter, additional information is reported on the sequence of electrical breakdowns that compose the attachment process. The analyses primarily focus on the dE/dt and x ray emissions of four dart stepped leaders preceding trigg ered lightning return strokes. Similar measurements are reported for one natural lightning dart stepped leader event. Leader and post

PAGE 429

429 leader processes evident in the dE/dt data are analyzed using the TOA technique and compared to those discussed in Howar d et al. [2010]. The present study also incorporates sensitive (sub ampere resolution) channel base current measurements during the attachment phase. Such sensitive channel base current measurements were not available in past studies. The dE/dt and chan nel base current measurements were digitized on the HBM digitization system. As a result, there is no timing ambiguity when comparing the measurements directly. Recall that the dE/dt measurements recorded on the HBM digitization system are digitized with an additional six bits of resolution compared to past measurements on the ICLRT DSO network and are essentially free of digitization noise. The dE/dt measurements are also configured with sensitivity a factor of four greater than the measurements used by Howard et al. [2008] and Howard et al. [2010], providing considerably greater resolution of the smaller amplitude field changes during the leader and post leader phase. The only tradeoff of the HBM digitization system is that the fast transition pulses f or triggered lightning dart stepped leader events typically saturate most dE/dt sensors in the network, rendering the computation of the TOA location of the fast transition pulse impossible. At times, the fast transition pulse can be located with the dE/d t measurements digitized on the ICLRT DSO network, which is configured with system sensitivity a factor of two less than the HBM digitization system. Finally, the timing of the peak of the dE/dt fast transition, obtained from an insensitive dE/dt measurem ent that is not part of the TOA network, is compared with the peak of the numerical derivative of the high level channel base current (dI/dt). This measurement provides insight to the possible extent and speed of the upward connecting leader.

PAGE 430

430 8.2 Analysi s of Triggered and Natural Lightning Dart Stepped Leaders In the following sections, four dart stepped leaders that preceded triggered lightning return strokes and one dart stepped that preceded a natural lightning return stroke during summer 2011 are anal yzed in detail. The events are discussed in the following order : 1) in Section 8.2.1, the tenth stroke of flash UF 11 15 on July 7, 2011, 2) in Section 8.2.2, the first stroke of flash UF 11 25 on August 5, 2011, 3) in Section 8.2.3, the second stroke of flash UF 11 32 on August 18, 2011, 4) in Section 8.2.4, the second stroke of flas h UF 11 34 on August 18, 2011, and 5) in Section 8.2.5, t he dart stepped leader preceding the second stroke of natural flash MSE 11 01 on July 7, 2011. 8.2.1 Tenth Return Stro ke of Flash UF 11 15 Flash UF 11 15 was triggered on July 7, 2011 at 19:02:19.528147 (UT). The flash was triggered with quasi static electric field at ground of about 4.8 kV/m and had a total of 11 return strokes. The tenth stroke, with peak current of 10.2 kA, was recorded at 19:02:20.044338 (UT) and was preceded by a dart stepped leader. The tenth stroke occurred following a 22.4 ms inter stroke interval. A total of 49 dE/dt pulses were located using the TOA technique described in Chapter 3. The loc ated pulses occurred within 200 m of the ground and within 80 s of the return stroke. The sustained UPL initiated when the triggering wire ascended to an altitude of 224 m, therefore, all TOA located pulses occurred while the dart stepped leader propagat ed within the height of the exploded triggering wire. A three dimensional plot of the TOA located dE/dt pulses for this dart stepped leader event is shown in Figure 8 1 A The view is looking in a northeasterly direction. The location of the launcher and intercepting wire ring are annotated and the points are color coded according to the key in 8 s time windows. I n Figure 8 1 B the altitude versus time projection of the dE/dt source locations is plotted. A regression line is fitted to the points and is indicative of the average downward leader speed. The average speed of this

PAGE 431

431 dart stepped leader was 2.2 x 10 6 m/s with a correlation coefficient of 0.97. The intercepting wire ring is located at an altitude of about 8.5 m with respect to the local coordi nate system origin, or about 5.5 m with respect to local ground level. A 26 s dE/dt waveform recorded at Station 3, located 138.3 m from the Field (Ground) Launcher is plotted in Figure 8 2. Note the dE/dt waveform is plotted on a truncated amplitude sc ale to better show the dart stepped leader pulses. The measurement saturates at an amplitude of about 8 kV/m/s. The propagation delay has been removed from the dE/dt waveform, assuming the return stroke initiated at an altitude of 10 m directly above th e launcher. dE/dt pulses that were located via the TOA technique are annotated with arrows and numbered with increasing time. The three dimensional spatial an d the emission times (T) of the sources relative to the return stroke are provided in Table 8 1. For reference, the lateral location of the Field (Ground) Launcher in the ICLRT local coordinate system is at position (X,Y) = (308.1 m, 450.9 m). The final column of Table 8 1 provides a designator for each located pulse. In the designator column, "LS" refers to a dart stepped leader step pulse, "LB" refers to a leader burst pulse, "SF" refers to a slow front pulse, and "FT" refers to a fast transition pulse In Figure 8 3 A a 36 s dE/dt waveform measured at Station 3, a distance of 138.3 m from the Field (Ground) Launcher, is plotted versus the II Very Low channel base current measurement. For reference in the following discussion, note that the II Very Low channel base current waveform is also shown in Figure 8 2 (dashed green trace). The dE/dt waveform between 30 s and 2 s in Figure 8 3 A is characterized by bi polar dart stepped leader step pulses that occur at time intervals of about 1 3 s (Pulse s 1 11 in Table 8 1). Significant positive deviations in the measured channel base current are evident beginning at about 15 s

PAGE 432

432 immediately following a leader step with three located pulses (Pulses 2 4 in Table 8 1). The pulses were located at altitudes of 59.5 61.1 m. These positive deviations in the current are consistent with the upward movement of positive charge through the current measurement (shunt). The current returns to a level at or near zero (within the resolution of the measurement) follow ing the leader step at 15 s. The subsequent leader step at about 13 s (Pulse 5), located at an altitude of 61 m, produces a similar positive deviation in the channel base current, though following the step pulse, the current does not decay completely to zero. The leader steps between 10 s and 6 s produce pulses in the channel base current with amplitudes from about 10 20 A super imposed on a background current flow of 1 2 A. Pulse 7 and Pulse 8 during this time interval were radiated from altitud es of 45.9 m and 41.7 m, respectively. The leader step pulses between 6 s and 2 s produce similar positive pulses in the channel base current, and the background current level continues to slowly rise to a level of about 20 A following the large step p ulse at 2 s. The large step pulse at 2 s appears to be associated with the final downward leader step, and occurs at an altitude of 34.1 m. The characteristics of Pulses 12 14 are similar to those classified by Howard et al. [2010] as the leader burs t. These three pulses were located at altitudes of 24.2 m, 20.5 m, and 27.2 m and were emitted in the time span of about 630 ns. Interestingly, the three leader burst pulses, which occurred at similar altitudes, were separated laterally by as much as 11 m (Pulse 13 and Pulse 14). Considering each leader burst pulse is associated with a significant rise in the channel base current, which begins at an amplitude of about 18 A at the time of Pulse 12 and rises to a level of about 60 A following Pulse 14, it is perhaps reasonable that the leader burst pulses are associated with the initial connections of the streamer zones of the downward negative and upward positive leaders, a process termed the "break through" phase by Rakov and Uman [2003]. The dE/dt wavef orm

PAGE 433

433 shown in Figures 8 2 and 8 3 does not exhibit a pronounced slow front following the leader burst, though there does appears to a be a slight rise in the background level coincident with Pulse 15. Pulses 15 18 are classified as slow front pulses in Tab le 8 1 based on their very low source location altitudes, their temporal proximity to the following fast transition, and the coincident increase of the channel base current, findings consistent with Howard et al. [2010] for a dart stepped leader preceding a triggered lightning stroke. Pulse 15 was located at an altitude of 12.6 m and Pulses 17 18 were located in the same spatial location at an altitude of 13.6 m. These pulses were radiated from positions only 4 5 m above the intercepting wire ring. In Fi gure 8 3 B a 10 s window of the dE/dt waveform plotted in Figure 8 3A (propagation time removed as stated previously) is compared with the II High channel base current measurement. Following Pulse 15, the channel base current rose from an amplitude of 60 A to about 440 A at the beginning of the fast transition in the time period of about 1 s. The beginning of the fast transition is annotated in Figure 8 3 B with a dotted vertical line. The channel base current in Figure 8 3 A exhibited steady flow for a bout 14.4 s prior to the beginning of the fast transition with super imposed pulses coincident with each dE/dt leader step pulse with amplitudes of the order of 10 20 A. A logical question is whether the steady current flow and super imposed current puls es are associated with a propagating upward connecting leader or leaders, and if so, what are the properties of the discharge(s) (length, duration, speed, etc.). To help answer this question, an analysis was performed of the timing relationship between t he peak of the dE/dt fast transition pulse and the associated derivative of the current (dI/dt) measured at ground. Based on the two wave model of the attachment process proposed by Jerauld et al. [2007], in which current waves propagate bi directionally away from a junction point above the ground surface, one would expect the peak dI/dt measured at ground

PAGE 434

434 to lag the peak dE/dt radiated from the junction point by the distance from junction point to the current measurement divided by the propagation speed o f the downward current wave. As previously stated, Jerauld et al. [2007] modeled the attachment process using a downward current wave propagation speed of 1.55 x 10 8 m/s. In Figure 8 4, a dE/dt waveform measured 109.9 m from the Field (Ground) Launcher i s plotted against the numerical derivative of the II High channel base current waveform. The amplitudes of each waveform have been normalized to their maximum values to enable easy timing comparison of the respective peaks. The dE/dt waveform was obtaine d from an independent sensor provided by Dr. Carlos Mata and is configured with less sensitivity than the dE/dt antennas that comprise the TOA network in order to resolve the fast transition peak. The dE/dt waveform is shifted to account for propagation d elay from an assumed junction altitude of 10 m. The numerical dI/dt was calculated using a current waveform that was bandwidth limited to 8 MHz by the 1 m current shunt. At the beginning of summer 2012, triggered lightning return stroke dI/dt waveforms were directly measured with sensor bandwidth of 25 MHz (the bandwidth of the dE/dt measurements used in this chapter) and were compared with the numerical dI/dt waveforms derived from the shunt measurement. It was found that the lower bandwidth of the nu merical dI/dt introduced an average of 30 ns worth of time lag in the peak dI/dt compared to the directly measured dI/dt. Another timing consideration when comparing the dE/dt and dI/dt measurements is the attenuation effect on the high frequency radiatio n components of the propagating dE/dt waveform from the relatively poor soil conductivity at the ICLRT, which is about 2.5 x 10 4 S/m according to Rakov et al. [1998]. To test the severity of the propagation effects, dE/dt waveforms from multiple return s trokes during summer 2011 were compared from stations at distances of 109.9 m and 391.1 m from the Field (Ground) Launcher. The waveforms from both

PAGE 435

435 sensors were shifted to account for propagation delay assuming a junction point height of 10 m. It was fou nd that the peak dE/dt of the fast transition pulse from the station at 109.9 m lead the corresponding peak dE/dt measured at 391.1 m by a maximum value of 30 ns, and was often in agreement to within 10 20 ns. The small variations in the observed time lag are likely due to varying degrees of high frequency content (and corresponding attenuation) among the different events. Considering the relatively small timing uncertainty introduced by propagation effects over a ground path length of about 280 m, propag ation effects will not be accounted for in the timing comparison of the dE/dt and dI/dt measurements with the dE/dt measurement recorded for all triggered lightning events at 109.9 m from the Field (Ground) Launcher. In Figure 8 4, the dE/dt (red trace) l eads the numerical dI/dt (blue trace) by seven sample points (70 ns) for the first fast transition peak and five sample points (50 ns) for the second fast transition peak. Taking into account the 30 ns time lag in the numerical dI/dt from the bandwidth li mitation discussed above, the dE/dt leads the dI/dt by 40 ns for the first peak and 20 ns for the second peak. The time lag between the dE/dt and dI/dt measurements establishes an upper bound on the distance between the initiation altitude of the return s troke (the junction point between the upward and downward leaders) and the current measuring device. For this event and all subsequent cases presented in this chapter, the propagation path length from the intercepting wire ring to the current measuring de vice is assumed to be 6 m. The current wave is assumed to propagate at the speed of light over this distance and at a speed of 1.55 x 10 8 m/s [e.g., Jerauld, 2007] in the air above the intercepting wire ring. Using the stated constraints and the maximum time lag between the dE/dt and dI/dt peaks for this event of 40 ns, the junction point could have been, at most, 3.1 m above the intercepting wire ring. From the discussion above, if there are errors in the timing comparison of the directly measured dE/dt and numerical dI/dt due to bandwidth

PAGE 436

436 differences and/or propagation effects, they are expected to be less than 30 ns and would result in an increased time difference between the dE/dt and dI/dt peaks. A worse cast scenario of 30 ns error would increase t he altitude of the junction point by 4.7 m. As previously stated, the lowest altitude slow front pulses (Pulses 15 and 17 18 in Figure 8 2) were located 4 5 m above the wire intercepting ring, and were associated with significant increases in the measured channel base current. Based on these measurements, it is reasonable to assume that an upward connecting leader propagated to an altitude of 3.1 m above the intercepting wire ring, reaching a peak cu rrent of about 440 A (Figure 8 3B ), before it connected with the downward propagating dart stepped leader. Based on an upward leader propagation duration of 14.4 s and a final length of 3.1 m, the upward connecting leader would have propagated at an average speed of about 2.15 x 10 5 m/s. At this point in t he analysis, it is interesting to consider the slow front and fast transition features of the electric field record obtained by numerically integrating the dE/dt waveform measured 109.9 m from the lightning channel base. A 26 s record of the electric fie ld is shown in Figure 8 5 A The corresponding dE/dt record is shown in Figure 8 5 B on a vertical scale designed to show the full extent of the fast transition peaks. While the dE/dt records shown in Figures 8 2 and 8 3 did not show prominent slow fronts, the ele ctric field record shown in Figure 8 5 A demonstrates a gradual rise beginning about 4 s prior to the fast transition. A horizontal red line is plotted for reference in Figure 8 5 A to better show the rising portion of the electric field waveform. Interestingly, the slow front appears to begin at almost exactly the same time the channel base current began to rise significantly, which was also coincident with the final downward leader step (dE/dt Pulse 11 in Table 8 1, which was radiated from an alt itude of 34.1 m). The duration of the slow front in this case is about a factor of two longer than the typical

PAGE 437

437 duration observed by Weidman and Krider [1978] for dart stepped leaders preceding natural subsequent strokes. High speed video frames of the d escending dart stepped leader preceding the tenth stroke of flash UF 11 15 are shown in Figure 8 6. The video frames were acquired by the Photron SA1.1 high speed camera at a frame rate of 300 kfps (3.33 s frame integration). The camera was located 300 m from the Field (Ground) Launcher and viewed an altitude of 141 m (at the location of the launcher) relative to the local coordinate system origin. The camera was equipped with a 14 mm fixed focal length lens set to an aperture of F/4. In order to bette r resolve lower amplitude optical processes, the gamma has been increased in post processing for all images by a factor of two. A total of 18 frames (60 s) are shown with the leader entering the field of view in Frame 1 and the return stroke occurring i n Frame 18. Between Frame 1 and Frame 16 (50 s), the dart stepped leader traversed a total distance of about 110 m, giving an average propagation speed of 2.2 x 10 6 m/s, an identical value with the TOA calculated vertical leader speed of 2.2 x 10 6 m/s. There were no clearly evident space stems/leaders imaged below the propagating dart stepped leader tip in the recorded high speed video frames, though Frame 4 and Frame 6 each contain luminous segments of the order of 1 m in length that may be separated sl ightly from the primary leader tip. If these short luminous segments are indeed space stems/leaders, they are not as well defined as those shown by Biagi et al. [2010] for a triggered lightning dart stepped leader. There is no detectable luminosity above the intercepting wire in the five frames (16.7 s) prior to the return stroke indicative of a propagating upward leader (recall that the upward leader is inferred to have duration of 14.4 s), though in Frame 17, there is a narrow luminous structure immed iately above the intercepting wire ring that appears different from the leader tips in previous frames. It is also worth noting that the leader

PAGE 438

438 propagation speed between Frame 16 and Frame 17 increased relative to the average speed in the prior 50 s. Du e to the 3.3 s frame integration time, it is unclear if Frame 17 contains the initial portion of the return stroke in addition to the final downward movement of the dart stepped leader, a scenario that would give the appearance of increased downward leade r speed. The increased optical blooming effect of the channel below 30 m in Frame 17 might indicate that the initial portion of the return stroke was indeed captured in the frame interval. In Figure 8 7, dE/dt ( Figure 8 7A) and x ray ( Figure 8 7B) wavef orms measured during the leader and attachment phase of the tenth stroke of flash UF 11 15 are shown on a 60 s time scale. The measurements were approximately co located at Station 3, with the dE/dt antenna and plastic scintillator located 138.3 m and 14 2.1 m from the Field (Ground) Launcher, respectively. This particular dart stepped leader was a weak emitte r of x rays. A small number of x ray photons were recorded associated with leader steps corresponding to Pulse 5 (Figure 8 2 and Table 8 1), Pulse 8, and Pulse 11 (the final leader step pulse). A final small burst of x rays appears to be associated with Pulses 16 17, which were classified as slow front pulses. Interestingly, there were no x rays detected in association with the leader burst pulses, a contrasting observation to that of Howard et al. [2010], who found the leader burst process to be the most prolific emitter of x rays during the leader and attachment phases. None of the x ray bursts were observed on a sufficient number of detectors fo r TOA calculations to be performed. The lack of observed x rays associated with leader step and post leader processes for this event is likely due to a combination of the low return stroke peak current (10.2 kA), which is indicative of comparatively low c harge volume per unit length on the lower part of the descending leader, and also the relatively high attenuation of x rays in air (about a 100 m e folding distance). Lower energy x rays may have been emitted at altitude from locations near the descending

PAGE 439

439 leader tip, but experienced sufficient atmospheric attenuation to be undetected by sensors on the ground. 8.2.2 First Return Stroke of Flash UF 11 25 Flash UF 11 25 was the second of three triggered lightning discharges that occurred on August 5, 2011. Detailed LMA observations of the IS process of flash UF 11 25 were presented in Section 7.2.2. The one return stroke of flash UF 11 25 occurred at 19:43:31.4899735 (UT) and had peak current of 12.1 kA. The flash was triggered with quasi static electric field at ground of 4.6 kV/m. The return stroke followed a zero current duration of about 91 ms after the cessation of the ICC. A total of 182 dE/dt sources were TOA located during the final 400 s of the dart stepped leader preceding the return stroke. The highest source occurred at an altitude of 318 m, which was about 132 m below the height of the triggering wire that exploded at an altitude of 450 m. A three dimensional plot of the dE/dt TOA source locations for the dart stepped leader is shown in F igure 8 8 A The plotted sources are color coded according to the key in 40 s time windows. I n Figure 8 8 B the downward leader speed was estimated by fitting a regression line to a plot of the altitude versus time projection of the dE/dt TOA source loca tions. The average downward leader speed was 6.9 x 10 6 m/s with a correlation coefficient of 0.99. In Figure 8 9, a 34 s plot is shown of the dE/dt waveform measured during the dart stepped leader of flash UF 11 25. The dE/dt waveform was measured at Station 3. The propagation delay has been removed from the dE/dt waveform assuming the return stroke initiated at an altitude of 10 m. The dE/dt waveform is plotted on a truncated amplitude scale to better show the fine pulse structure prior to the retur n stroke. dE/dt pulses that were located via the TOA technique are annotated and numbered sequentially with increasing time. The three dimensional spatial locations, emission times, three dimensional location uncertainties, and pulse

PAGE 440

440 designations for eac h labeled pulse in Figure 8 9 are provided in Table 8 2. The pulse designations follow the same convention stated previously in Section 8.2.1. The dart stepped leader preceding the single return stroke of flash UF 11 25 exhibited different characteristic s than the other triggered lightning dart stepped leaders recorded during summer 2011: 1) the interstep intervals of the dart stepped leader of flash UF 11 25 ranged from typical values of about 4 10 s, in good agreement with values reported by Orville a nd Idone [1982] and Krider [1977] for natural lightning dart stepped leaders, whereas interstep intervals for the additional triggered lightning dart stepped leaders in this study are of the order of 1 3 s, and 2) the estimated propagation speed of the da rt stepped leader preceding flash UF 11 25 was at least 68% faster than the remaining dart stepped leaders recorded during this study, and also faster, in general, than previously reported speeds in the literature for both triggered and natural lightning d art stepped leaders. The physical basis of these observed propagation differences is undetermined, though one might infer from the measurements that the channel conditioning following the IS process was unique in some regard. A 50 s plot of the dE/dt ( blue trace) measured at Station 3 is shown in Figure 8 10 A versus the II Very Low channel base current (green trace). The dE/dt waveform has been shifted identically to the waveform in Figure 8 9. Note that the same channel base current waveform is plott ed in Figure 8 9 with a dashed green line. The first significant current pulse in Figure 8 10 A occurs following the leader step at 0 s. The channel base current returns to the zero level following the pulse. About 10 s later, a second current pulse wa s recorded in association with the next dart stepped leader step. The current pulse had amplitude of about 12 A. TOA locations were obtained for six dE/dt pulses during the leader step at 10 s. The first pulse (Pulse 1) occurred at an altitude of 57.7 m. The dominant step pulse (Pulse 5) was radiated from

