<%BANNER%>

Analysis of airborne laser-scanning system configurations for detecting airport obstructions

University of Florida Institutional Repository

PAGE 1

ANALYSIS OF AIRBORNE LASER-SCANNING SYSTEM CONFIGURATIONS FOR DETECTING AIRPORT OBSTRUCTIONS By CHRISTOPHER E. PARRISH A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Christopher E. Parrish

PAGE 3

To Deborah

PAGE 4

iv ACKNOWLEDGMENTS I wish to express my gratitude to Dr. Grady Tuell, chair of my supervisory committee, for his significant contributions to this thesis and his continued guidance and support. I thank Drs. Bill Carter and Ramesh Shrestha for serving on my committee and for their helpful advice and input. In addition, I am indebted to the follow ing people who assisted me in various aspects of this work: Jim Lucas, Dr. Brent Smith, Michael Sartori, Dr. Ramu Ramaswamy, and Stu Kuper. The following members of the data collection team deserve thanks and recognition for their hard work : Bill Gutelius, Bill Kalbfleisch, Warwick Hadley, and Butch Miller. I thank Captain Jon Bailey and Steve Matu la at the National Geodetic Survey for providing me the opportunity to attend gradua te school. Finally, I thank Tom Accardi and Fred Anderson at the Federal Aviation Administration, Aviati on System Standards for funding the data collection for this research.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii LIST OF FIGURES...........................................................................................................ix ABSTRACT......................................................................................................................x ii CHAPTER 1 INTRODUCTION........................................................................................................1 Airport Obstruction Surveying.....................................................................................1 Airborne Laser Scanning..............................................................................................5 Background and Motivation.........................................................................................8 Organization of this Work..........................................................................................11 2 THEORY AND PREDICTIONS...............................................................................13 Laser Equation............................................................................................................13 Geometric Considerations in Obstruction Detection..................................................14 Radiometric Considerations in Obstruction Detection...............................................22 3 EXPERIMENTS.........................................................................................................27 Airborne Laser Data Collection..................................................................................27 Calibration..................................................................................................................30 Data Processing..........................................................................................................31 Field Spectrometer Data Collection............................................................................32 4 DATA ANALYSIS....................................................................................................36 Preliminary Analysis..................................................................................................36 Obstruction Detection Analysis..................................................................................42 Automated Obstruction Detection Analysis...............................................................44 Visual Analysis...........................................................................................................50 Analysis of Return Sign al Strength Calculations.......................................................52

PAGE 6

vi 5 CONCLUSIONS AND RECOMMENDATIONS.....................................................56 APPENDIX A DERIVATION OF RANGE EQUATION .................................................................61 B REFLECTANCE SPECTRA FOR OBSTRUCTIONS AND OTHER OBJECTS WITHIN THE SURVEY AREAS..............................................................................65 C PHOTOGRAPHS OF FIELD-S URVEYED OBSTRUCTIONS...............................71 D OUTPUT OF AUTOMATED OBSTRUCTION DETECTION ANALYSIS SOFTWARE...............................................................................................................79 Configuration 1...........................................................................................................79 Configuration 2...........................................................................................................80 Configuration 3...........................................................................................................82 Configuration 4...........................................................................................................84 Configuration 5...........................................................................................................86 Configuration 6...........................................................................................................87 Configuration 7...........................................................................................................89 Configuration 8...........................................................................................................91 Configuration 9...........................................................................................................93 Configuration 10.........................................................................................................94 Configuration 11.........................................................................................................96 Configuration 12.........................................................................................................98 Configuration 13.......................................................................................................100 Configuration 14.......................................................................................................101 LIST OF REFERENCES.................................................................................................104 BIOGRAPHICAL SKETCH...........................................................................................108

PAGE 7

vii LIST OF TABLES Table page 2-1 Narrow and wide beam divergences for the system used in this study, based on three different definiti ons of beam diameter............................................................19 3-1 The 14 data collection configurations used in this study and the predicted vertical and horizontal point spacing for each......................................................................27 3-2 Reflectance values at 1064 nm for fiel d-surveyed obstructions and other objects in the survey areas....................................................................................................34 3-3 Reflectance values at 1064 nm for thr ee horizontal surfaces in Survey Zone 1......35 4-1 Results of testing the air borne laser data sets using an independent data set of NGS kinematic GPS runway points.........................................................................37 4-2 Percent of obstructions detected in each airborne laser data set and the RMS difference in elevation between the fiel d-surveyed points and “matching” laser points........................................................................................................................4 7 4-3 Analysis of return signal st rength calculations for SPN 452...................................53 4-4 Analysis of return signal st rength calculations for SPN 449...................................54 4-5 Analysis of return signal st rength calculations for SPN 454...................................54 D-1 Tilt: 0; Div: N; FH:750.............................................................................................79 D-2 Tilt: 0; Div: W; FH: 750...........................................................................................80 D-3 Tilt: 10; Div: N; FH: 750..........................................................................................82 D-4 Tilt: 10; Div: W; FH: 750.........................................................................................84 D-5 Tilt: 20; Div: N; FH: 750..........................................................................................86 D-6 Tilt: 20; Div: W; FH: 1050.......................................................................................87 D-7 Tilt: 20; Div: W; FH: 1150.......................................................................................89 D-8 Tilt: 20; Div: W; FH: 750.........................................................................................91

PAGE 8

viii D-9 Tilt: 20; Div: W; FH: 850.........................................................................................93 D-10 Tilt: 20; Div: W; FH: 950.........................................................................................94 D-11 Tilt: 30; Div: N; FH: 750..........................................................................................96 D-12 Tilt: 30; Div: W; FH: 750.........................................................................................98 D-13 Tilt: 40; Div: N; FH: 750........................................................................................100 D-14 Tilt: 40; Div: W; FH: 750.......................................................................................101

PAGE 9

ix LIST OF FIGURES Figure page 1-1 Federal Aviation Regulation (FAR) Part 77 obstruction identification surfaces (OIS).......................................................................................................................... .2 1-2 Airborne laser scanning system s produced by the two leading commercial manufacturers.............................................................................................................5 1-3 Simplified illustration of air borne laser scanning principles.....................................6 1-4 Growth in commercial use of air borne laser scanners from 1995 to 2000................8 1-5 Results of comparing the three airbor ne laser data sets collected during the 2001 study against field-surveyed obstruction data..........................................................10 2-1 Vertical Point Spacing..............................................................................................15 2-2 Plot of vertical point spacing versus tilt angle based on the following settings: v = 55 m/s and = 0.019 sec....................................................................................16 2-3 Calculation of vertic al footprint diameter, Av..........................................................17 2-4 Illustration of vertical point spacing (VPS) and effective vertical spacing (EVS)..18 2-5 Profile of laser beam for the University of Florida airborne laser scanning system and a fitted gaussian.................................................................................................19 2-6 Effective vertical spacing versus tilt angle based on the following parameters: H = 750 m, v = 55 m/s, = 0.019 s, and = 0.60 mrad...........................................20 2-7 Definition of horizont al point spacing (HPS)..........................................................21 2-8 Schematic Illustration of the detection and measurement system............................23 2-9 Received power vs. tilt angle...................................................................................25 3-1 Survey project areas overlaid on a digital orthophoto and USGS quadrangles.......28 3-2 Variable-tilt sensor moun t designed for this study...................................................29

PAGE 10

x 3-3 Obtaining reflectance measurements for a guywire of one of the towers in Survey Zone 1 using the ASD LabSpec Pro portable spectrometer.....................................33 4-1 Plot of average elevation bias for each tilt angle setting on th e ordinate vs. tilt angle on the abscissa................................................................................................38 4-2 Average elevation bias vs. tilt angle and propagated systematic error in the elevation of a laser point..........................................................................................40 4-3 NGS field survey of obstructions at GNV...............................................................43 4-4 Obstruction detecti on analysis algorithm.................................................................45 4-5 Visual obstruction analysis............................. .........................................................51 4-6 Photograph of SPN 460, and data poin ts on this object based on laser returns obtained using configurations 5, 8, and 12...............................................................51 4-7 A potentially more rigorous method of performing the return signal strength computations involving modeling the intera ction of the incident laser radiation with a target as a convolution...................................................................................55 B-1 Reflectance spectra for SPN 445 strobe lighted tower, a guywire for the strobe lighted tower, and SPN 446 tower........................................................................65 B-2 Reflectance spectra for SPN 449 pole, SPN 452 antenna, and SPN 453 transmission pole......................................................................................................66 B-3 Reflectance spectra for SPN 454 fla gpole, SPN 456 pole, and a pine tree........67 B-4 Reflectance spectra for a palm tree, a pole, and a generator....................................68 B-5 Reflectance spectra for gr ass, concrete, and asphalt................................................69 C-1 Photographs of SPN 414 tree, and SPN 415 tree...............................................71 C-2 Photographs of SPN 418 tree, a nd SPN 431 obstruction light on pole..............72 C-3 Photographs of SPN 446 tower, SPN 445 antenna on strobe lighted tower, and SPN 448 antenna on strobe lighted tower......................................................73 C-4 Photographs of SPN 449 pole, SPN 453 transmission pole, and SPN 452 antenna...................................................................................................74 C-5 Photographs of SPN 454 flagpol e, SPN 456 pole, SPN 455 sign, and SPN 457 transmission pole...................................................................................75

PAGE 11

xi C-6 Photographs of SPN 457 – tran smission pole, and SPN 459 – pole........................76 C-7 Photograph of SPN 460 – pole.................................................................................77

PAGE 12

xii Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science ANALYSIS OF AIRBORNE LASER-SCANNING SYSTEM CONFIGURATIONS FOR DETECTING AIRPORT OBSTRUCTIONS By Christopher E. Parrish May 2003 Chair: Grady Tuell Major Department: Civil and Coastal Engineering Airborne laser scanning is a relatively new remote sensing technology that is finding use in an increasing number of su rveying and mapping applications. The strengths of airborne laser scanning, includ ing high data density and geometric accuracy, indicate promise in airport obstruction su rveying. The primary objective in this application is to accurately pos ition discrete point features that penetrate imaginary 3D survey surfaces around airfields. Early st udies revealed, however, that many airport obstructions, particularly poles, antennas and other small-diamet er objects, were often not detected using commercial airborne laser scanning systems. The systems employed in the early studies utilized standard data collection and system parameter configurations, which may be better suited for bare-earth terrain mapping than detection of airport obstruc tions. It is hypothesi zed that obstruction detection can be substantially improved thr ough modification of certain parameters. The objective of this research is to investigate, both analytically and empi rically, the ability to

PAGE 13

xiii improve obstruction detection capability with an airborne laser scanning system through modification of key parameters. The main pa rameters investigated include tilt angle, laser footprint, and flying height, although the effects of flying speed, scan angle and frequency, transmitted power, and receiver sensitivity are also discussed. The analytical analysis involves investig ation of both geomet ric and radiometric considerations in obstruction de tection. It is shown that tr adeoffs exist between the two; by improving the geometry for obstruction dete ction, the return signa l from targets is weakened, and vice versa. The optimum config uration is that which yields the best geometry possible (i.e., the highest density of laser pulses incident on a vertical feature), while still permitting a detectable return signal from targets of interest. We present results of test flights over th e Gainesville Regional Airport (GNV) and portions of the runway 10 approach usi ng fourteen different data collection configurations. The airborne laser data are compared agai nst field surveyed obstruction data obtained by an NGS field crew to asse ss each of the fourteen configurations. Analysis of the data reveals that signi ficant improvement in obstruction detection capability can be achieved with suitable confi gurations. It is shown that 100% detection (based on predefined criteria) with submeter vertical RMSE is attainable. We conclude with a discussion of potential future enhancements in obstruction detection capability.

PAGE 14

1 CHAPTER 1 INTRODUCTION Airport Obstruction Surveying To navigate safely into airports in re duced-visibility weather conditions, pilots follow published instrument approach procedur es that specify flight courses, turns, minimum altitudes, and so forth. Similarl y, departure procedures are followed in executing safe departures from airports. Becau se these procedures are critical to flight safety, it is essential that they be base d on accurate and up-to-date source data. A prerequisite step in designing an approach or departure procedure is to conduct an airport obstruction survey. The main objective in obstruction surv eying is to obtain accurate survey coordinates for vertical object s (both natural and manmade) in specified zones on and around the airfield and in the approach paths. Objects that lie within these zones and that penetrate (i.e., are of greater height than) mathematically-d efined 3D surfaces enveloping the airport (Figure 1-1) are termed obstructions or “obstacles.” Examples of obstructions include, but are by no means limited to, trees buildings, towers, poles, antennae, and terrain. In addition to suppor ting procedure development, obstruction survey data are used by airport and government authorities in planning, meeting or verifying compliance with airport operating certificate requirements determining maximum weights of aircraft for takeoff, and conducting accident investigat ions (U.S. Department of Transportation, 1996).

PAGE 15

2 Figure 1-1. Federal Aviation Regulation (FAR) Part 77 obstruction identification surfaces (OIS) (courtesy of FAA, ATA-100). The shape and dimensions of the surfaces for a particular obstruction survey will vary depending on the regulating agency, type of survey, ru nway end positions and designations, and other factors. Within the United States and its territories, the Federal Aviation Administration (FAA), Aviation System Standards (AVN) is responsible for developing and publishing approach procedures for all civil airports. Under an interagency agreement, airport obstruction surveys supporting the FAA are conducted by the National Geodetic Survey (NGS). These surveys are performed in accordance with FAA No. 405: Standards for Aeronautical Surveys and Related Products (U.S. Department of Transportation, 1996). Similarly, the National Imagery and Mapping Ag ency (NIMA) is ta sked with obtaining survey data and developing procedures fo r approximately 10,000 ai rports throughout the world in support of U.S. military operati ons (Harris and Johnson, 2001). Obstruction

PAGE 16

3 surveys performed by or for NIMA must m eet the specifications contained in the Airfield Initiative Document (National Imagery and Mapping Agency, 2001). In addition to meeting the standards pub lished by government agencies at the national level, airport obstruction surveys must also adhere to applicable international specifications. The International Civil Aviation Organization (ICAO) establishes and publishes international standards to be fo llowed by each of its 188 member nations (or “contracting states”), including the United Stat es. Specifications pertaining to airport obstruction surveying and charting are contai ned in two annexes to the Convention on International Civil Aviation (also known as the Chicago Convention of 1944): Annex 14 Aerodromes, Volume I Aerodrome Design and Operations (International Civil Aviation Organization, 1999) and Annex 4 Aeronautical Charts (International Civil Aviation Organization, 1995). Currently, airport obstruc tion surveys are most often completed through a combination of photogrammetry and field su rveying. Photogrammetry is a mature remote sensing technology, and the pro cedures and achievable accuracy are well documented. Field surveys offer the highest accuracy and reliability because experienced field crews visually inspect the survey areas to identify and locate small obstructions (Tuell, 1985). Based on records kept at NGS, the co mbination of photogrammetry and field surveying has been utilized successfully in airport obstruction surveying for over half a century. During this time, however, the mappi ng procedures used in NGS’s Aeronautical Survey Program have been continually upda ted as new technologies have become available. We can recognize cert ain major transitions (Tuell, 1987):

PAGE 17

4 In the early 1960s, several analog photogr ammetric plotters (Wild B-8s) were purchased to support the program. During this period, field-surveyed obstructions and control points were plot ted by hand, and measurements from the stereoplotters were used to position new obstructi ons and compile planimetry. Throughout the 1970s, innovations were focu sed on the integration of computers and computer-driven precision plotters into the program. In the mid-1980s, the transition from anal og to analytical photogrammetry and the initial design and implementation of a relational database to function as the data warehouse were accomplished. At the sa me time, the field survey teams implemented total station technology and developed th e capability to transfer obstruction data through the production system electronically In addition, the field teams began to use GPS for establishing control points on airports. In the 1990s, significant improvements were ma de in the ability to log field survey data and to compute positions while on-site. Hand-held la sers were introduced for quick measurement of distances to obstructions. In 2000, the analytical photogrammetric ster eoplotters were replaced with softcopy (digital) photogrammetric wo rkstations. In addition, CAD and GIS systems were introduced to facilitate the storage, ed iting, analysis, and distribution of digital obstruction data, incl uding digital charts. While field techniques and phot ogrammetry will continue to set the standard for high-accuracy obstruction surv eying, several factors motivat e continued investigation into new technologies for obstruction survey ing. First, the demand for survey data already exceeds production capability, and this demand is certain to increase as the FAA implements GPS-based navigation and landing systems, such as the Wide Area Augmentation System (WAAS) (Anderson et al., 2002). Second, new FAA and National Aeronautics and Space Administ ration (NASA) initiatives, such as Synthetic Vision Systems (SVS) (Prinzel et al., 2002), are fu rther increasing the de mand for high-accuracy digital terrain and obstacle databases. Th ird, because different types of obstruction surveys have different requirements for accu racy, cost and completion time, agencies would benefit from greater flex ibility in tailo ring the survey methods to the requirements.

PAGE 18

5 Of particular interest are remote sensing technologies that could potentially fulfill the need for rapid, inexpensive, medium-to-high-accuracy surveys. Airborne Laser Scanning Airborne laser scanning (also referred to as lidar) is an active remote sensing technology that is quickly gain ing recognition as an efficien t and cost-effective approach to a variety of surveying and mapping app lications. The primary components of an airborne laser-scanning system include 1) the laser scanner, 2) an inertial measurement unit (IMU), and 3) an airborne GPS recei ver and antenna. Figure 1-2 shows systems produced by the two leading commercial manufacturers. Figure 1-2. Airborne lase r-scanning systems produced by the two leading commercial manufacturers. Top: Optech, Inc. ALTM 2050 (photo courtesy of Opetch, Inc.). Bottom: Leica Geosystems, Inc. ALS40 (photo courtesy of Leica Geosystems, Inc ).

PAGE 19

6 Although airborne laser scanners are comple x instruments requiring integration of numerous subsystems, the basic concepts of the technology are relatively straightforward, as illustrated by the cartoon in Figure 1-3. Ranges are accurately computed from the round trip travel time of lase r pulses that are reflected by either terrain or elevated features on the Earth’s surface and return to the sensor. By combining range, sensor orientation, and scanner angle data, 3D vectors from the airb orne sensor to points on the reflective surface illuminated by the laser can be computed. These 3D vectors are then utilized in conjunction with post-processe d airborne GPS data and offset vectors describing the relative positions of the vari ous system components to compute accurate XYZ positions of terrain and feat ures in the mapping frame. Figure 1-3. Simplified illustration of airborne laser-scanning principles.

PAGE 20

7 While at least one operational system employs a continuous-wave (CW) laser (Wehr and Lohr, 1999), most airborne laser scan ners utilize pulsed lasers. Typically, the lasers are Q-switched to produce short (~ 10 ns) pulses and high peak transmitted power. Current state-of-the-art system s have pulse repetition frequencies (PRFs) of 30 to 50 kHz, and systems with even higher PRFs are in de velopment. Various types of scanners are used to produce a swath. For example, th e system employed in this research uses a single-axis, cross track scanning mirror that produces a saw-tooth pattern on the ground, as depicted in Figure 1-3. The past decade has seen significant a dvancement in airborne laser-scanning technology. Although laser altimetry dates back to the early 1970s (Blair et al., 1999), commercial airborne laser scan ners were not readily available until the mid-1990s. As illustrated in Figure 1-4 (adapted from Maune 2001), the growth in commercial adoption over the five-year period from 1995 to 2000 wa s nearly 2,000%. The increasing demand for new systems has naturally precipitated te chnological advances, such as higher pulse repetition and scan frequencies, improved re liability, more robust data collection and processing software, and so forth. Two of the often-cited streng ths of airborne laser sca nning are the high density of data points and the achievable geometric accura cy. As noted above, PRFs of 50 kHz or greater are currently attainable. Assuming continuous operation of the laser and a 97% probability of a good return from each pulse, one hour of data collection with a 50 kHz system will generate nearly 175 million da ta points. Several researchers have demonstrated vertical ac curacy of 15 cm (1 ) or better on terrain (see, e.g., Shrestha et al., 1999; Vaughn et al., 1996b). Horizontal accur acy of airborne laser data is harder to

PAGE 21

8 quantify and has been less rigorously investig ated. In general, it is expected that horizontal accuracy will be wo rse than vertical accuracy (B altsavias, 1999b; Maas, 2002), but at least one study has indica ted horizontal accuracy of bette r than half a meter (Tuell, 2002). Growth in Commercial Systems Use0 10 20 30 40 50 60 70 199519961997199819992000 YearTotal Systems Figure 1-4. Growth in commer cial use of airborne laser scanners from 1995 to 2000 (adapted from Manue, 2001). Background and Motivation Over the past few years, the high point density and geometric accuracy achievable with airborne laser scanning have brought this technology to the attention of several government agencies and private firms involved in obstruction surveying. The technology seems of obvious be nefit for the mapping of obs tructing areas and buildings, but because of the unique challenges involved in surveying discrete point features, as well as the critical nature of the data in flight safety, its performance for obstruction detection must be carefully analyzed. One of the first studies of this type was conducted jointly by the University of Florida (UF), NGS, FAA, and Optech, Inc. at Gainesville Regional Airport (GNV) in 2001.

PAGE 22

9 During the 2001 study, three data sets were collected using two different Optech Airborne Laser Terrain Mapper (ALTM) system s. One of these systems had a PRF of 10 kHz and was flown in a Cessna Skymaster owned and operated by UF. The flying height for the data collected with th e UF system was 600 m. The second system had a PRF of 33 kHz and was flown in a NOAA Cessna Citation at flying heights of 700 and 1200 m. In the NOAA Citation, a 7o tilt angle was used, meani ng the sensor was tilted 7o forward of nadir. The UF system did not use a tilte d sensor. Both systems had a constant beam divergence of approximately 0.18 mrad (full angle), based on the full width at half maximum (FWHM) points of the beam (or, equivalently, 0.22 mrad, based on the 1/ e points of the beam). Concurrent with the airborne laser data collection, an NGS survey crew conducted an obstruction survey using GPS and conven tional field techniques to provide a highaccuracy reference data set. To assess how well obstructions were detected using the airborne laser-scanning systems, researchers at UF and NGS compared the three airborne laser data sets against the field-surveyed obstruction data. Figure 1-5 (adapted from Tuell, 2002) shows the percent of obstructions detected in each of the three data sets. In performing this analysis, UF researchers measured the 3D distance from each fieldsurveyed obstruction to the cl osest point in the laser point cloud. A 3D distance of less than 20 feet was defined as the detection criterion in generating Figure 1-5. As illustrated here, at best, 94% of the obstructions we re detected based on this criterion. A more interesting observation, however, is that the detection pe rcentage drops off significantly with flying height. In fact, this study revealed that certain small targets (poles, antennae, etc.) were of ten not detected at all. Cl early, the ability to hit and

PAGE 23

10 measure a small target is a function of the su rvey geometry. What is not clear is how much of the loss at higher flying height results from the geometric effect of increasing the pulse spacing and how much of it originates w ill a fall-off in received signal strength due to increased laser range. Percent of Obstructions Detected to Within 20 feet of the Field-Surveyed Point (2001 Study)75 80 85 90 95 100 UFNOAA 700 mNOAA 1200 mAirborne Laser Data SetPercent Detected Figure 1-5. Results of comparing the three ai rborne laser data sets collected during the 2001 study against field-surveyed obstru ction data (adapted from Tuell, 2002). The 2001 study was not designed to systema tically investigat e the effects of various data collection and system parameters on the results. The experiment utilized commercial systems with configurations si milar to those employed in topographic mapping projects. Because commercial air borne laser-scanning systems are most frequently utilized for production of bare-earth data sets, it is likely that, during the rapid developments of the past decade, systems have been either intentionally or unintentionally optimized for th is type of work. However, obstruction surveying is a fundamentally different and more difficult ta sk; the parameters that work well for one application might not be well-suited for the othe r. It was hypothesize d, therefore, that the

PAGE 24

11 capability to detect obstructions with an airborne laser-scanning system could be enhanced by modifying certain data collecti on and system parameters. This hypothesis, combined with the valuable information gained from the earlier study, provided the foundation and motivation for the re search presented here. Organization of this Work The goal of this research is to investigate the effect of laser data collection geometry on the detection of small targets and to unders tand the tradeoffs between geometric and radiometric issues. In cont rast to the 2001 study, we have designed an experiment using data collected with a single instrument. This allows us to systematically investigate the effects of certain parameters without having to account for inter-instrument variability. Specifically, th e effects of sensor tilt (or “forward-look”) angle, laser footprint area, and flying height on obstruc tion detection are rigorously examined through both analytical and empiri cal methods. Other parameters, such as flying speed over ground, scan angle and fr equency, pulse repetition frequency, and transmitted power, are also examined mathem atically, though not experimentally. In Chapter 2, the underlying analytical c onsiderations are addressed. The problem of identifying the optimum conf iguration of the laser system is shown to be nontrivial, due to the number of variables and tradeoffs involved. The experiments are presented in Chapter 3. These include an airborne lase r data collection using fourteen different configurations, as well as collection of reflectance sp ectra using a portable field spectrometer. The results of the analysis ar e presented in Chapter 4. It is shown that significant improvements in obstruction detec tion capability can, in fact, be achieved through careful selection of the data colle ction parameters. We conclude with a

PAGE 25

12 discussion of the potential for further im provements and suggestions for continued research.

PAGE 26

13 CHAPTER 2 THEORY AND PREDICTIONS Laser Equation A quantitative description of airborne laser scanning starts with the relationship between measured quantities and the desired X, Y,Z coordinates of terra in or features in the mapping frame. The equation that gives the location of the laser footprint in the mapping frame, alternately referred to as the “georeferencing equation,” the “laser geolocation equation” or simply the “laser equation,” is given in various forms in, for example, Lindenberger (1989), Vaughn et al. (1 996a), Filin (2001), and Schenk (2001). Some form of the laser equation is used in all software packages that process airborne laser-scanning system measurements to output surface point coordinates. In this work, we will examine the effect of geometric parameters on the detection of small targets. In Equati on (2-1), we show a simplifie d form of the laser equation which explicitly addresses the pos sibility of a tilted sensor: X Y Z X Y Z xRst yRs zRstf f f GPS GPS GPS lx ly lz M cossin sin coscos (2-1) Here, [ Xf Yf Zf]T is the position of a lase r footprint in the mappi ng frame (e.g., State Plane coordinates); [ XGPS YGPS ZGPS]T is the position of the aircraft GPS antenna in the same mapping frame; s is the instantaneous scan angle; t is the tilt angle; R is the range; xl, yl, and zl are the coordinates of the laser beam origination point in the body frame; and xyandz are the sensor-to-antenna offset vector (“lever arm”) components. For the

PAGE 27

14 body frame, we have used a photogrammetric rather than aeronautical, convention: positive x is in the direction of flight, positive y is towards the left wing of the aircraft, and positive z is towards zenith. The rotation matrix M is given by M cossincoscossinsinsincossinsinsincos coscossincoscossinsinsinsincossincos sincossincoscosprprrpr prprrpr pprpr hhhhh hhhhh (2-2) where, r (roll), p (pitch), and h (heading) are the attitude angles reported by the IMU (Lucas, personal correspondence, 2002). It should be noted that Equation (2 -1) assumes that the tilt angle, t is measured independently from the attitude angles, r p and h If the tilt angle is incorporated into the attitude angles, then Equa tion (2-1) still holds, provided t is set equal to zero. Another assumption made in Equation (2-1) is that the offset distance from the IMU to the laser is negligible. This assumption is likely to introduce a small systematic error in the computed positions of laser points. However, in this work, the laser equation is used only in examining how errors are propagate d (see Chapter 4) and not in computing positions of points. Geometric Considerations in Obstruction Detection One metric by which the strength of the geometry for detecting vertical obstructions can be measured is the verti cal point spacing (Opt ech, unpublished data, 2002). The vertical point spacing (VPS) is defi ned as the vertical di stance between laser points from consecutive scan lines on the face of a vertical surface. The smaller the VPS, the better the geometry for detecting vertical obstructions. The VPS will vary depending on the point in the scan cycle at which the beam “catches” the obstruction. Specifically, the VPS will be greatest if th e obstruction lies on the outer ed ge of the scan and smallest

PAGE 28

15 if the obstruction lies directly on the flight path. Assuming the most conservative case (i.e., the obstruction lies on the outermost edge of the scan), the VPS is given by VPSvt[tan()] 900 vt cot (2-3) where v is the flying speed over ground, is the period of the scanner, and t is the tilt angle (see Figure 2-1). Figure 2-1. Vertical Point Spacing. Figure 2-2 shows a plot of VPS versus tilt angle using the following data collection parameters: v = 55 m/s (107 knots), and = 0.019 sec (corresponding to a frequency of 53 Hz). These are the parameters used in the data collection for this study, as detailed in Chapter 3. It should be noted that increasing the tilt angle is not th e only possible method of reducing the VPS. For example, the VPS could be decreased by reducing the aircraft speed, increasing the frequency of th e scanner (i.e., reducing the period, ), and/or by flying repeat passes. However, these met hods will not be considered here, since the

PAGE 29

16 flying speed and scanner frequency listed a bove are based on the actual data collection parameters and since repeat passes increase costs. Figure 2-2. Plot of vertical poi nt spacing versus tilt angle based on the following settings: v = 55 m/s and = 0.019 sec. In determining the desired VPS for obstru ction detection, it is beneficial to introduce another quantity, the “ver tical footprint.” The vertical footprint is defined here as the area illuminated by the laser beam on th e face of a vertical object, based on the full width at half maximum (FWHM) points of th e beam. With refere nce to Figure 2-3, the vertical footprint diameter, Av, is given approximately by A R t R tvo cos() sin 90 (2-4) In Equation (2-4), R is the range to the target and is the beam divergence. With simplifying assumptions of flat terrain, obs truction height much smaller than flying height, and an instantaneous scan angle of zero, the range can be expressed as

PAGE 30

17 R H tcos (2-5) where H is the flying height. Substituting Equation (2-5) into E quation (2-4) gives: A H ttv sincos (2-6) Figure 2-3. Calculation of ve rtical footprint diameter, Av. We define a new term, the effective vert ical spacing (EVS), as the VPS minus the vertical footprint diameter, Av (see Figure 2-4). The EVS provides a metric for the extent to which the vertical face of an obstructi on is illuminated by lase r radiation (i.e., how

PAGE 31

18 well the obstruction is “painted” by the lase r). An EVS of zero is interpreted as completely painting the face of the obs truction in the vertical dimension. Figure 2-4. Illustration of ver tical point spacing (VPS) and effective vertical spacing (EVS). Four footprints on the face of a box-shaped obstruction are shown. EVS and Av are both based on the FWHM points of the beam. The airborne laser-scanning system used in this study has two beam divergence settings: wide and narrow. A flip-in lens provides the mechanism for switching from one setting to the other. Table 2-1 shows the wide and narrow beam divergences for this system based on three different beam di ameter definitions: FWHM, 1/e, and 1/e2. Throughout this study, the FWHM definition will be used. However, because the beam (TEM00 mode) is very nearly gaussian (see Figu re 2-5), it is possible to convert from one beam diameter definition to either of the others through Equations (2-7) and (2-8).

PAGE 32

19 Table 2-1. Narrow and wide beam divergences for the system used in this study, based on three different definiti ons of beam diameter. Setting Divergence based on 1/e2 pts of beam (mrad) Divergence based on 1/e pts of beam (mrad) Divergence based on FWHM pts of beam (mrad) Wide 1.02 0.72 0.60 Narrow 0.34 0.24 0.20 1/ 12e beam diameter FWHM beam diamete r (Gaussian beam) (2-7) 1/ 172e beam diameter FWHM beam diamete r (Gaussian beam) (2-8) Figure 2-5. Profile of laser beam for the Univ ersity of Florida ai rborne laser-scanning system (green lines) and a fitted gaussian (orange lines). Although this is not the system used in this study, the profile is likely similar. (Image courtesy of Optech, Inc.) Figure 2-6 shows a plot of EVS vs. tilt angle using the wide beam divergence and the flying speed and scan period listed above. As illustrated in the graph, a tilt angle of

PAGE 33

20 15.6o produces an EVS of 2.0 m, a tilt angle of 26.0o produces an EVS of 1.0 m, and a tilt angle of 49.0o produces an EVS of zero (i.e., 100% c overage in the vertical dimension). Figure 2-6. Effective vertical spacing versus tilt angle based on th e following parameters: H = 750 m, v = 55 m/s, = 0.019 s, and = 0.60 mrad. In examining the ability to detect small obstructions, it is important to consider the horizontal point spacing (HPS) in addition to the vertical point spacing. The HPS is defined as the distance between laser points inci dent on the face of a vertical surface in a direction perpendicular to the vertical (s ee Figure 2-7). Due to the motion of the scanning mirror, the HPS is not uniform; point s are more tightly bunched near the outer edges of the scan. In the following discussion, therefore, HPS will be assumed to refer to the average spacing. Assuming, further, a flat ve rtical surface whose nor mal is parallel to the direction of flight, the HPS is given approximately by

PAGE 34

21 HPS Swath Width Number of Points per Scan Line PRF 22 2R Stan (2-9) where S is the full scan angle (note: lower case s refers to the instantaneous scan angle), and, as before, R refers to range and to the period of the scanner. As was done for the vertical point spacing, it is possible to define an eff ective horizontal spacing (EHS) by taking into account th e footprint diameter, Ah. Assuming, again, a fl at vertical surface whose normal is parallel to the direction of fli ght and an instantaneous scan angle of zero, Ah is given approximately by ARh (2-10) Figure 2-7. Definition of horiz ontal point spacing (HPS). Using Equations (2-9) and (2-10) and data collection parameter values that are valid for the current study ( H = 750 m, S = 30o, t = 20o, PRF = 50 kHz, = 0.019 s, and = 0.6 mrad), the HPS and EHS are approxi mately 0.90 m and 0.42 m, respectively. It can be seen, therefore, that the point spacing in the horizontal direc tion is typically better (smaller) than that in the vertical direction for the system used in this study. The HPS

PAGE 35

22 and EHS can be reduced further with relativel y minor modifications to the data collection parameters. For example, cutting the scan angle in half while keeping all other parameters the same produces a negative EHS, meaning that the pulse s will overlap in the horizontal dimension, based on the FWHM points of the beam. Although horizontal spacing cannot be neglected, vertical spac ing is currently a greater concern than horizontal spacing for obstruction detection. Radiometric Considerations in Obstruction Detection Figure 2-8 illustrates schematically how radiation emitted from the laser and reflected from a surface below is detected and used to determine range. As shown in the figure, reflected radiation that is incide nt on the receiver optics is passed through a narrow bandpass filter to remove background radiation (Optech, 1998) Next, a squarelaw detector converts the optical signal into a current that is proportional to the incident optical power (or to the squa re of the electric field). The output from the detector is next fed in to an amplifier and then into a constant fraction discriminator (CFD). The purpose of a CFD is to provide accurate triggering that is nearly independent of the amplitude of the input pulse (see, e.g., Binkley and Casey, 1988). Digital pulses output by the CFD are then fed into the actual timing mechanism, known as a Time Interval Meter (TIM) (Optech, 1998). Essentially the same detection and measurement mechanisms are used for both the transmitted and received pulses, with the primary difference being that scattered laser light within the system is captured and used to detect the transmitted pulse (Optech, 1998). The temporal difference between the corresponding points on the transmitted and received pulses, combined with the value for the group velocity of the laser light in the atmosphere, allows ranges to be determined.

PAGE 36

23 Figure 2-8. Schematic Illustration of th e detection and measurement system. For the purposes of this study, it is important to note that if the signal is below the detection threshold, the target wi ll not be detected. The first step in estimating the return signal strength involves calculating, Ea, the irradiance (W m 2) incident on the receiver.

PAGE 37

24 Derivations of expressions for Ea and the received power, Pr, are contained in Appendix A. For convenience, equations (A-7) and (A-8) are restated below: E P R Ta T ATM 224 2 (2-11) where PT is the transmitted power, TATM is the atmospheric transmittance, and is the effective target cross section given by Equati on (A-6). The power received can then be computed from: PAETrraSYS (2-12) where Ar is the receiver area and TSYS is the system transmittance, which is limited primarily by the transmittance of the bandpass filter. Next, the photocurrent generated by the detector can be computed from: IPphr (2-13) where is the responsivity (AW 1) of the photodetector. The exact value of TSYS for the system used in this research was not disclosed by the manufacturer, but a “typical” value of 0.6 was used in th e calculations. The value of was also not disclosed by the manufacturer, so it was not possible to directly calculate Iph using Equation (2-13). However, the re sponsivity of the p hotodetector can be assumed to be constant over the course of a project. In this study, return signal strength calculations were performed usi ng Equations (2-11) and (2-12). Figure 2-9 shows a plot of recei ved power vs. tilt angle. The target in this example is an antenna that was surveyed by the fiel d crew in 2001. The methods used to obtain the reflectance of this object are descri bed in Chapter 3. The value of the peak transmitted power for the system used in this study was obtained from the manufacturer. Other parameters used in the calculations were based on an actual data collection

PAGE 38

25 configuration used in the study (s ee Chapter 3). It is interest ing to note in Figure 2-9 that the received power drops off very rapidly with increasing tilt angle. This is because R increases with t and Pr decreases as R-4. Figure 2-9. Received power vs. tilt angl e based on the following parameters: PT = 11.3 kW (peak, not average); = 0.6 mrad; TATM = 0.87; H = 750 m; = 0.318 (reflectance of SPN 452); d = 0.305 m (diameter of SPN 452); Ar = 1.79x10-3 m2; and TSYS = 0.60. From Figures 2-6 and 2-9, we note that a tr adeoff exists between the geometric and radiometric considerations. Specifically, it is desirable to increase the tilt angle as much as possible to reduce the effective verti cal spacing (improving the geometry), but increasing the tilt angle also has the undesira ble effect of reduci ng the received power from a target. For example, for a given system we can achieve a zero EVS, but the received power would only be about one third of the power received with a nadir-viewing instrument. Likewise, increasing the beam divergence improves the geometry (larger footprint) but reduces the received signal strength, as seen in Equation (2-11). In

PAGE 39

26 configuring an airborne la ser-scanning system for obs truction detection, the goal, therefore, is to choose parameters that optimize the geometry while still enabling a detectable return signal fr om targets of interest.

PAGE 40

27 CHAPTER 3 EXPERIMENTS Airborne Laser Data Collection Based on the geometric and radiometric cons iderations discusse d in Chapter 2, the experiment was designed to test fourteen diffe rent data collection configurations. These fourteen configurations consisted of diffe rent combinations of tilt angle, beam divergence, and flying height, as listed in Table 3-1. The survey project areas for the study consisted of three zones covering the airfield and portions of the runway 10 approach at Gainesville Regional Air port (GNV), as shown in Figure 3-1. Table 3-1. The 14 data collection configuratio ns used in this study and the predicted vertical and horizontal point spacing for each. Note: by definition, VPS and EVS apply only to configurations that employ a tilted sensor. Config. # Tilt (deg) Divergence (Wide/ Narrow) Flying Height (m) Predicted VPS (m) Predicted EVS (m) Predicted HPS (m) Predicted EHS (m) 1 0 N 750 N/A N/A 0.8 0.7 2 0 W 750 N/A N/A 0.8 0.4 3 10 N 750 5.9 5.0 0.9 0.7 4 10 W 750 5.9 3.3 0.9 0.4 5 20 N 750 2.9 2.4 0.9 0.7 6 20 W 1050 2.9 0.9 1.3 0.6 7 20 W 1150 2.9 0.7 1.4 0.6 8 20 W 750 2.9 1.5 0.9 0.4 9 20 W 850 2.9 1.3 1.0 0.5 10 20 W 950 2.9 1.1 1.1 0.5 11 30 N 750 1.8 1.5 1.0 0.8 12 30 W 750 1.8 0.8 1.0 0.5 13 40 N 750 1.2 0.9 1.1 0.9 14 40 W 750 1.2 0.3 1.1 0.5

PAGE 41

28 Figure 3-1. Survey project areas overlaid on a digital orthophoto and USGS quadrangles. Zone 3 encompasses the airfield. Th e GPS reference station, NATASHA, is located on the UF campus. Although airborne laser data were collected over a significantly larger area during the 2001 study, the three zones shown in Fi gure 3-1 contain 90% of the obstructions surveyed by the field crew. Because of the la rge amount of data required for this study, it was not possible to collect data over the entire runway 10 approach. By limiting the data collection to these three zones, significant sa vings in cost, data storage, and processing time were achieved with minimal im pact on the obstruc tion analysis. Acquisition of the airborne laser data took place from June 10 through June 15, 2002. The system used in the data collec tion was an Optech ALTM 2050 mounted in a Cessna Skymaster. An Ashtech ZXII recei ver and 700936 D choke ring antenna located at NATASHA UF, an NGS Cooperative Base Network (CBN) control station on the UF

PAGE 42

29 campus, served as the GPS reference stati on. The average flying speed over ground for all fourteen configurations was approximately 55 m/s (107 knot s), and all configurations used a scan frequency of 53 Hz, a scan angle of 15o, and a PRF of 50 kHz. To enable variable tilt angles of 0 to 40o, a custom sensor mount (Figure 3-2) was designed and built by Optech. During the five-day ac quisition period, over 378 million laser data points were collected in the th ree zones shown in Figure 3-1. Figure 3-2. Variable-tilt sensor mount designed for this study. The left and right images show the 20o and 30o tilt angle positions, respectively. The use of variable tilt angles resulted in a few added complications over typical airborne laser data collecti on. Most importantly, the sens or-to-antenna offset vector (“lever arm”) components required separate measurements for each tilt angle setting. Clearly, in determining the offset vector co mponents, it could not be assumed that the zaxis of the sensor was aligned with the local vertical. To acquire these offsets, a least squares adjustment of a 3D trilateration was used. Di stances were measured from the bottom of the GPS antenna (specifically, th e bottom of the TNC female connector) to each of the four corners of the top of the A LTM sensor box and also to the screw hole in the handle. These measurements were re peated for each tilt angle setting. The coordinates of the box corners and handl e in the sensor frame are known from

PAGE 43

30 engineering diagrams provided by Optech. The additional offs ets from the screw hole in the handle to the center of the scan mirror and from the TNC connector on the GPS antenna to the antenna phase center are also known. Since there are three unknown sensor-to-antenna offset vector components, more than three measured distances permit a least squares solution. Custom software was written to perform the least squares adjustment of the data and output the final offset vector component s and their standard deviations, which averaged approximately 1.5 cm. Calibration Careful calibration of an airborne laser-s canning system is essential to obtaining high positional accuracy. The system used in this study was calibrated by the manufacturer prior to the da ta collection in Gainesville. In-flight calibration was performed on June 15 to refine the calibra tion parameters. Calibration flights were performed across (i.e., perpendicular to) runway 6-24 at GNV and also over a large, flatroofed building on the UF campus that had been accurately surveyed using GPS. For the flights over the building, bot h profile-mode (zero scan angle) and scan mode (4o scan angle) were used, while the runway flights utilized a scan angle of 20o. The data collected during these flights we re used to determine corrections to the pitch, roll, and scale calibration parameters. The calibration data were analyzed in Su rfer (Golden Software, Inc.) Version 8.00. The pitch correction value was obtained usi ng the profile-mode data captured over the field-surveyed building. By comparing the surveyed edges of the building with the locations at which the correspond ing changes in elevation in the laser data occurred, the amount by which the pitch was over/underreported could be determined. The roll

PAGE 44

31 correction was obtained in a similar manner using the 4o scan angle data and finding locations at which the edge of the building wa s captured at the outer edge of the scan. The mirror angle scale factor was determined by examining elevation profiles from the 20o scan angle data over the runway. Th e slope along a portion of the runway is presumed to be constant, so an upward curv e (“smile”) or downward curve (“frown”) in the elevation profiles indicates a nece ssary correction to the scale. Data Processing The airborne laser data were pro cessed using REALM (TopScan, GmbH, and Optech, Inc.) Version 3.0.3d. The processing was completed following standard procedures used by researchers at UF (see Shre stha et al., 1999; Cart er et al., 2001) with three notable exceptions. First, due to the very short baselines (approximately 6 to 11 km), it was not deemed necessa ry to utilize the Kinematic and Rapid Static (KARS) software (Mader, 1992) in processing the G PS trajectories; the GPS processing was done directly in REALM. Second, no filtering or gridding of the data was performed. And third, a few default processing parameters we re changed for the reasons listed below: The default setting for the REALM V. 3.0.3d parameter “Max. FL Diff” serves to eliminate “suspicious” data points by excluding any laser shot for which the distance between the first and last retu rns is greater than 100 m. Since many obstructions have Above Ground Level (AGL ) heights of over 100 m, this default setting cannot be used in obstruction detection. The default for the REALM V. 3.0.3d “Min. In tensity” setting is 1. The intensity value is a unitless di gital number that is proportional to the st rength of the return signal, and, hence, to the effective targ et cross section (see Appendix A). The “Min. Intensity” parameter is used to exclude laser shots whenever the intensity value is below the defined threshold. Optech recommends against setting this parameter below the default value of 1, as a value of zero indicates a potentially bad range measurement (Tickle, personal correspondence, 2003). In this work, however, the parameter was set to zero to avoid excluding obstructions with small effective target cross sections from the output.

PAGE 45

32 Yet another default in REALM V. 3.0.3d is to output only last return laser data. While this default setting is suitable fo r bare-earth terrain mapping, for obstruction detection, first returns are more important. Therefore, the setting was changed to output the first return data. As noted in Chapter 1, an underlying hypothesi s in this research is that standard data collection configurations are not well-su ited for obstruction detection. Interestingly, this statement was found to be equally applicab le to the processing software; as described above, several of the default parameter settin gs are unsuitable for obstruction detection. All of the settings can be ch anged with little difficulty, but unfortunately, the software does not have the capability to save the user’s settings from one se ssion to the next. In short, although the “off-the-shelf” processing software was utilized successfully in this project, it was clearly not de signed for this application. The output laser data points were projected in UTM (WGS84) Zone 17 North. Elevations were referenced to the WGS84 e llipsoid. In order to compare the airborne laser data sets against the NGS kinematic GPS runway data and field-surveyed obstruction data (see Chapter 4) the ellipsoid elevations we re converted to orthometric heights by the analysis soft ware using GEOID99. Field Spectrometer Data Collection In addition to the airborne laser data co llection, field work was also required to obtain reflectance measurements for obstructions and other objects in the survey areas. These reflectance measurements are needed in calculating the return signal strength from obstructions (see Equations (A-6 ) through (A-8)). Comparisons of the calculated return signal strength values with the empirical re sults allow the minimum detectable return signal to be estimated (Chapter 4).