PAGE 441

441 an altitude of 46 m, and the final pulse (Pulse 6) was located at an altitude of 41.9 m. Following the leader step pulse, the channel base current does not fall to zero, but instead flows steadily at a level of 1 3 A for about the next 7 s. The beginning of the steady current flow is annotated in the top panel of Figure 8 10 with a dotted vertical line. Pulse 7 occurs at about 13.5 s and is located at 45.8 m in altitude, similar to the prior domin ant step pulse. Pulse 7 did not produce a significant deviation in the low amplitude background channel base current, and hence, may not be associated with significant downward movement of negative charge. The next dart stepped leader step occurs at abou t 18 s and is associated with a 10 A current pulse. Pulse 8 and Pulse 10 were located at altitudes of 30.4 m and 34.1 m, respectively. Pulse 9 was located at 45.9 m, again at approximately the same altitude as the dominant step pulse (Pulse 5) from the leader step at 10 s. The physical significance of this observation is unclear. Following the current pulse at 18 s, the channel base current flowed at a level of about 5 A until the next leader step at about 20.5 s. Four dE/dt pulses were located ass ociated with this step with altitudes ranging from 32.7 m (Pulse 12) to 28.8 m (Pulse 14). The step at 20.5 s produced a pulse in the channel base current with peak amplitude of about 30 A. Following the large current pulse, the current resumed to a ste ady amplitude of about 7 A. The final downward leader step occurred at about 26.5 s. The two dE/dt pulses associated with the final step (Pulses 15 and 16) were radiated from altitudes of 27.4 m and 27.3 m, respectively, and produced a current pulse wit h amplitude of about 28 A. A relatively steady current of about 18 A flowed following the final leader step pulse. Pulse 17 was emitted about 1.3 s after Pulse 16, and was located at a similar altitude of 29 m. Pulse 17 appears to produce a sharp pulse in the channel base current with amplitude of about 30 A, but does not result in an elevation of the background current level. A very pronounced leader burst begins at about 28.5 s. The leader burst consists of four large

PAGE 442

442 dE/dt pulses (Pulse 18 21) rad iated in a time span of about 1.35 s. The leader burst pulses were radiated from within a lateral area less than 3 m 2 and occur at altitudes ranging from 27.4 m (Pulse 19) to 21.4 m (Pulse 21). Each leader burst pulse is associated with a significant ri se in the channel base current. The leader burst effectively elevated the background current level from 18 A to about 120 A. The final leader burst pulse (Pulse 21) produced a fast current pulse with peak amplitude of about 150 A. Unlike the dart steppe d leader preceding the tenth stroke of flash UF 11 15 (Section 8.2.1), the dE/dt waveform of this dart stepped leader exhibits a very pronounced slow front which begins about 1.3 s following the final leader burst pulse. TOA locations were obtained for a total of eight slow front pulses (Pulses 22 29). Interestingly, these eight pulses increase in altitude nearly monotonically from altitudes of 14.3 m (Pulse 22) to 56.8 m (Pulse 29). The dE/dt waveform in Figure 8 10 A is plotted on a 24 s time scale in the Figure 8 10 B versus the II High channel base current. The channel base current is observed to rise from 120 A immediately prior to Pulse 22 to about 2190 A at the time of the initial rise of the fast transition pulse, which occurs 100 ns following th e peak of Pulse 29. The time of the initial rise of the dE/dt fast transition pulse and the corresponding current level are annotated in Figure 8 10 B with a dotted vertical line. In Figure 8 9, the steady current was observed to flow for 24 s prior to the initial rise of the fast transition. Similar to the case of flash UF 11 15 presented in Section 8.2.1, the timing relationship of the measured dE/dt and numerical dI/dt can be used to obtain a upper bound for the initiation height of the return stroke and hence, the maximum length of an upward connecting leader. A 3 s record of the measured dE/dt (red trace) and numerical dI/dt (blue trace) are plotted in Figure 8 11. For the two fast transition peaks, the dE/dt leads the dI/dt by 70 ns and 50 ns, respectively, accounting for the 30 ns negative time shift in the dI/dt discussed

PAGE 443

443 above in Section 8.2.1. With the constraints stated in Section 8.2.1 (downward current wave propagation speed in air of 1.55 x 10 8 m/s from the junction height, and speed of light propagation in the 6 m distance between the intercepting wire and the current measuring device), and assuming a maximum current wave propagation time difference between the junction point and the current measuring device of 70 ns, the height of the junction point could be, at most, 7.8 m. With the assumption that the upward leader initiated when the steady current began flowing, the upward leader would have propagated for 24 s at an average speed of 3.25 x 10 5 m/s, reaching a final altitude of 7.8 m above the intercepting wire and a maximum peak current of 2190 A. The first three slow front pulses (Pulses 18 20) were located at altitudes of 14.3 17.3 m, or 5.8 8.8 m above the altitude of the intercepting wire. The calculated upward leader length o f 7.8 m covers the range of the first three slow front pulses. A 20 s dE/dt waveform measured 109.9 m from the lightning channel base is plotted in Figure 8 12 B The time integral of the dE/dt waveform is plotted in Figure 8 12 A In the numerically integrated dE/dt waveform, the slow front is observed to begin about 4.6 s prior to the field change associated with the first fast transition peak, a value about a factor of 2.2 longer than reported by Weidman and Krider [1978]. Unlike the case of the d art stepped leader presented in Section 8.2.1 (flash UF 11 15), the start of the slow front does not correspond to the time of the final downward leader step. In this case, the start of the flow front coincides nearly exactly in time with the first pulse of the leader burst (Pulse 18), which was located at an altitude of 27.4 m. Perhaps not by coincidence, the first pulse of the leader burst is associated with the first large increase in the steady current flow observed at ground (recall that the final do wnward leader step of flash UF 11 15 discussed above produced the first large increase in steady current flow prior to the leader burst). The leader burst pulses in flash UF 11 25 increased the

PAGE 444

444 background current level to 120 A, a factor of two larger tha n the current increase due to the leader burst pulses in flash UF 11 15. Similarly, the slow front pulses in flash UF 11 25 produced large scale increases in the channel base current. The current at the time of the initial rise of the fast transition was a factor of five larger than the same current observed in flash UF 11 15. These combined observations suggest that the leader burst pulses, which were located at altitudes of 21.9 27.4 m (13.4 18.9 m above the intercepting wire), were likely associated w ith the initial interactions of the streamer zones of the downward and upward leaders, and the initial three slow front pulses corresponded to more thorough connections of the two leaders. The physical significance of the final five slow front pulses (Pul ses 25 29), which propagated from altitudes of 33.8 56.8 m, is not clear. The source altitudes of these pulses are above the leader burst pulses and eventually reach altitudes comparable to the first leader step pulse shown in Figure 8 9, which occurred o ver 24 s prior to the return stroke. Unfortunately, there were no high speed video observations of this event with which to compare the prior observations. The x ray emissions of the dart stepped leader preceding the single return stroke of flash UF 11 25 were extremely weak. None of the eight plastic detectors recorded any x ray photons associated with the descending leader. The LaBr 3 detector located 20 m from the lightning channel base recorded two single photons associated with the final two leade r steps (Pulse 13 and Pulse 16 in Figure 8 9 and Table 8 2). No x ray were observed in time correlation with the leader burst or slow front pulses. 8.2.3 Second Stroke of Flash UF 11 32 Flash UF 11 32 was the first of four triggered lightning discharges on August 18, 2011. The flash had two return strokes following the IS period, with peak currents of 14.5 kA and 19.8 kA, respectively. The flash was triggered with quasi static electric field at ground of 6.4 kV/m. The second return stroke, which is the topic of this section, occurred at 20:37:31.078047 (UT)

PAGE 445

445 and was preceded by a dart stepped leader. It is worth noting that the IS of flash UF 11 32 was unique among the triggered lightning events discussed in this dissertation from 2009 2011. The ICC had total duration of 945 ms, transferring 225 C of negative charge to ground. Both the ICC duration and charge transfer were the largest values recorded for any triggered lightning event from 2009 2011. Detailed LMA observations of the IS of flash UF 1 1 32 were discussed in Section 7.3. A total of 44 dE/dt pulses were TOA located in the final 60 s of the dart stepped leader preceding the second return stroke from altitudes as high as 234 m. The triggering wire exploded at an altitude of about 280 m, therefore, all TOA located dE/dt sources occur within the triggering wire height. A three dimensional plot of the dE/dt source locations for this dart stepped leader are shown in Figure 8 13 A The points are color code according to the key in 6 s window s. Source locations of x ray pulses are plotted as black diamonds. Similar to the previously analyzed events, the average leader speed was obtained by fitting a regression line to the altitude versus time projection of the dE/dt source locations (shown i n Figure 8 13 B ). The average downward leader speed was 4.0 x 10 6 m/s with a correlation coefficient of 0.96. An 18 s dE/dt waveform of the dart stepped leader preceding the second stroke of flash UF 11 32 is plotted in Figure 8 14. The waveform was re corded at Station 3. Propagation delay has been removed from the dE/dt waveform according to the method previously described. A total of 14 dE/dt pulses are annotated in Figure 8 14 with pulse numbers increasing sequentially with time. The three dimensi onal spatial locations, emission times, location uncertainties, and pulse designations (same convention used in previous sections) are given for the 14 pulses in Table 8 3. In Figure 8 15 A a 26 s dE/dt waveform from Station 3 is plotted versus the II Ve ry Low channel base current. The propagation delay of the dE/dt waveform has also been removed in an identical manner as the waveform in Figure 8 14 to align the waveform with the measured

PAGE 446

446 current. For the following discussion, note that the current wave form plotted in Figure 8 15 A is also shown for reference in Figure 8 14 with a dashed green line. The first considerable deviation of the channel base current from the zero level occurs at about 10.8 s. This current pulse is associated closely in time with Pulse 1 and Pulse 2 (Figure 8 14 and Table 8 3). Pulses 1 and 2 were located at altitudes of 115.7 m and 77.5 m, respectively. The current does not fall to zero following Pulse 2. The calculated velocity between Pulses 1 and 2 is nearly twice the s peed of light, indicating that Pulse 2, which is a classic bipolar dart stepped leader step pulse, may have been the only pulse associated with the formation of the new leader step. The fact that Pulse 3, which occurs at about 9 s in Figure 8 14, is also located at higher altitude (89.9 m) than the previous step pulse, and also does not appear to alter the steadily flowing current, suggests that Pulse 1 and Pulse 3 (which share similar waveform characteristics) may be physically related to a separate proc ess. Pulses 4 and 5 occur between 8 s and 7 s and are located at altitudes of 78.2 m and 75.5 m. The pulses are radiated from nearly the same altitude as the previous leader step (Pulse 2), though they are separated laterally by 6 8 m from Pulse 2. Pulse 4 and Pulse 5 also produce a negligible change in the measured channel base current, signifying that they may not be associated with the downward propagation of the leader channel. The channel base current remains relatively constant at a level of 1 3 A between Pulse 2 and the next dominant leader step pulse (Pulse 6), which occurs at 6 s. Pulse 6 is located at an altitude of 50.1 m and produces a measured current pulse of about 11 A. The background current level increases slightly following Puls e 6 to a level of 3 5 A. Pulses 7 and 8 occur between 4.4 s and 3.6 s at altitudes of 57.4 m and 46.3 m, respectively. Both pulses appear to produce relatively small impulsive changes in the channel base current, which elevates to a steady level of a bout 7 A following Pulse 8. The final clear leader step pulse (Pulse 9) occurs at about 2 s

PAGE 447

447 and was emitted from an altitude of 40.9 m. There is a 21 A current pulse associated with Pulse 9, after which the background current level decreases to an ampl itude of about 10 A. Pulse 10 and Pulse 11, which occur between 1 s and 0 s are difficult to classify. The pulses together appear like a more or less normal dart stepped leader step, though the temporal proximity to the beginning of the fast transitio n (less than 2 s) and the fact that Pulse 10 and 11 produce the first significant increase in the background level of the channel base current (50 A following Pulse 11), might suggest, based on the analysis of the two previous events, that Pulses 10 and 1 1 are actually associated with a leader burst process. Pulses 10 and 11 were located at altitudes of 43.0 m and 38.0 m, respectively. The slow front of the dE/dt waveform appears to begin about 500 ns following the peak of Pulse 11. The first slow front pulse (Pulse 12) is the lowest located source at an altitude of 22.8 m (14.3 m above the intercepting wire). As expected from the analysis of the previous two events, the first slow front pulse is associated with a large increase in the channel base curr ent (90 A following Pulse 12). Interestingly, the subsequent slow front pulses (Pulse 13 and Pulse 14) were both located at an altitude of 53.4 m, over 30 m higher in altitude that the first slow front pulse. This rapid increase in altitude of the slow f ront pulses following the initial pulse(s) was also observed in flash UF 11 25 (Section 8.2.2). There were several additional small slow front pulses prior to the initial rise of the fast transition that were not resolved on a sufficient number of station s to be TOA located. In Figure 8 15 B the II High channel base current is plotted against a 10 s window of the dE/dt waveform shown in the top panel. The current rises to a level of nearly 300 A following Pulse 14, before increasing further over the nex t 670 ns to an amplitude of 1990 A at the beginning of the fast transition. The channel base current flowed steadily beginning about 12.7 s prior to the initial rise of the fast transition pulse. The upward leader is assumed to propagated for the entir e 12.7 s the

PAGE 448

448 current deviated steadily from zero. The measured dE/dt (red trace) and numerical dI/dt (blue trace) are plotted in Figure 8 16 and are again compared to determine the maximum altitude of the junction point between the upward and downward le aders. In this case, the peaks of the fast transition pulses are separated by only 40 ns. By applying the same geometrical and downward current wave speed constraints discussed in the previous sections, the junction of the upward and downward leaders cou ld have occurred no more than 3.1 m above the top of the intercepting wire. Considering the length of the upward leader and the duration of its propagation (12.7 s), the leader would have propagated at an average speed of 2.4 x 10 5 m/s. The lowest dE/dt source location for this event was 14.3 m above the intercepting wire, over 11 m above the maximum allowable altitude of the junction point. It is possible that the observed separation of the lowest dE/dt source location and the altitude of the junction point are also influenced by timing error in the comparison of the dE/dt and dI/dt peaks (which, as stated in Section 8.2.1 could account for the junction height being under estimated by as much as 4.7 m) and/or a faster downward current wave speed than th e value of 1.55 x 10 8 m/s. Unfortunately, correlated high speed video images, which may have helped explain the altitude difference in the in the calculated length of the upward connecting leader and the lowest dE/dt source, were not acquired of the atta chment region of flash UF 11 32. A 12 s record of the dE/dt and numerically integrated dE/dt measured 190.9 m from the lightning channel base are shown in Figure 8 17 B and Figure 8 17A, respectively The duration of the slow front for the second stroke of flash UF 11 32 was about 640 ns, considerably shorter than the slow fronts observed for the two previously analyzed events. The slow front in the electric field record begins about 200 ns following the peak of Pulse 14, the final located dE/dt pulse o f this event, and well after the first pronounced rise in the channel base current due to the

PAGE 449

449 apparent leader burst (Pulse 11). The duration of the slow front of this particular dart stepped leader is in much better agreement with the statistics reported by Weidman and Krider [1978] for natural subsequent strokes preceded by dart leaders. The dart stepped leader preceding the second stroke of flash UF 11 32 was the first triggered lightning dart stepped leader in summer 2011 that emitted significant x ra ys. In Figure 8 18 B a 30 s plot of the x ray emission recorded at Station 3 is plotted against the waveform from the co located dE/dt antenna ( Figure 8 18A ). Note that the time scale in Figure 8 18 does not correspond to the time scales plotted in the previous figures of this section. In Figure 8 18, the data point that corresponds to the initial rise of the dE/dt fast transition is set to 0 s. This is done to facilitate the alignment of the dE/dt waveforms recorded on the HBM digitization system wit h the x ray waveforms recorded on the ICLRT DSO network. Within 25 s of the start of the fast transition, there were at least a dozen x ray bursts associated with leader and post leader processes. X rays associated with two dart stepped leader steps and the leader burst were recorded on a sufficient number of stations for TOA calculations to be performed. In Figure 8 18 A the dE/dt leader step pulses with corresponding TOA located x ray pulses are annotated with sequentially increasing integers. The th ree dimensional spatial locations (X, Y, Z) ray pulses are given in Table 8 4 for each of the three cases. Additional statistics included in Table 8 4 for each case ar e the total distance from the dE/dt source to the x difference in position between the dE/dt and x ray source for each of the three individual spatial the dE/dt and x ray located x ray pulse occurred in conjunction with the dart stepped leader step at about 16 s in Figure 8 18 A The dE/dt pulse was located at a 111.9 m in altitude

PAGE 450

450 while the corresponding x ray source was l ocated at 85.7 m in altitude. The x ray source occurred a total distance of 26.8 m from the dE/dt source. The x ray source was emitted 26.2 m below, 0.9 to the west, and 5.7 m north of the dE/dt source. The x ray source was emitted 90 ns following the dE /dt peak. The spatial location uncertainties of the x ray source were very large in comparison to the uncertainties of the dE/dt source location. The second x ray location was associated with the final downward leader step at about 4 s (Pulse 9 in Figu re 8 14 and Table 8 3). In this case, the dE/dt and x ray sources occurred at altitudes of 40.9 m and 24.9 m, respectively. The sources were separated by 17.4 m, with the x ray source occurring 16 m below, 6.8 m west, and 1.5 m north of the dE/dt source. Similar to the first case, the spatial location uncertainties of the x ray source are large compares to those of the dE/dt source. The emission time of the x ray source was only 20 ns following the dE/dt source. The final TOA located x ray source occur red in conjunction with the first leader burst pulse (Pulse 10 in Figure 8 14 and Table 8 3). The dE/dt and x ray sources were radiated from altitudes of 43.0 m and 23.6 m, respectively, with a total separation of 23.9 m. The source occurred at exactly t he same easting (longitudinal) location, though the x ray source occurred 13.9 north of and 19.4 m below the dE/dt source. In this case, the spatial coordinates of the dE/dt and x ray sources were both well determined. The emission time of the x ray sour ce followed the dE/dt source by 340 ns. For the three located x rays sources, the average total separation from the dE/dt source was 22.7 m. The sources were separated in the easting (longitudinal) direction by an average of 2.6, in the northing (latitud inal) direction by an average of 7.0 m, and in altitude by an average of 20.5 m. Clearly, the vertical displacement of the dE/dt and x ray sources comprises the majority of the total separation. The x rays sources lagged the dE/dt sources in time by an average of 150 ns (GM of 85 ns). To better illustrate the spatial proximity of the x ray sources to the propagating

PAGE 451

451 leader channel, in Figure 8 19, three projection views (altitude versus easting, altitude versus northing, and altitude versus time) are pl otted of the dE/dt and x ray source locations for this dart stepped leader event. The dE/dt sources are plotted with the same color convention as previously used in Figure 8 13. X ray sources are plotted as black diamonds. 8.2.4 Fourth Stroke of Flash UF 11 35 The final triggered lightning dart stepped leader analyzed in this chapter occurred preceding the fourth return stroke of flash UF 11 35 on August 18, 2011. Flash UF 11 35 was the fourth and final triggered lightning discharge of August 18, and w as triggered with quasi static electric field at ground of 7.0 kV/m. The flash contained a total of seven return strokes, with the current of the fourth stroke reaching a peak of 27.4 kA. The fourth stroke had the largest peak current of any stroke init iated by a dart stepped leader during summer 2011. The fourth stroke occurred at 20:58:11.958275 (UT). This event had several unique characteristics: 1) dE/dt TOA source locations were obtained for sources exceeding 720 m in altitude, over a factor of tw o higher than any other triggered lightning event during summer 2011, 2) dE/dt sources were obtained at altitudes over 500 m above the top of the triggering wire, allowing the comparison of the leader propagation characteristics in the presence of differen t channel conditioning, and 3) copious x rays were detected at ground level in association with the propagating dart stepped leader for nearly 170 s prior to the return stroke. I n Figure 8 20 A a three dimensional view is shown of the 148 dE/dt source lo cations for the dart stepped leader preceding the fourth stroke of flash UF 11 3 5 The dE/dt sources are color coded according to the key at right in 24 s windows. TOA located x ray sources are shown as black diamonds (independent of emission time) in t he three dimensional view of Figure 8 20 A The dart stepped leader channel above the top of the triggering wire followed the path of the UPL, as expected. The leader descended in a east southeasterly direction before turning generally vertical at an

PAGE 452

452 alti tude of about 240 m (the hot pink points in Figure 8 20 A ). The top of the triggering wire was located at an altitude of 202 m. This level is annotated in the altitude versus time projection of the dE/dt source locations shown in Figure 8 20 B Separate r egression lines were fitted to the dE/dt source altitude coordinates for points above (blue line) and below (black line) the top of the triggering wire. The average speed of the dart stepped leader above the height of the triggering wire was 2.8 x 10 6 m/s with a correlation coefficient of 0.91, and the average leader speed below the height of the triggering wire was 4.8 x 10 6 m/s with a correlation coefficient of 0.95. The leader appears to have accelerated by a factor of about 71% upon entering the path to ground left behind by the exploded triggering wire. This is not unexpected considering the likely higher values of conductivity and temperature of the plasma channel near the location of the exploded copper triggering wire versus the atmosphere above t he triggering wire. An 18 s dE/dt waveform of the dart stepped leader measured at Station 3 is plotted in Figure 8 21. The propagation delay has been removed from the dE/dt waveform as described in the previous sections. In Figure 8 21, there are 26 T OA located dE/dt pulses labeled with sequentially increasing integers. The three dimensional source locations, location uncertainties, emission times, and designations for the 26 pulses are given in Table 8 5. In Figure 8 22 A a 30 s waveform of the dE/ dt measured at Station 3 is plotted against the II Very Low current. Figure 8 22 B shows a 12 s window of the dE/dt waveform plotted in Figure 8 22A versus the II High channel base current waveform. The dE/dt waveform in Figure 8 22 A exhibits very rapid stepping with leader step pulses occurring of the order of every 1 s. Small deviations in the channel base current are observed coincident with leader step pulses as much as 25 s prior to the return stroke, though the current does not flow steadily unti l the leader step pulse at about 11.7 s in Figure 8 22 A The current maintains a typical amplitude of 1 2 A with some small

PAGE 453

453 super imposed fluctuations for the next 6.5 s. Pulse 2 in Figure 8 21 produces an 11 A current pulse and subsequently, the firs t notable increase in the background level of the channel base current. Pulse 2 was located at an altitude of 88 m (Table 8 5). Following Pulse 2, the current flows with a typical amplitude between 5 10 A. Between Pulse 3 and Pulse 7 (time span of about 2 s), the leader descends from 86 m to 76.5 m in altitude, though Pulses 3 7 do not produce significant variations in the channel base current. The large leader step pulse at 2 s (Pulse 8) is located at an altitude of 69.0 m and is associated with a 1 4 A current pulse. The current elevates to a level of about 12 A following Pulse 8. Pulses 9 14 have similar pulse characteristics to the leader burst pulses observed in previous events, though they do not produce large increases in the channel base curr ent. Pulses 9 14 descend from an altitude of 74.8 m to 45.5 m in a time span of 810 ns, giving an average downward propagation speed of 3.5 x 10 7 m/s. This rapid descent of the leader burst dE/dt sources is characteristic of the events analyzed in Howard et al. [2010] and is unlike the leader burst characteristics of the previous events in this study analyzed in Sections 8.2.1 8.2.3. Following Pulse 14, which was located at 45.5 m in altitude, the channel base current rose to a level of about 32 A before gradually declining to a level of 20 A over the next 1.2 s. Pulse 15 and Pulse 16 occur during the period of the current decline and do not appear to alter the current flow. Pulse 15 was radiated from an altitude of 28 m, well below the final leader bu rst pulse. The physical significance of Pulse 15 is not clear. Pulse 16 was radiated from the same location as Pulse 18, which occurred 200 ns later. Pulse 17 and Pulse 18 were emitted from altitudes of 44.3 m and 51.1 m, respectively, and are associate d with a sharp increase in the channel base current to a level of about 67 A. Pulse 19 follows Pulse 18 by about 890 ns and is likewise associated with a large current pulse (90 A) and an elevation of the background current level to about 78 A. Pulse 19 was located at an altitude of 34.1 m.