PAGE 46

33 The spectrometer used for obtaining reflectan ce measurements in this study was an Analytical Spectral Devices, Inc. (ASD) LabS pec Pro. This instrument has a spectral range of 350 to 2500 nm. The spectral resoluti on of the LabSpec Pro is 3 nm at 700 nm and 10 nm at 1400 nm and 2100 nm. The sampling interval is 1.4 nm for the spectral region 350-1000 nm and 2 nm for the spect ral region 1000-2500 nm. Figure 3-3 shows the ASD LabSpec Pro being used to obtain a sp ectrum for a guywire of one of the towers in Survey Zone 1. Figure 3-3. Obtaining reflectance measurements for a guywire of one of the towers in Survey Zone 1 using the ASD La bSpec Pro portable spectrometer. The process of obtaining reflectance spectra involves first acquiring a reference spectrum with the instrument probe pointed at a calibrated refere nce panel, typically made of Spectralon (Labsphere). The refe rence panel should be highly Lambertian and have a known reflectance close to unity. Ne xt, data are collected while pointing the probe at the desired target. Finally, the reflectance values of the target are computed

PAGE 47

34 from the ratio of the measurements made over the target to those made over the reference panel for each band (Lillesand and Kiefer, 2000). Data collection with the field spectrometer took place from July 22 to July 24, 2002. During this three-day period, weather co nditions ranged from overcast to mostly sunny. Because of the varying conditions, a reference spectrum was obtained for every measurement, and each spectrum was acquire d within one minute of its corresponding reference spectrum. Data collection was lim ited to the hours of 10:00 AM to 2:45 PM each day, since the sun was used as the light source. Spectral measurements were acquired for twelve different vertical objects, including eight field-surveyed obstructions, a nd also for three horizontal surfaces within the survey areas. Appendix B contains the reflectance spectra. Water absorption bands and excessively-noisy regions have been removed from the spectra. Table 3-2 summarizes the reflectance values at 1064 nm for the twelve objects. Table 3-3 summarizes the reflectance values at 1064 nm for the three surfaces. Table 3-2. Reflectance values at 1064 nm for field-surveyed obstructions and other objects in the survey areas. Survey Point # Description 1064 445 Strobe Lighted Tower 0.607 445 Strobe Lighted Tower (Guywire) 0.107 446 Tower 0.128 449 Pole (wood) 0.309 452 Antenna 0.318 453 Transmission Pole 0.148 454 Flagpole 0.639 456 Pole (wood) 0.332 N/A Pine Tree 0.595 N/A Palm Tree 0.414 N/A Pole (wood) 0.368 N/A Generator (metal painted green) 0.256

PAGE 48

35 Table 3-3. Reflectance values at 1064 nm for th ree horizontal surfaces in Survey Zone 1. Survey Point # Description 1064 N/A Grass 0.567 N/A Concrete 0.402 N/A Asphalt 0.208 Some of the objects listed in Table 3-2 were noticeably more weathered on one side than on the other sides. In addition, a few of the manmade objects contained both painted and unpainted sections. In these cases, spectra were acquired for two or more different parts of the object, and a mean of the reflect ance values was taken. In obtaining spectra for the palm and pine trees, the probe was aime d at the leaves or needles, since the first laser return is likely to be from the top of a tree, rather than its trunk. The mean value of 1064 for the objects in Table 32 is 0.352. Twenty-five percent of the objects in Table 3-2 have 1064 values of less than 0.2, indicating that many obstructions are relatively poor reflectors of 1064 nm light. The significance of these low reflectance values is examined in Chapter 4.

PAGE 49

36 CHAPTER 4 DATA ANALYSIS Preliminary Analysis Before evaluating how well the field-survey ed obstructions were captured in the airborne laser data, the fourteen data sets we re checked for blunders and elevation biases. This analysis was performed using an indepe ndent data set consisting of 46 check points positioned by the National Geodetic Survey (NGS) using van-mounted GPS receivers and kinematic (KGPS) processing techniques. All of the points are located in relatively flat areas on or around the airf ield. Software was written to locate the laser data point closest in horizontal distance to each point in the independent data set. If the distance to the closest point was less than the mean laser footprint radius, the laser point was selected as a match to the NGS check point, and the difference in elevati on between the laser point and the NGS point was computed. A non-zero mean difference in elevation betw een the airborne laser data points and the NGS points was interpreted by the software as an elevation bias in the airborne laser data. After removing elevation biases, the RM SE and estimated vertical accuracy at the 95% confidence level were calculated for each of the fourteen airborne laser data sets (Table 4-1). The calculations were performe d in accordance with the National Standard for Spatial Data Accuracy (Federal Ge ographic Data Committee, 1998) using the following equations:

PAGE 50

37 RMSE ZZ n NGSLaser i n1 2 (4-1) Accuracy95% CL = 1.96(RMSE) (4-2) Table 4-1. Results of testing the airborne laser data sets using an i ndependent data set of NGS kinematic GPS runway points. Config # Parameters (tilt in deg; wide/narrow divergence; flying height in m) Day(s) of data collection Mean Difference in Elevation (m) RMSE, after removing elevation bias (m) Accuracy at 95% CL (m) 1 Tilt: 0, Div.: N, FH: 750 June 15 -0.06 0.08 0.15 2 Tilt: 0, Div.: W, FH: 750 June 15 -0.04 0.07 0.14 3 Tilt: 10, Div.: N, FH: 750 June 14-15 -0.01 0.05 0.09 4 Tilt: 10, Div.: W, FH: 750 June 14 0.00 0.05 0.10 5 Tilt: 20, Div.: N, FH: 750 June 12 0.09 0.12 0.23 6 Tilt: 20, Div.: W, FH: 1050 June 14 0.20 0.08 0.15 7 Tilt: 20, Div.: W, FH: 1150 June 14 0.12 0.11 0.21 8 Tilt: 20, Div.: W, FH: 750 June 12 0.20 0.12 0.23 9 Tilt: 20, Div.: W, FH: 850 June 13 -0.01 0.09 0.18 10 Tilt: 20, Div.: W, FH: 950 June 14 0.14 0.08 0.15 11 Tilt: 30, Div.: N, FH: 750 June 10-11 0.18 0.22 0.43 12 Tilt: 30, Div.: W, FH: 750 June 10-11 0.27 0.18 0.34 13 Tilt: 40, Div.: N, FH: 750 June 10 0.23 0.12 0.23 14 Tilt: 40, Div.: W, FH: 750 June 10 0.35 0.15 0.29 Several researchers (e.g., Vaughn et al., 1996b; Shrestha et al., 1999; Krabill et al., 1995) have demonstrated that vertical RMSE s of 5 to 15 cm on te rrain are achievable through airborne laser mapping. Although most of the RMSEs listed in Table 4-1 are within this range, the RMSEs for the data collected with the 30o tilt angle are notably poorer. Also noteworthy from Table 4-1 are the mean differences in elevation between the airborne laser data and the NGS points. These values appear to increase with tilt angle and become quite large for the data sets collected with 30o and 40o tilt angles. This correlation of height bias with tilt angle is clearly illustrated in Figure 4-1. Here, the

PAGE 51

38 biases shown were calculated by averaging the biases for the various deployment modalities of the airborne laser system. Figure 4-1. Plot of average elev ation bias for each tilt angle setting on the ordinate vs. tilt angle on the abscissa, based on the values from Table 4-1. Although careful calibration of an airborne laser-scanning system should reduce systematic error, elevation biases in airbor ne laser data are not uncommon. Vaughn et al. (1996b) reported an elevation bias of 54 cm in their data and Shrestha et al. (1999) reported elevation biases that varied from pass to pass and ranged from –10 to +20 cm. After removing these biases, both groups determin ed the vertical accura cy of their data to be 10 cm (1 ) or better. The apparent correlation between elevati on bias and tilt angle shown in Figure 4-1 merits investigation. One possible explanation lies in the relationship between the attitude parameters (pitch in pa rticular) and the computed elev ations of laser points. By considering the geometry involved, it can be intu ited that an error in pitch will have little effect on the computed elevation of a laser point with a nadir-poin ting beam, but as the

PAGE 52

39 tilt angle is increased, the effect on the comput ed elevation will increase. A mathematical analysis of this effect in volves first expanding Equation (2-1) to give the following expression for the elevation of a laser point: ZZpxRstpryRs przRstfGPSlxly lz sin(cossin)cossin(sin) coscos(coscos) (4-3) Using Equation (4-3), it is possible to examine how errors in attitude angles propagate to errors in computed elevations of laser points. Because we are considering systematic (as opposed to random) errors, th e methods applicable to propagation of systematic errors (see, e.g., Mikhail and Ackermann, 1976; Van ek and Krakiwsky, 1986) must be applied. Letting r p and h be small, known syst ematic errors in orientation parameters (which can be either pos itive or negative), the propagated error in the elevation of a la ser point is given by h h Z p p Z r r Z Zf f f f (4-4) Removal of systematic errors from airbor ne laser data has become the focus of widespread research as the airborne lase r scanning community strives towards everincreasing positional accuracy (e.g., Filin, 2002 ; Toth et al., 2002). Unknown systematic errors in various parameters, in cluding attitude, are likely to exist for any airborne laser data set. While rigorous treatment of syst ematic errors is beyond the scope of this research, we found through numerical curve-fittin g techniques that the following errors in attitude propagate to elevation erro rs that fit the empirical curve: p = 0.043o, r = -0.040o, and h = 0. These values were obtained by conducting a systematic search in 2D parameter space to find the ( p r ) pair that would yield the best fit to the empirical

PAGE 53

40 curve. The search interval for both p and rwas [-2o, +2o], and the sampling step was 0.001o. Figure 4-2 shows a plot of propagated erro r in elevation vs. tilt angle using the errors listed above and the follo wing additional parameter values: R = 750 m (the flying height for ten of the fourteen configurations), s = 7.5o (half of the maximum instantaneous scan angle fo r all configurations), and p = r = 0.5o (typical values). The blue curve is the experimental data curve shown in Figure 4-1, and the red curve is the propagated systematic error as a function of tilt angle calculated us ing Equation (4-4). Although, as noted above, the re d curve was calculated using ad hoc methods, the close agreement with the experimental data indicat es that uncorrected systematic errors in orientation could account for the observed trend. Figure 4-2. Blue curve: average elevation bias vs. tilt angle, based on the data shown in Table 4-1. Red curve: propagated systema tic error in the elevation of a laser point due to the following system atic errors in orientation: p = 0.043o, r = -0.040o, h = 0.

PAGE 54

41 Errors in GPS (with possible day-to-day va riability) could also help explain the effects seen in Table 4-1. Based on the expe rience of researchers at UF and NGS, it is reasonable to expect errors of approximatel y 4 cm horizontal and 8 cm vertical in the trajectories, given the lengths of the baselines and the me thods used in processing the GPS data (Sartori, personal correspondence, 2002). In addition, atmospheric effects should also be considered. For purposes of this study, it was not deemed necessary to perform a rigorous analysis of atmospheric eff ects. Nevertheless, increased tilt angle will clearly magnify any error due to the atmos phere for two reasons: 1) the optical path length of the laser increases with tilt angle; and 2) refraction increases with increase in optical path length. Yet another factor that may have contributed to the trend seen in Figure 4-1 is the possible reduction in range accur acy with a tilted sensor. Th e tilted sensor results in an elongated pulse, which, in turn, leads to a long er rise time for the re turn pulse. The CFD may experience difficulty with the longer rise time, leading to reduced range accuracy (Liadsky, personal correspondence, 2003). Lastly, height error (bia s) may have been introduced when changing the configuration of the laser syst em on board the aircraft. In this work, in-flight sensor calibration was performed for only one configuration: 0o tilt, narrow divergence, and 1200 m flying height. (See Chapter 3 for an overview of the calibration procedures.) While the importance of a separate calibra tion for each configuration was recognized, time constraints permitted only one calibration flight. It is not surprising to find systematic errors for configurations that are very different from th at used in calibration.

PAGE 55

42 Although the main focus of this study is on the detection of airport obstructions, rather than the absolute vertical accuracy, th e results of this analysis provide valuable information for the implementation of air borne laser mapping technology in airport obstruction surveying programs. First, to ac hieve the highest verti cal accuracy, in-flight calibrations should be performed for each configur ation of the system to be used in data collection. Second, with a tilted sensor, addi tional steps should be taken to reduce or eliminate errors in attitude angles to the greatest extent possible. Obstruction Detection Analysis The primary objective in analyzing th e data was to determine how well obstructions were detected in each of the fourte en airborne laser data sets. The reference data set used in the obstruc tion detection analysis consisted of 52 field-surveyed obstructions. These obstructions were positi oned by an NGS survey crew in February 2001, using GPS and conventional survey techni ques (Figure 4-3). Many of the objects surveyed by the field crew did not actually obstruct the FAR Part 77 or ANA surfaces at GNV, but were selected as be ing representative of “typical ” types of obstructions. The procedures followed in surveying these obj ects were identical to those commonly followed by NGS surveyors in performing FAR Part 77 and ANA obstruction surveys in accordance with FAA No. 405, Standards for Aeronautical Surveys and Related Products (U.S. Department of Transportation, 1996). P hotographs of many of the obstructions can be found in Appendix C. For purposes of this study, “detection” wa s defined as satisfying the following two conditions: 1) laser returns were received from the object surveyed by the field crew; and 2) the difference in elevation between the fiel d-surveyed point and the closest laser point on the object was within a predefined limit. It was not deemed necessary for the laser

PAGE 56

43 point to have hit the same part of the object surveyed by the field crew. For example, if the survey crew positioned an obstruction light on the west side of the top of a tower while the closest laser point hit the east si de of the tower, this would qualify as a detection, provided the difference in elevation was within tolerance. Furthermore, for clusters of trees that are in such close proxi mity to one another that the branches overlap, it was not considered necessary (or even possible) to verify that the closest laser point to the field-surveyed point actually hit the same tree and not its neighbor. Figure 4-3. NGS field survey of obstructions at GNV. Although the NGS data set originally contai ned 55 obstructions in the project areas, three of the obstructions were excluded from the analysis based on visual inspection of the survey area and discussions with air port authorities. Two of the excluded obstructions, Survey Point Numbers (SPNs) 418 and 438, were trees that were either burned or blown down between the dates of th e NGS field survey and the airborne laser data collection. The third excluded obstruc tion was an antenna (SPN 450) that was found to have been removed. Since the 52 remain ing obstructions include d seven antennae and

PAGE 57

44 fifteen trees, the absence of these three obstructions was not considered detrimental to the overall analysis. Automated Obstruction Detection Analysis The first step in the analysis entailed writing software to determine the number of field-surveyed obstructions detected in each airborne laser data set and the RMS difference in elevation between the field-su rveyed points and closest laser points on the detected obstructions. The software used a search cylinder centered on each fieldsurveyed obstruction (Figure 44) as the detection criteri on. If no laser point was found within the search cyli nder for a particular obstruction, th en that obstruction was reported as “not detected” by the softwa re. If the search cylinder contained one or more laser points, the laser point in the cylinder and cl osest in 3D distance to the field-surveyed point was located and used as the “matching” point. The output of the software included the number of obstructions detected in each la ser data set, the hor izontal and vertical distance from each field-surveyed obstruction to its matching laser point, and the RMS differences in elevation. The search cylinder was adopted based on th e definition of “detection” and served two purposes: first, to provide some measure of assurance that the laser point selected by the software hit the same object surveyed by the field crew; and second, to impose a limit on the maximum difference in elevation betw een the field-surveyed point and closest laser point that would qualify as a detection of that object. The radius of the search cylinder used in the analysis software wa s based on three factors: 1) the estimated horizontal position error in the field-surveyed data; 2) the estima ted horizontal position error in the airborne laser da ta; and 3) an allowance for th e movement of the tops of obstructions due to wind.

PAGE 58

45 Figure 4-4. Obstruction detection analysis algorithm. The red laser point is in the search cylinder and closest in 3D distance to the field-surveyed point, so it is selected as the “matching” point. If the search cylinder contains no laser points, the obstruction is reported to be “not detected.” Although no accuracy assessment was perfor med on the field-surveyed data, the survey party chief estimated the horizontal a nd vertical accuracy of the obstruction data to be no worse than 0.76 m (2.50 ft) and 0.15 m (0.50 ft), respectively, at the 95% confidence level. These estimates were ba sed on the survey methods and instruments used and on checks provided by redundant obs ervations during the fi eld survey (Kuper, personal correspondence, 2002). Tuell (2002) de termined the horizontal accuracy of the airborne laser data collected with Optech ALTM systems during the 2001 study to be 0.15 m at the 1level. The corresponding 95%-conf idence-level value, 0.26 m, was used as the estimated horizontal accur acy of the airborne laser data.

PAGE 59

46 Based on weather data from the Southeast Regional Climate Center in Columbia, South Carolina, wind speeds at GNV during the airborne laser data collection averaged just 13 km/hr (8 mph), but gusts of up to 37 km/hr (23 mph) were recorded. The wind conditions during the 2001 field survey ar e unknown but significantly less important, since an experienced field surveyor will take steps to minimize the effect of wind on the surveyed-position of an obstruction. Since ma ny of the trees surr ounding the airport are 20 to 25-m pines whose tops tend to sway in even mild breezes, 2 m was selected as a reasonable (perhaps slightly conservative) allowance for horizontal movement due to wind. Summing the estimated horizontal posit ion errors and the wind allowance gives a 3-m radius for the search cylinder. The horizontal accuracy requireme nt for obstructions specified by FAA No. 405 is 6.1 m (20 feet) or 15.2 m (50 f eet), depending on the locatio n of the object within the Obstruction Identification Surfaces (OIS). Thes e values were not used for the radius of the search cylinder because they were deemed too large to provide any assurance that the laser point hit the same object surveyed by th e field crew. The height of the search cylinder, on the other hand, wa s selected based on the vertical accuracy requirement specified in FAA No. 405 The vertical accuracy require ment is 0.91 m (3 feet) or 6.1 m (20 feet), again depending on th e location of the object within the OIS. The less-stringent value, 6.1 m, was used as the maximum differen ce in elevation (i.e., half the height of the cylinder) in the analysis software. Vertical accuracy of 6.1 m is an extr emely significant criterion to approach procedure developers. Specifi cally, obstruction data that m eet this standard can be designated as vertical code “C,” thus satisfy ing one of the key mini mum vertical accuracy

PAGE 60

47 requirements specified in FAA Order 8260.19, Flight Procedures and Airspace (U.S. Department of Transportation, 1993). Data th at do not meet this standard are given a vertical accuracy code of “D” (50 feet) or worse, which can affect minimum altitudes and lead to operational restri ctions (U.S. Department of Transportation, 1993). Table 4-2 summarizes the results of the au tomated analysis. The table has been sorted based on how well obstructions were detected using each configuration. Configuration 5 at the top of the table has the highest percent of obstructions detected (100%) and the lowest RMSE (0.88 m). At th e bottom of the table is Configuration 12 with only 63% detection and an RMSE of 2.17 m. The actual output of the analysis software showing the horizontal, vertical, a nd 3D distances from each obstruction to its matching laser point for each of the fourteen data sets is contained in Appendix D. Table 4-2. Percent of obstructions detected in each airborne laser data set and the RMS difference in elevation between the fiel d-surveyed points and “matching” laser points. Config # Parameters (tilt in deg; wide/narrow divergence; flying height in m) Percent of Obstructions Detected RMSE (m) Accuracy at 95% CL (m) 5 Tilt: 20, Div.: N, FH: 750 100 0.88 1.73 1 Tilt: 0, Div.: N, FH: 750 100 1.04 2.04 8 Tilt: 20, Div.: W, FH: 750 100 1.26 2.46 9 Tilt: 20, Div.: W, FH: 850 98 1.81 3.55 13 Tilt: 40, Div.: N, FH: 750 96 1.14 2.23 2 Tilt: 0, Div.: W, FH: 750 96 1.24 2.42 3 Tilt: 10, Div.: N, FH: 750 94 1.23 2.42 4 Tilt: 10, Div.: W, FH: 750 94 1.27 2.49 10 Tilt: 20, Div.: W, FH: 950 87 1.99 3.91 11 Tilt: 30, Div.: N, FH: 750 85 1.83 3.58 14 Tilt: 40, Div.: W, FH: 750 77 2.13 4.17 7 Tilt: 20, Div.: W, FH: 1150 77 2.16 4.22 6 Tilt: 20, Div.: W, FH: 1050 73 2.02 3.97 12 Tilt: 30, Div.: W, FH: 750 63 2.17 4.25

PAGE 61

48 Several interesting observations can be made from the results shown in Table 4-2: With constant flying height and tilt angl e, narrow divergence was consistently better than wide divergen ce for obstruction detection. For the 10o and 20o tilt angles, the numb er of obstructions detected was the same for the narrow and wide divergence setti ngs, with the only difference being in the RMSE. For the 30o and 40o tilt angles, the difference in the percent of obstructions detected between the narrow and wide di vergence setting increases to over 20%. For the 20o tilt angle and wide divergence, the pe rcentage of obstructions detected decreases rapidly with flying height. In Chapter 2, it was shown mathematically that obstruction dete ction depends on an interplay between geometry and return signa l strength (radiometry). The results and observations above clearly illustrate the tr adeoff that exists between geometry and radiometry. Based purely on geometric c onsiderations, confi gurations employing 30o to 40o tilt angles and wide divergence, such as numbers 12 and 14, should have been very well suited for obstruction detection, whereas configurations with near nadir-pointing beams and narrow divergence, such as number 1, should have been poor. Instead, however, configurations 12 and 14 are both near the bottom of the table, while configuration 1 is near the t op. The ability to detect obs tructions with a particular configuration clearly cannot be predicted based on geometry alone. The importance of radiometric considerations can be understood through examination of the reflectance data obtained with the field spectrometer (see Table 3-2 and Appendix B). These data indicate that many obstructions are poor reflectors at the laser wavelength of 1064 nm. Low reflectance at this wavelength combined with small cross-sectional area leads to very small eff ective target cross sect ions (see Equation (A6)). Regardless of the number of laser pulses incident on these objects, they will not be detected unless the irradiance is sufficiently high to result in a detectable return signal.

PAGE 62

49 It is interesting to note in Ta ble 4-2 that three of the four best configurations used a 20o tilt angle. With constant flying height and beam divergence, a nadir-pointing beam will minimize the range, giving the highest ir radiance on a target (see Equation (A-3)), whereas a large tilt angle will produce better geometry (Equation (2-3) and Figure 2-6). The results presented in Table 4-2 show that in this study, a tilt angle of 20o provided the best geometry while still en abling a detectable return si gnal from the obstructions. The somewhat anomalous results for confi guration 1 merit further investigation. Although, as noted above, the nadir-pointing beam used in configuration 1 should enable high return signal strength, the analysis pres ented in Chapter 2 shows that the obstruction detection geometry is poor with this configuration. In fact, configuration 1 is very similar to the configurations used in the 2001 st udy, which produced relatively poor results. Nevertheless, this configuration produced th e second best results in this study. A possible explanation for the results achieve d with configuration 1 lies in the fact that fairly substantial over lap between strips (approxima tely 25%) was used throughout this study. Apparently, the higher PRF a nd good strip overlap improved the geometry enough to enable configuration 1 to outperfor m the similar configurations used in the 2001 study. Because it is impossible to pred ict where within the scan pattern an obstruction will fall, however, increasing the tilt angle is a more reliable method of improving the detection geometry than increas ing the strip overlap. Further tests are needed to determine whether or not the favor able results achieved with configuration 1 can be replicated. Despite the steps taken to minimize uncont rolled variables duri ng the study, daily variations in weather conditions, GPS geomet ry, and even the performance of the laser

PAGE 63

50 are unavoidable. In at least one instance, these uncontrolled variables may have had a noticeable effect on the results. Specifical ly, reports generated by the data processing software showed that the data collected with the 30o tilt angle (configurations 11 and 12) on June 10th had a significantly lower percenta ge of returns than any of the other data sets collected during this st udy. It is suspected that th e atmospheric conditions were relatively poor on the eveni ng of June 10th, causing the 30o tilt angle configurations not to perform as well as they might have under better conditions. Visual Analysis The automated obstruction detection anal ysis was supplemented with a visual analysis performed using TerraScan Viewer (Terrasolid, Ltd.). A 0.02 km2 subset around each obstruction of interest was taken from each of the original airbor ne laser data sets. Using the TerraScan Viewer, the laser points we re displayed in profile mode to determine how well the obstruction was detected with each configuration. For example, Figure 4-5 shows the data points on an antenna (SPN 452) computed from the laser returns obtained using configurations 5 and 14. From Table 3-2, the reflectance of this object at 1064 nm is 0.32. The automated analysis software re ported SPN 452 to be “not detected” with configuration 14. From Figure 45 it can be seen that, in fa ct, with configuration 14, no detectable laser returns were rece ived from this obstruction. Figure 4-6 shows the results of a similar analysis performed on a pole (SPN 460). The automated analysis software reported th at this object was detected to within 0.34 meters vertical of the field-surveyed point with configuration 5 and to within 0.69 meters with configuration 8. The only difference be tween configurations 5 and 8 is that the former used narrow divergence and the latter wide. The obstruction was reported to be

PAGE 64

51 “not detected” with configuration 12. Again, the visual analysis s upports and clarifies the results of the automated analysis. It should be noted that the colors of the laser points in Figures 4-5 and 4-6 correspond to a classification by elevation range, but the classes were not standardized between the two figures. Figure 4-5.Visual obstruction analysis. The image on the left is a photograph of SPN 452. The middle image shows the data points computed from laser returns obtained using configuration 5. The image on the right shows that with configuration 14, no detectab le laser returns were received from the object. Figure 4-6. From left to right: photograph of SPN 460, and data points on this object based on laser returns obtained using c onfigurations 5, 8, a nd 12, respectively. Although it is difficult to discern from the photograph on the far left, the pole is in front of the tree. The tree was not included in the TerraScan profiles.

PAGE 65

52 Analysis of Return Sign al Strength Calculations Using Equations (2-11) and (2-12) and the reflectance data given in Table 3-2, the received power from an obstruction for each configuration can be estimated. By comparing the calculated return signal strength values against the results of the automated obstruction detection analysis (Appendix D) it is then possibl e, in theory, to determine the value of the minimum detectable return signal. This is some what of an inexact science because of the dayto-day (or even pulse-to-pul se) variability in certain parameters, such as the output power of the la ser and atmospheric transmittance. Further, the reflectance of an obstruction and its cros s-sectional area are typically not constant over its entire surface. Lastly, it is nearly impossible to pred ict, for any given laser pulse, where within the footprint an obstruction will fall. Based on the above arguments, the return si gnal strength calculat ions give, at best, a sort of estimated average value. It can be countered, however, that with good geometry, numerous laser pulses (perhaps even hundreds of pulses) will be incident on an obstruction. Therefore, even an estimated average value of th e return signal strength is useful, under the assumption that at least one pu lse will lead to a retu rn signal equal to or greater than the calculated value. Table 4-3 shows the calculated received power for SPN 452 (the antenna shown in Figure 4-5) for each configuration, along with the results of the automated obstruction detection analysis. Tables 4-4 and 4-5 show the results of similar analyses for SPNs 449 and 454, respectively. SPN 449 is a wood pole, while SPN 454 is a flagpole (see photographs in Appendix C). The value of the atmospheric transmittance, TATM, used in the calculations was 0.87.

PAGE 66

53 The data in Tables 4-3 through 4-5 indicat e that the minimum received power for detection is approximately 0.5 W. However, the following simplifying assumptions have been made in the return signal strength calculations: 1) all targets are lambertian; 2) all surfaces are flat; and 3) th e power distribution within the footprint is uniform. The calculated received power values would be sm aller if these assumptions had not been made. Hence, the values shown here should not be interpreted as an accurate portrayal of the performance characteristics of the Optech system. Nevertheless, the good relative agreement between the three tables suggests th at the analytical methods presented here may be used to examine the ability to de tect targets whose reflectance has been measured. Table 4-3. Analysis of return signal streng th calculations for SPN 452. Configurations 11 and 12 have been excluded from this analysis because, as noted above, the atmospheric conditions were poor during the collection of those two data sets. Configuration # Parameters (tilt in deg; wide/narrow divergence; flying height in m) Calculated Received Power from SPN 452 ( W) Detected (Y/N) 1 Tilt: 0, Div.: N, FH: 750 1.65 Yes 3 Tilt: 10, Div.: N, FH: 750 1.60 Yes 5 Tilt: 20, Div.: N, FH: 750 1.46 Yes 2 Tilt: 0, Div.: W, FH: 750 1.12 Yes 4 Tilt: 10, Div.: W, FH: 750 1.07 Yes 13 Tilt: 40, Div.: N, FH: 750 0.97 Yes 8 Tilt: 20, Div.: W, FH: 750 0.93 Yes 9 Tilt: 20, Div.: W, FH: 850 0.64 Yes 14 Tilt: 40, Div.: W, FH: 750 0.50 No 10 Tilt: 20, Div.: W, FH: 950 0.46 Yes 6 Tilt: 20, Div.: W, FH: 1050 0.34 No 7 Tilt: 20, Div.: W, FH: 1150 0.26 No

PAGE 67

54 Table 4-4. Analysis of return signal strength calculations for SPN 449. Again, configurations 11 and 12 have been excluded from the analysis for the reasons mentioned above. Configuration # Parameters (tilt in deg; wide/narrow divergence; flying height in m) Calculated Received Power from SPN 449 ( W) Detected (Y/N) 1 Tilt: 0, Div.: N, FH: 750 1.61 Yes 3 Tilt: 10, Div.: N, FH: 750 1.56 Yes 5 Tilt: 20, Div.: N, FH: 750 1.42 Yes 2 Tilt: 0, Div.: W, FH: 750 1.04 Yes 4 Tilt: 10, Div.: W, FH: 750 0.99 Yes 13 Tilt: 40, Div.: N, FH: 750 0.94 Yes 8 Tilt: 20, Div.: W, FH: 750 0.86 Yes 9 Tilt: 20, Div.: W, FH: 850 0.59 Yes 14 Tilt: 40, Div.: W, FH: 750 0.47 No 10 Tilt: 20, Div.: W, FH: 950 0.42 No 6 Tilt: 20, Div.: W, FH: 1050 0.31 No 7 Tilt: 20, Div.: W, FH: 1150 0.24 No Table 4-5. Analysis of return signal strength calculations for SPN 454. Configuration # Parameters (tilt in deg; wide/narrow divergence; flying height in m) Calculated Received Power from SPN 454 ( W) Detected (Y/N) 1 Tilt: 0, Div.: N, FH: 750 3.32 Yes 3 Tilt: 10, Div.: N, FH: 750 3.22 Yes 5 Tilt: 20, Div.: N, FH: 750 2.93 Yes 13 Tilt: 40, Div.: N, FH: 750 1.43 Yes 2 Tilt: 0, Div.: W, FH: 750 1.06 Yes 4 Tilt: 10, Div.: W, FH: 750 1.01 Yes 8 Tilt: 20, Div.: W, FH: 750 0.88 Yes 9 Tilt: 20, Div.: W, FH: 850 0.61 Yes 14 Tilt: 40, Div.: W, FH: 750 0.48 No 10 Tilt: 20, Div.: W, FH: 950 0.43 No 6 Tilt: 20, Div.: W, FH: 1050 0.32 No 7 Tilt: 20, Div.: W, FH: 1150 0.24 No Depending on the diameter of the laser foot print and, thus, on the configuration, the obstructions were sometimes treated as area ta rgets and sometimes as linear targets in the calculations. In the linear target cases, rather than assuming that the target fell directly in the center of the laser footprint, the expected value for the extent of the target in the

PAGE 68

55 footprint was used. The expected (average) va lue for the extent of a linear target in a circular footprint is obtained by dividing a qua drant of the footprint (i.e., the area under the curve) by the radius, giving 78.5% of the maximum value. A potentially more rigorous method of pe rforming the return signal strength calculations would be to model the interaction of the incident laser radiation with the target as a convolution, as depi cted in Figure 4-7. In this method, for each position of the kernel (i.e., each laser footprint position) a re turn signal strength value can be obtained. One obvious advantage of this method is a more rigorous modeling of both the laser footprint and the target Although this method was not used in the return signal strength calculations in this study due to the obvious complexity of modeling targets in this manner, it may be worth investig ating in future research. Figure 4-7. A potentially more rigorous method of performing the return signal strength computations involving modeling the intera ction of the incident laser radiation with a target as a convolution.

PAGE 69

56 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS The results of this research provide strong indication of the capability to detect and position airport obstructions requiring 6.1-m ve rtical accuracy using an airborne laserscanning system that has been appropriately configured. Three of the fourteen data collection configurations test ed in this study resulted in 100% of the field-surveyed obstructions being detected to within predefined tolerances that were established based on requirements set forth in FAA specifications documents. Particul arly encouraging is the vertical RMSE of 0.88 m ach ieved with configuration 5. While these results attest to the potential of airborne laser scanning in airport obstruction surveying, we may have only begun to scratch the surfac e in terms of the ability to detect and accurate ly position discrete point f eatures. Clearly, improvements above and beyond those demonstrated in this study are feasible. However, there is a limit to the improvement that can be achieve d through modification of data collection parameters by the user of the system; additional enhancements will require the cooperation of the system manufacturer. Further improvements to geometry will require increasing the pulse repetition frequency and, even more importantly, the s can frequency. One method of achieving a higher scan frequency is to ut ilize a dual-axis scanner. A side benefit of the dual-axis scanner would be a more regular scan pattern. Improvements in parameters relating to radiometric considerations are also possible. Several of the data collection configurations used in this study resulted in a

PAGE 70

57 high density of laser pulses incident on the obstructions, but the obstructions were not detected because the received signal was be low the detection threshold. This suggests that obstruction detection capability could be further improved by increasing the sensitivity of the receiver. This could be achieved, at least in th eory, through any of the following methods: Lowering the preset signal threshold in the CFD Utilizing a photodetector with a higher responsivity Increasing the area of the receiving optics Cooling the detector Modifying the bandpass filter to optimize SNR Changing any other parameters necessa ry to reduce noise in the system Although the modifications listed above s hould improve the detection capability, some of them could have significant drawback s. For example, reducing the preset signal threshold would increase the prob ability of false returns. Only the system manufacturer, or someone with intimate knowledge of the various system components and performance characteristics, would be able to evaluate the potential benefits and drawbacks associated with each. There is little doubt, ho wever, that it is possible to better tune the system for obstruction detection. While many applic ations demand the highest possible data accuracy, for obstruction detection it would be permissible to sacrifice a small amount of geometric accuracy for increased receiver sensitivity. The possible use of array detectors fo r further improving obstruction detection capability also merits investigation. Degnan (2002) proposes a spaceborne system employing a highly-pixellated (e.g., 10x10) array detector and a dual wedge optical scanner. Although the intended application of this proposed system is the mapping of planets, such as Mars, from orbiting spacecra ft, the use of array detectors could also prove beneficial in airport obstruction detection from air borne platforms. Potential

PAGE 71

58 advantages include increased spatial resolu tion and, depending on the type of detectors used, increased sensitivity. Although this study has focused on detec tion of airport obstructions, other important research topics invol ve extraction and classification of obstructions in airborne laser data sets. In completing an obstru ction survey through ai rborne laser scanning, detection of obstructions is onl y the first step. The obstructio n data must then be subset from the larger data set and classified. For example, it is important to know whether an object that penetrates the FAA survey surfaces is a manmade feature, such as a pole, or a natural feature, such as a tree. Most m odern airborne laser-scanning systems output “intensity” data in addition to the range m easurements. These data consist of digital numbers proportional to the gene rated photocurrent in the rece iver. The photocurrent is, in turn, proportional to the received optical power and, hence, to the reflectance of the target at the laser wavelength. This intensity data combined with aerial photography, if available, will undoubtedly prove beneficial in classification. Neve rtheless, the problem of classification is nontrivial and wi ll require continued research. Lastly, necessary enhancements to the pro cessing software should not be ignored. In this study, it was found that just as commercia l airborne laser mapping systems have been optimized for bare-earth terrain mappi ng, so too have data processing software packages. For airborne laser-scanning systems to be used in production airport obstruction surveying programs, software that is better tailored to this application would be extremely beneficial. Exis ting software could be used wi th the simple addition of a specialized airport obs truction processing module. This module would disable functions that serve to eliminate “suspect” last returns or interpolate or filter the data in any way.

PAGE 72

59 In addition, the module would include tools to allow better visualiza tion of obstructions, such as in profile mode. Through further experimentation and possibl e incorporation of the ideas mentioned above, it is likely that airborne laser sca nning will continue to b ecome an increasingly effective technology for airport obstruction surveying. The density of laser points incident on the vertical faces of obstructi ons will continue to increase, as will the probability of detecting obstructions with sma ll effective target cross sections. Improved software will give data pr ocessors greater ability to capture, detect, and visualize obstructions in the laser data set. It should be kept in mind, however, that the goal of these efforts is to supplement, rather than replace, existing technologies. It is unrealistic to expect, for example, that airborne lase r scanning could eliminate the need for field surveys or entirely replace photogrammetric procedures. However, expanding the number of viable airport obstruction survey ing technologies will give survey planners increased options and allow the survey methods to be better tailored to specific project requirements. The implementation of airborne laser scanning could represent the next major step in the technological evol ution of airport obs truction surveying.

PAGE 73

APPENDIX A DERIVATION OF RANGE EQUATION

PAGE 74

61 The following is a derivation of an e quation for the received signal power, commonly referred to as the “range equation” or “laser radar range equation.” The range equation can be found in various forms in numerous journal articles, including the following: Wyman (1968), Jelali an (1992), and Baltsavias (1999a). To start, it is noted that the irradiance incident on the receiver, Ea, measured inW m 2, can be expressed as E P R Ta refl s ATM 2 (A-1) where Prefl is the reflected power from a target, s is scattering solid angle of the target, R is the range, and TATM is the atmospheric transmittance. Prefl is given by PEArefltartar (A-2) where is the target reflectance, Etar is the irradiance on the target, and Atar is the target area. Next, Etar can be expressed as E P R Ttar T t ATM 2 (A-3) where PT is the transmitted power, and t is the solid angle into which the transmitted power is radiated (i.e., the so lid angle subtended by the lase r footprint), which is given by t FA R2 (A-4) In Equation (A-4), AF is the area of the footpr int, given approximately by ARF 422 (A-5) where is the beam divergence in radians. Fi nally, adapting the defi nition of effective target cross section, from Jelalian (1992): 4s tarA (A-6)

PAGE 75

62 and combining equations (A-1) through (A-6) gives the following expression for Ea: E P R Ta T ATM 224 2 (A-7) Using Equation (A-7), the received power can then be calculated from PAET PA R TTrraSYS Tr ATMSYS 224 2 (A-8) where Ar is the receiver area and TSYS is the system transmittance, which is limited primarily by the transmittance of the bandpa ss filter. By considering the size and position of the target in rela tion to the size and position of the laser footprint, three different types of targets can be defined: 1) an “area target” f ills the entire footprint; 2) a “linear target” extends the entire length of th e footprint but has a width that is small in comparison with the footprint diameter; and 3) a “point target” has an area much smaller than that of the footprint. Jelalian (1992) gives the following expressions for for area, linear and point targets, respectively: 2 2 Rarea (A-9) d Rlinear 4 (A-10) tar poA 4int (A-11) In Equation (A-10), d is the diameter of the linear target. The targets are assumed to be Lambertian ( s = The significance of equations (A-9) through (A-11) is that Ea is inversely proportional to R2 for an area target, R3 for a linear target, and R4 for a point target. In the special case that the target is lambertian and fills the entire footprint, and the system transmittance can be neglected, Equation (A-8) reduces to

PAGE 76

63 2 2 ATM r T rT R A P P (A-12) which is identical to the range e quation given in Baltsavias (1999a).

PAGE 77

APPENDIX B REFLECTANCE SPECTRA FOR OBSTRUCTIONS AND OTHER OBJECTS WITHIN THE SURVEY AREAS

PAGE 78

65 Figure B-1. Reflectance spectra for SPN 445 – strobe lighted tower (top), a guywire for the strobe lighted tower (middl e), and SPN 446 – tower (bottom).

PAGE 79

66 Figure B-2. Reflectance spectra for SPN 449 – pole (top), SPN 452 – antenna (middle), and SPN 453 – transmission pole (bottom).

PAGE 80

67 Figure B-3. Reflectance spectra for SPN 454 – flagpole (top), SPN 456 – pole (middle), and a pine tree (bottom).

PAGE 81

68 Figure B-4. Reflectance spectra for a palm tree (top), a pole (middle), and a generator (bottom).

PAGE 82

69 Figure B-5. Reflectance spectra for grass (top), concrete (mi ddle), and asphalt (bottom).

PAGE 83

APPENDIX C PHOTOGRAPHS OF FIELD-SURVEYED OBSTRUCTIONS

PAGE 84

71 Figure C-1. Photographs of SPN 414 – tree (top), a nd SPN 415 – tree (bottom). SPN 414 TREE SPN 415 TREE

PAGE 85

72 Figure C-2. Photographs of SPN 418 – tree (top), and SPN 431 – obstruction light on pole (bottom). SPN 418 TREE SPN 431 OL ON POLE

PAGE 86

73 Figure C-3. Photographs of SPN 446 – tower, SPN 445 – antenna on strobe lighted tower (top), and SPN 448 – antenna on strobe lighted tower (bottom). SPN 445 ANT ON STROBE LTD TW R SPN 446 TWR SPN 448 ANT ON STROBE LTD TW R

PAGE 87

74 Figure C-4. Photographs of SPN 449 – pole, SPN 453 – tran smission pole (top), and SPN 452 – antenna (bottom). SPN 449 POLE SPN 453 TRMSN POLE SPN 452 ANT

PAGE 88

75 Figure C-5. Photographs of SPN 454 – flagpole (top), SPN 456 – pole, SPN 455 – sign, and SPN 457 – transmission pole (bottom). SPN 454 FLGPL SPN 456 POLE SPN 455 SIGN SPN 457 TRMSN POLE

PAGE 89

76 Figure C-6. Photographs of SPN 457 – transmission pole (top), and SPN 459 – pole (bottom). SPN 457 TRMSN POLE SPN 459 POLE

PAGE 90

77 Figure C-7. Photograph of SPN 460 – pole. SPN 460 POLE

PAGE 91

APPENDIX D OUTPUT OF AUTOMATED OBSTRUCTIO N DETECTION ANALYSIS SOFTWARE

PAGE 92

79 Configuration 1 Table D-1. Tilt: 0; Div: N; FH:750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] 1.2328 0.8823 0.8610 436 OBST# 436 TREE 0.6160 0.5282 0.3170 437 OBST# 437 TREE 0.3563 0.1586 0.3190 439 OBST# 439 FENCE 0.4524 0.4524 -0.0010 440 OBST# 440 ANT ON HGR 0.6055 0.2659 0.5440 444 ANT ON STROBE LTD TWR!444 0.6229 0.1159 0.6120 445 ANT ON STROBE LTD TWR!445 0.7897 0.2314 0.7550 446 OBST# 446 TWR 0.1283 0.1209 -0.0430 447 ROD ON STROBE LTD TWR!447 0.9752 0.9195 0.3250 448 ANT ON STROBE LTD TWR!448 5.0592 2.0324 4.6330 449 OBST# 449 POLE 0.2608 0.2563 0.0480 451 OBST# 451 BLDG 0.8668 0.8668 -0.0030 452 OBST# 452 ANT 0.3562 0.3354 0.1200 453 OBST# 453 TRMSN POLE 0.2430 0.2384 0.0470 454 OBST# 454 FLGPL 0.2069 0.1436 -0.1490 455 OBST# 455 SIGN 0.9851 0.7070 0.6860 456 OBST# 456 POLE 0.9750 0.9470 0.2320 457 OBST# 457 TRMSN POLE 0.2118 0.2105 0.0230 458 OBST# 458 BLDG 0.3248 0.2875 0.1510 459 OBST# 459 POLE 0.2281 0.2045 0.1010 460 OBST# 460 POLE 1.0108 0.8408 0.5610 461 OBST# 461 ROD ON OL AMOM 1.1850 0.5257 -1.0620 462 OL VORTAC!462 [GNV] "NCM" 0.2879 0.2851 0.0400 463 OBST# 463 LT POLE 0.7598 0.1086 0.7520 464 OBST# 464 LT POLE 0.2711 0.2708 -0.0130 465 OBST# 465 FENCE 0.4637 0.3949 0.2430 466 OBST# 466 FENCE 0.3325 0.3069 0.1280 467 OBST# 467 TREE 0.4545 0.2971 0.3440 468 OBST# 468 TREE 0.2478 0.2297 0.0930 469 OBST# 469 TREE 2.8884 2.0830 2.0010 470 OBST# 470 TREE 0.7550 0.0871 0.7500 471 OBST# 471 TREE 0.4776 0.0497 0.4750 472 OBST# 472 TREE 0.4138 0.2912 0.2940

PAGE 93

80 Table D-1—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 473 OBST# 473 TREE 0.1700 0.1697 0.0090 474 OBST# 474 TREE 0.3912 0.1329 0.3680 475 OBST# 475 HGR 0.3616 0.3563 -0.0620 476 OBST# 476 SIGN 1.0931 0.1869 1.0770 477 OBST# 477 FENCE 0.7263 0.6391 0.3450 478 OBST# 478 FLGPL 1.1430 0.7848 0.8310 479 OBST# 479 TREE 0.7573 0.3032 0.6940 480 OBST# 480 ANT ON BLDG 2.4871 0.5566 2.4240 302 ANT ON OL ATCT!ATCT FLOOR164 4.6637 2.7009 3.8020 306 OL ON LTD WSK 0.5995 0.5846 0.1330 25 ROD ON OL APBN!APBN 1.0998 0.3240 1.0510 412 OL ON LOC!(28) 0.4656 0.4656 0.0040 415 TREE 0.2852 0.2517 0.1340 423 TREE 0.6122 0.5364 -0.2950 430 BLDG 0.1711 0.1711 0.0000 431 OL ON POLE 1.8813 1.8813 0.0070 402 ROD ON OL GS!(28) 0.8102 0.3993 0.7050 414 TREE 0.4778 0.1100 0.4650 425 TREE 0.4921 0.4865 -0.0740 RMSE: 1.04065 Accuracy: 2.03968 Percent Detected: 100. Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt0_DivN_Fh750.alf Configuration 2 Table D-2. Tilt: 0; Div: W; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] 1.2247 0.3455 1.1750 436 OBST# 436 TREE 0.2734 0.1252 0.2430 437 OBST# 437 TREE 0.4934 0.1860 0.4570 439 OBST# 439 FENCE 0.6672 0.3533 0.5660 440 OBST# 440 ANT ON HGR 5.2558 1.0090 5.1580 444 ANT ON STROBE LTD TWR!444 3.0710 2.4787 1.8130