PAGE 454

454 Pulses 17 18 and Pulse 19 have the appearance of consecutive typical dart stepped leader steps, though their close temporal proximity to the fast transition and their association with a large channel base current incr eases indicate that they, along with the previous leader burst pulses, may be involved in the initial interactions of the streamer zones of downward and upward leaders. The fact that this dart stepped leader was associated with a high peak current return stroke might suggest the presence of longer than normal streamer zones, such as the case of the "chaotic" dart leader discussed in Section 6.4.2, which exhibited a downward leader streamer zone length of 25 m, and a return stroke peak current (28.3 kA) sim ilar to this event. Dart stepped leaders that carry more negative charge at their tips may inherently exhibit more complex interactions in the attachment process to the strike object. Considering Pulses 16 19 are of an undetermined nature, they are not g iven a classification in Table 8 5. Following Pulse 19, there is an observed slow front in the dE/dt waveform. Pulses 20 26 are therefore classified as slow front pulses in Table 8 5, and occur over a time span of about 1.7 s. Like all slow front pulse s discussed in the previous sections, Pulses 20 26 are associated with rapid increases in the channel base current, from 78 A at the time of Pulse 20 to about 1070 A at the time of Pulse 26. As shown in Figure 8 22 B the current further elevated to a leve l of 1360 A at the time of the initial rise of the fast transition. The dE/dt source locations of the slow front pulses are very difficult to interpret. The sources essentially "bounce" in altitude over several tens of meters from as high as 47.1 m (Puls e 20 and Pulse 22) to as low as 18.2 m (Pulse 25). This type of behavior has not been observed in the previously analyzed events where the slow front pulses either occurred at low altitude immediately above the intercepting wire (dart stepped leader of f lash UF 11 15) or began at low altitude and ascended rapidly (dart stepped leaders of flashes UF 11 25 and UF 11 32). For more energetic leaders, perhaps the bi directional current waves

PAGE 455

455 that Jerauld et al. [2007] associated with the radiation of the slow front can also produce impulsive field variations that are radiated nearly simultaneously from different sections of the channel above and below the junction height, giving rise to dE/dt sources that appear to fluctuate in altitude. In Figure 8 21, the channel base current was observed to deviate steadily from zero about 17 s prior to the initial rise of the fast transition. The timing relationship of the peaks of fast transition peaks of the measured dE/dt and numerical dI/dt are compared in Figure 8 23 to determine the maximum altitude of the junction point of the downward and upward leaders. For the two fast transition peaks, the numerical dI/dt lagged the dE/dt by 60 ns and 20 ns, respectively. With the maximum time difference of 60 ns, and assumi ng the same stated geometrical and leader speed constraints of the previous analyses, the junction height could have been, at most, 6.2 m above the intercepting wire ring. With the assumption that the upward connecting leader propagated for the full 17 s time duration of the observed steady current flow, the upward leader would have propagated at an average speed of about 3.6 x 10 5 m/s between the intercepting wire ring and the junction height of 6.2 m. While no high speed video images were obtained of t he attachment process of this event, a still image (Figure 8 23) taken from Launch Control does show two clear unconnected upward leaders from the intercepting wire ring. The leftmost upward leader is the longer of the two and has total length of about 1 m. According to Rakov and Uman [2003], the presence of an unconnected upward leader from the strike object indicates that the upward leader that successfully connected to the downward propagating leader was longer than the unconnected upward leader. A 1 6 s plot of the measured dE/dt in shown in Figure 8 24B and the numerically integrated dE/dt is shown in Figure 8 24 A The observed slow front in the numerical electric

PAGE 456

456 field waveform began at about 4.3 s, coincident in time with Pulse 22 (Figure 8 21 and Table 8 5), the largest amplitude slow front pulse. Pulse 22 was located at an altitude of 47.1 m. The slow front had total duration of 1.7 s. Similar to the dart stepped leader of flash UF 11 32 discussed in Section 8.2.3, the slow front duration for this event is in better agreement with the statistics reported by Weidman and Krider [1978] for dart leaders than for dart stepped leaders. The dart stepped leader preceding the fourth stroke of flash UF 11 35 was the most prolific emitter of x rays of any triggered lightning dart stepped leader recorded at the ICLRT for the 2009 2011 dataset. As stated previously, pronounced x ray bursts were recorded as much as 170 s prior to the return stroke, when the dart stepped leader was at an altitude of a bout 550 m. In Figure 8 25, 110 s waveforms of the dE/dt ( Figure 8 25A ) and x ray ( Figure 8 25B ) emission recorded during the dart stepped leader are shown. This time window does not show the full extent of the x ray emission, but does include all TOA l ocated x ray pulses, which occurred within about 84 s of the return stroke. The dE/dt waveform in Figure 8 25 A clearly shows the dramatic difference in the propagation characteristics of the dart stepped leader above and below the height of the triggerin g wire. Above the triggering wire, dart stepped leader steps occurred with interstep intervals of the order of 4 6 s. There was a "quiet" period where no leader step pulses were emitted for a time of about 8 s following the last leader step above the t op of the triggering wire. The final leader step above the triggering wire is located at about 40 s in Figure 8 25 A Below the height of the triggering wire, leader step pulses were recorded about every 1 s, and as discussed at the beginning of this se ction, the downward leader propagation speed increased by about 71%. There were a total of 13 x ray sources located during the descent of the dart stepped leader. Four sources were associated with leader steps above the height of the triggering wire and nine sources were associated with steps below the

PAGE 457

457 height of the triggering wire. The dE/dt pulses associated with the 13 x ray sources are annotated in Figure 8 25 A with increasing integers. The three dimensional source locations, location uncertainties, and emission times of the dE/dt and x rays sources for each of the 13 cases are listed in Table 8 6. The total total distance from the dE/dt source to the x ray source, the relative differences between the three spatial coordinates of the dE/dt and x ray sources, and the emission time difference between the dE/dt and x ray sources are also provided. For eight of the 13 cases, the x ray source is potentially associated with at least two different dE/dt pulses, and for two cases, with as many as four diffe rent dE/dt pulses. For completeness, the aforementioned TOA parameters are given in Table 8 6 for all of the dE/dt pulses that could have been associated with the x ray emission. X ray sources were located associated with leader steps that occurred from altitudes of 336 m (Case 1) to 69 m (Case 12). The final located x ray source (Case 13) was emitted by the leader burst process. For the 13 cases (taking into account all possible dE/dt pulses that could have been associated with the x ray emission) the dE/dt sources were separated from the x rays sources by an absolute average total distance of 29 m (GM of 26.4 m). The sources were separated in the easting (longitudinal) direction by an absolute average of 11.7 m (GM of 6.9 m) and mean of 4.8 m, in the northing (latitudinal) direction by an absolute average of 11.2 m (GM of 8.7 m) and mean of 6.9 m, and in altitude by an absolute average of 19.3 m (GM of 13.6 m) and mean of 18.2 m. While there appears to be no systematic trend in the sign of the later al position differences (i.e., the x ray sources are distributed around the leader channel without bias for a particular direction), the x ray sources occurred below the dE/dt sources in 11 of the 13 cases (85%). In Case 8, the x ray source was emitted 6. 2 m above the dE/dt source, though the altitude uncertainty in the TOA calculation for the x ray source was more than 16 m. For one additional case (Case 12), the altitude uncertainty

PAGE 458

458 in the x ray source location is larger than the separation of the dE/dt and x ray sources for all possible associated dE/dt sources. The average time difference between the dE/dt and x ray sources was 290 ns (GM of 170 ns), with a range from 20 ns to 2.16 s. The results of this study are in good agreement with the findings of Howard et al. [2010], who reported that x rays were emitted within 50 m of the dE/dt source and occurred from 0.1 s to 1.3 s after the dE/dt source emission time. In Figure 8 26, three projections views (altitude versus easting, altitude versus nor thing, and altitude versus time) are plotted of the dE/dt and x ray source locations for the dart stepped leader preceding the fourth stroke of flash UF 11 35. Recall that the top of the triggering wire occurred at an altitude of 202 m. The dE/dt sources are plotted with the same color convention used in the three dimensional plot of Figure 8 20 A and the x ray sources are plotted as black diamonds. The plots clearly show the time and spatial correlation of the x ray sources with the descending leader cha nnel. Further, the random lateral distribution of the x ray sources is evident from the easting and northing projections of Figure 8 26. 8.2.5 Second Stroke of Flash MSE 11 01 Natural flash MSE 11 01 terminated on the southwest quadrant of the ICLRT at 19:37:34.737534 (UT) on July 7, 2011. The flash contained a total of four return strokes. The second stroke, which struck ground at 19:37:34.794163 (UT), was preceded by a dart stepped leader. The NLDN reported peak current of the second stroke was 22.1 kA. A total of 180 dE/dt source locations were determined during the leader and post leader phase of the second return stroke. A three dimensional plot of the dE/dt source locations is shown in Figure 8 28 A The view is looking approximately north nort heast. The dE/dt sources are color coded according to the key at right in 21 s time windows. The x ray source locations are shown as black diamonds. The location of the Field (Ground) Launcher is shown for reference. Source s are

PAGE 459

459 plotted within 210 s of the return stroke and 500 m of the ground. Within 500 m of ground, the leader propagated from east to west at approximately a 22 degree angle from the vertical and with very little deviation in the north south plane. I n Figure 8 28 B the altitude vers us time projection of the dE/dt source locations is plotted. A regression line is fitted to the points to obtain an estimate for the average downward leader speed. The average leader speed was 1.9 x 10 6 m/s with a correlation coefficient of 0.98. In Fig ure 8 29, the easting versus altitude projection of the dE/dt source locations is overlaid on a still photograph of the flash MSE 11 01. The discharge struck approximately 395 m southwest of the camera location in IS2, which borders the road on the northe rn boundary of the ICLRT. The photograph is a six second time exposure recorded at a focal length of 10 mm. The dE/dt sources follow the discharge very well to about 275 m in altitude where the points begin to diverge. The separation of the dE/dt source s from the lightning channel in the photograph at higher altitude is due to two factors, 1) the optical distortion at the edges of the frame due to the "fisheye" effect of a wide focal length lens that is angled upward, and 2) the camera view angle is not looking exactly due south, and hence, the easting projection of the dE/dt sources is not an exact mapping into the plane of the image at the distance of the discharge. In Figure 8 30, a 35 s dE/dt waveform is shown of the dart stepped leader preceding the second stroke of flash MSE 11 01. The waveform was measured at Station 3, a distance of about 345 m from the strike point. A total of 31 dE/dt pulses are annotated with increasing integers. The three dimensional TOA locations, location uncertainties and emission times of the 31 pulses are given in Table 8 7. Considering there is no available channel base current with which to compare the dE/dt TOA locations as there were for the previous analyses of triggered lightning dart stepped leaders, the cla ssifications of the pulses in the following discussion are based

PAGE 460

460 primarily on pulse shape and temporal proximity to the return stroke. The initial leader dart stepped leader step in Figure 8 30 at about 68 s has three TOA located pulses (Pulses 1 3). Th e three pulses were located in close spatial proximity and the dominant step pulse (Pulse 3) was located at an altitude of 78 m. There was no impulsive leader activity for about 3.5 s following Pulse 3. Pulse 4 occurred about 15 m lower than Pulse 3 and was radiated from a point about 12 m to the west. Pulse 5 occurred at almost exactly the same spatial location as Pulse 3. Pulse 6 and Pulse 7 appears to be associated with the final downward leader step. Pulse 6 occurs at an altitude of 59.6 m and Pul se 7 was emitted about 560 ns later at an altitude of 47.2 m. Pulse 7 was also located about 6 m to the east and 8.5 m to the south of Pulse 6. Following Pulse 7, the characteristics of the dE/dt pulses do not strongly resemble typical dart stepped leade r steps. Indeed, the next located pulse (Pulse 8) is radiated from an altitude of only 25.3 m. Without high speed video evidence or measured current, it is impossible to state definitively the physical significance of Pulse 8, though one might infer that it could be due to a stepping mechanism in an unconnected upward positive leader from a grounded object The area where the flash terminated is populated with pine trees that have heights comparable to the emission altitude of Pulse 8. Pulses 9 15 appea rs very similar to the leader burst pulses recorded in association with triggered lightning dart stepped leaders discussed in the previous sections. The pulses do not demonstrate significant vertical propagation, covering an altitude range from 59.4 m to 47.2 m in the time span of about 2.75 s. The general motion of the sources is slightly to the east. Pulse 16, which at casual glance could easily be classified as part of the previous pulse group, occurs at an altitude of 27.4 m. Pulse 16 occurs about 6 m to the east and about 2 m above the location of Pulse 8. One possible reason is that the increase in the electric field due to the burst of pulses (Pulse 9 15) immediately prior to Pulse 16 prompted the further extension of an existing upward

PAGE 461

461 leader c hannel that began with Pulse 8. Following Pulse 16, the next three source locations (Pulses 17 19) occur at higher altitude from 43.6 m to 58.8 m. Pulse 17 19 are radiated from a very similar lateral areas as the pulse group that contained Pulses 9 15. Pulses 20 is radiated about 460 ns following Pulse 19, but from an altitude of only 23.3 m. Pulse 19 and Pulse 20 are separated by a total distance of about 36 m, giving a calculated straight line propagation speed between the sources of greater than 7.7 x 10 7 m/s, an unreasonably high propagation speed for a dart stepped leader. The fact that Pulse 21, which is emitted 180 ns after Pulse 20, also occurs at an altitude of only 23.9 m provides further evidence that the dE/dt pulses within about 15 s of th e return stroke in Figure 8 30 are a superposition of two (or possibly more if multiple upward propagating leaders are present) independent leader processes. A reasonably pronounced slow front is observed in the dE/dt waveform following Pulse 21. Not sur prisingly, Pulses 22 30 all occur at low altitude (below 32.5 m). The large collection of sources between 19.7 m and 30 m in a time span of about 6 s may indicate the presence of multiple upward connecting discharges with varying levels of connectivity t o the streamers below the dart stepped leader tip. From the observations of triggered lightning slow front pulses and the correlated channel base current, one might infer that the large dE/dt pulses at about 89 s (Pulse 26 and Pulse 27) are likely associ ated with major connections between the upward and downward leaders. A close examination of the lateral coordinates of the slow front pulses reveals that the sources are grouped to some extent, though without photographic evidence, an attempt to explain t he geometry of the connection region would be quite speculative. The first fast transition pulse was located for this event at an altitude of 36.2 m and in the same lateral region as the final two slow front pulses.

PAGE 462

462 The 35 s dE/dt waveform plotted in F igure 8 30 is numerically integrated and plotted in Figure 8 31 A The original dE/dt waveform is shown in Figure 8 31 B for reference. The electric field waveform exhibits a very pronounced slow front in comparison to the triggered lightning events previo usly analyzed. The slow front has duration of about 6.9 s, a factor of about 3.3 longer than the typical duration reported by Weidman and Krider [1978] for natural dart stepped leaders. The slow front initiates coincident with Pulse 19 (Figure 8 30 and Table 8 7), which was located at an altitude of 58.8 m, but also occurred immediately prior to Pulse 20 and Pulse 21, both of which were radiated from altitudes of about 24 m. Like the dart stepped leader preceding the fourth stroke of triggered lightnin g flash UF 11 35 (Section 8.2.4), the dart stepped leader that initiated the second stroke of MSE 11 01 produced a large flux of x rays measured at ground. Bursts of x rays associated with dart stepped leader steps were recorded more than 250 s prior to the return stroke, indicating, from Figure 8 28, that the source altitudes were likely above 500 m. In Figure 8 32 B a 275 s waveform is shown of the x ray emission recorded at Station 3, located about 345 m from the ground strike point. The correspondi ng dE/dt waveform measured at the same station is shown in Figure 8 32 A for reference. A total of 14 x ray sources were TOA located during the descent of the dart stepped leader. The dE/dt pulses that correspond to the TOA located x rays sources are anno tated in Figure 8 32 A with increasing integers. The located x ray sources occurred between 225 s and 40 s prior to the return stroke. Though x ray photons were recorded in association with the leader burst pulses and with the largest slow front pulses, these events could not be TOA located. A discussion of the issues involved with determining the source locations of low altitude x ray sources is provided in Section 8.3. The source locations, location uncertainties, and emission times of the 14 x rays sources and the related dE/dt sources are given

PAGE 463

463 in Table 8 8. Similar to the data provided in the previous sections, Table 8 8 also includes statistics for the differences in the TOA determined spatial coordinates and emission times of the dE/dt and x ray sources. For this event, it was found that the dE/dt and x ray sources for the 14 cases were separated by an average total distance of 39.2 m (GM of 35.5 m). The dE/dt and x ray sources were separated by an absolute average in the easting (longitudinal ) direction of 17.2 m (GM of 12.5 m) and mean of 6.7 m, an absolute average in the northing (latitudinal) direction of 7.7 m (GM of 4.4 m) and mean of 0.2 m, and an absolute average in altitude of 29.2 m (GM of 19.6 m) and mean of 28.4 m. From Table 8 8, it is clear that the signs of the easting and northing differences between the dE/dt and x ray source locations are not systematic. The cause of the factor of two larger absolute difference in the easting source coordinates of the dE/dt and x rays source s versus the northing coordinates is undetermined. Based on the fact that the absolute averages of the easting and northing coordinate differences were nearly identical for the fourth stroke of triggered lightning flash UF 11 35, which exhibited a general ly vertical channel below the height of the triggering wire, it is possible that the tilted channel geometry of the dart stepped leader preceding the second stroke of flash MSE 11 01 in some way impacted the spatial emission pattern of the x rays. Data fr om additional natural lightning events with significant x ray emission are needed to test this hypothesis. The altitude coordinates of the x rays occurred below the associated dE/dt source altitude in 12 of the 14 cases (86%). In the two cases where the x ray source was emitted above the location of the dE/dt source (Case 7 and Case 13), the total separation of the dE/dt and x ray sources was dominated by the lateral coordinates, indicating that the leader may have been propagating more horizontally, in t urn producing streamers that may have propagated in outward directions instead of downward. In all 14 cases,

PAGE 464

464 the x ray source was emitted following the dE/dt source(s). The average timing difference between the peak of the dE/dt pulse and the x ray emis sion was 280 ns (GM of 150 ns). Three projection views of the dE/dt and x ray source locations of the dart stepped leader associated with the second stroke of flash MSE 11 01 are shown in Figure 8 33. From the altitude versus easting and altitude versus northing projections shown in Figure 8 33, it is clear that the x ray sources follow the general propagation path of the leader channel and are not systematically biased towards a particular direction. As discussed above, there is more spread in the x ra y sources in the easting directions, though that spread appears to be random about the channel geometry. 8.3 Discussion of Results Measured parameters of the four triggered lightning dart stepped leaders discussed in Section 8.2.1 8.2.4 are given in Tabl e 8 9. Table 8 9 also include similar parameters for three additional triggered lightning dart stepped leader events recorded during summer 2011 that were not discussed in the above sections (second stroke of flash UF 11 34 on August 18, 2011, and the six th and seventh strokes of flash UF 11 35 on August 18, 2011). A subset of the statistics not pertaining to measured channel base current are also provided for natural flash MSE 11 01. Tabulated statistics include the peak current of the subsequent return stroke, the number of calculated dE/dt and x ray TOA source locations, the estimated average downward dart stepped leader speed, the measured duration of the steady current flow (in this case, a direct proxy for the duration of the upward connecting leade r), the calculated altitude of the junction point between the downward and upward leaders, the calculated average vertical speed of the upward connecting leader, the estimated peak current of the upward connecting leader, and measured duration of the slow front.