PAGE 94

81 Table D-2—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 445 ANT ON STROBE LTD TWR!445 4.1782 0.5015 4.1480 446 OBST# 446 TWR 0.7792 0.6001 0.4970 447 ROD ON STROBE LTD TWR!447 0.8820 0.4154 0.7780 448 ANT ON STROBE LTD TWR!448 0.9346 0.8788 -0.3180 449 OBST# 449 POLE 0.6199 0.5664 0.2520 451 OBST# 451 BLDG 0.4753 0.4739 -0.0360 452 OBST# 452 ANT 0.4710 0.4541 0.1250 453 OBST# 453 TRMSN POLE 0.4136 0.3998 0.1060 454 OBST# 454 FLGPL 0.3527 0.3413 -0.0890 455 OBST# 455 SIGN 0.3905 0.3887 -0.0370 456 OBST# 456 POLE 0.3595 0.3527 0.0700 457 OBST# 457 TRMSN POLE 0.2098 0.1981 -0.0690 458 OBST# 458 BLDG 0.2669 0.2647 0.0340 459 OBST# 459 POLE 0.4305 0.2548 0.3470 460 OBST# 460 POLE 0.5633 0.4960 0.2670 461 OBST# 461 ROD ON OL AMOM 1.2677 0.8150 -0.9710 462 OL VORTAC!462 [GNV] "NCM" 0.2436 0.2433 0.0110 463 OBST# 463 LT POLE 0.4518 0.4464 0.0700 464 OBST# 464 LT POLE 0.2907 0.2885 -0.0360 465 OBST# 465 FENCE 2.1580 0.4754 2.1050 466 OBST# 466 FENCE 2.0950 0.4854 2.0380 467 OBST# 467 TREE 0.5921 0.1529 0.5720 468 OBST# 468 TREE 0.6561 0.5885 0.2900 469 OBST# 469 TREE 3.1639 1.9871 2.4620 470 OBST# 470 TREE 0.8718 0.2959 0.8200 471 OBST# 471 TREE 0.6156 0.2974 0.5390 472 OBST# 472 TREE 0.2945 0.2942 0.0120 473 OBST# 473 TREE 0.2388 0.2319 0.0570 474 OBST# 474 TREE 0.5225 0.2786 0.4420 475 OBST# 475 HGR 0.1443 0.1000 -0.1040 476 OBST# 476 SIGN 1.1316 0.1392 1.1230 477 OBST# 477 FENCE 2.0282 0.1145 2.0250 478 OBST# 478 FLGPL 0.3542 0.2417 0.2590 479 OBST# 479 TREE 1.0034 0.2819 0.9630 480 OBST# 480 ANT ON BLDG ND ND ND

PAGE 95

82 Table D-2—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 0.7204 0.6439 0.3230 25 ROD ON OL APBN!APBN 1.1052 0.0963 1.1010 412 OL ON LOC!(28) 0.1876 0.1715 0.0760 415 TREE 0.7163 0.1267 0.7050 423 TREE 0.4329 0.2991 -0.3130 430 BLDG 0.0873 0.0830 0.0270 431 OL ON POLE 1.1387 1.1322 0.1220 402 ROD ON OL GS!(28) 0.7321 0.2950 0.6700 414 TREE 0.4697 0.4649 -0.0670 425 TREE 0.4524 0.3084 0.3310 RMSE: 1.23575 Accuracy: 2.42208 Percent Detected: 96.15 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt0_DivW_Fh750.alf Configuration 3 Table D-3. Tilt: 10; Div: N; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] 1.1040 0.5417 0.9620 436 OBST# 436 TREE 0.5105 0.4229 0.2860 437 OBST# 437 TREE 0.6040 0.5781 0.1750 439 OBST# 439 FENCE 0.1095 0.1007 0.0430 440 OBST# 440 ANT ON HGR 5.6905 1.2025 5.5620 444 ANT ON STROBE LTD TWR!444 2.7029 1.9863 1.8330 445 ANT ON STROBE LTD TWR!445 0.3955 0.3112 0.2440 446 OBST# 446 TWR 0.6567 0.5810 0.3060 447 ROD ON STROBE LTD TWR!447 ND ND ND 448 ANT ON STROBE LTD TWR!448 ND ND ND 449 OBST# 449 POLE 1.4932 0.6689 1.3350 451 OBST# 451 BLDG 0.3812 0.3608 -0.1230

PAGE 96

83 Table D-3—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 452 OBST# 452 ANT 0.3050 0.2302 0.2000 453 OBST# 453 TRMSN POLE 0.6971 0.6507 -0.2500 454 OBST# 454 FLGPL 1.9684 0.7760 1.8090 455 OBST# 455 SIGN 0.4176 0.4117 0.0700 456 OBST# 456 POLE 0.4989 0.4201 0.2690 457 OBST# 457 TRMSN POLE 0.3086 0.3005 0.0700 458 OBST# 458 BLDG 0.3189 0.2759 0.1600 459 OBST# 459 POLE 0.5197 0.5179 -0.0430 460 OBST# 460 POLE 0.5767 0.5582 0.1450 461 OBST# 461 ROD ON OL AMOM 1.5547 0.6780 -1.3990 462 OL VORTAC!462 [GNV] "NCM" 0.4395 0.4306 0.0880 463 OBST# 463 LT POLE 0.6831 0.6300 0.2640 464 OBST# 464 LT POLE 0.4883 0.4879 0.0180 465 OBST# 465 FENCE 0.4544 0.3726 0.2600 466 OBST# 466 FENCE 0.3149 0.2441 0.1990 467 OBST# 467 TREE 0.5216 0.4710 0.2240 468 OBST# 468 TREE 0.4369 0.2401 0.3650 469 OBST# 469 TREE 3.5618 1.5864 3.1890 470 OBST# 470 TREE 1.0985 0.9451 0.5600 471 OBST# 471 TREE 1.1680 1.0357 0.5400 472 OBST# 472 TREE 0.1022 0.0918 0.0450 473 OBST# 473 TREE 0.3937 0.3317 0.2120 474 OBST# 474 TREE 0.5500 0.1703 0.5230 475 OBST# 475 HGR 0.2818 0.2816 -0.0110 476 OBST# 476 SIGN 0.7289 0.5342 0.4960 477 OBST# 477 FENCE 2.1520 0.4879 2.0960 478 OBST# 478 FLGPL 0.2563 0.2147 -0.1400 479 OBST# 479 TREE 0.8971 0.3124 0.8410 480 OBST# 480 ANT ON BLDG ND ND ND 302 ANT ON OL ATCT!ATCT FLOOR164 4.5412 2.9329 3.4670 306 OL ON LTD WSK 0.9616 0.9113 0.3070 25 ROD ON OL APBN!APBN 1.1466 0.8629 0.7550 412 OL ON LOC!(28) 0.2434 0.2420 0.0260 415 TREE 0.8733 0.5832 0.6500 423 TREE 0.9269 0.8912 0.2550 430 BLDG 0.3583 0.3559 0.0420 431 OL ON POLE 1.7611 1.5335 0.8660 402 ROD ON OL GS!(28) 0.9555 0.3951 0.8700

PAGE 97

84 Table D-3—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 414 TREE 0.3709 0.3443 0.1380 425 TREE 0.6098 0.4334 0.4290 RMSE: 1.23274 Accuracy: 2.41618 Percent Detected: 94.23 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt10_DivN_Fh750.alf Configuration 4 Table D-4. Tilt: 10; Div: W; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] 1.4240 0.2658 1.3990 436 OBST# 436 TREE 0.3808 0.2281 0.3050 437 OBST# 437 TREE 0.4292 0.1792 0.3900 439 OBST# 439 FENCE 0.4839 0.2351 0.4230 440 OBST# 440 ANT ON HGR 5.7306 1.5677 5.5120 444 ANT ON STROBE LTD TWR!444 2.6042 1.8733 1.8090 445 ANT ON STROBE LTD TWR!445 3.9322 0.7131 3.8670 446 OBST# 446 TWR 0.4808 0.1185 0.4660 447 ROD ON STROBE LTD TWR!447 ND ND ND 448 ANT ON STROBE LTD TWR!448 0.5122 0.4936 -0.1370 449 OBST# 449 POLE 0.3567 0.3561 0.0200 451 OBST# 451 BLDG 0.1087 0.1087 0.0020 452 OBST# 452 ANT 0.3851 0.3378 0.1850 453 OBST# 453 TRMSN POLE 0.1687 0.1646 -0.0370 454 OBST# 454 FLGPL 0.4268 0.3651 -0.2210 455 OBST# 455 SIGN 0.4655 0.3958 0.2450 456 OBST# 456 POLE 0.2723 0.1376 0.2350 457 OBST# 457 TRMSN POLE 0.2206 0.2180 -0.0340 458 OBST# 458 BLDG 0.2848 0.2474 0.1410 459 OBST# 459 POLE 0.3931 0.2211 0.3250 460 OBST# 460 POLE 0.5908 0.4311 0.4040

PAGE 98

85 Table D-4—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 461 OBST# 461 ROD ON OL AMOM 1.5205 1.2753 -0.8280 462 OL VORTAC!462 [GNV] "NCM" 0.4436 0.4428 0.0270 463 OBST# 463 LT POLE 0.4003 0.3989 0.0330 464 OBST# 464 LT POLE 0.4193 0.4163 -0.0500 465 OBST# 465 FENCE 1.9764 0.2461 1.9610 466 OBST# 466 FENCE 2.0964 0.1759 2.0890 467 OBST# 467 TREE 0.5538 0.2745 0.4810 468 OBST# 468 TREE 0.4668 0.1085 0.4540 469 OBST# 469 TREE 2.6613 0.9781 2.4750 470 OBST# 470 TREE 0.7176 0.4026 0.5940 471 OBST# 471 TREE 0.5720 0.4222 0.3860 472 OBST# 472 TREE 0.2171 0.1900 0.1050 473 OBST# 473 TREE 0.4238 0.4195 0.0600 474 OBST# 474 TREE 0.6157 0.1088 0.6060 475 OBST# 475 HGR 0.2662 0.2483 -0.0960 476 OBST# 476 SIGN 1.1811 0.3163 1.1380 477 OBST# 477 FENCE 2.1547 0.5161 2.0920 478 OBST# 478 FLGPL 0.6021 0.3248 0.5070 479 OBST# 479 TREE 0.9774 0.1670 0.9630 480 OBST# 480 ANT ON BLDG ND ND ND 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 1.6682 0.6917 1.5180 25 ROD ON OL APBN!APBN 1.0391 0.3125 0.9910 412 OL ON LOC!(28) 0.2251 0.2080 -0.0860 415 TREE 0.3916 0.3163 0.2310 423 TREE 0.5273 0.5206 0.0840 430 BLDG 0.2182 0.2146 0.0390 431 OL ON POLE 1.2977 1.2662 0.2840 402 ROD ON OL GS!(28) 0.8698 0.3366 0.8020 414 TREE 0.0364 0.0232 0.0280 425 TREE 0.5761 0.3345 0.4690 RMSE: 1.27157 Accuracy: 2.49227 Percent Detected: 94.23 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt10_DivW_Fh750.alf

PAGE 99

86 Configuration 5 Table D-5. Tilt: 20; Div: N; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] 1.4770 0.4645 1.4020 436 OBST# 436 TREE 0.2998 0.1561 0.2560 437 OBST# 437 TREE 0.3465 0.3084 0.1580 439 OBST# 439 FENCE 0.8246 0.7516 0.3390 440 OBST# 440 ANT ON HGR 1.1372 0.8141 0.7940 444 ANT ON STROBE LTD TWR!444 2.1425 0.1665 2.1360 445 ANT ON STROBE LTD TWR!445 2.3963 0.7251 2.2840 446 OBST# 446 TWR 0.1925 0.1916 0.0180 447 ROD ON STROBE LTD TWR!447 0.6692 0.2838 0.6060 448 ANT ON STROBE LTD TWR!448 0.6050 0.5348 -0.2830 449 OBST# 449 POLE 1.0022 0.6659 0.7490 451 OBST# 451 BLDG 0.5406 0.5002 -0.2050 452 OBST# 452 ANT 0.4515 0.4488 0.0500 453 OBST# 453 TRMSN POLE 0.1002 0.0707 -0.0710 454 OBST# 454 FLGPL 0.3412 0.2574 -0.2240 455 OBST# 455 SIGN 0.6802 0.6736 0.0940 456 OBST# 456 POLE 0.8395 0.8281 0.1380 457 OBST# 457 TRMSN POLE 0.5876 0.5758 -0.1170 458 OBST# 458 BLDG 0.5294 0.5285 0.0310 459 OBST# 459 POLE 0.7662 0.7295 0.2340 460 OBST# 460 POLE 0.9116 0.8446 0.3430 461 OBST# 461 ROD ON OL AMOM 1.0323 0.4967 -0.9050 462 OL VORTAC!462 [GNV] "NCM" 0.2815 0.2752 0.0590 463 OBST# 463 LT POLE 0.8736 0.8734 0.0190 464 OBST# 464 LT POLE 0.2213 0.1589 0.1540 465 OBST# 465 FENCE 0.5452 0.5380 0.0880 466 OBST# 466 FENCE 0.3493 0.3149 0.1510 467 OBST# 467 TREE 0.8078 0.3778 0.7140 468 OBST# 468 TREE 0.5312 0.2638 0.4610 469 OBST# 469 TREE 2.7547 2.1485 1.7240 470 OBST# 470 TREE 0.8340 0.3808 0.7420 471 OBST# 471 TREE 0.7738 0.6037 0.4840

PAGE 100

87 Table D-5—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 472 OBST# 472 TREE 0.5147 0.3312 0.3940 473 OBST# 473 TREE 0.1867 0.0771 0.1700 474 OBST# 474 TREE 0.6312 0.4866 0.4020 475 OBST# 475 HGR 0.6480 0.6480 0.0000 476 OBST# 476 SIGN 1.2029 0.1372 1.1950 477 OBST# 477 FENCE 0.8258 0.6720 0.4800 478 OBST# 478 FLGPL 0.4588 0.3687 0.2730 479 OBST# 479 TREE 0.9437 0.8010 0.4990 480 OBST# 480 ANT ON BLDG 1.1148 0.8460 0.7260 302 ANT ON OL ATCT!ATCT FLOOR164 4.9239 2.9598 3.9350 306 OL ON LTD WSK 0.4589 0.3718 0.2690 25 ROD ON OL APBN!APBN 1.2111 0.5001 1.1030 412 OL ON LOC!(28) 0.8603 0.8425 -0.1740 415 TREE 0.4401 0.4302 -0.0930 423 TREE 1.1912 0.9351 0.7380 430 BLDG 0.1622 0.1458 -0.0710 431 OL ON POLE 1.4131 1.3294 0.4790 402 ROD ON OL GS!(28) 0.9245 0.4948 0.7810 414 TREE 0.3673 0.3400 0.1390 425 TREE 0.6229 0.5864 -0.2100 RMSE: 0.881212 Accuracy: 1.72718 Percent Detected: 100. Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt20_DivN_Fh750.alf Configuration 6 Table D-6. Tilt: 20; Div: W; FH: 1050 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] ND ND ND 436 OBST# 436 TREE 0.3413 0.0790 0.3320 437 OBST# 437 TREE 0.5860 0.2414 0.5340 439 OBST# 439 FENCE 0.6164 0.5565 0.2650 440 OBST# 440 ANT ON HGR 5.6915 0.3621 5.6800 444 ANT ON STROBE LTD TWR!444 5.6582 1.2995 5.5070

PAGE 101

88 Table D-6—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 445 ANT ON STROBE LTD TWR!445 4.5828 0.7063 4.5280 446 OBST# 446 TWR 1.4308 0.4476 1.3590 447 ROD ON STROBE LTD TWR!447 1.2171 0.7259 0.9770 448 ANT ON STROBE LTD TWR!448 ND ND ND 449 OBST# 449 POLE ND ND ND 451 OBST# 451 BLDG 0.5908 0.5706 -0.1530 452 OBST# 452 ANT ND ND ND 453 OBST# 453 TRMSN POLE ND ND ND 454 OBST# 454 FLGPL ND ND ND 455 OBST# 455 SIGN 1.0435 1.0405 0.0790 456 OBST# 456 POLE ND ND ND 457 OBST# 457 TRMSN POLE ND ND ND 458 OBST# 458 BLDG 0.1503 0.1240 0.0850 459 OBST# 459 POLE 0.6130 0.4309 0.4360 460 OBST# 460 POLE 6.0378 2.3564 5.5590 461 OBST# 461 ROD ON OL AMOM ND ND ND 462 OL VORTAC!462 [GNV] "NCM" 0.1990 0.1507 0.1300 463 OBST# 463 LT POLE ND ND ND 464 OBST# 464 LT POLE 0.3378 0.3370 0.0230 465 OBST# 465 FENCE 1.9994 0.1165 1.9960 466 OBST# 466 FENCE 2.1648 0.2254 2.1530 467 OBST# 467 TREE 0.7824 0.3536 0.6980 468 OBST# 468 TREE 0.5077 0.1791 0.4750 469 OBST# 469 TREE 3.5163 2.3881 2.5810 470 OBST# 470 TREE 0.8980 0.3355 0.8330 471 OBST# 471 TREE 1.2478 0.8319 0.9300 472 OBST# 472 TREE 0.2610 0.1801 0.1890 473 OBST# 473 TREE 0.1534 0.1375 0.0680 474 OBST# 474 TREE 0.8027 0.1580 0.7870 475 OBST# 475 HGR 0.4914 0.4242 -0.2480 476 OBST# 476 SIGN 1.2336 0.7400 0.9870 477 OBST# 477 FENCE 2.2138 0.5854 2.1350 478 OBST# 478 FLGPL ND ND ND 479 OBST# 479 TREE 0.9373 0.0248 0.9370 480 OBST# 480 ANT ON BLDG ND ND ND

PAGE 102

89 Table D-6—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 2.7312 1.3574 2.3700 25 ROD ON OL APBN!APBN 1.1498 0.1340 1.1420 412 OL ON LOC!(28) 0.7298 0.2751 0.6760 415 TREE 1.9302 1.4974 1.2180 423 TREE 0.4551 0.4376 -0.1250 430 BLDG 0.4200 0.0764 0.4130 431 OL ON POLE ND ND ND 402 ROD ON OL GS!(28) 2.1831 0.7198 2.0610 414 TREE 0.2251 0.1531 0.1650 425 TREE 0.3359 0.3148 0.1170 RMSE: 2.02307 Accuracy: 3.96522 Percent Detected: 73.08 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt20_DivW_Fh1050.alf Configuration 7 Table D-7. Tilt: 20; Div: W; FH: 1150 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] ND ND ND 436 OBST# 436 TREE 0.4077 0.2462 0.3250 437 OBST# 437 TREE 0.4440 0.3189 0.3090 439 OBST# 439 FENCE 0.7205 0.6696 0.2660 440 OBST# 440 ANT ON HGR 5.3857 0.7584 5.3320 444 ANT ON STROBE LTD TWR!444 5.3988 0.5066 5.3750 445 ANT ON STROBE LTD TWR!445 4.9737 0.7011 4.9240 446 OBST# 446 TWR 1.7117 0.9330 1.4350 447 ROD ON STROBE LTD TWR!447 1.1967 0.5611 1.0570 448 ANT ON STROBE LTD TWR!448 ND ND ND 449 OBST# 449 POLE ND ND ND 451 OBST# 451 BLDG 0.8886 0.7825 -0.4210

PAGE 103

90 Table D-7—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 452 OBST# 452 ANT ND ND ND 453 OBST# 453 TRMSN POLE ND ND ND 454 OBST# 454 FLGPL ND ND ND 455 OBST# 455 SIGN 4.6751 0.5470 4.6430 456 OBST# 456 POLE ND ND ND 457 OBST# 457 TRMSN POLE ND ND ND 458 OBST# 458 BLDG 0.3882 0.2060 0.3290 459 OBST# 459 POLE 0.8606 0.4834 0.7120 460 OBST# 460 POLE 5.6833 2.5463 5.0810 461 OBST# 461 ROD ON OL AMOM ND ND ND 462 OL VORTAC!462 [GNV] "NCM" 0.6545 0.6450 0.1110 463 OBST# 463 LT POLE ND ND ND 464 OBST# 464 LT POLE 0.7507 0.6886 -0.2990 465 OBST# 465 FENCE 1.9107 0.8026 1.7340 466 OBST# 466 FENCE 2.0265 0.3078 2.0030 467 OBST# 467 TREE 0.5305 0.3821 0.3680 468 OBST# 468 TREE 0.5458 0.4370 0.3270 469 OBST# 469 TREE 4.4601 1.6913 4.1270 470 OBST# 470 TREE 0.7164 0.3607 0.6190 471 OBST# 471 TREE 1.4352 1.3252 0.5510 472 OBST# 472 TREE 0.2392 0.1007 0.2170 473 OBST# 473 TREE 0.5429 0.4167 0.3480 474 OBST# 474 TREE 0.7703 0.6631 0.3920 475 OBST# 475 HGR 0.3403 0.1237 -0.3170 476 OBST# 476 SIGN 1.3447 0.3283 1.3040 477 OBST# 477 FENCE 1.7517 0.6128 1.6410 478 OBST# 478 FLGPL 1.0378 0.3146 0.9890 479 OBST# 479 TREE 0.9885 0.2003 0.9680 480 OBST# 480 ANT ON BLDG ND ND ND 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 2.6094 1.4172 2.1910 25 ROD ON OL APBN!APBN 1.0277 0.2933 0.9850 412 OL ON LOC!(28) 0.8258 0.6970 0.4430 415 TREE 2.1079 1.9329 0.8410 423 TREE 0.4492 0.3027 -0.3320 430 BLDG 0.2565 0.2508 0.0540 431 OL ON POLE 0.8523 0.8040 0.2830 402 ROD ON OL GS!(28) 3.7867 0.7126 3.7190

PAGE 104

91 Table D-7—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 414 TREE 0.0843 0.0622 0.0570 425 TREE 0.4658 0.3194 0.3390 RMSE: 2.15529 Accuracy: 4.22436 Percent Detected: 76.92 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt20_DivW_Fh1150.alf Configuration 8 Table D-8. Tilt: 20; Div: W; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] 1.6623 0.6133 1.5450 436 OBST# 436 TREE 0.4583 0.4423 0.1200 437 OBST# 437 TREE 0.5452 0.4399 0.3220 439 OBST# 439 FENCE 0.4430 0.3305 0.2950 440 OBST# 440 ANT ON HGR 4.6767 0.9116 4.5870 444 ANT ON STROBE LTD TWR!444 1.3423 0.5466 1.2260 445 ANT ON STROBE LTD TWR!445 2.5176 0.5353 2.4600 446 OBST# 446 TWR 0.6597 0.6101 0.2510 447 ROD ON STROBE LTD TWR!447 0.9074 0.3822 0.8230 448 ANT ON STROBE LTD TWR!448 0.7191 0.7074 0.1290 449 OBST# 449 POLE 0.5469 0.4073 0.3650 451 OBST# 451 BLDG 0.4044 0.3520 0.1990 452 OBST# 452 ANT 0.2296 0.2294 0.0100 453 OBST# 453 TRMSN POLE 0.1997 0.1997 0.0050 454 OBST# 454 FLGPL 5.7185 0.4210 5.7030 455 OBST# 455 SIGN 0.5982 0.5972 0.0340 456 OBST# 456 POLE 0.6039 0.5387 0.2730 457 OBST# 457 TRMSN POLE 0.1110 0.0827 0.0740 458 OBST# 458 BLDG 0.3131 0.2295 0.2130 459 OBST# 459 POLE 0.5283 0.2894 0.4420 460 OBST# 460 POLE 0.7865 0.3793 0.6890