PAGE 465

465 The average downward leader speed of the seven triggered lightning dart stepped leaders below the final height of the triggering wire was about 3.5 x 10 6 m/s (GM of 3.2 x 10 6 m/s), a speed in good agreement with values reported by Wang et al. [19 99], Howard et al. [2010], and Biagi et al. [2010] for triggered lightning dart stepped leaders. Estimated speeds ranged from 2.0 to 6.9 x 10 6 m/s. The estimated leader speed of the natural lightning dart stepped leader was 1.9 x 10 6 m/s, also in good ag reement with the optical measurements of natural dart stepped leader speed reported by Schonland [1956] and Orville and Idone [1982]. The average duration of the measured steady current flow prior to the initial rise of the fast transition for the seven t riggered lightning dart stepped leaders was 14.1 s (GM of 13.3 s) with a range from 8.7 24 s. In this study, the upward connecting leader is assumed to have the same duration as the measured steady current flow. The altitudes of the junction heights b etween the upward and downward leaders, calculated from the timing difference between the fast transition peaks of the dE/dt and numerical dI/dt waveforms, were found to occupy a narrow range from 0.0 7.8 m with an average altitude of 3.6 m above the heigh t of the intercepting wire. The calculated upward connecting leader lengths are somewhat shorter than those reported by Idone [1990], who used photographic measurements to estimate an upward boundary for upward connecting leader lengths of 10 20 m for tri ggered lightning subsequent strokes. The upward leader lengths in this study are in better agreement with observations reported by Wang [1999], who used the ALPS optical imaging system to infer the existence of two upward connecting leader preceding trigg ered lightning return strokes with lengths of 7 11 m and 4 7 m, respectively. From 2009 2011, there has been frequent discussion of why upward connecting leaders are rarely imaged in the Photron high speed video records of the attachment region. The fact that the upward connecting leaders appear to only extend to an average length of 3.6 m, which corresponds to a

PAGE 466

466 pixel region with height of only 8 pixels at the location of the Field (Ground) Launcher assuming a focal length of 14 mm, might help to explain this observation. Further, the average upward leader duration of 14.1 s occupies four frame integrations at a frame rate of 300 kfps (3.33 s exposure), meaning the upward connecting leader typically propagates less than 1 m per frame integration. It i s also not impossible that the upward connecting leaders are relatively dim in comparison to the downward propagating dart stepped leader channel, particularly when viewed at a distance of 300 m. The short upward connecting leaders should be better resolv ed with the Cordin 550 high speed camera that presently operates from the Optical Building, a distance of 200 m from the Field (Ground) Launcher, especially considering the improvement in spatial resolution gained from the 1 mega pixel sensor of the Cordin camera. The speeds of the upward connecting leaders were calculated from the duration of the steady current flow and the junction height. The average upward connected leader speed for the six triggered lightning dart stepped events that exhibited junct ion heights above the intercepting wire ring was 2.6 x 10 5 m/s (GM of 2.5 x 10 5 m/s). This average speed is in good agreement with the optical measurements of the speeds of upward connecting leaders reported by Yokoyama et al. [1990] for six discharges to an 80 m telecommunications tower in Japan. Yokoyama et al. [1990] used the ALPS optical imaging system to measured upward connecting leader speeds ranging from 0.8 to 2.7 x 10 5 m/s. The estimated upward connecting leader speeds in this study are also in fairly good agreement with measurements of UPL propagation speeds from the top of the ascending wire in triggered lightning experiments. From high speed video measurements, Biagi et al. [2010] reported a two dimensional upward propagation speed of 5.6 x 10 4 m/s for the first 100 m of propagation of an UPL at the ICLRT, and Jiang et al. [2011] reported a two dimensional speed of 1.0 x 10 5 m/s for an UPL in China propagating between 130 730 m above

PAGE 467

467 ground. The maximum upward connecting leader currents in t his study, inferred from the amplitude of the channel base current at the time of the initial rise of the fast transition pulses, ranged from 270 2190 A, with an average amplitude of 1 kA (GM of 730 A). Upward leader currents have traditionally been thoug ht to be of the order of 100 A. Finally, the slow front durations for the seven triggered lightning dart stepped leaders measured from the numerically integrated dE/dt waveform at a distance of 109.9 m from the lightning channel base were found to range from about 600 ns to 4.6 s with an average duration of 2.2 s (GM of 1.7 s). The average slow front duration is in excellent agreement with the value of 2.1 s reported by Weidman and Krider [1978] for natural lightning dart stepped leaders, and equal t o the value of 2.2 s reported by Jerauld et al. [2007] for a triggered lightning dart stepped leader at the ICLRT. The measured slow front duration for the dart stepped leader of natural flash MSE 11 01 was 6.9 s, about a factor of 3.3 longer than the a verage duration reported by Weidman and Krider [1978] for natural dart stepped leaders. The TOA locations of dE/dt pulses within 20 s of the return stroke for the four triggered lightning dart stepped leaders analyzed in Sections 8.2.1 8.2.4, together with the time aligned, low level channel base current provided insight to the sequence of electrical breakdown processes that occur prior to the fast transition pulse(s). A leader burst process, similar to those reported by Murray et al. [2005], Jerauld et al. [2007], and Howard et al. [2010], was observed in each of the four triggered lightning events. In the four events, the initial leader burst pulse occurred from 2.3 6.6 s prior to the return stroke (average of 4.1 s), and initiated at altitudes ra nging from 24.2 74.8 m (average of 42.4 m). Recall that Murray et al. [2005] reported leader burst pulses occurring from 1 9 s preceding the return stroke. In this study, each leader burst was composed of 2 7 individual dE/dt pulses. The leader burst p ulses were found to occur below

PAGE 468

468 the spatial position of the final preceding leader step, a finding in agreement with Howard et al. [2010]. In contrast to the results of Howard et al. [2010], who reported that the leader burst was typically characterized b y rapid downward (and in one case, horizontal) propagation, three of the four leader bursts analyzed in this study occurred within a relatively small source volume with minimal altitude fluctuation. The lone exception was the leader burst associated with the fourth stroke of flash UF 11 35, which exhibited a downward propagation speed of 3.5 x 10 7 m/s during a time span of 810 ns. The natural dart stepped leader of flash MSE 11 01 also exhibited a group of pulses that shared some characteristics with the leader bursts observed in the triggered lightning events. The initial pulse occurred about 13.5 s prior to the return stroke at an altitude of 59.4 m. The seven pulses included in the leader burst of flash MSE 11 01 were also confined to a small source volume and did not exhibit rapid downward propagation. The leader burst pulses in three of the four triggered lightning events were associated with significant increases in the measured channel base current. The leader burst of flash UF 11 15 elevated th e background current amplitude from 18 A to 60 A; the leader burst of flash UF 11 25 resulted in the background current amplitude increasing from 18 A to 120 A; and the leader burst of flash UF 11 32 elevated the background current amplitude from 10 A to 5 0 A. The exception to this trend was the dart stepped leader preceding the fourth stroke of flash UF 11 35, which produced a comparatively small rise in the background current level from 12 A to 20 A. Considering that the leader burst pulses typically co rrespond to the initial large increase in the background level of the channel base current, it is likely, as discussed previously, that they are associated with the initial interaction of the streamer zones of the downward and upward leaders. Howard et al [2010] speculated that the leader burst may be a result of simultaneous activity involving the downward and upward leaders, but the authors did not have sensitive channel base current

PAGE 469

469 measurements to validate the hypothesis. The results presented here s uggest their reasoning is valid. It is also interesting to compare the observations of the leader burst in this study to the case presented by Wang et al. [2001], who observed a pronounced electric field pulse immediately prior to the slow front of a nega tive cloud to ground discharge that was correlated in time with a pulse of light viewed by the ALPS optical imaging system. At the time of the electric field pulse, the upward leader had already initiated. This scenario is analogous to the cases presente d here in which the channel base current is observed to flow (and hence, an upward connecting leader is assumed to be propagating), in each case, for 10 20 s prior to the time of the initial leader burst pulse. Pulses super imposed on the slow front we re observed in each of the four triggered lightning dart stepped leaders, with each event containing from 3 8 TOA located slow front pulses. The slow front pulses of these four events exhibited three distinctly different types of positional behavior. The four slow front pulses of the dart stepped leader of flash UF 11 15 were all located at low altitude (below 20 m), and three of the four slow front pulses were located very near the altitude of the calculated junction point. In contrast, the eight slow f ront pulses of the dart stepped leader preceding the first stroke of flash UF 25 began at low altitude (about 14 m), and then ascended rapidly to an altitude of about 57 m in the time span of about 1.2 s. The initial three slow front pulse were within th e altitude range of the calculated junction height. Similarly, the three slow front pulses of the dart stepped leader associated with the second stroke of flash UF 11 32 ascended rapidly from 22 m to about 53 m in the time span of about 830 ns. The lowes t slow front pulse was about 11 m above the calculated height of the junction point. Finally, the eight TOA located slow front pulses of the dart stepped leader preceding the fourth stroke of flash UF 11 35 were observed to fluctuate rapidly in altitude o ver

PAGE 470

470 several tens of meters from altitudes of about 18 47 m. Regardless of the spatial positioning of the slow front pulses, they are, without exception, associated with large and rapid increases in the measured channel base current. In each case, the slo w front pulses were associated with increases in the channel base current to near the maximum upward leader current (Table 8 9). Based on this observation, it is likely the slow front pulses correspond to significant connections between the downward and u pward propagating leader channels. Howard et al. [2010], based on a video image of a triggered lightning dart stepped leader that demonstrated a pronounced loop above the strike object in addition to several unconnected upward leaders, suggested that the slow front pulses each corresponded to individual connections. The correlated dE/dt and channel base current observations reported in this chapter support that line of reasoning. It is worth noting that the complexity of the dE/dt pulse structure during the slow front period appears to be positively correlated to the peak current of the subsequent return stroke. Leaders that transport more charge (and hence produce higher return stroke peak currents) likely exhibit longer and more wide spread streamer zo nes, and may also induce an increased number of upward connecting leaders from the strike object. Indeed, the dart stepped leader preceding the fourth stroke of flash UF 11 35, with return stroke peak current of 27.4 kA, exhibited a very complex pulse str ucture during the slow front period. As shown in the still image of Figure 8 24, flash UF 11 35 also exhibited multiple unconnected upward leaders. The triggered flash analyzed by Howard et al. [2010], which, as noted above, exhibited multiple unconnecte d upward leaders, also had return stroke peak current of 45 kA. The slow front period of the dart stepped leader preceding the second stroke of natural flash MSE 11 01 also exhibited a very complex pulse structure, with nine slow front pulses located belo w 32.5 m in altitude. As discussed in Section 8.2.5, the large collection of low altitude sources, which show evidence of

PAGE 471

471 being grouped spatially in at least three regions, are likely associated with the interaction of multiple upward connecting discharge s with the streamer zone of the downward propagating dart stepped leader. The fast transition pulse for the second stroke of flash MSE 11 01 was located at about 36 m altitude (above the locations of the slow front pulses, but in a similar lateral locatio n). The source locations of x rays associated with dart stepped leader steps, and in two cases, the leader burst pulses, were determined for two triggered lightning dart stepped leaders (second stroke of flash UF 11 32 and fourth stroke of flash UF 11 3 5) and one natural lightning dart stepped leader (second stroke of flash MSE 11 01). The dart stepped leaders that were associated with lower peak return stroke currents did not produce sufficient x ray emissions for source locations to be calculated. A total of 30 x ray source locations were determined in the three events. For the two triggered lightning dart stepped leaders recorded in flash UF 11 32 and flash UF 11 35, the average total separations between the dE/dt and x ray sources were 22.7 m and 2 9 m, respectively. For the dart stepped leader of natural flash MSE 11 01, the average total separation between the dE/dt and x ray sources was 39.2 m, a value 16.5 m and 10.2 m larger than the displacement of the two sources for the triggered lightning e vents. The larger total displacement between the dE/dt and x ray sources for the natural dart stepped leader implies that the streamers in front of the propagating leader tip, where the x rays are thought to be produced, are likely longer. The electric f ield magnitude at the tip of the leader is probably larger for the natural dart stepped leader event, which likely extends the high field region necessary for electrons to run away and emit x rays. For seven x ray source locations associated with a natura l stepped leader and a triggered lightning dart stepped leader, Howard et al. [2008] reported that the x ray sources were emitted within 50 m of the related dE/dt source. The results presented

PAGE 472

472 here fit within the upper boundary reported by Howard et al. [ 2008], though in this study, more x ray source locations have been computed and the total separation between the two sources has been better quantified. Howard et al. [2008] also reported a systematic trend for the x ray source to be displaced in the east direction from the related dE/dt source and that the easterly displacement often dominated the total separation. In this study, the dE/dt and x rays sources were separated in the easting direction by means of 2.6 m, 4.8 m, and 6.9 m for the three events (UF 11 32, UF 11 35, and MSE 11 01). Likewise, the sources were separated in the northing direction for the three events by means of 7.0 m, 6.9 m, and 0.2 m. Finally, the sources were separated in the altitude direction for the three events by means o f 20.5 m, 18.2 m, and 28.4 m. Obviously, in the present study, the altitude displacement dominates the total separation, accounting for 90.3%, 63%, and 72% of the total separation of the dE/dt and x ray sources. The relatively random distributions of x r ays sources about the lightning channel for each event were shown graphically in the two dimensional projections views of the dE/dt and x ray source locations in Figures 8 19, 8 27, and 8 33. For the three events in this study where x ray sources were d etermined, the x rays were emitted following the related dE/dt pulse peaks in all 30 cases. The emission times of the dE/dt pulse peaks led the x ray sources by an average of 150 ns (GM of 85 ns) for the second stroke of flash UF 11 32, by an average of 2 90 ns (GM of 170 ns) for the fourth stroke of flash UF 11 35, and by an average of 280 ns (GM of 150 ns) for the second stroke of flash MSE 11 01. X ray sources in flash UF 11 32 were emitted from 20 ns to 340 ns following the dE/dt pulse peak. In flash UF 11 35, x rays were emitted from 20 ns to 2.16 s after the dE/dt pulse peak. Finally, in flash MSE 11 01, x rays were emitted from 10 ns to 1.76 s following the dE/dt pulse peak. Spatial and temporal parameters of the x ray source locations for the t hree flashes (UF 11 32, UF

PAGE 473

473 11 35, and MSE 11 01) are provided in Table 8 10. About 57% (17 cases) of the x ray source locations were potentially associated with 2 4 dE/dt pulses, while the remaining 13 cases were associated with a single dE/dt pulse. For the x ray sources that were associated with multiple dE/dt pulses, there is not a clear method to determine which dE/dt pulse was exactly associated with the emission of the x ray. The relative emission times of all possible associated dE/dt pulses with respect to the x ray sources are included in the averages stated above. For the 13 cases with only a single dE/dt pulse, the emission times of the x rays followed the dE/dt sources by 20 ns to 340 ns, with an average time separation of about 120 ns. Howa rd et al. [2008] found that the x ray sources were emitted from 100 ns to 1.3 s following the dE/dt sources, which is within the range of values reported in this study. The fact that the x ray sources always follow the dE/dt sources for a common event su ggests that the generally downward movement of negative charge following the large current pulses associated with the formation of a leader step is likely associated with the propagation of streamers from the ending point in space of the step, and hence, t he necessary electric field conditions for x rays to be generated. As noted by Howard et al. [2008], the x rays are emitted coincident with the electrostatic field change following the step formation that results from the aforementioned charge movement. Of the 30 total x ray source locations, 28 x rays were located in association with dart stepped leader steps and two x rays were located in association with leader burst processes. X rays were detected in time correlation with the slow front pulses of a ll three flashes, but the x ray sources could not be determined. In Figure 8 34, the distribution of x ray source location altitudes is shown in histogram format for the 30 cases. The source altitudes are binned in 20 m increments. Calculated x ray sour ce altitudes ranged from 24 m to 395 m, with an average source altitude of 181 m (GM of 143 m) and standard deviation of 114 m. Twenty of the 30 x

PAGE 474

474 ray sources (~66.7%) were located at altitudes between 60 280 m. Four x ray sources (~13.3%) were located b elow 60 m, and six x ray sources (20%) were located above 300 m. Many x ray bursts were detected, but not successfully located when the dart stepped was above 300 m (flashes UF 11 35 and MSE 11 01) and below 60 m (all events). For x ray sources radiated from high altitudes (greater than 300 m), the propagation path lengths between the source and all ten TOA stations are obviously large, and hence, the x ray attenuation over the path length is significant (recall that x rays have an e folding distance in t he lower atmosphere of about 100 m). Lower energy x rays emitted during the step formation process are likely attenuated or scattered prior to reaching the ground based array of detectors. As a result, x ray source locations for leader steps at high alti tude can only be determined for steps that produce large fluxes of high energy x rays, which arrive at a sufficient number of ground based scintillators with adequate energy to be detected. Though the measurements reported by Saleh et al. [2009] indicate that the x ray emission is more or less isotropic from leader stepping processes, at times, high altitude leader steps produce x rays that are only detected strongly at a small subset of stations (indicating large, high energy x ray fluxes emitted by the s ource) with negligible detection of x ray photons at nearby sensors. This observation indicates there may be some degree of a beaming phenomena present, which could be related to the geometry of the propagating leader channel. The process of locating x r ay sources radiated when the leader channel is at low altitude (below 60 m) is more problematic than the high altitude case discussed above. At low altitudes, the path lengths between the source and the array of detectors can be similar to the high altitu de case for detectors that are located on the fringe of the network. The x rays that arrive at detectors located at large radial distances from the low altitude source experience similar atmospheric attenuation to the high altitude case described above. In contrast

PAGE 475

475 to the high altitude case, some of the path lengths between the source and the detectors, particularly for sources located centrally within the network, are very short (potentially only a few tens of meters). As a result, the emitted x rays ex perience comparatively little atmospheric attenuation before arriving at the close detectors. This scenario introduces a major problem for accurately selecting the arrival times of x rays at stations located different radial distances from the source. Th e closer stations often detect lower energy x rays that are emitted earlier in the step formation process while the distant stations often fail to detect these earlier emitted, lower energy x rays due to atmospheric attenuation. When the dominant x ray bu rst is emitted following the step formation, both the close and distant detectors record the signal. Unfortunately, it is nearly impossible to accurately distinguish the arrival time of the dominant x ray burst on the close detectors considering they have already been excited by lower energy photons. Recall that the x ray signal arrival times at each station are determined from the initial deflection from the system noise. The pulse structure is not well correlated across the sensor array following the i nitial deflection, and hence, later points on the waveform cannot be used as measured arrival times. When the relative arrival times of the dE/dt and x ray sources are examined at the same station for the low altitude case described above, the time differ ences between the sources are often much shorter in duration at the closer stations than for the more distant stations, effectively providing signal arrival times across the network that are not radiated from the same source. Not surprisingly, the TOA sol ution algorithm for such a scenario produces an unreasonable output. In order for the solution algorithm to produce a reasonable output, the selected arrival times at a sufficient number of stations must correspond to the same source, which for this study appears to occur most frequently when the leader channel is between 60 280 m.

PAGE 476

476 Figure 8 1. dE/dt TOA locations of the dart stepped leader preceding the tenth return stroke of triggered flash UF 11 15 on July 7, 2011. A) a three dimensional plot of the dE/dt TOA source locations The sources span 80 s and are color coded according to the key at far right in 8 s windows. Sources within 200 m of ground are plotted. The location of the Field (Ground) Launcher is annotated. B) the altitude project ion of the dE/dt TOA source locations is plotted versus source emission time. A regression line is fitted to the points to obtain an estimate for the leader velocity. The altitude of the intercepting wire is annotated.

PAGE 477

477 Figure 8 2. A 26 s plot of t he dE/dt measured at Station 3 in association with the dart stepped leader preceding the tenth stroke of flash UF 11 15. TOA located dE/dt pulses during the leader and post leader phases are annotated with increasing integers. The II Very Low channel bas e current is plotted in the background for reference.

PAGE 478

478 Table 8 1. dE/dt TOA source locations within 20 s of the return stroke for the dart stepped leader preceding the tenth return stroke of flash UF 11 15 on July 7, 2011. The pulse numbers correspond to the labeled pulses in Figure 8 2. The three dimensional spatial coordinates (X, Y, Z), emission time (T), and associated spatial uncertainties Pulse X (m) Y (m) Z (m) T (s) De signator 1 308.4 447.3 56.9 19.63 0.1 0.2 1.4 LS 2 309.3 453.2 59.5 16.96 0.1 0.1 0.1 LS 3 309.3 453.2 59.5 16.90 0.1 0.1 0.1 LS 4 305.2 451.7 61.1 16.88 0.1 0.1 0.2 LS 5 303.8 452.8 60.9 14.11 0.3 0.5 3.1 LS 6 305.8 455.1 64.4 13.41 0.2 0.3 1. 3 LS 7 307.3 451.8 45.9 11.46 0.1 0.1 0.2 LS 8 305.6 457.9 41.7 10.20 0.1 0.2 0.4 LS 9 300.0 455.9 48.8 6.18 0.1 0.1 0.4 LS 10 302.3 455.6 41.5 5.78 0.1 0.1 0.1 LS 11 303.2 453.1 34.1 3.90 0.4 0.5 4.2 LS 12 310.6 449.8 24.2 2.31 0.6 0.5 1.0 LB 13 303.8 456.3 20.5 1.84 0.1 0.1 0.7 LB 14 313.1 450.5 27.2 1.68 0.1 0.1 0.1 LB 15 306.6 453.0 12.6 1.18 0.6 0.6 4.0 SF 16 316.7 446.9 19.8 0.60 9.5 6.8 12.1 SF 17 300.3 454.3 13.6 0.45 0.4 0.1 0.9 SF 18 300.3 454.3 13.6 0.21 0.4 0.1 0.9 SF

PAGE 479

479 Figure 8 3. dE/dt and channel base current waveforms of the dart stepped leader preceding the tenth stroke of flash UF 11 15 on July 7, 2011. A) a 36 s dE/dt waveform measured at Station 3 plotted versus the II Very Low channel base current. The beginning of the steady current flow is annotated. B) a 10 s window of the dE/dt waveform in Figure 8 3A plotted versus the II High channel base current. The peak upward leader current is annotated.

PAGE 480

480 Figure 8 4. A 3 s plot of the measured dE /dt at NASA Station 1 plotted against the numerical dI/dt for the tenth stroke of flash UF 11 15.

PAGE 481

481 Figure 8 5. dE/dt and electric field waveforms of the dart stepped leader preceding the tenth stroke of flash UF 11 15 on July 7, 2011. A) a 26 s elec tric field waveform of the tenth stroke of flash UF 11 15 obtained by numerically integrating the dE/dt wavefo rm measured at NASA Station 1, and B) the 26 s dE/dt waveform measured at NASA Station 1 The beginning of the flow front, the duration of the s low front, and the locations of the fast transition field changes are annotated.

PAGE 482

482 Figure 8 6. A sequence of 18 Photron high speed video frames (60 s total) of the dart stepped leader preceding the tenth stroke of flash UF 11 15. The altitude of the intercepting wire ring is annotated. The frames span 138 m in altitude and about 14 m in horizontal extent. Photos courtesy of the author.

PAGE 483

483 Figure 8 7. dE/dt and x ray waveforms of the dart stepped leader preceding the tenth stroke of flash UF 11 15 on July 7, 2011. A) a 60 s dE/dt record measured at Station 3, and B) a corresponding 60 s record of the x ray emission recorded by the plastic scintillator at Station 3. There were no located x ray sources for this triggered lightning event.

PAGE 484

484 Figu re 8 8. dE/dt TOA locations of the d art stepped leader preceding the first stroke of triggered flash UF 11 25 on August 5, 2011. A) a three dimensional plot of the dE/dt TOA source locations The sources span 400 s and are color coded according to the key at far right in 40 s windows. Sources within 325 m of ground are plotted. The location of the Field (Ground) Launcher is annotated. B) the altitude projection of the dE/dt TOA source locations is plotted versus source emission time. A regression l ine is fitted to the points to obtain an estimate for the leader velocity. The altitude of the intercepting wire is annotated.

PAGE 485

485 Figure 8 9. A 34 s plot of the dE/dt measured at Station 3 in association with the dart stepped leader preceding the firs t stroke of flash UF 11 25. TOA located dE/dt pulses during the leader and post leader phases are annotated with increasing integers. The II Very Low channel base current is plotted in the background for reference.