PAGE 105

92 Table D-8—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 461 OBST# 461 ROD ON OL AMOM 0.3064 0.1728 0.2530 462 OL VORTAC!462 [GNV] "NCM" 0.4586 0.4524 0.0750 463 OBST# 463 LT POLE 0.8437 0.8256 0.1740 464 OBST# 464 LT POLE 0.6014 0.5328 0.2790 465 OBST# 465 FENCE 0.4635 0.3472 0.3070 466 OBST# 466 FENCE 0.7172 0.7042 0.1360 467 OBST# 467 TREE 0.2531 0.1637 0.1930 468 OBST# 468 TREE 0.3348 0.2470 0.2260 469 OBST# 469 TREE 2.6053 1.9820 1.6910 470 OBST# 470 TREE 0.9117 0.4242 0.8070 471 OBST# 471 TREE 1.1023 0.7389 0.8180 472 OBST# 472 TREE 0.5080 0.4948 0.1150 473 OBST# 473 TREE 0.3594 0.3564 0.0460 474 OBST# 474 TREE 0.6381 0.2227 0.5980 475 OBST# 475 HGR 0.4286 0.4084 -0.1300 476 OBST# 476 SIGN 0.4992 0.4809 0.1340 477 OBST# 477 FENCE 1.9759 1.9288 0.4290 478 OBST# 478 FLGPL 1.0470 0.8573 0.6010 479 OBST# 479 TREE 0.8132 0.4155 0.6990 480 OBST# 480 ANT ON BLDG 1.5923 1.3387 0.8620 302 ANT ON OL ATCT!ATCT FLOOR164 4.1660 2.9477 2.9440 306 OL ON LTD WSK 0.3448 0.3357 0.0790 25 ROD ON OL APBN!APBN 0.9638 0.4831 0.8340 412 OL ON LOC!(28) 0.0944 0.0868 0.0370 415 TREE 0.7914 0.7866 -0.0870 423 TREE 1.0592 1.0464 0.1640 430 BLDG 0.2556 0.2461 -0.0690 431 OL ON POLE 2.1898 2.1756 0.2490 402 ROD ON OL GS!(28) 1.1681 1.0051 0.5950 414 TREE 0.4196 0.4134 0.0720 425 TREE 0.4456 0.4174 -0.1560 RMSE: 1.25604 Accuracy: 2.46183 Percent Detected: 100. Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt20_DivW_Fh750.alf

PAGE 106

93 Configuration 9 Table D-9. Tilt: 20; Div: W; FH: 850 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] 1.4500 0.5682 1.3340 436 OBST# 436 TREE 0.5808 0.4854 0.3190 437 OBST# 437 TREE 0.3248 0.1759 0.2730 439 OBST# 439 FENCE 0.4462 0.3285 0.3020 440 OBST# 440 ANT ON HGR 5.4716 0.3988 5.4570 444 ANT ON STROBE LTD TWR!444 3.1993 2.6102 1.8500 445 ANT ON STROBE LTD TWR!445 4.8616 0.2119 4.8570 446 OBST# 446 TWR 0.2989 0.1322 0.2680 447 ROD ON STROBE LTD TWR!447 1.2319 0.6665 1.0360 448 ANT ON STROBE LTD TWR!448 0.7014 0.6758 -0.1880 449 OBST# 449 POLE 0.7928 0.1038 0.7860 451 OBST# 451 BLDG 0.2964 0.2961 -0.0140 452 OBST# 452 ANT 0.3079 0.2950 0.0880 453 OBST# 453 TRMSN POLE 2.8172 0.2510 2.8060 454 OBST# 454 FLGPL 5.5819 0.3932 5.5680 455 OBST# 455 SIGN 0.1970 0.1960 0.0200 456 OBST# 456 POLE 0.3438 0.3112 0.1460 457 OBST# 457 TRMSN POLE 0.3479 0.3231 0.1290 458 OBST# 458 BLDG 0.3035 0.3018 0.0320 459 OBST# 459 POLE 0.5345 0.4898 0.2140 460 OBST# 460 POLE 0.4835 0.4696 0.1150 461 OBST# 461 ROD ON OL AMOM 1.1645 0.9496 -0.6740 462 OL VORTAC!462 [GNV] "NCM" 0.4591 0.4067 0.2130 463 OBST# 463 LT POLE 0.2429 0.2389 -0.0440 464 OBST# 464 LT POLE 0.2255 0.2223 -0.0380 465 OBST# 465 FENCE 1.9079 0.3311 1.8790 466 OBST# 466 FENCE 2.0103 0.1932 2.0010 467 OBST# 467 TREE 0.5287 0.3613 0.3860 468 OBST# 468 TREE 0.4122 0.1523 0.3830 469 OBST# 469 TREE 3.4098 1.4375 3.0920 470 OBST# 470 TREE 1.0005 0.4815 0.8770 471 OBST# 471 TREE 0.7839 0.3701 0.6910

PAGE 107

94 Table D-9—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 472 OBST# 472 TREE 0.6210 0.4750 0.4000 473 OBST# 473 TREE 0.5069 0.4613 0.2100 474 OBST# 474 TREE 0.5904 0.1713 0.5650 475 OBST# 475 HGR 0.2883 0.2700 0.1010 476 OBST# 476 SIGN 1.0044 0.5414 0.8460 477 OBST# 477 FENCE 1.9296 0.3936 1.8890 478 OBST# 478 FLGPL 1.1272 0.6296 0.9350 479 OBST# 479 TREE 1.0117 0.3222 0.9590 480 OBST# 480 ANT ON BLDG ND ND ND 302 ANT ON OL ATCT!ATCT FLOOR164 6.1277 1.5167 5.9370 306 OL ON LTD WSK 2.3877 0.5925 2.3130 25 ROD ON OL APBN!APBN 1.0167 0.5245 0.8710 412 OL ON LOC!(28) 0.4757 0.4613 0.1160 415 TREE 0.6006 0.4895 0.3480 423 TREE 0.9821 0.8854 0.4250 430 BLDG 0.2014 0.1366 0.1480 431 OL ON POLE 1.3655 1.3509 0.1990 402 ROD ON OL GS!(28) 0.8257 0.5338 0.6300 414 TREE 0.4366 0.4365 -0.0080 425 TREE 0.7801 0.5954 0.5040 RMSE: 1.81365 Accuracy: 3.55476 Percent Detected: 98.08 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt20_DivW_Fh850.alf Configuration 10 Table D-10. Tilt: 20; Div: W; FH: 950 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] ND ND ND 436 OBST# 436 TREE 0.3449 0.2243 0.2620 437 OBST# 437 TREE 0.4888 0.1909 0.4500 439 OBST# 439 FENCE 0.8810 0.5926 0.6520 440 OBST# 440 ANT ON HGR 5.4783 0.9516 5.3950 444 ANT ON STROBE LTD TWR!444 5.5136 0.7948 5.4560

PAGE 108

95 Table D-10—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 445 ANT ON STROBE LTD TWR!445 4.6474 0.4849 4.6220 446 OBST# 446 TWR 1.1290 0.6960 0.8890 447 ROD ON STROBE LTD TWR!447 1.2046 0.8074 0.8940 448 ANT ON STROBE LTD TWR!448 5.3862 2.0519 4.9800 449 OBST# 449 POLE ND ND ND 451 OBST# 451 BLDG 0.1808 0.1807 0.0060 452 OBST# 452 ANT 0.6582 0.6535 -0.0780 453 OBST# 453 TRMSN POLE 4.5609 2.2239 3.9820 454 OBST# 454 FLGPL ND ND ND 455 OBST# 455 SIGN 0.6083 0.6028 0.0820 456 OBST# 456 POLE 0.8510 0.5850 0.6180 457 OBST# 457 TRMSN POLE ND ND ND 458 OBST# 458 BLDG 0.3941 0.3118 0.2410 459 OBST# 459 POLE 0.4225 0.3743 0.1960 460 OBST# 460 POLE 0.3819 0.0908 0.3710 461 OBST# 461 ROD ON OL AMOM ND ND ND 462 OL VORTAC!462 [GNV] "NCM" 0.3562 0.3504 -0.0640 463 OBST# 463 LT POLE 0.4155 0.4129 -0.0460 464 OBST# 464 LT POLE 0.2689 0.2676 0.0270 465 OBST# 465 FENCE 2.1667 0.6528 2.0660 466 OBST# 466 FENCE 2.2230 0.2747 2.2060 467 OBST# 467 TREE 0.6680 0.4440 0.4990 468 OBST# 468 TREE 0.3689 0.1904 0.3160 469 OBST# 469 TREE 5.1507 0.8566 5.0790 470 OBST# 470 TREE 0.8663 0.6352 0.5890 471 OBST# 471 TREE 0.3014 0.1960 0.2290 472 OBST# 472 TREE 0.4528 0.3980 0.2160 473 OBST# 473 TREE 0.1122 0.0627 0.0930 474 OBST# 474 TREE 0.8913 0.3493 0.8200 475 OBST# 475 HGR 0.6360 0.6355 -0.0240 476 OBST# 476 SIGN 1.2450 0.4029 1.1780 477 OBST# 477 FENCE 1.9286 0.1179 1.9250 478 OBST# 478 FLGPL 1.2483 0.6689 1.0540 479 OBST# 479 TREE 1.0199 0.3472 0.9590 480 OBST# 480 ANT ON BLDG ND ND ND

PAGE 109

96 Table D-10—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 3.1680 1.4724 2.8050 25 ROD ON OL APBN!APBN 1.0237 0.2753 0.9860 412 OL ON LOC!(28) 0.6572 0.6510 0.0900 415 TREE 1.5653 0.9782 1.2220 423 TREE 0.7909 0.7886 0.0600 430 BLDG 0.3598 0.2598 0.2490 431 OL ON POLE 1.9887 1.9839 0.1390 402 ROD ON OL GS!(28) 0.9713 0.2993 0.9240 414 TREE 0.2995 0.2785 0.1100 425 TREE 0.5006 0.4862 0.1190 RMSE: 1.9925 Accuracy: 3.90531 Percent Detected: 86.54 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt20_DivW_Fh950.alf Configuration 11 Table D-11. Tilt: 30; Div: N; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] ND ND ND 436 OBST# 436 TREE 0.4275 0.4260 0.0360 437 OBST# 437 TREE 0.6208 0.5953 0.1760 439 OBST# 439 FENCE 0.3345 0.1846 0.2790 440 OBST# 440 ANT ON HGR 5.2552 1.0365 5.1520 444 ANT ON STROBE LTD TWR!444 5.0175 0.5338 4.9890 445 ANT ON STROBE LTD TWR!445 4.8295 0.4660 4.8070 446 OBST# 446 TWR 2.2285 1.8836 1.1910 447 ROD ON STROBE LTD TWR!447 1.3082 0.1854 1.2950 448 ANT ON STROBE LTD TWR!448 1.2721 1.2405 -0.2820 449 OBST# 449 POLE 1.0113 0.6695 0.7580 451 OBST# 451 BLDG 0.8870 0.8762 0.1380

PAGE 110

97 Table D-11—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 452 OBST# 452 ANT ND ND ND 453 OBST# 453 TRMSN POLE ND ND ND 454 OBST# 454 FLGPL ND ND ND 455 OBST# 455 SIGN 5.8888 1.0414 5.7960 456 OBST# 456 POLE ND ND ND 457 OBST# 457 TRMSN POLE ND ND ND 458 OBST# 458 BLDG 0.5519 0.1232 0.5380 459 OBST# 459 POLE 2.3198 1.1406 2.0200 460 OBST# 460 POLE 1.2503 1.0353 0.7010 461 OBST# 461 ROD ON OL AMOM 1.6049 1.4651 0.6550 462 OL VORTAC!462 [GNV] "NCM" 0.5592 0.5008 0.2490 463 OBST# 463 LT POLE 0.7648 0.7114 0.2810 464 OBST# 464 LT POLE 0.5125 0.4368 0.2680 465 OBST# 465 FENCE 1.8613 0.8035 1.6790 466 OBST# 466 FENCE 2.1996 0.4691 2.1490 467 OBST# 467 TREE 0.6923 0.4901 0.4890 468 OBST# 468 TREE 0.8353 0.5953 0.5860 469 OBST# 469 TREE 3.7340 2.2539 2.9770 470 OBST# 470 TREE 1.2778 0.1101 1.2730 471 OBST# 471 TREE 1.1689 0.8270 0.8260 472 OBST# 472 TREE 0.2837 0.1111 0.2610 473 OBST# 473 TREE 0.4397 0.4111 -0.1560 474 OBST# 474 TREE 0.6049 0.4334 0.4220 475 OBST# 475 HGR 0.8806 0.8641 0.1700 476 OBST# 476 SIGN 0.7394 0.5650 0.4770 477 OBST# 477 FENCE 2.0390 0.5297 1.9690 478 OBST# 478 FLGPL 1.0603 0.9034 0.5550 479 OBST# 479 TREE 0.7753 0.6226 0.4620 480 OBST# 480 ANT ON BLDG ND ND ND 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 1.6025 1.5380 0.4500 25 ROD ON OL APBN!APBN 1.2084 0.4981 1.1010 412 OL ON LOC!(28) 0.3529 0.3409 -0.0910 415 TREE 1.1478 0.3773 1.0840 423 TREE 1.3440 1.2606 0.4660 430 BLDG 0.3550 0.2786 0.2200 431 OL ON POLE 2.7725 2.7381 0.4350 402 ROD ON OL GS!(28) 1.5006 0.7302 1.3110

PAGE 111

98 Table D-11—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 414 TREE 0.6765 0.6765 0.0080 425 TREE 0.6964 0.1714 0.6750 RMSE: 1.82691 Accuracy: 3.58073 Percent Detected: 84.62 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt30_DivN_Fh750.alf Configuration 12 Table D-12. Tilt: 30; Div: W; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] ND ND ND 436 OBST# 436 TREE 0.2654 0.1111 0.2410 437 OBST# 437 TREE 0.5104 0.4000 0.3170 439 OBST# 439 FENCE 0.4995 0.0863 0.4920 440 OBST# 440 ANT ON HGR 5.9864 0.3533 5.9760 444 ANT ON STROBE LTD TWR!444 6.0879 2.0509 5.7320 445 ANT ON STROBE LTD TWR!445 5.4885 0.2869 5.4810 446 OBST# 446 TWR 2.2392 1.7140 1.4410 447 ROD ON STROBE LTD TWR!447 1.1506 0.3830 1.0850 448 ANT ON STROBE LTD TWR!448 ND ND ND 449 OBST# 449 POLE ND ND ND 451 OBST# 451 BLDG 1.0177 0.8947 0.4850 452 OBST# 452 ANT ND ND ND 453 OBST# 453 TRMSN POLE ND ND ND 454 OBST# 454 FLGPL ND ND ND 455 OBST# 455 SIGN 2.0272 1.9676 0.4880 456 OBST# 456 POLE ND ND ND 457 OBST# 457 TRMSN POLE ND ND ND 458 OBST# 458 BLDG 0.4475 0.3417 0.2890 459 OBST# 459 POLE ND ND ND 460 OBST# 460 POLE ND ND ND

PAGE 112

99 Table D-12—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 461 OBST# 461 ROD ON OL AMOM ND ND ND 462 OL VORTAC!462 [GNV] "NCM" 0.4922 0.4186 0.2590 463 OBST# 463 LT POLE ND ND ND 464 OBST# 464 LT POLE ND ND ND 465 OBST# 465 FENCE 1.9144 0.2499 1.8980 466 OBST# 466 FENCE 2.0096 0.3818 1.9730 467 OBST# 467 TREE 0.7412 0.4628 0.5790 468 OBST# 468 TREE 0.7604 0.3463 0.6770 469 OBST# 469 TREE ND ND ND 470 OBST# 470 TREE 2.4087 2.0161 1.3180 471 OBST# 471 TREE 0.8825 0.3185 0.8230 472 OBST# 472 TREE 0.2716 0.2598 0.0790 473 OBST# 473 TREE 0.1126 0.1124 0.0070 474 OBST# 474 TREE 0.8674 0.3399 0.7980 475 OBST# 475 HGR 0.3110 0.2938 0.1020 476 OBST# 476 SIGN 0.2234 0.1579 0.1580 477 OBST# 477 FENCE 1.8350 0.3412 1.8030 478 OBST# 478 FLGPL ND ND ND 479 OBST# 479 TREE 0.8769 0.3080 0.8210 480 OBST# 480 ANT ON BLDG ND ND ND 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 2.5866 0.8266 2.4510 25 ROD ON OL APBN!APBN 0.9886 0.3268 0.9330 412 OL ON LOC!(28) 2.1114 0.2189 2.1000 415 TREE ND ND ND 423 TREE 1.0568 1.0113 0.3070 430 BLDG 0.4147 0.1931 0.3670 431 OL ON POLE ND ND ND 402 ROD ON OL GS!(28) 5.0315 0.6599 4.9880 414 TREE 0.6607 0.4972 0.4350 425 TREE 0.8148 0.4564 0.6750 RMSE: 2.1689 Accuracy: 4.25105 Percent Detected: 63.46 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt30_DivW_Fh750.alf

PAGE 113

100 Configuration 13 Table D-13. Tilt: 40; Div: N; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] 1.2577 0.3466 1.2090 436 OBST# 436 TREE 0.4251 0.4207 0.0610 437 OBST# 437 TREE 0.5924 0.3593 0.4710 439 OBST# 439 FENCE 0.5839 0.5021 0.2980 440 OBST# 440 ANT ON HGR 5.5039 1.9783 5.1360 444 ANT ON STROBE LTD TWR!444 3.7900 2.2127 3.0770 445 ANT ON STROBE LTD TWR!445 2.8985 0.6354 2.8280 446 OBST# 446 TWR 1.3101 1.1074 0.7000 447 ROD ON STROBE LTD TWR!447 1.2746 0.3485 1.2260 448 ANT ON STROBE LTD TWR!448 0.3412 0.3412 0.0070 449 OBST# 449 POLE 0.9047 0.4390 0.7910 451 OBST# 451 BLDG 0.5886 0.3231 -0.4920 452 OBST# 452 ANT 0.9434 0.9400 -0.0800 453 OBST# 453 TRMSN POLE 0.5313 0.4770 -0.2340 454 OBST# 454 FLGPL 0.7586 0.7470 -0.1320 455 OBST# 455 SIGN 0.9711 0.6527 0.7190 456 OBST# 456 POLE 0.3385 0.3015 0.1540 457 OBST# 457 TRMSN POLE 0.9631 0.8854 -0.3790 458 OBST# 458 BLDG 0.5976 0.3851 0.4570 459 OBST# 459 POLE 0.5985 0.5985 -0.0090 460 OBST# 460 POLE 0.7156 0.5499 0.4580 461 OBST# 461 ROD ON OL AMOM 0.5778 0.3231 -0.4790 462 OL VORTAC!462 [GNV] "NCM" 1.0543 0.9742 0.4030 463 OBST# 463 LT POLE 0.7260 0.6771 0.2620 464 OBST# 464 LT POLE 0.1181 0.1124 0.0360 465 OBST# 465 FENCE 1.9144 0.5073 1.8460 466 OBST# 466 FENCE 0.7212 0.5462 0.4710 467 OBST# 467 TREE 0.4678 0.4431 0.1500 468 OBST# 468 TREE 0.4607 0.3882 0.2480 469 OBST# 469 TREE 2.7465 1.8256 2.0520 470 OBST# 470 TREE 0.5073 0.4751 0.1780 471 OBST# 471 TREE 1.0373 0.8606 0.5790 472 OBST# 472 TREE 0.6738 0.1145 0.6640

PAGE 114

101 Table D-13—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 473 OBST# 473 TREE 0.2472 0.1942 -0.1530 474 OBST# 474 TREE 0.6973 0.4723 0.5130 475 OBST# 475 HGR 0.3118 0.3118 0.0030 476 OBST# 476 SIGN 0.6896 0.6819 0.1030 477 OBST# 477 FENCE 1.4872 1.3781 0.5590 478 OBST# 478 FLGPL 1.4992 0.5177 1.4070 479 OBST# 479 TREE 0.7508 0.3941 0.6390 480 OBST# 480 ANT ON BLDG ND ND ND 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 0.9690 0.9103 0.3320 25 ROD ON OL APBN!APBN 1.3585 0.2815 1.3290 412 OL ON LOC!(28) 0.4532 0.4529 0.0170 415 TREE 0.9072 0.9059 0.0500 423 TREE 1.4013 1.3997 -0.0670 430 BLDG 0.2333 0.1988 0.1220 431 OL ON POLE 1.5240 1.4727 0.3920 402 ROD ON OL GS!(28) 0.8578 0.4035 0.7570 414 TREE 0.4216 0.4152 -0.0730 425 TREE 0.2893 0.2711 0.1010 RMSE: 1.13574 Accuracy: 2.22604 Percent Detected: 96.15 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt40_DivN_Fh750.alf Configuration 14 Table D-14. Tilt: 40; Div: W; FH: 750 SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 410 ROD ON OL ASOS!SENSOR [GNV] ND ND ND 436 OBST# 436 TREE 0.3499 0.2125 0.2780 437 OBST# 437 TREE 0.5313 0.4256 0.3180 439 OBST# 439 FENCE 1.1324 0.5861 0.9690 440 OBST# 440 ANT ON HGR 5.5875 0.2246 5.5830 444 ANT ON STROBE LTD TWR!444 6.0607 1.8944 5.7570

PAGE 115

102 Table D-14—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 445 ANT ON STROBE LTD TWR!445 5.0697 0.3444 5.0580 446 OBST# 446 TWR 1.7696 0.7726 1.5920 447 ROD ON STROBE LTD TWR!447 1.1456 0.3496 1.0910 448 ANT ON STROBE LTD TWR!448 ND ND ND 449 OBST# 449 POLE ND ND ND 451 OBST# 451 BLDG 0.8698 0.7509 -0.4390 452 OBST# 452 ANT ND ND ND 453 OBST# 453 TRMSN POLE ND ND ND 454 OBST# 454 FLGPL ND ND ND 455 OBST# 455 SIGN 1.4969 1.4828 0.2050 456 OBST# 456 POLE ND ND ND 457 OBST# 457 TRMSN POLE ND ND ND 458 OBST# 458 BLDG 0.1957 0.1952 -0.0140 459 OBST# 459 POLE 0.7852 0.1679 0.7670 460 OBST# 460 POLE 1.2238 0.5260 1.1050 461 OBST# 461 ROD ON OL AMOM 5.9917 1.1816 5.8740 462 OL VORTAC!462 [GNV] "NCM" 0.2425 0.2177 0.1070 463 OBST# 463 LT POLE ND ND ND 464 OBST# 464 LT POLE 0.5024 0.3366 0.3730 465 OBST# 465 FENCE 2.1086 0.7114 1.9850 466 OBST# 466 FENCE 1.9395 0.2283 1.9260 467 OBST# 467 TREE 0.5900 0.4925 0.3250 468 OBST# 468 TREE 0.9447 0.7716 0.5450 469 OBST# 469 TREE 5.0882 2.4362 4.4670 470 OBST# 470 TREE 0.7356 0.4862 0.5520 471 OBST# 471 TREE 0.8963 0.6020 0.6640 472 OBST# 472 TREE 0.3124 0.2938 0.1060 473 OBST# 473 TREE 0.2651 0.2451 0.1010 474 OBST# 474 TREE 0.6415 0.3448 0.5410 475 OBST# 475 HGR 0.2584 0.1586 -0.2040 476 OBST# 476 SIGN 0.4001 0.3977 0.0440 477 OBST# 477 FENCE 1.3264 0.7616 1.0860 478 OBST# 478 FLGPL 0.8933 0.1894 0.8730 479 OBST# 479 TREE 0.8064 0.5286 0.6090 480 OBST# 480 ANT ON BLDG ND ND ND