PAGE 486

486 Table 8 2. dE/dt TOA source locations within 25 s of the return stroke for the dart stepped leader preceding the tenth return stroke of flash UF 11 25 on August 5, 2011. The pulse numbers correspond to the labeled pulses in Figure 8 9. The three dimensional spatial coordinates (X, Y, Z), em ission time (T), and associated spatial uncertainties Pulse X (m) Y (m) Z (m) T (s) Designator 1 304.1 459.3 57.7 24.43 0.02 0.04 0.06 LS 2 307.2 451.7 46.0 24.18 0.01 0.01 0.01 LS 3 307.2 451.7 46.0 23.99 0.01 0.01 0.02 LS 4 30 4.6 461.6 46.9 23.94 0.03 0.06 0.07 LS 5 307.2 451.7 46.0 23.73 0.01 0.01 0.01 ? 6 305.6 458.1 41.9 23.58 0.01 0.02 0.07 LS 7 307.2 451.7 45.8 20.53 0.03 0.03 0.19 LS 8 305.7 451.5 30.4 16.45 0.01 0.03 0.12 LS 9 307.2 451.7 45.9 16.24 0.02 0.02 0.06 ? 10 304.2 454.4 34.1 16.10 0.01 0.01 0.02 LS 11 305.7 451.5 30.4 13.97 0.01 0.03 0.12 LS 12 304.7 452.3 32.7 13.74 0.04 0.05 0.10 LS 13 306.9 451.1 31.9 13.57 0.47 0.36 0.86 LS 14 303.2 452.4 28.8 12.60 0.28 0.41 0.89 LS 15 305.9 450.8 27. 4 7.91 0.01 0.01 0.02 LS 16 306.1 454.9 27.3 7.31 0.09 0.18 0.32 LS 17 303.2 452.3 29.0 5.96 0.16 0.16 0.45 LS 18 305.9 450.8 27.4 4.70 0.01 0.02 0.02 LB 19 305.5 451.1 27.0 4.34 0.04 0.04 0.05 LB 20 304.5 451.9 21.9 3.82 0.01 0.01 0.03 LB 21 304.2 454.0 23.2 3.35 0.04 0.07 0.12 LB 22 301.3 453.9 14.3 1.75 0.24 0.23 0.65 SF 23 304.9 454.7 17.3 1.22 0.01 0.02 0.05 SF 24 304.3 454.0 16.1 0.96 0.15 0.25 9.38 SF 25 303.1 452.6 33.8 0.90 0.12 0.21 0.35 SF 26 303.2 452.8 33.7 0.87 0.03 0.0 4 0.07 SF 27 306.9 451.3 36.6 0.84 0.15 0.12 0.26 SF 28 307.2 451.7 45.8 0.58 0.03 0.03 0.19 SF 29 308.0 460.0 56.8 0.54 1.03 1.48 6.27 SF

PAGE 487

487 Figure 8 10. dE/dt and channel base current waveforms of the dart stepped leader preceding the first stro ke of flash UF 11 25. A) a 50 s dE/dt waveform measured at Station 3 plotted versus the I I Very Low channel base current. The beginning of the steady current flow is annotated. B) a 24 s window of the dE/dt waveform shown in Figure 8 10A plotted versu s the II High channel base current. The peak upward leader current is annotated.

PAGE 488

488 Figure 8 11. A 3 s plot of the measured dE/dt at NASA Station 1 plotted against the numerical dI/dt for the first stroke of flash UF 11 25.

PAGE 489

489 Figure 8 12. dE/dt and electric field waveforms of the dart stepped leader preceding the first stroke of flash UF 11 25 on August 5, 2011. A) a 20 s electric field waveform obtained by numerically integrating the dE/dt waveform measured at NASA Station 1. B) a 20 s dE/dt wav eform measured at NASA Station 1. The beginning of the flow front, the duration of the slow front, and the locations of the fast transition field changes are annotated.

PAGE 490

490 Figure 8 13. dE/dt TOA locations of the dart stepped leader preceding the seco nd stroke of triggered flash UF 11 32 on August 18, 2011. A) a three dimensional plot of the dE/dt TOA source locations The sources span 60 s and are color coded according to the key at far right in 6 s windows. Sources within 240 m of ground are plo tted. Locations of x ray sources are shown as red diamonds in the three dimensional plot at the left. The location of the Field (Ground) Launcher is annotated. B) the altitude projection of the dE/dt TOA source locations is plotted versus source emissio n time. A regression line is fitted to the points to obtain an estimate for the leader velocity. The altitude of the intercepting wire is annotated.

PAGE 491

491 Figure 8 14. An 18 s plot of the dE/dt measured at Station 3 in association with the dart stepped leader preceding the second stroke of flash UF 11 32. TOA located dE/dt pulses during the leader and post leader phases are annotated with increasing integers. The II Very Low channel base current is plotted in the background for reference.

PAGE 492

492 Table 8 3. d E/dt TOA source locations within 14 s of the return stroke for the dart stepped leader preceding the tenth return stroke of flash UF 11 32 on August 18, 2011. The pulse numbers correspond to the labeled pulses in Figure 8 14. The three dimensional spati al coordinates (X, Y, Z), emission time (T), and associated spatial Pulse X (m) Y (m) Z (m) T (s) Designator 1 296.9 442.6 115.7 13.05 0.04 0.06 0.20 ? 2 300.1 443. 8 77.5 12.56 0.07 0.07 0.36 LS 3 295.1 444.9 89.9 11.20 0.32 0.37 0.96 ? 4 291.2 441.8 78.2 10.25 0.03 0.06 0.14 LS 5 294.7 444.7 75.5 9.68 0.02 0.02 0.06 LS 6 293.0 447.8 50.1 8.30 0.07 0.08 0.36 LS 7 294.0 446.5 57.4 6.51 0.13 0.20 0.77 LS 8 294.6 442.8 46.3 5.97 0.07 0.14 0.51 LS 9 297.4 444.5 40.9 4.11 0.14 0.29 1.47 LS 10 296.1 445.2 43.0 2.87 0.30 0.26 1.62 LB 11 299.0 443.5 38.0 2.41 0.01 0.01 0.01 LB 12 299.3 448.4 22.8 1.61 0.58 0.83 2.22 SF 13 300.5 453.7 53.4 1.30 0.05 0.09 0.43 SF 14 300.5 453.7 53.4 0.78 0.05 0.09 0.43 SF

PAGE 493

493 Figure 8 15. dE/dt and channel base current waveforms of the dart stepped leader preceding the second stroke of flash UF 11 32. A) a 26 s dE/dt waveform measured at Station 3 plotted versus the I I Very Low channel base curren The beginning of the steady current flow is annotated. B) a 10 s window of the dE/dt waveform shown in Figure 8 15A plotted versus the II High channel base current. The peak upward leader current is annotated.

PAGE 494

494 Figure 8 16. A 3 s plot of the measured dE/dt at NASA Station 1 plotted against the numerical dI/dt for the second stroke of flash UF 11 32.

PAGE 495

4 95 Figure 8 17. dE/dt and electric field waveforms of the dart stepped leader preceding the second stroke of flash UF 11 32. A) a 12 s electric field waveform obtained by numerically integrating the dE/dt waveform measured at NASA Station 1. B) a 12 s dE/dt waveform measured at NASA Station 1. The beginning of the flow front, the duration of the slow front, and the l ocations of the fast transition field changes are annotated.

PAGE 496

496 Figure 8 18. dE/dt and x ray waveforms of the dart stepped leader preceding the second stroke of flash UF 11 32. A) a 30 s dE/dt record measured at Station 3. The dE/dt pulses associated with the three located x ray sources for the dart stepped leader are annotated with increasing integers. B) a corresponding 30 s record of the x ray emission recorded by the plastic scintillator at Station 3

PAGE 497

497 Table 8 4. dE/dt and X ray TOA source loc ations for the dart stepped leader preceding the second return stroke of flash UF 11 32 on August 18, 2011. The case numbers correspond to the labeled pulses in the top panel of Figure 8 18. The three dimensional spatial coordinates (X, Y, Z), emission t ime (T), and associated spatial ray pulse. In addition, for each case, the total distance from the dE/dt pulse to the X ray pulse e dE/dt pulse to the X ray ray Case Pulse X (m) Y (m) Z (m) T (s) (m) (m) (m) (m) (m) (m) (m) (s) 1 dE/dt 1 292.3 449.8 1 11.9 15.72 0.05 0.10 0.26 26.8 0.9 5.7 26.2 0.09 X ray 291.4 455.5 85.7 15.63 17.16 13.61 29.69 2 dE/dt 1 297.4 444.5 40.9 4.11 0.08 0.13 1.26 17.4 6.8 1.5 16.0 0.02 X ray 290.6 446.0 24.9 4.09 7.55 12.20 25.22 3 dE/dt 1 296.1 445.2 4 3.0 2.87 0.30 0.26 1.62 23.9 0.0 13.9 19.4 0.34 X ray 296.1 459.1 23.6 2.53 0.43 0.75 1.83

PAGE 498

498 Figure 8 19. Three projection views of the dE/dt and x ray source locations for the dart stepped leader preceding the second stroke of flash UF 11 32 A) the alti tude versus easting projection, B) the altitude ve rsus northing projection, and C) the altitude versus time projection. dE/dt sources are color coded according to the key at right in 6 s windows. X ray sources are plotted as black diamonds independent of emission time. The altitude of the intercepting wire is annotated.

PAGE 499

499 Figure 8 20. dE/dt TOA locations of the dart stepped leader preceding the fourth stroke of triggered flash UF 11 35 on August 18, 2011. A) a three dimensional plot of the dE/dt TOA source locations. The sources span 240 s and are color coded according to the key at far right in 24 s windows. Sources within 750 m of ground are plotted. The location of the Field (Ground) Launcher is annotated. Locations of x ray so urces are shown as black diamonds in the three dimensional plot at left. B) the altitude projection of the dE/dt TOA source locations is plotted versus source emission time. Regression lines are fitted to the points to obtain estimates for the leader vel ocity above and below the altitude of the triggering wire. The altitude of the intercepting wire is annotated.

PAGE 500

500 Figure 8 21. An 18 s plot of the dE/dt measured at Station 3 in association with the dart stepped leader preceding the tenth stroke of fl ash UF 11 35. TOA located dE/dt pulses during the leader and post leader phases are annotated with increasing integers. The II Very Low channel base current is plotted in the background for reference.

PAGE 501

501 Table 8 5. dE/dt TOA source locations within 12 s o f the return stroke for the dart stepped leader preceding the fourth return stroke of flash UF 11 35 on August 18, 2011. The pulse numbers correspond to the labeled pulses in Figure 8 21. The three dimensional spatial coordinates (X, Y, Z), emission time (T), and associated spatial Pulse X (m) Y (m) Z (m) T (s) Designator 1 300.0 461.6 100.3 11.39 0.09 0.19 0.35 LS 2 301.9 453.5 88.0 10.93 0.14 0.17 0.29 LS 3 297 .6 459.7 86.0 10.33 0.13 0.16 0.26 LS 4 297.9 456.1 87.3 10.21 0.00 0.00 0.02 LS 5 307.2 452.2 72.6 9.92 0.15 0.07 0.23 LS 6 303.6 454.7 76.6 9.19 0.19 0.17 0.59 LS 7 301.4 454.0 76.5 8.38 0.31 0.34 0.87 LS 8 302.8 452.1 69.0 7.24 0.64 0.53 1.61 LS 9 312.1 459.1 74.8 6.58 0.04 0.05 0.13 LB 10 298.6 451.0 64.4 6.46 0.23 0.46 1.16 LB 11 300.7 455.0 60.8 6.33 0.19 0.45 2.06 LB 12 300.0 451.2 47.2 6.20 0.04 0.09 0.52 LB 13 300.0 451.2 47.2 5.90 0.04 0.09 0.52 LB 14 305.2 451.7 61.1 5.84 0 .04 0.05 0.09 LB 15 304.1 449.8 45.5 5.77 0.01 0.01 0.05 LB 16 311.9 444.4 28.0 5.38 0.08 0.05 0.17 ? 17 300.1 455.8 51.1 5.14 0.01 0.01 0.04 ? 18 300.6 455.9 44.3 4.43 0.00 0.01 0.05 ? 19 300.1 455.8 51.1 4.34 0.01 0.01 0.04 ? 20 299.3 451.4 34 .3 3.45 0.82 1.12 13.69 SF 21 311.8 454.4 47.1 1.87 0.26 0.32 1.29 SF 22 301.4 451.5 22.3 1.83 0.00 0.00 0.02 SF 23 311.8 454.4 47.1 1.71 0.26 0.32 1.29 SF 24 298.7 451.7 34.6 1.14 0.18 0.17 1.89 SF 25 299.5 451.5 30.2 0.82 0.27 0.31 5.03 SF 2 6 298.8 452.8 18.2 0.33 0.05 0.06 0.92 SF 27 306.4 450.9 28.3 0.21 0.14 0.07 0.56 SF

PAGE 502

502 Figure 8 22. dE/dt and channel base current waveforms of the dart stepped leader preceding the fourth stroke of flash UF 11 35. A) a 30 s dE/dt waveform measure d at Station 3 plotted versus the I I Very Low channel base current. The beginning of the steady current flow is annotated. B) a 12 s window of the dE/dt waveform shown in Figure 8 22A plotted versus the II High channel base current. The peak upward lea der current is annotated.

PAGE 503

503 Figure 8 23. A 3 s plot of the measured dE/dt at NASA Station 1 plotted against the numerical dI/dt for the tenth stroke of flash UF 11 35.

PAGE 504

504 Figure 8 24. A still photograph of flash UF 11 35 taken from the Launch Control trailer. Two unconnected upward streamers were recorded propagating from the intercepting wire ring. The leftmost discharge (the longer of the two) has approximate length of 1 m. Photo courtesy of the author.

PAGE 505

505 Figure 8 25. dE/dt and electric field waveforms of the dart stepped leader preceding the fourth stroke of flash UF 11 35. A) a 16 s electric field waveform of the tenth stroke of flash UF 11 35 obtained by numerically integrating the dE/dt waveform measured at NASA Station 1. B) a 16 s dE/ dt waveform measured at NASA Station 1 The beginning of the flow front, the duration of the slow front, and the locations of the fast transition field changes are annotated.

PAGE 506

506 Figure 8 26. dE/dt and x ray waveforms of the dart stepped leader precedin g the fourth stroke of flash UF 11 35. A) a 90 s dE/dt record measured at Station 3. The dE/dt pulses associated with the 13 located x ray sources for the dart stepped leader are annotated in the top panel with increasing integers. B) a corresponding 9 0 s record of the x ray emission recorded by the plastic scintillator at Station 3

PAGE 507

507 Table 8 6. dE/dt and X ray TOA source locations for the dart stepped leader preceding the second return stroke of flash UF 11 35 on August 18, 2011. The case numbers correspond to the labeled pulses in the top panel of Figure 8 26. The three dimensional spatial coordinates (X, Y, Z), emission time (T), and associated spatial ray pulse. In addition, for each case, the total total distance from the dE/dt pulse to the X ray pulse ray ray ) are given. Case Pulse X (m) Y (m) Z (m) T (s) (m) (m) (m) (m) (m) (m) (m) (s) 1 dE/dt 1 203.8 463.0 336.4 83.52 0.38 0.37 1.18 17.7 8.8 10.1 11.6 0.16 dE/dt 2 208.6 465.7 333.5 83.46 5.75 5.72 18.01 12.1 4.0 7.4 8.7 0.10 X ray 212.6 473.1 324.8 83.36 0.61 0.63 2.13 2 dE/dt 1 228.1 457.2 314.6 76.37 0.19 0.39 1.11 12.9 5.6 3.1 11.2 0.06 X ray 222.5 460.3 303.4 76.31 0.03 0.05 0.13 3 dE/dt 1 257.9 451.9 280.9 67.30 4.42 5.15 17.68 38.2 21.5 29.7 10.6 0.18 X ray 279.4 481.6 270.3 67.12 1.75 2.54 4.73 4 dE/dt 1 307.0 471.8 211.4 40.14 0.01 0.01 0.02 52.9 4.9 7.8 52.1 2.16 dE/dt 2 309.0 470.2 216.8 38.84 0.22 0.14 0.36 58.2 6.9 6.2 57.5 0.86 dE/dt 3 319.2 459.9 183.8 38.28 0.03 0.04 0.06 30.2 17.1 4.1 24.5 0.30 dE/dt 4 300.1 464.3 199.7 38.15 0.01 0.01 0.01 40.5 2.0 0.3 40.4 0.17 X ray 302.1 464.0 159.3 37.98 0.25 0.46 1.25 5 dE/dt 1 299.3 461.8 174.2 28.65 0.08 0.15 0.33 18.1 9.7 6.3 13.9 0.13 dE/dt 2 310.6 462.2 16 1.5 28.54 0.05 0.09 0.21 21.8 21.0 5.9 1.2 0.02 X ray 289.6 468.1 160.3 28.52 0.92 2.07 4.97 6 dE/dt 1 319.3 452.7 138.5 24.58 0.04 0.05 0.08 40.6 35.5 13.3 14.6 0.16 dE/dt 2 319.3 452.7 138.5 24.48 0.04 0.05 0.08 40.6 35.5 13.3 14.6 0.06 X ray 283.8 466.0 123.9 24.42 0.89 0.98 2.32 7 dE/dt 1 300.1 459.8 140.0 22.72 0.05 0.11 0.29 20.3 4.5 8.8 17.7 0.23 dE/dt 2 313.9 457.6 124.2 22.55 0.33 0.43 1.21 21.4 18.3 11.0 1.9 0.06 X ray 295.6 468.6 122.3 22.49 0.71 0.50 1.09 8 dE/dt 1 302.9 455.6 122.3 20.55 0.13 0.17 0.47 12.0 7.1 7.4 6.2 0.09 X ray 295.8 448.2 128.5 20.46 5.82 5.24 16.14 9 dE/dt 1 309.4 454.8 110.4 17.82 0.06 0.04 0.07 20.0 10.7 16.4 4.0 0.50 dE/dt 2 309.4 454.8 110.4 17.55 0.06 0.04 0.07 2 0.0 10.7 16.4 4.0 0.23 X ray 298.7 471.2 114.4 17.32 0.74 0.69 1.77 10 dE/dt 1 297.3 460.3 104.9 14.95 0.31 0.43 1.82 37.6 21.0 14.8 27.4 0.27 dE/dt 2 299.3 462.2 101.4 14.89 0.03 0.08 0.17 33.1 19.0 12.9 23.9 0.21 X ray 318.3 475.1 77 .5 14.68 0.26 0.54 0.68 11 dE/dt 1 301.9 453.5 88.0 10.93 0.14 0.17 0.29 28.5 9.8 13.8 22.9 0.06 X ray 311.7 467.3 65.1 10.87 0.04 0.04 0.07 12 dE/dt 1 302.8 452.1 69.0 7.24 0.64 0.53 1.61 34.1 0.3 32.0 11.7 0.04 X ray 303.1 420.1 57 .3 7.20 10.60 10.99 20.73 13 dE/dt 1 312.1 459.1 74.8 6.58 0.04 0.05 0.13 41.1 13.7 4.8 38.4 0.49 dE/dt 2 298.6 451.0 64.4 6.46 0.23 0.46 1.16 30.8 0.2 12.9 28.0 0.37 dE/dt 3 300.7 455.0 60.8 6.33 0.19 0.45 2.06 26.1 2.3 8.9 24.4 0.24 dE /dt 4 300.0 451.2 47.2 6.20 0.04 0.09 0.52 16.7 1.6 12.7 10.8 0.11 X ray 298.4 463.9 36.4 6.09 1.08 0.36 1.27

PAGE 508

508 Figure 8 27. Three projection views of the dE/dt and x ray source locations for the dart stepped leader preceding the fourth str oke of flash UF 11 35. A) the altitude versus easting projection B) the altitude versus northing projection and C) the altitude versus time projection. dE/dt sources are color coded according to the key at right in 24 s windows. X ray sources are plo tted as black diamonds independent of emission time. The altitude of the intercepting wire is annotated.

PAGE 509

509 Figure 8 28. dE/dt TOA locations of the dart stepped leader preceding the second stroke of natural flash MSE 11 01 on July 7, 2011. A) a three d imensional plot of the dE/dt TOA source locations. The sources span 210 s and are color coded according to the key at far right in 21 s windows. Sources within 475 m of ground are plotted. X ray source locations are plotted in the three dimensional vi ew at left as black diamonds. The location of the Field (Ground) Launcher is annotated. B) the altitude projection of the dE/dt TOA source locations is plotted versus source emission time. A regression line is fitted to the points to obtain an estimate for the leader speed.

PAGE 510

510 Figure 8 29. The easting versus altitude projection of the dE/dt source locations for the second stroke of flash MSE 11 01 overlaid on a still photograph taken from IS2, which is located on the northern boundary of the ICLRT. T he strike point is about 395 m from the camera location. The sources span 210 s and are color coded according to the key at right in 21 s time windows. The deviation of the dE/dt sources from the lightning channel above 275 m is explained in Section 8. 2.5. Photo courtesy of the author.

PAGE 511

511 Figure 8 30. A 35 s plot of the dE/dt measured at Station 3 in association with the dart stepped leader preceding the second stroke of natural flash MSE 11 01. TOA located dE/dt pulses during the leader and post le ader phases are annotated with increasing integers.