PAGE 116

103 Table D-14—Continued SPN Description Dist to Closest LIDAR Pt (m) 2D Dist (m) Delta Elev (m) 302 ANT ON OL ATCT!ATCT FLOOR164 ND ND ND 306 OL ON LTD WSK 2.3891 0.6500 2.2990 25 ROD ON OL APBN!APBN 1.0781 0.4907 0.9600 412 OL ON LOC!(28) 1.4009 0.6383 1.2470 415 TREE 1.4711 1.1016 0.9750 423 TREE 1.1049 1.0998 0.1060 430 BLDG 0.1969 0.1395 -0.1390 431 OL ON POLE ND ND ND 402 ROD ON OL GS!(28) 3.1107 0.0650 3.1100 414 TREE 0.3318 0.2483 0.2200 425 TREE 0.4411 0.2461 -0.3660 RMSE: 2.12989 Accuracy: 4.17458 Percent Detected: 76.92 Search Radius: 3. LIDAR data file: C:\1Chris\University_of_Florida\New _AccuracyTest\Tilt40_DivW_Fh750.alf

PAGE 117

104 LIST OF REFERENCES Anderson, F., G. Tuell, and C. Parrish, 2002. Application of LIDAR for airport mapping and obstacle detection. The 3rd Inte rnational LIDAR Workshop: Mapping GeoSurficial Processes Using Laser Altimetry 07-09 October, 2002, Columbus, Ohio. Anderson, J.M, and C. Lee, 1975. Analytical in-flight calibration. Photogrammetric Engineering and Remote Sensing, Vol. 41, No. 11, pp. 1337-1348. Baltsavias, E.P., 1999a. Airborne laser s canning: basic relations and formulas. ISPRS Journal of Photogrammetry & Remote Sensing, Vol. 54, pp. 199-214. Baltsavias, E.P., 1999b. A comparison betw een photogrammetry and laser scanning. ISPRS Journal of Photogrammetry & Remote Sensing, Vol. 54, pp. 83-94. Binkley, D.M., and M.E. Casey, 1988. Perf ormance of fast monolithic ECL voltage comparators in constant-fraction discri minators and other timing circuits. IEEE Transactions on Nuclear Science, Vol. 35, No. 1, pp. 226-230. Blair, J.B., D.L. Rabine, and M.A. Hofton, 1999. The laser vegetation imaging sensor: a medium-altitude, digitisation–only, airborne laser altimeter for mapping vegetation and topography. ISPRS Journal of Photogrammetry & Remote Sensing, Vol. 54, pp. 115-122. Carter, W., R. Shrestha, G. Tuell, D. Bloo mquist, and M. Sartori, 2001. Airborne laser swath mapping shines new light on earth’s topography. Eos, Transactions, American Geophysical Union, Vol. 82, No. 46, pp. 549-555. Degnan, J.J., 2002. A conceptual design for a spaceborne 3D imaging lidar. Elektrotechnik und in formationstechnik, heft 4, pp. 99-106. Federal Geographic Data Committee, 1998. Geospatial Positioning Accuracy Standards Part 3: National Standard for Spatial Data Accuracy. Federal Geographic Data Committee, Reston, Virginia. Filin, S., 2001. Calibration of Airborne and Spaceborne Laser Altimeters Using Natural Surfaces, Ph.D. dissertation, The Ohio Stat e University, Columbus, Ohio, 128 p. Filin, S., 2002. A laser strip adjustment mode l for the removal of systematic errors in airborne laser data. The 3rd Intern ational LIDAR Workshop: Mapping GeoSurficial Processes Using Laser Altimetry 07-09 October, 2002, Columbus, Ohio.

PAGE 118

105 Harris, R.L., Jr., and A.R. Johnson, 2001. Detec ting airfield vertical obstructions using digital photogrammetry and GIS. Proceedings of the ASPRS 2001 Annual Convention, 23-27 April, St. Louis, Mi ssouri (American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland), unpaginated CDROM. International Civil Aviation Organization, 1995. Annex 4 to the Convention of International Civil Avia tion – Aeronautical Charts. ICAO, Montreal, Canada. International Civil Aviation Organization, 1999. Annex 14 to the Convention of International Civil Avia tionAerodromes, Volume I Aerodrome Design and Operations. ICAO, Montreal, Canada. Jelalian, A.V., 1992. Laser Radar Systems. Artech House, Boston, Massachusetts. Kasap, S.O., 2001. Optoelectronics and Photonics. Prentice-Hall, Inc., Upper Saddle River, New Jersey, 340 p. Kenefick, J.F., M.S. Gyer, and B.F. Ha rp, 1972. Analytical self-calibration. Photogrammetric Engineering, Vol. 38, No. 11, pp. 1117-1126. Krabill, W.B., R.H. Thomas, C.F. Martin, R.N. Swift, and E.B. Frederick, 1995. Accuracy of airborne laser altim etry over the Greenland ice sheet. International Journal of Remote Sensing, Vol. 16, No. 7, pp. 1211-1222. Lillesand, T.M., and R.W. Kiefer, 2000. Remote Sensing and Image Interpretation, Fourth Edition. John Wiley & Sons, Inc., New York, NY. Lindenberger, J., 1989. Test results of lase r profiling for topographic terrain survey. Proceedings 42nd Photogrammetric Week, Stuttgart, Germany. Maas, H., 2002. Methods for measuring height and planimetry discrepancies in airborne laserscanner data. Phogogrammetric Engineering & Remote Sensing, Vol. 68, No. 9, pp. 933-940. Mader, G.L., 1992. Rapid static and kinematic global positioning system solutions using the ambiguity function technique. Journal of Geophysical Research, Vol. 97, pp. 3271-3283. Maune, D.F., ed., 2001. Digital Elevation Model Tec hnologies and Applications: The DEM Users Manual. American Society for Photogrammetry and Remote Sensing, Bethesda, Maryland, Ch. 7, pp. 207-236 and Ch. 12, pp. 395-440. Mikhail, E.M., and F. Ackermann, 1976. Observations and Least Squares. IEPA DunDonnelley Publisher, New York, New York, 497 p.

PAGE 119

106 National Imagery and Mapping Agency, 2001. Airfield Initiative Document. URL: http://164.214.2.62/products/rbai/index.ht ml (last date accessed: 02 November 2002). Optech, 1998. Airborne Laser Terrain Mapper Training Manual. Optech, Inc., Toronto, Ontario, Canada. Prinzel, L.J., L.J. Kramer, J.R. Comstock, R.E. Bailey, M.F. Hughes, and R.V. Parrish, 2002. NASA synthetic vision EGE flight test. Human Factors and Ergonomics Society 46th Annual Meeting, 30 September – 04 October, 2002, Baltimore, Maryland. Schenk, T., B. Csath, and D.C. Lee, 1999. Quality control issues of airborne laser ranging data and accuracy study in an urban area. International Archives of Photogrammetry and Remote Sensing, Vol. 32, Part 3-W14, pp. 101-108. Schenk, T., 2001. Modeling and Analyzing Systematic Errors in Airborne Laser Scanners. Technical Notes in Photogrammetry No. 19, Department of Civil and Environmental Engineering and Geodetic Science, The Ohio State University, Columbus, Ohio, 42 p. Schenk, T, S. Seo, and B. Csath, 2001. Accu racy study of airborne laser scanning data with photogrammetry. International Archives of Photogrammetry and Remote Sensing, Vol. 34, Part 3-W4, pp. 113-118. Shrestha, R.L., W.E. Carter, M. Lee, P. Fine r, and M. Sartori, 1999. Airborne laser swath mapping: accuracy assessment for su rveying and mapping applications. Surveying and Land Information Systems Journal of the American Congress on Surveying and Mapping, Vol. 59, No. 2, pp. 83-94. Slama, C.C., ed., 1980. Manual of Photogrammetry, Fourth Edition. American Society of Photogrammetry, Falls Church, Virginia, Ch. II, pp. 37-101. Toth, C., D. Grejner-Brzezins ka, and N. Csanyi, 2002. Calibra ting airborne lidar systems: automating the boresight misalignment, The 3rd International LIDAR Workshop: Mapping Geo-Surficial Processes Using La ser Altimetry, 07-09 October, 2002, Columbus, Ohio. Tuell, G., 1985. Airport Surveys Field Handbook. U.S. Department of Commerce, NOAA, NOS, Norfolk, Virginia. Tuell, G., 1987. Technical Development Plan for the Modernization of the Airport Obstruction Charting Program. NOAA Charting Research and Development Laboratory, Office of Ch arting and Geodetic Services, Rockville, Maryland.

PAGE 120

107 Tuell, G., 2002. Airborne laser sw ath mapping (ASLM) for determining discrete airspace obstructions. Invi ted Presentation, Special Session on Airspace Obstruction Mapping, National Transportation Research Board (TRB), January 2002, Washington, DC. U.S. Department of Transportation, 1993. FAA Order 8260.19, Flight Procedures and Airspace. Federal Aviation Administ ration, Washington, DC. U.S. Department of Transportation, 1996. FAA No. 405, Standards for Aeronautical Surveys and Related Products, Fourth Edition. Federal Aviation Administration, Washington, DC. Van ek, P., and E.J. Krakiwsky, 1986. Geodesy: The Concepts, Second Edition. Elsevier Science Publishers B.V., Amsterdam, the Netherlands, 697 p. Vaughn, C.R., J.L. Bufton, W.B. Krabill, and D. Rabine, 1996a. Georeferencing of airborne laser altimeter measurements. International Journal of Remote Sensing, Vol. 17, No. 11, pp. 2185-2200. Vaughn, C.R., J.L. Bufton, and D. Rabine, 1996b. Progress in geor eferencing airborne laser altimeter measurements. Second In ternational Airborne Remote Sensing Conference and Exhibition, 24-27 June, 1996, San Francisco, California, pp. I-286294. Wehr, A., and U. Lohr, 1999. Airborne lase r scanning – an introduction and overview. ISPRS Journal of Photogrammetry & Remote Sensing, Vol. 54, pp. 68-82. Wyman, P.W., 1968. Definition of laser radar cross section. Applied Optics, Vol. 7, No. 1, p. 207.

PAGE 121

108 BIOGRAPHICAL SKETCH Christopher E. Parrish was born in Corv allis, Oregon, in 1970. He earned a Bachelor of Science degree with honors in physics from Bates College in 1993. In 1994, he accepted a commission in the NOAA Corps. He then served as a Junior Officer aboard the NOAA Ship WHITING, a hydrogr aphic survey vessel, until 1997. From 1997 through 2000, Mr. Parrish was assigned to the National Geodetic Survey (NGS) as Geodetic Operations and Liaison Officer. In that position, he participated in geodetic control, airp ort obstruction, runway profile, and NAVAID surveys, serving as Party Chief for three surveys. Additional duties included automation, software development, and remote sensing research in the NGS Aeronautical Survey Program. In 2000, Mr. Parrish resigned his commissi on in the NOAA Corps but remained in federal service. He is currently employed as a physical scientist in the NGS Remote Sensing Division. His responsibilities incl ude research and development in emerging remote sensing technologies to support NGS programs. After completing his master’s degree at the University of Florida, Mr. Parris h will return to Silver Spring, Maryland, to continue his career at NGS.


Permanent Link: http://ufdc.ufl.edu/UFE0000765/00001

Material Information

Title: Analysis of airborne laser-scanning system configurations for detecting airport obstructions
Physical Description: Mixed Material
Language: English
Creator: Parrish, Christopher E. ( Dissertant )
Tuell, Grady H. ( Thesis advisor )
Shrestha, Ramesh L. ( Thesis advisor )
Carter, Bill ( Reviewer )
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2003
Copyright Date: 2003

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000765:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000765/00001

Material Information

Title: Analysis of airborne laser-scanning system configurations for detecting airport obstructions
Physical Description: Mixed Material
Language: English
Creator: Parrish, Christopher E. ( Dissertant )
Tuell, Grady H. ( Thesis advisor )
Shrestha, Ramesh L. ( Thesis advisor )
Carter, Bill ( Reviewer )
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2003
Copyright Date: 2003

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000765:00001


This item has the following downloads:


Full Text












ANALYSIS OF AIRBORNE LASER-SCANNING SYSTEM CONFIGURATIONS
FOR DETECTING AIRPORT OBSTRUCTIONS
















By

CHRISTOPHER E. PARRISH


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2003

































Copyright 2003

by

Christopher E. Parrish




























To Deborah















ACKNOWLEDGMENTS

I wish to express my gratitude to Dr. Grady Tuell, chair of my supervisory

committee, for his significant contributions to this thesis and his continued guidance and

support. I thank Drs. Bill Carter and Ramesh Shrestha for serving on my committee and

for their helpful advice and input.

In addition, I am indebted to the following people who assisted me in various

aspects of this work: Jim Lucas, Dr. Brent Smith, Michael Sartori, Dr. Ramu

Ramaswamy, and Stu Kuper. The following members of the data collection team deserve

thanks and recognition for their hard work: Bill Gutelius, Bill Kalbfleisch, Warwick

Hadley, and Butch Miller.

I thank Captain Jon Bailey and Steve Matula at the National Geodetic Survey for

providing me the opportunity to attend graduate school. Finally, I thank Tom Accardi

and Fred Anderson at the Federal Aviation Administration, Aviation System Standards

for funding the data collection for this research.
















TABLE OF CONTENTS
page

A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TA BLE S .............. ............ ... ........... .. ........... .............. .. vii

LIST OF FIGURES ......... ......................... ...... ........ ............ ix

ABSTRACT .............. ..................... .......... .............. xii

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

A airport Obstruction Surveying ................................................... ....... ...............
A airborne L aser Scanning ..................................................................................... 5
B background and M otivation ........................................................... ............... 8
O organization of this W ork .......................................................... ............... 11

2 THEORY AND PREDICTION S ........................................ ......................... 13

Laser Equation .......................................................... ................... 13
Geometric Considerations in Obstruction Detection...............................................14
Radiometric Considerations in Obstruction Detection.............................................22

3 EXPERIM EN TS .................. .................. ................. .......... .. ............ 27

A airborne L aser D ata C ollection............................................ ........... ............... 27
C alib ratio n ................................................................ 3 0
Data Processing .................................................31
Field Spectrom eter D ata Collection........................................ ........................ 32

4 D A T A A N A L Y SIS .......................................................................... ....................36

Prelim inary A analysis ........................................ ... .... ........ ......... 36
O obstruction D election A nalysis............................................ ........... ............... 42
Automated Obstruction Detection Analysis............................................................44
V isual A analysis ............................................................. 50
Analysis of Return Signal Strength Calculations ....................................... .......... 52





v









5 CONCLUSIONS AND RECOMMENDATIONS............................................. 56

APPENDIX

A DERIVATION OF RANGE EQUATION ...............................................................61

B REFLECTANCE SPECTRA FOR OBSTRUCTIONS AND OTHER OBJECTS
W ITHIN THE SURVEY AREAS.................................... ............................. ....... 65

C PHOTOGRAPHS OF FIELD-SURVEYED OBSTRUCTIONS.............................71

D OUTPUT OF AUTOMATED OBSTRUCTION DETECTION ANALYSIS
SO FTW A RE ............... ........... ......................... ...........................79

C configuration 1 ........................ .. ........................ .. .... ........ ........ 79
Configuration 2 ........ .. .. .. ............................... ................ 80
C configuration 3 ........................ .. ........................ .. .... ........ ......... 82
Configuration 4 ........ .. .. .. ............................... ................ 84
C onfi guration 5 ........................ .. ......................... .... ........ ......... 86
Configuration 6.................... ................... ..... ............ 87
Configuration 7............. ........ ..................... ............ 89
Configuration 8 ............. ................................ ........ 91
Configuration 9 .................... ................... ..... ............ 93
Configuration 10..........................................................94
Configuration 11 ......................... ........... .. .. ......... ..... ..... 96
Configuration 12..........................................................98
Configuration 13 .................... ......................... ........ 100
Configuration 14 ................................... .. .. ........ .. ............101

LIST OF REFEREN CES ..................................................................... ............... 104

BIO GR A PH ICA L SK ETCH .................................... ........... ......................................108
















LIST OF TABLES


Table page

2-1 Narrow and wide beam divergences for the system used in this study, based on
three different definitions of beam diameter..................................................19

3-1 The 14 data collection configurations used in this study and the predicted vertical
and horizontal point spacing for each ........................................... ............... 27

3-2 Reflectance values at 1064 nm for field-surveyed obstructions and other objects
in the survey areas ............. ................... ................. .............. ... 34

3-3 Reflectance values at 1064 nm for three horizontal surfaces in Survey Zone 1......35

4-1 Results of testing the airborne laser data sets using an independent data set of
NGS kinematic GPS runway points .................. ............. .................37

4-2 Percent of obstructions detected in each airborne laser data set and the RMS
difference in elevation between the field-surveyed points and "matching" laser
p o in ts ................................................................................4 7

4-3 Analysis of return signal strength calculations for SPN 452 ...................................53

4-4 Analysis of return signal strength calculations for SPN 449 ...................................54

4-5 Analysis of return signal strength calculations for SPN 454. ...............................54

D -1 T ilt: 0; D iv : N ; FH :750............ ... .................................................... .... .... ....... 79

D -2 Tilt: 0; D iv: W ; FH : 750............................................... ................................ 80

D -3 T ilt: 10 ; D iv : N ; F H : 7 50 ............................................................... .....................82

D -4 T ilt: 10; D iv : W ; F H : 750 .............................................................. .....................84

D -5 T ilt: 2 0 ; D iv : N ; F H : 7 50 ............................................................... .....................86

D -6 T ilt: 20; D iv : W ; F H : 1050 ............................................................ .....................87

D -7 Tilt: 20; D iv: W ; FH : 1150........ ................. .................................. ............... 89

D -8 Tilt: 20; D iv: W ; FH : 750............................................... .............................. 91









D -9 T ilt: 20; D iv : W ; F H : 850 .............................................................. .....................93

D -10 T ilt: 20; D iv : W ; F H : 950 .............................................................. .....................94

D-11 Tilt: 30; Div: N; FH: 750............................................... ............................... 96

D -12 T ilt: 30; D iv : W ; F H : 750 .............................................................. .....................98

D-13 Tilt: 40; Div: N; FH: 750................................................ ............................ 100

D-14 Tilt: 40; Div: W ; FH: 750............................................... ............................ 101












































viii
















LIST OF FIGURES


Figure page

1-1 Federal Aviation Regulation (FAR) Part 77 obstruction identification surfaces
(O I S ) ................................................... ...................... ................ .. 2

1-2 Airborne laser scanning systems produced by the two leading commercial
m manufacturers ........................................... ................ ......... .............. .. 5

1-3 Simplified illustration of airborne laser scanning principles ...................................6

1-4 Growth in commercial use of airborne laser scanners from 1995 to 2000 ..............8

1-5 Results of comparing the three airborne laser data sets collected during the 2001
study against field-surveyed obstruction data................ .... .................10

2-1 V ertical Point Spacing .......... ...................................... ................ .. .... ...... 15

2-2 Plot of vertical point spacing versus tilt angle based on the following settings:
v = 55 m /s and = 0.019 sec. .............................. ... ................................... 16

2-3 Calculation of vertical footprint diameter, Av .............. .... .................17

2-4 Illustration of vertical point spacing (VPS) and effective vertical spacing (EVS) ..18

2-5 Profile of laser beam for the University of Florida airborne laser scanning system
and a fitted gaussian ........................................................ .. ............ 19

2-6 Effective vertical spacing versus tilt angle based on the following parameters:
H= 750 m, v = 55 m/s, r= 0.019 s, and y= 0.60 mrad. ......................................20

2-7 Definition of horizontal point spacing (HPS) ........................................................21

2-8 Schematic Illustration of the detection and measurement system............................23

2-9 Received pow er vs. tilt angle .............................................................................25

3-1 Survey project areas overlaid on a digital orthophoto and USGS quadrangles .......28

3-2 Variable-tilt sensor mount designed for this study ..................................................29









3-3 Obtaining reflectance measurements for a guywire of one of the towers in Survey
Zone 1 using the ASD LabSpec Pro portable spectrometer.................................33

4-1 Plot of average elevation bias for each tilt angle setting on the ordinate vs. tilt
angle on the abscissa ...................... ................ ................. .... .... ... 38

4-2 Average elevation bias vs. tilt angle and propagated systematic error in the
elevation of a laser point ........................ .......... ..........................40

4-3 NGS field survey of obstructions at GNV .....................................................43

4-4 Obstruction detection analysis algorithm ...... ......... ...................................... 45

4-5 Visual obstruction analysis................. ........................................ ........ 51

4-6 Photograph of SPN 460, and data points on this object based on laser returns
obtained using configurations 5, 8, and 12.................................... ............... 51

4-7 A potentially more rigorous method of performing the return signal strength
computations involving modeling the interaction of the incident laser radiation
w ith a target as a convolution ........................................................ ............... 55

B-l Reflectance spectra for SPN 445 strobe lighted tower, a guywire for the strobe
lighted tower, and SPN 446 tower. ............................................ ............... 65

B-2 Reflectance spectra for SPN 449 pole, SPN 452 antenna, and SPN 453 -
tran sm mission p ole ................................................. ................ 6 6

B-3 Reflectance spectra for SPN 454 flagpole, SPN 456 pole, and a pine tree........67

B-4 Reflectance spectra for a palm tree, a pole, and a generator............... ..............68

B-5 Reflectance spectra for grass, concrete, and asphalt........................ .............69

C-1 Photographs of SPN 414 tree, and SPN 415 tree .............................................71

C-2 Photographs of SPN 418 tree, and SPN 431 obstruction light on pole.............72

C-3 Photographs of SPN 446 tower, SPN 445 antenna on strobe lighted tower,
and SPN 448 antenna on strobe lighted tower ............................................... 73

C-4 Photographs of SPN 449 pole, SPN 453 transmission pole, and
SPN 452 antenna .................. ............................. ........ .. ........ .... 74

C-5 Photographs of SPN 454 flagpole, SPN 456 pole, SPN 455 sign, and
SPN 457 transm mission pole .............................................................................75









C-6 Photographs of SPN 457 transmission pole, and SPN 459 pole ......................76

C-7 Photograph of SPN 460 pole ........................................ ........................... 77















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

ANALYSIS OF AIRBORNE LASER-SCANNING SYSTEM CONFIGURATIONS
FOR DETECTING AIRPORT OBSTRUCTIONS

By

Christopher E. Parrish

May 2003

Chair: Grady Tuell
Major Department: Civil and Coastal Engineering

Airborne laser scanning is a relatively new remote sensing technology that is

finding use in an increasing number of surveying and mapping applications. The

strengths of airborne laser scanning, including high data density and geometric accuracy,

indicate promise in airport obstruction surveying. The primary objective in this

application is to accurately position discrete point features that penetrate imaginary 3D

survey surfaces around airfields. Early studies revealed, however, that many airport

obstructions, particularly poles, antennas and other small-diameter objects, were often not

detected using commercial airborne laser scanning systems.

The systems employed in the early studies utilized standard data collection and

system parameter configurations, which may be better suited for bare-earth terrain

mapping than detection of airport obstructions. It is hypothesized that obstruction

detection can be substantially improved through modification of certain parameters. The

objective of this research is to investigate, both analytically and empirically, the ability to









improve obstruction detection capability with an airborne laser scanning system through

modification of key parameters. The main parameters investigated include tilt angle,

laser footprint, and flying height, although the effects of flying speed, scan angle and

frequency, transmitted power, and receiver sensitivity are also discussed.