PAGE 512

512 Table 8 7. dE/dt TOA source locations within 23 s of the return stroke for the dart stepped leader preceding the second return stroke of natural flash MSE 11 01 on July 7, 2011. The pulse numbers cor respond to the labeled pulses in Figure 8 29. The three dimensional spatial coordinates (X, Y, Z), emission time (T), and associated spatial Pulse X (m) Y (m) Z (m) T (s) (m) Designator 1 159.4 314.9 72.6 22.56 0.06 0.06 0.20 LS 2 157.0 309.6 78.0 22.36 0.06 0.11 0.16 LS 3 157.4 309.9 78.0 21.95 0.01 0.00 0.02 LS 4 145.5 310.4 62.9 17.76 0.60 0.20 0.76 LS 5 157.3 309.9 78.0 17.43 0.02 0.01 0.03 LS 6 154.1 311.2 59.5 15.71 0.28 0.13 2.14 LS 7 147.8 302.7 47.2 15.15 0.07 0.11 0.27 LS 8 155.2 310.6 25.3 14.12 0.12 0.06 1.88 ? 9 152.2 309.6 59.4 13.48 0.07 0.03 0.70 LB 10 152.2 309.6 59.4 13.17 0.07 0.03 0.70 LB 11 147.8 302.7 47.2 12.69 0.07 0.11 0.27 LB 12 150.7 307.9 49.6 12.33 0.11 0.05 0.95 LB 13 149.7 309.1 55.2 11.57 0.34 0.19 0.48 LB 14 141.7 314.5 52.8 11.11 0.06 0.04 0.07 LB 15 148.5 307.9 53.6 10.74 0.14 0.07 0.25 LB 16 149.5 310.1 27.4 10.54 0.34 0.19 0.77 ? 17 145.1 306.3 43.6 8.55 0.07 0.04 0.14 ? 18 148.4 302.9 50.2 8.06 0.27 0.17 0.37 ? 19 146.7 304.9 58.8 7.09 0.00 0.00 0.04 ? 20 147.0 307.5 23.3 6.63 0.06 0.05 0.27 ? 21 136.4 299.9 23.9 6.45 0.18 0.06 0.54 ? 22 142.9 306.1 21.7 4.72 0.00 0.00 0.07 SF 23 144.0 301.6 3 2.5 4.10 0.15 0.09 0.41 SF 24 142.9 306.1 21.7 3.68 0.00 0.00 0.07 SF 25 151.8 293.8 29.1 3.44 0.14 0.07 0.11 SF 26 144.0 301.6 24.6 2.76 0.06 0.04 0.12 SF 27 135.9 299.8 25.2 2.57 0.38 0.23 3.45 SF 28 132.8 302.1 21.1 1.36 3.48 1.79 39.46 SF 2 9 146.5 313.5 19.7 0.59 0.27 0.14 7.24 SF 30 147.0 309.4 23.1 0.33 0.19 0.11 4.65 SF 31 143.8 311.5 36.2 0.00 0.20 0.14 1.31 FT

PAGE 513

513 Figure 8 31. dE/dt and electric field waveforms of the dart stepped leader preceding the second stroke of natural flash MSE 11 01. A) a 35 s electric field waveform obtained by numerically integrating the dE/dt waveform measured at Station 3. B) a 35 s dE/dt waveform measured at Station 3 The beginning of the flow front, the duration of the slow front, and the l ocations of the fast transition field changes are annotated.

PAGE 514

514 Figure 8 32. dE/dt and x ray waveforms of the dart stepped leader preceding the second stroke of natural flash MSE 11 01. A) a 275 s dE/dt record measured at Station 3. The dE/dt pulse s associated with the 14 located x ray sources for this dart stepped leader are annotated in the top panel with increasing integers. B) a corresponding 275 s record of the x ray emission recorded by the plastic scintillator at Station 3

PAGE 515

515 Table 8 8. dE /dt and x ray TOA source locations (X, Y, Z), location uncertainties ( stepped leader preceding the second return stroke of flash MSE 11 01 on July 7, 2011. Case Pulse X (m) Y (m) Z (m) T (s) (m) (m) ( m) (m) (m) (m) (m) (s) 1 dE/dt 1 296.1 329.2 446.1 222.33 0.05 0.03 0.11 55.6 15.5 10.6 52.3 0.12 X ray 280.6 339.8 393.8 222.21 6.24 6.16 30.29 2 dE/dt 1 274.6 318.5 419.3 201.18 0.13 0.10 0.58 38.0 27.4 9.5 24.6 0.11 dE/d t 2 269.6 322.5 430.7 201.16 0.02 0.02 0.07 42.8 22.4 5.5 36.0 0.09 X ray 247.2 328.0 394.7 201.07 0.15 0.14 0.93 3 dE/dt 1 262.2 319.0 405.2 190.45 0.72 1.10 2.04 32.2 20.6 16.8 18.2 0.12 X ray 241.6 302.2 387.0 190.33 3.39 2.77 11.94 4 dE/dt 1 253.9 312.2 405.0 185.75 0.03 0.02 0.09 64.4 15.1 1.0 62.6 0.31 X ray 269.0 313.2 342.4 185.44 0.93 1.12 2.84 5 dE/dt 1 248.3 301.2 319.1 152.86 0.03 0.06 0.21 46.0 4.1 9.4 44.8 0.29 dE/dt 2 243.0 297.1 326.8 152.83 0.14 0.14 0 .89 55.0 9.4 13.5 52.5 0.26 dE/dt 3 253.2 299.8 318.0 152.73 0.27 0.25 0.96 45.0 0.8 10.8 43.7 0.16 dE/dt 4 253.5 303.0 330.5 152.71 0.00 0.00 0.01 56.7 1.1 7.6 56.2 0.14 X ray 252.4 310.6 274.3 152.57 0.34 0.29 1.77 6 dE/dt 1 257.4 296. 9 309.9 146.32 0.03 0.08 0.08 70.1 18.2 2.1 67.7 0.30 dE/dt 2 250.5 310.1 305.6 146.22 0.15 0.12 0.31 65.3 11.3 11.1 63.4 0.20 dE/dt 3 247.0 288.7 297.0 146.10 0.03 0.02 0.19 56.3 7.8 10.3 54.8 0.08 X ray 239.2 299.0 242.2 146.02 0.39 0.33 1.13 7 dE/dt 1 243.9 309.9 282.1 124.34 0.98 0.56 1.27 47.3 45.8 2.4 11.5 0.07 dE/dt 2 240.4 278.2 264.9 124.28 0.31 0.23 0.79 51.8 42.3 29.3 5.7 0.01 X ray 198.1 307.5 270.6 124.27 1.21 0.90 5.13 8 dE/dt 1 218.7 306.8 253.8 106.35 0.03 0 .02 0.07 31.0 5.9 9.1 29.0 0.40 dE/dt 2 235.2 306.0 245.4 105.97 0.14 0.22 0.99 24.6 10.6 8.3 20.6 0.02 X ray 224.6 297.7 224.8 105.95 0.26 0.25 0.72 9 dE/dt 1 206.8 316.4 224.7 90.74 0.14 0.09 0.32 32.9 12.6 16.6 25.5 0.13 X ray 219.4 29 9.8 199.2 90.61 0.96 0.59 2.29 10 dE/dt 1 190.5 310.0 173.1 75.53 0.04 0.04 0.11 19.3 16.6 0.1 9.9 0.77 dE/dt 2 190.5 310.0 173.1 74.87 0.04 0.04 0.11 19.3 16.6 0.1 9.9 0.11 dE/dt 3 189.5 309.8 174.1 74.82 0.10 0.07 0.34 19.0 15.6 0.1 10.9 0 .06 dE/dt 4 185.5 313.9 189.6 74.80 0.01 0.01 0.02 29.1 11.6 4.0 26.4 0.04 X ray 173.9 309.9 163.2 74.76 1.69 0.95 2.84 11 dE/dt 1 175.0 304.7 142.2 57.72 0.12 0.05 0.45 25.9 2.0 9.0 24.2 0.05 X ray 177.0 313.7 118.0 57.67 0.15 0.11 0.39 12 dE/dt 1 167.1 305.4 145.6 53.34 0.04 0.03 0.15 41.6 33.3 8.2 23.5 1.22 dE/dt 2 176.8 309.3 124.2 52.21 0.18 0.14 0.45 44.7 43.0 12.1 2.1 0.09 X ray 133.8 297.2 122.1 52.12 8.32 1.49 29.56 13 dE/dt 1 172.3 302.6 130.9 48.79 0.07 0.04 0.51 13.6 13.1 2.2 2.7 1.76 dE/dt 2 169.1 302.7 135.3 47.29 0.01 0.01 0.06 16.5 16.3 2.3 1.7 0.26 dE/dt 3 165.6 307.2 130.2 47.15 0.01 0.01 0.01 21.2 19.8 6.8 3.4 0.12 X ray 185.4 300.4 133.6 47.03 4.19 3.05 20.51 14 dE/dt 1 163.7 313. 8 135.8 41.23 0.06 0.04 0.12 43.4 21.6 5.7 37.2 0.84 dE/dt 2 172.8 311.9 107.2 40.66 0.01 0.01 0.02 15.6 12.5 3.8 8.6 0.27 dE/dt 3 163.2 312.1 144.5 40.49 0.07 0.04 0.23 51.1 22.1 4.0 45.9 0.10 X ray 185.3 308.1 98.6 40.39 6.93 3.94 24.57

PAGE 516

516 Figure 8 33. Three projection views of the dE/dt and x ray source locations for the dart stepped leader preceding the second stroke of flash MSE 11 01. A) the alt itude versus easting projection, B) the altitude versus northing projection and C) th e altitude versus time projection. dE/dt sources are color coded according to the key at right in 21 s windows. X ray sources are plotted as black diamonds independent of emission time.

PAGE 517

517 Table 8 9. Measured and calculated statistics of the seven tri ggered lightning dart stepped leaders and one natural lightning dart stepped leader analyzed in this study. Flash Stroke Peak Current (kA) TOA Located Sources (dE/dt/X ray) Downward Leader Speed (x 10 6 m/s) Steady Current Interval (s) Junction Altitude d (m) Upward Leader Speed (x 10 5 m/s) Upward Leader Peak Current (A) Slow Front Duration (s) UF 11 15 10 10.2 49/0 2.2 14.4 3.1 2.2 440 4.0 UF 11 25 1 12.1 182/0 6.9 24 7.8 3.3 2190 4.6 UF 11 32 2 19.8 44/3 4.0 12.7 3.1 2.4 1990 0.6 UF 11 34 2 12.1 52 /0 2.7 12.5 0.0 270 1.2 UF 11 35 4 27.4 148/13 2.8 b 4.8 c 17.0 6.2 2.1 1360 1.7 UF 11 35 6 7.5 32/0 2.0 9.3 1.6 1.7 400 1.2 UF 11 35 7 8.4 28/0 3.2 8.7 3.1 3.6 380 MSE 11 01 2 22.1 a 180/14 1.9 6.9 a NLDN reported peak current b Downward lea der speed above the height of the triggering wire c Downward leader speed below the height of the triggering wire d Altitude above intercepting wire

PAGE 518

518 Table 8 10. Measured statistics of the 30 x ray source locations relative to the dE/dt sources Fla sh Stroke Peak Current (kA) TOA Located Sources (X ray) Total Average dE/dt & X ray Separation (m) Average Easting Separation (m) Average Northing Separation (m) Average Altitude Separation (m) Average Time Separation (dE/dt X ray) (ns) UF 11 32 2 19.8 3 22.7 2.6 7.0 20.5 150 ns (GM 85 ns) UF 11 35 4 27.4 13 29 4.8 6.9 18.2 290 ns (GM 170 ns) MSE 11 01 2 22.1 14 39.2 6.9 0.2 28.4 280 ns (GM 150 ns)

PAGE 519

519 Figure 8 34. Distribution of x ray source altitudes for the 30 sources that were TOA located i n flashes UF 11 32, UF 11 35, and MSE 11 01.

PAGE 520

520 CHAPTER 9 SUMMARY OF RESULTS A ND RECOMMENDATIONS F OR FUTURE RESEARCH 9 .1 Summary of Experimental Results The measurement network at the ICLRT underwent significant modifications and upgrades from 2009 to 2011. Many of these changes were described in detail in Chapter 2. Primary to work presented in this dissertation, the eight station TOA network described in Howard et al. [2008, 2010] was upgraded to include ten sensors, including at least one station within 45 m of the launching facility. The addition of more sensors, and the re location of rocket launching operations from the Tower Launcher to the Field (Ground) Launcher, which is more centrally located within the TOA network, improved the low altitude accu racy of the dE/dt TOA systems by a significant margin. The addition of the parallel HBM dE/dt TOA network in 2011 provided an improvement in sensitivity over the ICLRT DSO dE/dt system by a factor of two, and also provided substantially better quality wav eforms (six additional bits of resolution and essentially no system noise) than those transmitted over the analog fiber optic links and digitized on LeCroy DSOs. The eight station energetic radiation TOA network described in Howard et al. [2008, 2010], wh ich utilized NaI scintillation detectors with inherently long light decay times, was also substantially upgraded. Eight plastic scintillation detectors and two LaBr 3 scintillation detectors were added at the locations of the ten dE/dt sensors. Both the p lastic and LaBr 3 detectors have short light decay times compared to the NaI detectors, and hence, mitigate much of the photon pile up issues experienced with the slower NaI detectors. The addition of high speed cameras, particularly the Photron SA1.1, to the ICLRT measurement network provided a method to optically image leader processes at very high time resolution. These images, coupled with correlated measurements of the channel base current and dE/dt, provide valuable insight to the physics and mechani sms of leader propagation and attachment. Finally, the seven station LMA

PAGE 521

521 system installed prior to summer 2011 provided high resolution three dimensional mappings of the initial stages and subsequent leader/return stroke processes of triggered lightning e vents. The coordinated source location data from the dE/dt TOA networks and the LMA provide a relatively complete picture of the lightning propagation path from within about 20 m of ground to more than 12 km altitude. While not discussed in this disserta tion, the LMA also recorded all natural lightning activity surrounding the ICLRT during summer 2011. The three ICLRT TOA networks (the ICLRT DSO dE/dt and x ray networks and the HBM dE/dt network) were described in detail in Chapter 3 In 2009 and again in 2011, the locations of all TOA sensors were determined by a professional group of surveyors. The quality of the surveyed locations, which are accurate to within about 1 cm, contribute to the quality of the TOA solutions described in this dissertation. Detailed descriptions were also provided in Chapter 3 for determining cabling and fiber optic delays between the outputs of each TOA sensor and the digitizer inputs. The delays were measured with accuracy greater than the sampling resolution of the DSO d igitization system (4 ns). The determination of time delays through the photomultiplier tubes (PMT) mounted to the plastic scintillation detectors were also discussed. PMT delays have not previously been accounted for in TOA measurements of energetic rad iation at the ICLRT. Detailed descriptions and illustrations were provided regarding the author's procedure for aligning both dE/dt and energetic radiation waveforms acquired at the ten TOA stations to select commonly detected pulses, and then subsequentl y selecting the signal arrival times in a systematic and accurate manner. The details of the TOA solution algorithm, which utilizes a non linear least squares optimization technique based on the Levenburg Marquardt algorithm, were discussed as was the met ric used to determine the best

PAGE 522

522 solution. Finally, a discussion was provided of the TOA spatial location errors of the present network with comparison to the previous generation TOA network at the ICLRT. In Chapter 4, the procedures for cataloguing and d ocumenting collected lightning data were outlined Information was provided on the preparation of documentation files pertaining to measurement amplitude calibration factors, waveform file names, and general characteristics of each recorded event. Chapte r 4 also presented tables including information on all natural and triggered lightning events recording fro 2009 2011. For natural lightning events, parameters were provided including the NLDN reported peak current, multiplicity, and ground strike locatio n. For triggered lightning events, parameters of the initial stage were provided including the total charge transfer, the average current amplitude, and the full duration of the UPL/ICC process. The time duration between the initiation of the UPL and the ICV was reported for events where the ICV was clearly evident. For each triggered flash, the peak return stroke current and multiplicity were provided. A statistical analysis of the previously mentioned parameters for triggered lightning events between 2009 2011 was also given in Chapter 4 with data plotted graphically in histogram format for each individual year of study and for the entire dataset. The geometric mean (GM) of the UPL/ICC durations for the 2009 2011 dataset was 387 ms. This value was 27 % larger than the value of 305 ms reported by Miki et al. [2005] for 45 flashes triggered at the ICLRT, and 39% larger than the value of 279 ms reported by Wang et al. [1999] for 37 flashes triggered at Fort McClellan and at the ICLRT. UPL/ICC durations w ere recorded ranging from 160 945 ms. Similarly, the GM value of the UPL/ICC charge transfer for the 2009 2011 dataset was 50 C, 64% larger than the 30.4 C value given my Miki et al. [2005] and 85% larger than the 27 C value given by Wang et al. [1999]. Charge transfers of the UPL/ICC process were measured ranging from 8 225 C. The GM value of the average current

PAGE 523

523 amplitude during the UPL/ICC for the 2009 2011 dataset was 130 A, also 31% higher than the value of 99.6 A reported by Miki et al. [2005] and 3 5% higher than the value of 96 A reported by Wang et al. [1999]. Average current amplitudes of the UPL/ICC process were measured ranging from 49 834 A. Interestingly, the two events exh i biting abnormally large average current amplitudes during the UPL/I CC process (834 A for flash UF 09 30 and 328 A for flash UF 11 26) also exhibited current polarity reversals during the ICC. Flash UF 11 26 was discussed in detail in Chapter 7 of this dissertation and flash UF 09 30 was described in Yoshida et al. [2012] The GM time of the ICV relative to the initiation of the UPL for the 2009 2011 dataset was 7.5 ms, a value in good agreement with the previous study of Wang et al. [1999], who reported a GM value of 8.6 ms for 22 flashes at Fort McClellan and the ICLR T. Time durations between the UPL and the ICV were measured ranging from 2.1 69.4 ms. Finally, the action integral (or specific energy) was also measured between the initiation of the UPL and the ICV for the 37 flashes where the ICV was clearly evident. The GM action integral was calculated to be 119 A 2 s, a value in good agreement with the previous study of Wang et al. [1999], who reported a GM action integral of 110 A 2 s for 22 flashes at Fort McClellan and the ICLRT. The GM value of triggered lightni ng return stroke peak currents for the 2009 2011 dataset was 10.9 kA for 156 total return strokes. This value was found to agree well with past studies at the ICLRT and at the Kennedy Space Center in Florida, Fort McClellan in Alabama, and Saint Privat d' Allier in France. Return stroke peak currents were measured ranging from 1.5 46.5 kA. GM return stroke peak currents for strokes preceded by 121 dart leaders, 18 dart stepped leaders, and 17 "chaotic" dart leaders were 9.5 kA, 17.5 kA, and 1.7 kA, respec tively. The return stroke currents associated with dart stepped leaders and "chaotic" dart leaders were 84% and 86%

PAGE 524

524 higher, respectively, than those associated with dart leaders. In addition, dart stepped leaders were found to precede the first stroke fo llowing the IS in 32% of measured cases, while "chaotic" dart leaders were found to precede the first stroke in 47% of measured cases. The previous two statistical parameters relating leader type to triggered lightning return stroke peak current and strok e order are not found elsewhere in the literature to the best of the author's knowledge. For the 2009 2011 data, the GM flash multiplicity for triggered flashes was 3 return strokes. Eight flashes (21% of the total number) had seven or more return strok es following the IS. In Chapter 5, high speed video observations were presented of a natural lightning stepped leader (also discussed in Hill et al. [2011]) recording during summer 2010. The leader was photographed at a frame rate of 300 kfps ( an exposu re time of 3.33 s ) and represented the fastest time resolution images recorded of a natural lighting discharge to date. Stepped leader parameters were measured from the video including step length, interstep interval, leader speed, and the characteristic s of space stems/leaders for a total of three primary branches and five secondary branches. Step lengths for the primary branches and secondary branches ranged from 4.8 5.1 m and 5.4 7.1 m, respectively. Interstep intervals for primary and secondary bran ches ranged from 13.7 15.1 s and 12.2 40.0 s, respectively. Average two dimensional leader propagation speeds for the three primary branches were from 4.4 to 4.6 x 10 5 m/s, and for two secondary branches were 2.7 x 10 5 m/s and 6.2 x 10 5 m/s. For 82 le ader steps, it was found that the measured step lengths and interstep intervals were shorter than those reported in the literature from previous optical measurements employing the streak photography technique [e.g., Schonland et al. 1935; Schonland, 1956; Berger, 1967] and in better agreement with statistics measured for dart stepped leader steps [e.g., Schonland, 1956; Krider, 1977; Orville and Idone, 1982, 1984; Davis, 1999]. The leader step formation process was analyzed for the 82 individual

PAGE 525

525 steps. Si xteen instances of space stems/leaders were recorded preceding the formation of the leader step, the first examples of these optical phenomena recorded in association with stepped leader steps. The lengths of the space stems/leaders (average of 3.9 m) and the separation of the space stems/leaders from the previous leader channel (average of 2.1 m) were found to agree well with statistics measured for similar optical phenomena associated with triggered lightning dart stepped leaders at the ICLRT [e.g., Biag i et al., 2010]. For 28 leader steps, a luminosity wave was photographed propagating back up the existing leader channel following the step formation in either one, two, or three successive 3.33 s frames at average speed of about 7.5 x 10 6 m/s, a new obs ervation for stepped leader steps consistent with reports in the literature for optical (ALPS photodiode array) measurements of dart stepped leader steps associated with triggered lightning return strokes reported by Wang et al. [1999]. Chapter 6 presen ted observations of four "chaotic" dart leaders recorded preceding triggered lightning return strokes during summer 2010, two of which are analyzed in significant detail (flashes UF 10 13 and UF 10 24). Some of these observations are also discussed in Hil l et al. [2012a]. Prior to this study, t here were no previous reports in the literature of "chaotic" dart leaders associated with triggered lightning return strokes. The dE/dt signature of "chaotic" dart leaders within about 200 m of the ground and about 10 s of the return stroke is found to exhibit large amplitude high frequency variations not recorded for typical dart leaders. These pulses of radiation, with typical widths of several tens of nanoseconds, are superimposed on slower background field cha nges termed "bursts", which have typical widths of several hundred nanoseconds. The TOA locations of the individual pulses were determined for two "chaotic" dart leaders. The TOA locations of successive pulses were found to "bounce" up and down over vert ical ranges of often several tens of meters. Often, the calculated speeds between the

PAGE 526

526 locations of successive pulses exceeded the speed of light, indicating that the pulses were likely being radiated nearly simultaneously from multiple (perhaps many) spat ial locations near the tip of the descending leader. High speed video observations indicated that "chaotic" dart leaders often exhibit longer streamer zones than do dart or dart stepped leaders. In one case (flash UF 11 24), the downward "chaotic" dart l eader with streamer zone length of 25 m and an upward connecting leader with length of 11 m were imaged in the same frame. "Chaotic" dart leaders preceding triggered strokes were found to emit nearly continuous pulses of energetic radiation up to about 13 s prior to the return stroke, with some single photon energies in excess of 2 MeV. In the "chaotic" dart leaders of flashes UF 11 13 and UF 11 24, the energetic radiation emission was observed to continue for about 1 s following the return stroke, a re sult inconsistent with previous observations of energetic radiation emitted by dart and dart stepped leaders. Analyses were also conducted for two natural "chaotic" dart leaders preceding the third and fourth return strokes of a flash recorded on July 7 2011 (flash MSE 11 01). The natural "chaotic" dart leaders exhibited the same high frequency variations in the dE/dt signature as the triggered "chaotic" dart leaders, but for a significantly longer duration (up to 100 s prior to the return stroke). B ursts of energetic radiation were observed for both natural "chaotic" dart leaders with emission recorded up to 45 s prior to the return stroke, similar to the triggered "chaotic" dart leaders. Single photon energies were measured up to about 1.76 MeV. F or both natural "chaotic" dart leaders, the energetic radiation was observed to continue for over 1 s following the return stroke as was the case for the triggered "chaotic" dart leaders. In Chapter 7, o bservations of the initial stage processes of nine triggered lightning discharges during summer 2011 were analyzed using a combination of data from a seven station LMA, channel base currents, and vertical scan RHI images taken with a C band dual

PAGE 527

527 polarimeteric SMART radar. The initial stage processes for four triggered flashes (UF 11 24 through UF 11 26 on August 5, 2011, and UF 11 32 on August 18, 2011) were analyzed and discussed in detail. Statistical parameters of the initial UPL geometry, IS branch geometry, and correlated channel base current were g iven for all nine triggered lightning flashes with accompanying LMA data. The IS was found to not branch extensively in Florida triggered lightning compared to triggered flashes at higher altitude sites in New Mexico and France. IS branches were observed from altitudes as low as 580 m, but more typically occurred between about 700 m and 5 km. The times of IS branches, determined from the LMA source locations, were correlated with the measured channel base current. Unexpectedly, the channel base current was not found to change significantly at the time the IS branches occur. IS branches initiated with channel base current amplitudes ranging from 9 203 A. For the nine triggered flashes, the IS was found to transition from vertical to horizontal propagat ion at typical altitudes between 3 6 km, near the 0 C level of 4 5 km, and several kilometers below the expected center of the negative charge region in the cloud. As indicated by the LMA source locations and time coincident vertical scan radar images, I S branches often propagated horizontally for many kilometers along and above the general contour of the 0 C level. For one triggered flash (UF 11 25 on August 5, 2011) a natural cloud to ground discharge appears to have initiated in conjunction with the propagating IS channels. The discharge occurred 5 7 km northwest of the launching facility. A second triggered flash (UF 11 26) initiated a more or less naturally appearing bi level intracloud discharge, causing a 57 ms current polarity reversal measured at ground during the ICC process. LMA sources indicate that an upward negative leader initiated at about 5.6 km altitude and propagated upwards to about 9.3 km altitude, where widespread negative breakdown ensued from about 7.5 10.5 km altitude. Widespr ead positive breakdown

PAGE 528

528 associated with the IS occurred simultaneously at about 4 5.5 km altitude. The upward negative leader propagated for a time duration of 11 ms at an average three dimensional speed of 4.1 x 10 5 m/s. Current polarity reversals durin g the ICC processes of triggered lightning discharges have been rarely observed at the ICLRT, having occurred in only 2 out of 51 events (~ 4%) from 2008 2011. For several flashes, LMA source locations were obtained for positive polarity impulsive current s during the triggering wire ascent (precursor current pulses) with amplitudes less than 10 A, in contrast to expectations published in the literature of higher thresholds. In Chapter 8, analyses were performed of the propagation characteristics and atta chment processes of triggered and natural lightning dart stepped leaders using TOA source locations of dE/dt and x ray pulses, channel base currents, and photographic data. Four triggered lightning dart stepped leaders were analyzed in detail. In each ca se, the dE/dt and sensitive channel base current (II Very Low) waveforms were time aligned and compared to determine the physical significance of the leader burst pulses and slow front pulses that occurred, for each event, following the final dart stepped leader step. With the exception of the fourth stroke of flash UF 11 35, t he leader burst pulses were found to be emitted from relatively small source volumes at low altitude (from about 13 35 m above the intercepting wire ring) and were also found to coi ncide with the initial significant change in the background level of the measured channel base current, suggesting they are perhaps related to the initial interactions between the streamer zones of the downward and upward connecting leaders. dE/dt pulses superimposed on the slow front period prior to the fast transition were found to occur at lower altitudes than the leader burst pulses, and were, in all cases, associated with large increases in the background level of the channel base current to amplitude s near the maximum upward leader current. For the seven triggered lightning dart stepped leaders, the maximum upward leader currents ranged from about

PAGE 529

529 270 2190 A. Slow front pulses are likely associated with significant connections of the downward and up ward leaders given their low altitude source locations and timing correlation with large increases in the channel base current. For each triggered lightning dart stepped leader, t he timing of the fast transition peaks of the dE/dt and dI/dt waveforms were compared to establish an upper bound on the propagation time of the downward moving current wave from the junction height of downward and upward leaders. As expected, there was a time lag between dE/dt and dI/dt peaks of typically 20 70 ns. Given the st ated time lags, an assumed downward propagating current wave speed in air of 1.55 x 10 8 m/s, and propagation at the speed of light within the 6 m path length from the intercepting wire ring to the current measurement device, the maximum junction heights we re calculated to be no more than 8 m above the intercepting wire ring. With the assumption that the junction height is the final length of the upward connecting leader, and taking the upward leader duration to be the time between the initial steady deviat ion from zero of the background channel base current and the beginning of the fast transition (8.7 24 s) the upward leader speeds were calculated to range from 1.7 to 3.6 x 10 5 m/s. Though no channel base current was measured for the natural lightning d art stepped leader, the source altitudes and pulse characteristics of the leader burst and slow front dE/dt pulses following the final leader step suggests the attachment process to ground was similar to the triggered lightning events. A total of 30 x ray source locations were determined for two triggered dart stepped leaders and one natural dart stepped leader. For the triggered events, the x ray sources were located within 30 m of the source locations of the corresponding dE/dt pulse peaks. The x ray s ources were located within 40 m of the source locations of the dE/dt pulse peaks for the natural dart stepped leader. The total separation between in the dE/dt and x ray sources was dominated by vertical displacement and the x ray sources occurred beneath the dE/dt sources in 26 of the 30

PAGE 530

530 cases. The lateral distribution of the x ray sources about the leader channel was found to be r andom. The emission times of the x ray sources were found to lag the emission times of the associated dE/dt pulse peaks in a ll 30 cases by averages of 150 ns (GM 85) ns for flash UF 11 32, 290 ns (GM 170 ns) for flash UF 11 35, and 280 ns (GM 150 ns) for flash MSE 11 01. 9 .2 Recommendations for Future Research The author proposes the following topics and improvements for futu re study. Many of these improvements have been implemented, or will likely be implemented during summer 2012 or in future summers. The network of ten fast energetic radiation detectors (eight plastic detectors and two LaBr 3 detectors) should be digitized on the HBM digitization system in parallel with the ICLRT DSO system. Many of the difficulties in performing TOA locations of energetic radiation pulses associated with leader stepping processes are related to the high level of system noise with the anal og fiber optic transmission system, and the lack of bit depth available on the LeCroy DSOs. The HBM digitization system alleviates both issues. In addition, having the ten dE/dt measurements and the ten energetic radiation measurements digitized on the s ame time base with synchronized sampling would remove any timing uncertainty related to the measurement system, especially considering the HBM digitization system automatically calculates and removes cabling induced delays each time the system is armed. During summer 2011, the Cordin 550 high speed camera was operated with strictly the electrically driven rotating mirror system, which permits the camera to operate at a maximum frame rate of about 800 kfps. Based on the high speed video data of triggered lightning dart stepped leaders and one natural lightning stepped leader (Chapter 5), it is the author's opinion that the Cordin camera will need to operate at or near its maximum frame rate of 4 Mfps (250 ns frame integration) in order to begin to resolve the individual processes that form a single leader step, which is believed to form in about 1 s. During summer 2011, the Cordin high speed camera always triggered on the first return stroke following the IS. While the statistics shown in Chapter 4 indi cate that dart stepped leaders often occur preceding the first stroke following the IS, the author believes it is necessary to implement a trigger circuit that will differentiate between dart leaders and dart stepped leaders, producing a trigger output onl y when a dart stepped leader is detected. The trigger circuit could utilize either a dE/dt signal input or possibly a photodiode input, coupled with the ICLRT master trigger signal to insure a return stroke current has been measured at the lightning chann el base. A multi channel vertical photodiode array should be constructed to analyze the sub microsecond optical characteristics of the final few dart stepped leader steps, the upward

PAGE 531

531 connecting leader, and the attachment process for triggered lightning flashes. Unlike prior optical experiments where the photodiodes were spaced to view a large vertical section of the triggered lightning channel, the diodes should be aimed through narrow slits to view only the bottom 50 m or so of the channel with very h igh spatial resolution. In order to resolve the lower luminosity processes at the tips of lightning leaders and during the onset of the attachment phase, the system should be configured with increased sensitivity (using avalanche photodiodes) compared to prior photodiode systems. The output of such a photodiode array could ideally be compared directly with images from the Cordin high speed camera and dE/dt TOA locations from the HBM dE/dt system to yield a quite complete view of the leader step formation process and the subsequent attachment process. The LMA system, which operated with 80 s acquisition windows for the duration of summer 2011, should be converted to run in the 10 s acquisition window mode. The LMA system at Langmuir Lab in New Mexico presently operates in the 10 s mode and produces considerably more clear VHF images of the branching structure of both the IS in triggered lightning discharges and of intracloud activity associated with natural lightning. The 80 s acquisition windows ar e a limiting factor when multiple branches are propagating simultaneously, particularly considering the small area of the LMA network at the ICLRT and the small number of stations. Within the past year, there has been much discussion of flying small, i nstrumented aircraft over the ICLRT during storm conditions in an attempt to measure the electric field aloft prior to triggering lightning. Perhaps these same instrumented aircraft, which are already equipped with GPS systems for tracking their positions could also be instrumented with narrowband VHF radiation transmitters within the frequency band of the LMA system. With proper permission from Camp Blanding Range Control, a VHF transmitter flown on a small plane in clear weather could provide a very go od estimate of the three dimensional errors of the ICLRT LMA system as a function of altitude and range from the network. A similar test was conducted for the New Mexico LMA system with a VHF transmitter attached to a balloon. The small plane offers the advantage of controlling the flight path in a systematic manner.

PAGE 532

532 LIST OF REFERENCES Bailey, J., J.C. Willett, E.P. Krider, and C. Leteinturier (1988), Submicrosecond structures of the radiation fields from multiple events in lightning flashe s. Proceedings of the International Conference on Atmospheric Electricity, pp. 458 463. Bazelyan, E.M., and Yu. P. Raizer (1998), Spark Discharge 294 pp., Boca Raton, Florida: CRC Press. Beasley, W.H., M.A. Uman, and P.L. Rustan (1982), Electric fields preceding cloud to ground lightning flashes, J. Geophys. Res 87: 4883 902. Behnke, S. A., R. J. Thomas, P. R. Krehbiel, and W. Rison (2005), Initial leader velocities during intracloud lightning: Possible evidence for a runaway breakdown effect J. Geop hys. Res. 110 D10207, doi:10.1029/2004JD005312. Bent, R.B., and Lyons, W.A. (1984), Theoretical evaluations and initial operational experiences of LPATS (Lightning Positioning and Tracking System) to monitor lightning strikes using a time of arrival (TOA ) technique. In Proc. 7th Int. Conf. on Atmospheric Electricity, Albany, New York, pp. 317 24. Berger, K. (1955a), Die Messeinrichtungen fur die Blitzforschung auf dem Monte San Salvatore, Bull. Schweiz. Elektrotech Ver., 46, 193 204. Berger, K. (1955b), Resultate der Blitzmessungen der Jahre 1947 1954 auf dem Monte San Salvatore, Bull. Schweiz. Elektrotech Ver., 46, 405 424. Berger, K. (1962), Gas Discharges and the Electricity Supply Industry chap. Front duration and current steepness of lightning str okes to Earth, pp. 63 73, Butterworths. Berger, K. (1967a), Novel observations on lightning discharges: results of research on Mount San Salvatore, J. Franklin Inst ., 283, 478 525. Berger, K. (1967b), Gewitterforschung auf dem Monte San Salvatore, Elektrot echnik (Z A), 82, 249 260. Berger, K. (1972), Methoden und Resultate der Blitzforschung auf dem Monte San Salvatore bei Lugano in den Jahren 1963 1971, Bull. Schweiz. Elektrotech Ver., 63, 1403 1422. Berger, K. (1980), Extreme Blitzstrome und Blitzschutz, Bull. Schweiz. Elektrotech Ver., 71, 460 464. Berger, K., and E. Garbagnati (1984), Lightning current parameters. results obtained in Switzerland and Italy, uRSI Commission E, Florence, Italy, 13 pp. 320. Berger, K., and E. Vogelsanger (1965), Messungen und Resultate der Blitzforschung der Jahre 1955 1963 auf dem Monte San Salvatore, Bull. Schweiz. Elektrotech Ver., 56, 2 22.

PAGE 533

533 Berger, K., and E. Vogelsanger (1969), Planetary Electrodynamics chap. New results of lightning observations, pp. 489 510, Gordon and Breach. Berger, K., R. B. Anderson, and H. Kroninger (1975), Parameters of lightning flashes, Electra, 80, 223 37. Bevington, P.R. (1969), Data Reduction and Error Analysis for the Physical Sciences, McGraw Hill, New York, 336. Biagi, C. J., D. M. Jor dan, M. A. Uman, J. D. Hill, W. H. Beasley, and J. Howard (2009), High speed video observations of rocket and wire initiated lightning, Geophys. Res. Lett .,doi:10.1029/2009GL038525. Biagi, C. J., M. A. Uman, J. D. Hill, and D. M. Jordan (2011b), Observatio ns of the initial, upward propagating, positive leader steps in a rocket and wire triggered lightning discharge, Geophys. Res. Lett. 38, L24809, doi:10.1029/2011GL049944. Biagi, C. J., M. A. Uman, J. D. Hill, D. M. Jordan, V. A. Rakov, and J. Dwyer (2010) Observations of stepping mechanisms in a rocket and wire triggered lightning flash, J. Geophys. Res. 115, D23215, doi:10.1029/2010JD014616. Biagi, C. J., M. A. Uman, J. D. Hill, V. A. Rakov, and D. M. Jordan (2012), Transient current pulses in rocket ex tended wires used to trigger lightning, J. Geophys. Res. 117, D07205, doi:10.1029/2011JD016161. Biagi, C. J., M. A. Uman, J. Gopalakrishnan, J. D. Hill, V. A. Rakov, T. Ngin, and D. M. Jordan (2011a), Determination of the electric field intensity and spac e charge density versus height prior to triggered lightning, J. Geophys. Res. 116, D15201, doi:10.1029/2011JD015710. Biggerstaff, M. I., L. J. Wicker, J. Guynes, C. Ziegler, J. M. Straka, E.N. Rasmussen, A. Dogget IV, L. D. Carey, J. L. Schroeder, and C. Weiss, 2005: The Shared Mobile Atmospheric Research and Teaching (SMART) Radar: A collaboration to enhance research and teaching. Bull. Amer. Meteor. Soc. 86, 1263 1274. Boccippio, D. J., S. Heckman, and S. J. Goodman (2001), A diagnostic analysis of the Kennedy Space Center LDAR network 1. Data characteristics, J. Geophys. Res., 106 (D5), 4769 86. Boccippio, D. J., S. Heckman, and S. J. Goodman (2001), A diagnostic analysis of the Kennedy Space Center LDAR network 2. Cross sensor studies J. Geophys. Res. 106 (D5), 4787 4796, doi:10.1029/2000JD900688. Bringi, V. N., K. Knupp, A. Detwiler, L. Liu, I. J. Caylor, R. A. Black, (1997), Evolution of a Florida Thunderstorm during the Convection and Precipitation/Electrification Experiment: The Case of 9 August 19 91. Mon. Wea. Rev. 125 2131 2160.

PAGE 534

534 Byrne, C. J., A. A. Few, and M. E. Weber (1983), Altitude, thickness and charge concentration of charge regions of four thunderstorms during TRIP 1981 based upon in situ balloon electric field measurements, Geophys. Res Lett 10, 39 42. Casper, P. W., and R. B. Bent (1992), Results from the LPATS USA national lightning detection and tracking system for the 1991 lightning season, in Proc. 21st Int. Conf. on Lightning Protection, Berlin, Germany, 339 42. Chen, M., N. Tak agi, T. Watanabe, D. Wang, Z. I. Kawasaki, and X. Liu (1999), Spatial and temporal properties of optical radiation produced by stepped leaders, J. Geophys. Res ., 104, 27,573 27,584. Cooray, V., and S. Lundquist (1985), Characteristics of the radiation fiel ds from lightning in Sri Lanka in the tropics, J. Geophys. Res. 90: 6099 109. Cooray, V., R. Montano, and V. Rakov (2004), A model to represent negative and positive lightning first strokes with connecting leaders, J. Electrostatics 60 97 109. Crawfo rd, D. E., V. A. Rakov, M. A. Uman, G. H. Schnetzer, K. J. Rambo, M. V. Stapleton, and R. J. Fisher (2001), The close lightning electromagnetic environment: Dart leader electric field change versus distance, J. Geophys. Res. 106(D14), 14,909 14,917, doi:1 0.1029/2001JD900106. Crum, T. D., and R. L. Alberty, 1993: The WSR 88D and the WSR 88D Operational Support Facility. Bull. Amer. Meteor. Soc. 74, 1669 1687. Cummins, K. L., J. A. Cramer, C. Biagi, E. P. Krider, J. Jerauld M. A. Uman, and V. A. Rakov (200 6), The U.S. National Lightning Detection Network: Post upgrade status, in Proceedings of the Second Conference on Meteorological Applications of Lightning Data, 86th AMS Annual Meeting American Meteorological Society, Atlanta, GA. Cummins, K. L., M. J. M urphy, E. A. Bardo, W. L. Hiscox, R. B. Pyle, and A. E. Pifer (1998), A combined TOA/MDF technology upgrade of the U.S. National Lightning Detection Network, J. Geophys. Res. 103 (D8), 9035 9044. Davis, S.M. (1999), Properties of lightning discharges from multiple station wideband electric field measurements. Ph.D. dissertation, University of Florida, Gainesville, 228 pp. Depasse, P. (1994), Statistics on artificially triggered lightning J. Geophys. Res. 99 (D9), 18,515 18,522, doi:10.1029/94JD00912. Dwye r, J. R., et al. (2003), Energetic radiation produced during rocket triggered lightning. Science 299, 694 697. Dwyer, J. R., et al. (2004), Measurements of x ray emission from rocket triggered lightning, Geophys. Res. Lett ., 31, L05118, doi:10.1029/2003GL 018770.

PAGE 535

535 Dwyer, J. R., et al. (2005), X ray bursts associated with leader steps in could to ground lightning, Geophys. Res. Lett., 32, L01803, doi:10.1029/2004GL021782. Dwyer, J. R., M. Schaal, H. K. Rassoul, M. A. Uman, D. M. Jordan, and D. Hill (2011), Hi gh speed X ray images of triggered lightning dart leaders, J. Geophys. Res. 116, D20208, doi:10.1029/2011JD015973. Eriksson, A. J. (1978), Lightning and tall structures, Trans. South African IEE 69 238 252. Fieux, R., C. Gary, B. Hutzler, A Eybert Berar d, P. Hubert, A Meesters, P. Perroud, J. Hamelin, and J. Person, Research on artificially triggered lightning in France IEEE Trans. Power Appar. Syst ., 97, 725 733, 1978. Fisher, R. J., G. H. Schnetzer, R. Thottappillil, V. A. Rakov, M. A. Uman, and J. D Goldberg (1993), Parameters of Triggered Lightning Flashes in Florida and Alabama J. Geophys. Res. 98 (D12), 22,887 22,902, doi:10.1029/93JD02293. Gallimberti, I., G. Bacchiega, A. Bondiou Clergerie, and P. Lalande (2002), Fundamental processes in long air gap discharges, C.R. Phys ., 3, 1335 1359. Gomes, C., V. Cooray, M. Fernando, R. Montano, and U. Sonndara (2004), Characteristics of chaotic pulse trains generated by lightning flashes, Journal of Atmospheric and Solar Terrestrial Physics Volume 66, Is sue 18, p. 1733 1743 Gorin, B.N., V.I. Levitov, and A.V. Shkilev (1976), Some principles of leader discharge of air gaps with a strong non uniform field, Gas Discharges IEE Conf. Publ. 143, pp. 274 8. Gremillion, Michael S., Richard E. Orville, (1999), Th understorm Characteristics of Cloud to Ground Lightning at the Kennedy Space Center, Florida: A Study of Lightning Initiation Signatures as Indicated by the WSR 88D. Wea. Forecasting 14 640 649. Gringel, W., Rosen, J.M., and Hoffman, D.J. (1986), Electr ical structure from 0 to 30 km. In eds, pp. 166 82, Washington DC: National Academy Press. Guo, C., and E. P. Krider (1982), The Optical and Radiation Field Signatures Produced by Lightning Return Strokes J. Geophys. R es. 87 (C11), 8913 8922, doi:10.1029/JC087iC11p08913. Hansen, A. E., H. E. Fuelberg, and K. E. Pickering (2010), Vertical distributions of lightning sources and flashes over Kennedy Space Center, Florida J. Geophys. Res. 115 D14203, doi:10.1029/2009JD01 3143. Harris Jr., G.N., K.P. Bowman, and D. B. Shin, (2000), Comparison of freezing level altitudes from the NCEP reanalysis with TRMM Precipitation Radar brightband data, J. Climate, 13 4137 4148.

PAGE 536

536 Hendry, A., G. C. McCormick, B. L. Barge, 1976: The Degree of Common Orientation of Hydrometeors Observed by Polarization Diversity Radars. J. Appl. Meteor. 15 633 640. Hill, J. D., J. Pilkey, M. A. Uman, D. M. Jordan, W. Rison, and P. R. Krehbiel (2012), Geometrical and electrical characteristics of the initi al stage in Florida triggered lightning, Geophys. Res. Lett. 39, L09807, doi:10.1029/2012GL051932. Hill, J. D., M. A. Uman, and D. M. Jordan (2011), High speed video observations of a lightning stepped leader, J. Geophys. Res. 116, D16117, doi:10.1029/20 11JD015818. leaders in triggered lightning: Electric fields, X rays, and source locations, J. Geophys. Res. 117, D03118, doi:10.1029/2011JD016737. Howard, J. (2009), Lightning propagation and ground attachment processes from multiple station electric field and x ray measurements, University of Florida, Gainesville, FL. Howard, J., M. A. Uman, C. Biagi, D. Hill, J. Jerauld, V. A. Rakov, J. Dwyer, Z. Saleh, and H. Rasso ul (2010), RF and X ray source locations during the lightning attachment process, J. Geophys. Res. 115, D06204, doi:10.1029/2009JD012055. Howard, J., M. A. Uman, J. R. Dwyer, D. Hill, C. Biagi, Z. Saleh, J. Jerauld, and H. K. Rassoul (2008), Co location o f lightning leader x ray and electric field change sources, Geophys. Res. Lett. 35, L13817, doi:10.1029/2008GL034134. Hubert, P., and G. Mouget (1981), Return Stroke Velocity Measurements in Two Triggered Lightning Flashes J. Geophys. Res. 86 (C6), 5253 5261, doi:10.1029/JC086iC06p05253. Hubert, P., P. Laroche, A. Eybert Berard, and L. Barret (1984), Triggered Lightning in New Mexico J.Geophys. Res. 89 (D2), 2511 2521, doi:10.1029/JD089iD02p02511. Idone, V. P., and R. E. Orville (1984a), Three Unusual St rokes in a Triggered Lightning Flash J. Geophys. Res. 89 (D5), 7311 7316, doi:10.1029/JD089iD05p07311. Idone, V. P., and R. E. Orville (1985), Correlated Peak Relative Light Intensity and Peak Current in Triggered Lightning Subsequent Return Strokes J. G eophys. Res. 90 (D4), 6159 6164, doi:10.1029/JD090iD04p06159. Idone, V. P., R. E. Orville, P. Hubert, L. Barret, and A. Eybert Berard (1984b), Correlated Observations of Three Triggered Lightning Flashes J. Geophys. Res. 89 (D1), 1385 1394, doi:10.1029/JD 089iD01p01385. Idone, V. P. (1990), Length bounds for connecting discharges in triggered lightning, J. Geophys. Res. 95 20,409 20,416. Idone, V.P. (1992), The luminous development of Florida triggered lightning, Res. Lett. Atmos. Electr : 12: 23 8.

PAGE 537

537 Jera uld, J. (2007), Properties of natural cloud to ground lightning inferred from multiple station measurements of close electric and magnetic fields and field derivatives, Ph.D. dissertation, University of Florida, Gainesville, FL. (Available at http.//purl.f cla.edu/fcla/etd/UFE0021279) Jerauld, J., M. A. Uman, V. A. Rakov, K. J. Rambo, and D. M. Jordan (2004), A triggered lightning flash containing both negative and positive strokes, Geophys. Res. Lett. 31 L08,104, doi:10.1029/2004GL019457. Jerauld, J., M. A. Uman, V. A. Rakov, K. J. Rambo, and G. H. Schnetzer (2007), Insights into the ground attachment process of natural lightning gained from an unusual triggered lightning, J. Geophys. Res ., 112, D13113, doi:10.1029/2006JD007682. Jerauld, J., V. A. Rakov, M A. Uman, K. J. Rambo, and D. M. Jordan (2005), An evaluation of the performance characteristics of the U.S. National Lightning Detection Network using triggered lightning in Florida, J. Geophys. Res., 110, D19106, doi:10.1029/2005JD005924. Jordan, D. M., V. P. Idone, V. A. Rakov, M. A. Uman, W. H. Beasley, and H. Jurenka (1992), Observed dart leader speed in natural and triggered lightning, J. Geophys. Res ., 97, 9951 9957. Kita gawa, N. (1957), On the electric field change due to the leader processes and s ome of their discharge mechanism, Pap Meteor: Geophys (Tokyo) 7: 400 14. Kodali, V., V. A. Rakov, M. A. Uman, K. J. Rambo, G. H. Schnetzer, J. Schoene, and J. Jerauld (2005), Triggered lightning properties inferred from measured currents and very close e lectric fields, Atmos. Res ., 76, 355 376, doi:10.1016/j.atmosres.2004.11.036. Koshak, W. J., and R. J. Solakiewicz (1996), On the retrieval of lightning radio sources from the time of arrival data, J. Geophys. Res., 101, 26,631 26,639. Koshak, W. J., et al (2004), North Alabama Lightning Mapping Array (LMA): VHF source retrieval algorithm and error analyses, J. Atmos. Oceanic Technol., 21, 543 558. Krehbiel, P. R. (1986), chap. The electrical structure of thunderstorms, pp. 90 113, National Academy Press., Washington, DC. Krehbiel, P. R., R. J. Thomas, W. Rison, T. Hamlin, J. Harlin, and M. Davis (2000), Lightning mapping observations in central Oklahoma, Eos Trans AGU 81 (3), 21 25. Krider, E.P. (1974), The relative li ght intensity produced by a lightning stepped leader, J. Geophys. Res. 79: 4542 4. Krider, E.P., and G.J. Radda (1975), Radiation field waveforms produced by lightning stepped leaders, J. Geophys. Res 80: 2653 7. Krider, E.P., C.D. Weidman, and R.C. Nogg le (1977), The electric field produced by lightning stepped leaders J. Geophys. Res 82: 951 60.

PAGE 538

538 Lalande, P., A. Bondiou Clergerie, P. Laroche, A. Eybert Berard, J. P. Berlandis, B. Bador, A. Bonamy, M. A. Uman, and V. A. Rakov (1998), Leader properties determined with triggered lightning techniques, J. Geophys. Res. 103(D12), 14,109 14,115, doi:10.1029/97JD02492. Lan, Y., Y. Zhang, W. Dong, W. Lu, H. Liu, and D. Zheng (2011), Broadband analysis of chaotic pulse trains generated by negative cloud to gro und lightning discharge J. Geophys. Res ., 116, D17109, doi:10.1029/2010JD015159. Laroche, P., A. Eybert Berard, L. Barret, and J.P. Berlandis (1998), Observations of preliminary discharges initiating lightning flashes triggered by the rocket and wire tech nique, paper presented at the 8th International Conference on Atmospheric Electricity, Uppsala Univ., Swed. Nat. Sci. Res. Counc., Uppsala, Sweden, June 13 16. Lee, A. C. L. (1986), An experimental study of the remote location of lightning flashes using a VLF arrival time difference technique. Q. J. R. Meteor. Soc. 112, 203 29. Lennon, C. L., and H.A. Poehler (1982), Lightning detection and ranging, Astronaut. Aeronaut., 20 29 31. Les Renardieres Group (1978), Negative discharges in long air gaps at Les Re nardieres, Electra No. 74. Lewis, E. A., R. B. Harvey, and J. E. Rasmussen (1960), Hyperbolic direction finding with sferics of transatlantic origin, J. Geophys. Res., 65, 1879 1905. Lu, W., D. Wang, N. Takagi, V. Rakov, M. Uman, and M. Miki (2008), Chara cteristics of the optical pulses associated with a downward branched stepped leader, J. Geophys. Res ., 113, D21206, doi:10.1029/2008JD010231. MacGorman, D.R., Rust, W.D. (1998), The Electrical Nature of Thunderstorms, 422 pp., New York: Oxford University P ress. Mach, D. M., and W. D. Rust (1989), Photoelectric return stroke velocity and peak current estimates in natural and triggered lightning, J. Geophys. Res ., 94, 13,237 13,247. Maier, L., C. Lennon, T. Britt, and S. Schaefer (1995), LDAR system performan ce and analysis, paper presented at the 6th Conference on Aviation Weather Systems, Am. Meteorol. Soc., Boston, Mass. Mkel, J. S., Edirisinghe, M., Fernando, M., Montao, R., and Cooray, V (2007), HF radiation emitted by chaotic leader processes, Journal of Atmospheric and Solar Terrestrial Physics Volume 69, Issue 6, p. 707 720. Marshall, T. C., and W. D. Rust (1991), Electric Field Soundings Through Thunderstorms, J. Geophys. Res. 96(D12), 22,297 22,306, doi:10.1029/91JD02486.

PAGE 539

539 Marshall, T. C., W. D. R ust, W. P. Winn, and K. E. Gilbert (1989), Electrical Structure in Two Thunderstorm Anvil Clouds, J. Geophys. Res. 94(D2), 2171 2181, doi:10.1029/JD094iD02p02171. Miki, M., V. A. Rakov, T. Shindo, G. Diendorfer, M. Mair, F. Heidler, W. Zischank, M. A. Uma n, R. Thottappillil, and D. Wang (2005), Initial stage in lightning initiated from tall objects and in rocket triggered lightning, J. Geophys. Res. 110, D02109, doi:10.1029/2003JD004474. Moore, C. B., K. B. Eack, G. D. Aulich, and W. Rison (2001), Energet ic radiation associated with lightning stepped leaders, Geophys. Res. Lett. 28(11), 2141 2144, doi:10.1029/2001GL013140. Murray, N. D., E. P. Krider, and J. C. Willett (2005), Multiple pulses in dE/dt and the fine structure of E during the onset of first return strokes in cloud to ocean lightning, Atmos. Res. 76 455 480. Nag, A., et al. (2011), Evaluation of U.S. National Lightning Detection Network performance characteristics using rocket triggered lightning data acquired in 2004 2009 J. Geophys. Re s. 116 D02123, doi:10.1029/2010JD014929. Newman, M (1965), Use of triggered lightning to study the discharge channel, Problems of Atmospheric and Space Electricity ed. L.G. Smith, ppp. 482 90, New York: Elsevier. Newman, M., J.R. Stahmann, J.D. Robb, E .A. Lewis, S.G. Martin, and S.V. Zinn (1967), Triggered lightning strokes at very close range, J. Geophys. Res 72: 4761 4. Nucci, C. A., C. Mazzetti, F. Rachidi, and M. Ianoz (1988), On lightning return stroke models for LEMP calculations, in Proc. 19th Int. Conf. Lightning Protection Graz, Austria. Olsen, R. C., III, D. M. Jordan, V. A. Rakov, M. A. Uman, and N. Grimes (2004), Observed one dimensional return stroke propagation speeds in the bottom 170 m of a rocket triggered lightning channel, Geophys. Res. Lett. 31, L16107, doi:10.1029/2004GL020187. Olsen, R. C., III, V. A. Rakov, D. M. Jordan, J. Jerauld, M. A. Uman, and K. J. Rambo (2006), Leader/return stroke like processes in the initial stage of rocket triggered lightning, J. Geophys. Res. 111, D13202, doi:10.1029/2005JD006790. Ortega, P. P. Domens, A. Gilbert, B. Hutzler, and G. Riquel (1994), Performance of a 16.7 m air rod plane gap under a negative switching impulse J. Phys. D: Appl. Phys ., 27, 2379 2387. Orville, R. E., and V. P. Idone (198 2), Lightning leader characteristics in the Thunderstorm Research International Program (TRIP), J. Geophys. Res ., 87, 11,177 11,192. Poehler, H. A. (1977), An accuracy analysis of the LDAR system, NASA Contract. Rep., CR 154631, 83.

PAGE 540

540 Proctor, D. E. (1971), A hyperbolic system for obtaining VHF radio pictures of lightning, J. Geophys. Res., 76, 1478 1489. Proctor, D. E. (1981), VHF radio pictures of cloud flashes, J. Geophys. Res., 86, 4041 4071. Proctor, D. E. (1983), Lightning and precipitation in a small m ulticellular thunderstorm, J.Geophys. Res., 88, 5421 5440. Proctor, D. E., R. Uytenbogaardt, and B. M. Meredith (1988), VHF RADIO PICTURES OF LIGHTNING FLASHES TO GROUND J. Geophys. Res. 93 (D10), 12,683 12,727, doi:10.1029/JD093iD10p12683. Rakov, V. A., and M. A. Uman (1990), SOME PROPERTIES OF NEGATIVE CLOUD TO GROUND LIGHTNING FLASHES VERSUS STROKE ORDER, J. Geophys. Res. 95(D5), 5447 5453, doi:10.1029/JD095iD05p05447. Rakov, V. A., and M. A. Uman (2003), Lightning: physics and effects, Cambridge Uni v. Press, Cambridge, United Kingdom. Rakov, V. A., D. E. Crawford, V. Kodali, V. P. Idone, M. A. Uman, G. H. Schnetzer, and K. J. Rambo (2003), Cutoff and reestablishment of current in rocket triggered lightning, J.Geophys. Res ., p. 4747, doi:10.1029/2003J D003694. Rakov, V. A., and A. A. Dulzon (1987), Calculated electromagnetic fields of lightning return stroke, Tekh. Elektrodinam 1 87 89. Rakov, V. A., M. A. Uman, and R. Thottappillil (1994), Review of lightning properties from electric field and TV ob servations J. Geophys. Res. 99 (D5), 10,745 10,750, doi:10.1029/93JD01205. Rakov, V. A., V. Kodali, D. E. Crawford, J. Schoene, M. A. Uman, K. J. Rambo, and G. H. Schnetzer (2005), Close electric field signatures of dart leader/return stroke sequences in rocket triggered lightning showing residual fields, J. Geophys. Res. 110, D07205, doi:10.1029/2004JD005417. Reess, T., P. Ortega, A. Gilbert, P. Domens, and P. Pignolet (1995), An experimental study of negative discharge in a 1.3 m point plane air gap: t he function of the space stem in the propagation mechanisms, J. Phys. D: Appl. Phys ., 28, 2306 2313. Reid, G. C. (1986), The chap. Electrical structure of the middle atmosphere, pp. 183 194, National Academy Press, Washingt on, DC. Rison, W., R. J. Thomas, P. R. Krehbiel, T. Hamlin, and J. Harlin (1999), A GPS based three dimensional lightning mapping system: Initial observations in central New Mexico, Geophys. Res. Lett 26 3573 3576. Roble, R. G., and I. Tzur (1986), The chap. The global atmospheric electrical circuit, pp. 206 231, National Academy Press, Washington, DC.

PAGE 541

541 Rub i nstein, M., F. Rachidi, M. A. Uman, R. Thottappillil, V. A. Rakov, and C. A. Nucci (1995), Characterization of vertic al electric fields 500 m and 30 m from triggered lightning, J. Geophys. Res. 100(D5), 8863 8872, doi:10.1029/95JD00213. Rubin Jiang, Xiushu Qie, Caixia Wang, Jing Yang (2011) Step like characteristics of an upward positive leader in triggered lightning, Lightning (APL), 2011 7th Asia Pacific International Conference on vol., no., pp.614 617, 1 4 Nov. 2011. doi: 10.1109/APL.2011.6110200. Saleh, Z., J. Dwyer, J. Howard, M. Uman, M. Bakhtiari, D. Concha, M. Stapleton, D. Hill, C. Biagi, and H. Rassoul (2 009), Properties of the X ray emission from rocket triggered lightning as measured by the Thunderstorm Energetic Radiation Array (TERA), J. Geophys. Res. 114, D17210, doi:10.1029/2008JD011618. Schoene, J., M. A. Uman, V. A. Rakov, K. J. Rambo, J. Jerauld, C. T. Mata, A. G. Mata, D. M. Jordan, and G. H. Schnetzer (2009), Characterization of return stroke currents in rocket triggered lightning, J. Geophys. Res. 114, D03106, doi:10.1029/2008JD009873. Schoene, J., M. A. Uman, V. A. Rakov, V. Kodali, K. J. Ram bo, and G. H. Schnetzer (2003a), Statistical characteristics of the electric and magnetic fields and their time derivatives 15 m and 30 m from triggered lightning J. Geophys. Res ., 108, 4192, doi:10.1029/2002JD002698. Schonland, B. F. J. (1956 ), Handbuch der Physik vol. 22, chap. The lightning discharge, pp. 576 628, Springer Verlag, Berlin. Schonland, B. F. J., D. J. Malan, and H. Collens (1935), Progressive lightning II, Proc. Roy. Soc A, 152, 595 625. Shao, X. M., and P. R. Krehbiel (1996), The spatia l and temporal development of intracloud lightning J. Geophys. Res. 101 (D21), 26,641 26,668, doi:10.1029/96JD01803. Starr, S., D. Sharp, F. Merceret, J. Madura, and M. Murphy (1998), LDAR, a three dimensional lightning warning system: its development and use by the government, and transition to public availability, in Proc. 35th Space Congress, Horizons Unlimited, Cocoa Beach, Florida pp. 299 305, Canaveral Council of Technical Societies. Stolzenburg, M., T. C. Marshall, W. D. Rust, and D. L. Bartels (20 02), Two simultaneous charge structures in thunderstorm convection J. Geophys. Res. 107 (D18), 4352, doi:10.1029/2001JD000904. Stolzenburg, M., W. D. Rust, and T. C. Marshall (1998b), Electrical structure in thunderstorm convective regions 2. Isolated s torms, J. Geophys. Res. 103(D12), 14,079 14,096, doi:10.1029/97JD03547. Thomas, R. J., P. R. Krehbiel, W. Rison, S. J. Hunyady, W. P. Winn, T. Hamlin, and J. Harling (2004), Accuracy of the Lightning Mapping Array, J. Geophys. Res., 109, D14207, doi:10.10 29/2004JD004549.

PAGE 542

542 Thomson, E.M. (1980), Characteristics of the Port Moresby ground flashes, J. Geophys. Res 85: 1027 36. Thottappillil, R., and M. A. Uman (1993), Comparison of return stroke models, J. Geophys. Res. 98 (D12), 22,903 22,914. Thottappilli l, R., V. A. Rakov, and M. A. Uman (1997), Distribution of charge along the lightning channel: relation to remote electric and magnetic fields and to return stroke models, J. Geophys. Res. 102 (D6), 6987 7006. Uman, M. A., D. K. McLain (1969), Magnetic field of the lightning return stroke, J. Geophys. Res., 74, 6899 6910. Uman, M. A. (1974), The earth and its atmosphere as a leaky spherical capacitor, Am. J. Phys ., 42, 1033 35. Uman, M. A., J. Schoene, V. A. Rakov, K. J. Rambo, and G. H. Schnetzer (2002 ), Correlated time derivatives of current, electric field intensity, and magnetic flux density for triggered lightning at 15 m, J. Geophys. Res. 107, 4160, doi:10.1029/2000JD000249. Uman, M. A., V. A. Rakov, G. H. Schnetzer, K. J. Rambo, D. E. Crawford, a nd R. J. Fisher (2000), Time derivative of the electric field 10, 14, and 30 m from triggered lightning strokes, J. Geophys. Res. 105(D12), 15,577 15,595, doi:10.1029/2000JD900046. Visacro, S., A. Soares Jr., M. A. O. Schroeder, L. C. Cherchiglia, and V. J. de Sousa (2004), Statistical analysis of lightning current parameters: measurements at Morro do Cachimbo Station, J. Geophys. Res. 109 D01,105, doi:10.1029/2003JD003662. Wang, D., N. Takagi, T. Watanabe, V. A. Rakov, and M. A. Uman (1999a), Observe d leader and return stroke propagation characteristics in the bottom 400 m of the rocket triggered lightning channel, J. Geophys. Res ., 104, 14,369 14,376. Wang, D., V. A. Rakov, M. A. Uman, M. I. Fernandex, K. J. Rambo, G. H. Schnetzer, and R. J. Fisher ( 1999c), Characteristics of the initial stage of negative rocket triggered lightning, J. Geophys. Res ., 104(D4), 4213 4222. Wang, D., V. A. Rakov, M. A. Uman, N. Takagi, T. Watanabe, D. E. Crawford, K. J. Rambo, G.H. Schnetzer, R. J. Fisher, and Z. I. Kawas aki (1999b), Attachment process in rocket triggered lightning strokes, J. Geophys. Res ., 104, 2143 2150. Wang, D., N. Takagi, T. Watanabe, V.A. Rakov, and M.A. Uman (2000), Luminosity Waves in Branched Channels of Two Negative Lightning Flashes Journal of Atmospheric Electricity, 20, 91 97. Wang, D.,N. Takagi, T. Watanabe, V.A. Rakov, M.A. Uman, and K.J. Rambo, M.V. Stapleton (2005), A comparison of channel base currents and optical signals for rocket triggered lightning strokes Atmos. Res., vol. 76, issue 1 4, pp. 412 422.

PAGE 543

543 Weidman, C. D., and E. P. Krider (1978), The fine structure of lightning return stroke wave f orms, J. Geophys. Res. 83 (C12), 6239 6247. Weidman, C.D. (1982), The submicrosecond structure of lightning radiation fields, Ph.D Dissertation, University of Arizona. Willett, J. C., D. A. Davis, P. Laroche (1999), An experimental study of positive leader s initiating rocket triggered lightning, Atmos. Res ., 51,189 219. Willett, J., J. Bailey, C. Leteinturier, and E. Krider (1990), Lightning Electromagnetic Radiation Field Spectra in the Interval From 0.2 to 20 MHz J. Geophys. Res ., 95(D12), 20367 20387. W illett J., V. Idone, R. Orville, C. Leteinturier, A. Eybert Berard, L. Barret, and E. Krider ( 1988 Radiation From Triggered Lightning Return Strokes, J. Geophys. Res ., 93(D4), 386 7 3878. Wilson, C. T. R. (1920), Investigations on lightning discharges and on the electric field of thunderstorms, Phil. Trans. Roy. Soc. A, 221, 73 115. Winn, W. P., E. M. Eastvedt, J. J. Trueblood, K. B. Eack, H. E. Edens, G. D. Aulich, S. J. Hunyady, and W. C. Murray (2012), Luminous pulses during triggered lightning J. Geophys. Res. 117 D10204, doi:10.1029/2011JD017105. Yoshida, S., C. J. Biagi, V. A. Rakov, J. D. Hill, M. V. Stapleton, D. M. Jordan, M. A. Uman, T. Morimoto, T. Ushio, and Z. I. Kaw asaki (2010), Three dimensional imaging of upward positive leaders in triggered lightning using VHF broadband digital interferometers, Geophys. Res. Lett. 37, L05805, doi:10.1029/2009GL042065. Yoshida, S., et al. (2012), The initial stage processes of roc ket and wire triggered lightning as observed by VHF interferometry, J. Geophys. Res. 117, D09119, doi:10.1029/2012JD017657.

PAGE 544

544 BIOGRAPHICAL SKETCH Jonathan Dustin Hill was born in Jacksonville, FL in 1984. He attended Hilliard Middle Senior High School in H illiard, FL. Dustin received a Bachelor of Science degree in 2007, and a Master of Science degree in 2009, both in electrical and computer engineering from the University of Florida. Dustin began working in the Lightning Research Group in January of 2006 as an undergraduate student. He has worked at the International Center for Lightning Research and Testing for seven summers, and collected data towards a Ph.D from 200 9 to 2011. Dustin earned a Ph.D in electrical engineering from the University of Flori da in August of 2012. Dustin has been an author or co author on 20 peer reviewed journal publications, 2 7 conference proceedings, and two technical reports.