The analytical analysis involves investigation of both geometric and radiometric

considerations in obstruction detection. It is shown that tradeoffs exist between the two;

by improving the geometry for obstruction detection, the return signal from targets is

weakened, and vice versa. The optimum configuration is that which yields the best

geometry possible (i.e., the highest density of laser pulses incident on a vertical feature),

while still permitting a detectable return signal from targets of interest.

We present results of test flights over the Gainesville Regional Airport (GNV) and

portions of the runway 10 approach using fourteen different data collection

configurations. The airborne laser data are compared against field surveyed obstruction

data obtained by an NGS field crew to assess each of the fourteen configurations.

Analysis of the data reveals that significant improvement in obstruction detection

capability can be achieved with suitable configurations. It is shown that 100% detection

(based on predefined criteria) with submeter vertical RMSE is attainable. We conclude

with a discussion of potential future enhancements in obstruction detection capability.














CHAPTER 1
INTRODUCTION

Airport Obstruction Surveying

To navigate safely into airports in reduced-visibility weather conditions, pilots

follow published instrument approach procedures that specify flight courses, turns,

minimum altitudes, and so forth. Similarly, departure procedures are followed in

executing safe departures from airports. Because these procedures are critical to flight

safety, it is essential that they be based on accurate and up-to-date source data. A

prerequisite step in designing an approach or departure procedure is to conduct an airport

obstruction survey.

The main objective in obstruction surveying is to obtain accurate survey

coordinates for vertical objects (both natural and manmade) in specified zones on and

around the airfield and in the approach paths. Objects that lie within these zones and that

penetrate (i.e., are of greater height than) mathematically-defined 3D surfaces enveloping

the airport (Figure 1-1) are termed obstructions or "obstacles." Examples of obstructions

include, but are by no means limited to, trees, buildings, towers, poles, antennae, and

terrain. In addition to supporting procedure development, obstruction survey data are

used by airport and government authorities in planning, meeting or verifying compliance

with airport operating certificate requirements, determining maximum weights of aircraft

for takeoff, and conducting accident investigations (U.S. Department of Transportation,

1996).




































2 A Runway Centerlines

Figure 1-1. Federal Aviation Regulation (FAR) Part 77 obstruction identification surfaces
(OIS) (courtesy ofFAA, ATA-100). The shape and dimensions of the
surfaces for a particular obstruction survey will vary depending on the
regulating agency, type of survey, runway end positions and designations, and
other factors.

Within the United States and its territories, the Federal Aviation Administration

(FAA), Aviation System Standards (AVN) is responsible for developing and publishing

approach procedures for all civil airports. Under an interagency agreement, airport

obstruction surveys supporting the FAA are conducted by the National Geodetic Survey

(NGS). These surveys are performed in accordance with FAA No. 405: Standards for

Aeronautical Surveys and Related Products (U.S. Department of Transportation, 1996).

Similarly, the National Imagery and Mapping Agency (NIMA) is tasked with obtaining

survey data and developing procedures for approximately 10,000 airports throughout the

world in support of U.S. military operations (Harris and Johnson, 2001). Obstruction


SConical Surface
SPrecision Instrument Approach
* Visual or Non Precision Approach
S1/2 C (Slope E)









surveys performed by or for NIMA must meet the specifications contained in the Airfield

Initiative Document (National Imagery and Mapping Agency, 2001).

In addition to meeting the standards published by government agencies at the

national level, airport obstruction surveys must also adhere to applicable international

specifications. The International Civil Aviation Organization (ICAO) establishes and

publishes international standards to be followed by each of its 188 member nations (or

"contracting states"), including the United States. Specifications pertaining to airport

obstruction surveying and charting are contained in two annexes to the Convention on

International Civil Aviation (also known as the Chicago Convention of 1944): Annex 14 -

Aerodromes, Volume I Aerodrome Design and Operations (International Civil Aviation

Organization, 1999) and Annex 4 Aeronautical Charts (International Civil Aviation

Organization, 1995).

Currently, airport obstruction surveys are most often completed through a

combination of photogrammetry and field surveying. Photogrammetry is a mature

remote sensing technology, and the procedures and achievable accuracy are well

documented. Field surveys offer the highest accuracy and reliability because experienced

field crews visually inspect the survey areas to identify and locate small obstructions

(Tuell, 1985).

Based on records kept at NGS, the combination of photogrammetry and field

surveying has been utilized successfully in airport obstruction surveying for over half a

century. During this time, however, the mapping procedures used in NGS's Aeronautical

Survey Program have been continually updated as new technologies have become

available. We can recognize certain major transitions (Tuell, 1987):









* In the early 1960s, several analog photogrammetric plotters (Wild B-8s) were
purchased to support the program. During this period, field-surveyed obstructions
and control points were plotted by hand, and measurements from the stereoplotters
were used to position new obstructions and compile planimetry.

* Throughout the 1970s, innovations were focused on the integration of computers
and computer-driven precision plotters into the program.

* In the mid-1980s, the transition from analog to analytical photogrammetry and the
initial design and implementation of a relational database to function as the data
warehouse were accomplished. At the same time, the field survey teams
implemented total station technology and developed the capability to transfer
obstruction data through the production system electronically. In addition, the field
teams began to use GPS for establishing control points on airports.

* In the 1990s, significant improvements were made in the ability to log field survey
data and to compute positions while on-site. Hand-held lasers were introduced for
quick measurement of distances to obstructions.

* In 2000, the analytical photogrammetric stereoplotters were replaced with softcopy
(digital) photogrammetric workstations. In addition, CAD and GIS systems were
introduced to facilitate the storage, editing, analysis, and distribution of digital
obstruction data, including digital charts.

While field techniques and photogrammetry will continue to set the standard for

high-accuracy obstruction surveying, several factors motivate continued investigation

into new technologies for obstruction surveying. First, the demand for survey data

already exceeds production capability, and this demand is certain to increase as the FAA

implements GPS-based navigation and landing systems, such as the Wide Area

Augmentation System (WAAS) (Anderson et al., 2002). Second, new FAA and National

Aeronautics and Space Administration (NASA) initiatives, such as Synthetic Vision

Systems (SVS) (Prinzel et al., 2002), are further increasing the demand for high-accuracy

digital terrain and obstacle databases. Third, because different types of obstruction

surveys have different requirements for accuracy, cost and completion time, agencies

would benefit from greater flexibility in tailoring the survey methods to the requirements.









Of particular interest are remote sensing technologies that could potentially fulfill the

need for rapid, inexpensive, medium-to-high-accuracy surveys.

Airborne Laser Scanning

Airborne laser scanning (also referred to as lidar) is an active remote sensing

technology that is quickly gaining recognition as an efficient and cost-effective approach

to a variety of surveying and mapping applications. The primary components of an

airborne laser-scanning system include 1) the laser scanner, 2) an inertial measurement

unit (IMU), and 3) an airborne GPS receiver and antenna. Figure 1-2 shows systems

produced by the two leading commercial manufacturers.


Figure 1-2. Airborne laser-scanning systems produced by the two leading commercial
manufacturers. Top: Optech, Inc. ALTM 2050 (photo courtesy of Opetch,
Inc.). Bottom: Leica Geosystems, Inc. ALS40 (photo courtesy of Leica
Geosystems, Inc.).









Although airborne laser scanners are complex instruments requiring integration of

numerous subsystems, the basic concepts of the technology are relatively straightforward,

as illustrated by the cartoon in Figure 1-3. Ranges are accurately computed from the

round trip travel time of laser pulses that are reflected by either terrain or elevated

features on the Earth's surface and return to the sensor. By combining range, sensor

orientation, and scanner angle data, 3D vectors from the airborne sensor to points on the

reflective surface illuminated by the laser can be computed. These 3D vectors are then

utilized in conjunction with post-processed airborne GPS data and offset vectors

describing the relative positions of the various system components to compute accurate

XYZ positions of terrain and features in the mapping frame.


Figure 1-3. Simplified illustration of airborne laser-scanning principles.


GPS Antenna


Range


GPS Reference
Nk Station









While at least one operational system employs a continuous-wave (CW) laser

(Wehr and Lohr, 1999), most airborne laser scanners utilize pulsed lasers. Typically, the

lasers are Q-switched to produce short (- 10 ns) pulses and high peak transmitted power.

Current state-of-the-art systems have pulse repetition frequencies (PRFs) of 30 to 50 kHz,

and systems with even higher PRFs are in development. Various types of scanners are

used to produce a swath. For example, the system employed in this research uses a

single-axis, cross track scanning mirror that produces a saw-tooth pattern on the ground,

as depicted in Figure 1-3.

The past decade has seen significant advancement in airborne laser-scanning

technology. Although laser altimetry dates back to the early 1970s (Blair et al., 1999),

commercial airborne laser scanners were not readily available until the mid-1990s. As

illustrated in Figure 1-4 (adapted from Maune, 2001), the growth in commercial adoption

over the five-year period from 1995 to 2000 was nearly 2,000%. The increasing demand

for new systems has naturally precipitated technological advances, such as higher pulse

repetition and scan frequencies, improved reliability, more robust data collection and

processing software, and so forth.

Two of the often-cited strengths of airborne laser scanning are the high density of

data points and the achievable geometric accuracy. As noted above, PRFs of 50 kHz or

greater are currently attainable. Assuming continuous operation of the laser and a 97%

probability of a good return from each pulse, one hour of data collection with a 50 kHz

system will generate nearly 175 million data points. Several researchers have

demonstrated vertical accuracy of 15 cm (1 o) or better on terrain (see, e.g., Shrestha et

al., 1999; Vaughn et al., 1996b). Horizontal accuracy of airborne laser data is harder to










quantify and has been less rigorously investigated. In general, it is expected that

horizontal accuracy will be worse than vertical accuracy (Baltsavias, 1999b; Maas, 2002),

but at least one study has indicated horizontal accuracy of better than half a meter (Tuell,

2002).


Growth in Commercial Systems Use

70-
60
50
S40
30
O 20
10 -
0
1995 1996 1997 1998 1999 2000
Year

Figure 1-4. Growth in commercial use of airborne laser scanners from 1995 to 2000
(adapted from Manue, 2001).

Background and Motivation

Over the past few years, the high point density and geometric accuracy achievable

with airborne laser scanning have brought this technology to the attention of several

government agencies and private firms involved in obstruction surveying. The

technology seems of obvious benefit for the mapping of obstructing areas and buildings,

but because of the unique challenges involved in surveying discrete point features, as

well as the critical nature of the data in flight safety, its performance for obstruction

detection must be carefully analyzed. One of the first studies of this type was conducted

jointly by the University of Florida (UF), NGS, FAA, and Optech, Inc. at Gainesville

Regional Airport (GNV) in 2001.









During the 2001 study, three data sets were collected using two different Optech

Airborne Laser Terrain Mapper (ALTM) systems. One of these systems had a PRF of 10

kHz and was flown in a Cessna Skymaster owned and operated by UF. The flying height

for the data collected with the UF system was 600 m. The second system had a PRF of

33 kHz and was flown in a NOAA Cessna Citation at flying heights of 700 and 1200 m.

In the NOAA Citation, a 70 tilt angle was used, meaning the sensor was tilted 70 forward

of nadir. The UF system did not use a tilted sensor. Both systems had a constant beam

divergence of approximately 0.18 mrad (full angle), based on the full width at half

maximum (FWHM) points of the beam (or, equivalently, 0.22 mrad, based on the 1/e

points of the beam).

Concurrent with the airborne laser data collection, an NGS survey crew conducted

an obstruction survey using GPS and conventional field techniques to provide a high-

accuracy reference data set. To assess how well obstructions were detected using the

airborne laser-scanning systems, researchers at UF and NGS compared the three airborne

laser data sets against the field-surveyed obstruction data. Figure 1-5 (adapted from

Tuell, 2002) shows the percent of obstructions detected in each of the three data sets. In

performing this analysis, UF researchers measured the 3D distance from each field-

surveyed obstruction to the closest point in the laser point cloud. A 3D distance of less

than 20 feet was defined as the detection criterion in generating Figure 1-5. As illustrated

here, at best, 94% of the obstructions were detected based on this criterion.

A more interesting observation, however, is that the detection percentage drops off

significantly with flying height. In fact, this study revealed that certain small targets

(poles, antennae, etc.) were often not detected at all. Clearly, the ability to hit and










measure a small target is a function of the survey geometry. What is not clear is how

much of the loss at higher flying height results from the geometric effect of increasing the

pulse spacing and how much of it originates will a fall-off in received signal strength due

to increased laser range.


Percent of Obstructions Detected to Within 20 feet
of the Field-Surveyed Point (2001 Study)

100

S95

90







UF NOAA 700 m NOAA 1200 m
Airborne Laser Data Set

Figure 1-5. Results of comparing the three airborne laser data sets collected during the
2001 study against field-surveyed obstruction data (adapted from Tuell,
2002).

The 2001 study was not designed to systematically investigate the effects of

various data collection and system parameters on the results. The experiment utilized

commercial systems with configurations similar to those employed in topographic

mapping projects. Because commercial airborne laser-scanning systems are most

frequently utilized for production of bare-earth data sets, it is likely that, during the rapid

developments of the past decade, systems have been either intentionally or

unintentionally optimized for this type of work. However, obstruction surveying is a

fundamentally different and more difficult task; the parameters that work well for one

application might not be well-suited for the other. It was hypothesized, therefore, that the









capability to detect obstructions with an airborne laser-scanning system could be

enhanced by modifying certain data collection and system parameters. This hypothesis,

combined with the valuable information gained from the earlier study, provided the

foundation and motivation for the research presented here.

Organization of this Work

The goal of this research is to investigate the effect of laser data collection

geometry on the detection of small targets and to understand the tradeoffs between

geometric and radiometric issues. In contrast to the 2001 study, we have designed an

experiment using data collected with a single instrument. This allows us to

systematically investigate the effects of certain parameters without having to account for

inter-instrument variability. Specifically, the effects of sensor tilt (or "forward-look")

angle, laser footprint area, and flying height on obstruction detection are rigorously

examined through both analytical and empirical methods. Other parameters, such as

flying speed over ground, scan angle and frequency, pulse repetition frequency, and

transmitted power, are also examined mathematically, though not experimentally.

In Chapter 2, the underlying analytical considerations are addressed. The problem

of identifying the optimum configuration of the laser system is shown to be nontrivial,

due to the number of variables and tradeoffs involved. The experiments are presented in

Chapter 3. These include an airborne laser data collection using fourteen different

configurations, as well as collection of reflectance spectra using a portable field

spectrometer. The results of the analysis are presented in Chapter 4. It is shown that

significant improvements in obstruction detection capability can, in fact, be achieved

through careful selection of the data collection parameters. We conclude with a







12


discussion of the potential for further improvements and suggestions for continued

research.














CHAPTER 2
THEORY AND PREDICTIONS

Laser Equation

A quantitative description of airborne laser scanning starts with the relationship

between measured quantities and the desired X,Y,Z coordinates of terrain or features in

the mapping frame. The equation that gives the location of the laser footprint in the

mapping frame, alternately referred to as the "georeferencing equation," the "laser

geolocation equation" or simply the "laser equation," is given in various forms in, for

example, Lindenberger (1989), Vaughn et al. (1996a), Filin (2001), and Schenk (2001).

Some form of the laser equation is used in all software packages that process airborne

laser-scanning system measurements to output surface point coordinates.

In this work, we will examine the effect of geometric parameters on the detection

of small targets. In Equation (2-1), we show a simplified form of the laser equation

which explicitly addresses the possibility of a tilted sensor:


Xf XGPS x Jx + Rcosssint
Y, = YGPS +M y,-y+Rsins (2-1)
Zi t ZGPS Zl ; Rcosscost

Here, [Xf Yf]T is the position of a laser footprint in the mapping frame (e.g., State Plane

coordinates); [XGPS YGPS ZGPS]T is the position of the aircraft GPS antenna in the same

mapping frame; s is the instantaneous scan angle; t is the tilt angle; R is the range; xi, yi,

and zI are the coordinates of the laser beam origination point in the body frame; and

,x, 5y, and z are the sensor-to-antenna offset vector ("lever arm") components. For the









body frame, we have used a photogrammetric, rather than aeronautical, convention:

positive x is in the direction of flight, positive y is towards the left wing of the aircraft,

and positive z is towards zenith. The rotation matrix M is given by

cospsinh -coshcosr- sinhsinpsinr coshsinr- sinhsinpcosr
M= cospcosh sinhcosr- coshsinpsinr sinhsinr- coshsinpcosr (2-2)
sinp cospsinr cospcosr

where, r (roll), p (pitch), and h (heading) are the attitude angles reported by the IMU

(Lucas, personal correspondence, 2002).

It should be noted that Equation (2-1) assumes that the tilt angle, t, is measured

independently from the attitude angles, r,p, and h. If the tilt angle is incorporated into

the attitude angles, then Equation (2-1) still holds, provided t is set equal to zero.

Another assumption made in Equation (2-1) is that the offset distance from the IMU to

the laser is negligible. This assumption is likely to introduce a small systematic error in

the computed positions of laser points. However, in this work, the laser equation is used

only in examining how errors are propagated (see Chapter 4) and not in computing

positions of points.

Geometric Considerations in Obstruction Detection

One metric by which the strength of the geometry for detecting vertical

obstructions can be measured is the vertical point spacing (Optech, unpublished data,

2002). The vertical point spacing (VPS) is defined as the vertical distance between laser

points from consecutive scan lines on the face of a vertical surface. The smaller the VPS,

the better the geometry for detecting vertical obstructions. The VPS will vary depending

on the point in the scan cycle at which the beam "catches" the obstruction. Specifically,

the VPS will be greatest if the obstruction lies on the outer edge of the scan and smallest









if the obstruction lies directly on the flight path. Assuming the most conservative case

(i.e., the obstruction lies on the outermost edge of the scan), the VPS is given by

VPS= vr[tan(900 t)]
= vrcott (2-3)

where v is the flying speed over ground, ris the period of the scanner, and t is the tilt

angle (see Figure 2-1).


--VT >

900- t

VPSVPS














Figure 2-1. Vertical Point Spacing.

Figure 2-2 shows a plot of VPS versus tilt angle using the following data collection

parameters: v = 55 m/s (107 knots), and r= 0.019 sec (corresponding to a frequency of

53 Hz). These are the parameters used in the data collection for this study, as detailed in

Chapter 3. It should be noted that increasing the tilt angle is not the only possible method

of reducing the VPS. For example, the VPS could be decreased by reducing the aircraft

speed, increasing the frequency of the scanner (i.e., reducing the period, r), and/or by

flying repeat passes. However, these methods will not be considered here, since the









flying speed and scanner frequency listed above are based on the actual data collection

parameters and since repeat passes increase costs.



10 -


8


S6


4


2



0 10 20 30 40 50 60 70 80 90
Tilt Angle (deg)
Figure 2-2. Plot of vertical point spacing versus tilt angle based on the following settings:
v = 55 m/s and r= 0.019 sec.

In determining the desired VPS for obstruction detection, it is beneficial to

introduce another quantity, the "vertical footprint." The vertical footprint is defined here

as the area illuminated by the laser beam on the face of a vertical object, based on the full

width at half maximum (FWHM) points of the beam. With reference to Figure 2-3, the

vertical footprint diameter, Av, is given approximately by

Ry Ry
A, R1 R
Scos(90 t) sint (2-4)

In Equation (2-4), R is the range to the target and yis the beam divergence. With

simplifying assumptions of flat terrain, obstruction height much smaller than flying

height, and an instantaneous scan angle of zero, the range can be expressed as









H
cost

where H is the flying height. Substituting Equation (2-5) into Equation (2-4) gives:

Hy
Ssintcost
vsmtcost


(2-5)




(2-6)


We define a new term, the effective vertical spacing (EVS), as the VPS minus the

vertical footprint diameter, Av (see Figure 2-4). The EVS provides a metric for the extent

to which the vertical face of an obstruction is illuminated by laser radiation (i.e., how









well the obstruction is "painted" by the laser). An EVS of zero is interpreted as

completely painting the face of the obstruction in the vertical dimension.


Figure 2-4. Illustration of vertical point spacing (VPS) and effective vertical spacing
(EVS). Four footprints on the face of a box-shaped obstruction are shown.
EVS and A, are both based on the FWHM points of the beam.

The airborne laser-scanning system used in this study has two beam divergence

settings: wide and narrow. A flip-in lens provides the mechanism for switching from one

setting to the other. Table 2-1 shows the wide and narrow beam divergences for this

system based on three different beam diameter definitions: FWHM, l/e, and 1/e2.

Throughout this study, the FWHM definition will be used. However, because the beam

(TEMOO mode) is very nearly gaussian (see Figure 2-5), it is possible to convert from one

beam diameter definition to either of the others through Equations (2-7) and (2-8).









Table 2-1. Narrow and wide beam divergences for the system used in this study, based on
three different definitions of beam diameter.
Setting Divergence based on Divergence based on Divergence based on
1/e2 pts of beam 1/e pts of beam FWHM pts of beam
(mrad) (mrad) (mrad)
Wide 1.02 0.72 0.60
Narrow 0.34 0.24 0.20


1/e beam diameter
FWHM beam diameter

1/e2 beam diameter
FWHM beam diameter


1.2 (Gaussian beam)


1.7 (Gaussian beam)


Figure 2-5. Profile of laser beam for the University of Florida airborne laser-scanning
system (green lines) and a fitted gaussian (orange lines). Although this is not
the system used in this study, the profile is likely similar. (Image courtesy of
Optech, Inc.)

Figure 2-6 shows a plot of EVS vs. tilt angle using the wide beam divergence and

the flying speed and scan period listed above. As illustrated in the graph, a tilt angle of


(2-7)


(2-8)










15.60 produces an EVS of 2.0 m, a tilt angle of 26.00 produces an EVS of 1.0 m, and a tilt

angle of 49.00 produces an EVS of zero (i.e., 100% coverage in the vertical dimension).

6


4 -


2



01..
L.



-4

15.60 26.00 49.00
-6 I I ,

0 10 20 30 40 50 60 70 80 90
Tilt Angle (deg)
Figure 2-6. Effective vertical spacing versus tilt angle based on the following parameters:
H= 750 m, v = 55 m/s, r= 0.019 s, and y= 0.60 mrad.

In examining the ability to detect small obstructions, it is important to consider the

horizontal point spacing (HPS) in addition to the vertical point spacing. The HPS is

defined as the distance between laser points incident on the face of a vertical surface in a

direction perpendicular to the vertical (see Figure 2-7). Due to the motion of the

scanning mirror, the HPS is not uniform; points are more tightly bunched near the outer

edges of the scan. In the following discussion, therefore, HPS will be assumed to refer to

the average spacing. Assuming, further, a flat vertical surface whose normal is parallel to

the direction of flight, the HPS is given approximately by









Swath Width
HPS
Number of Points per Scan Line
2Rtan(2) (2-9)
PRF(/2)

where S is the full scan angle (note: lower case s refers to the instantaneous scan angle),

and, as before, R refers to range and rto the period of the scanner. As was done for the

vertical point spacing, it is possible to define an effective horizontal spacing (EHS) by

taking into account the footprint diameter, Ah. Assuming, again, a flat vertical surface

whose normal is parallel to the direction of flight and an instantaneous scan angle of zero,

Ah is given approximately by

Ah Ry (2-10)

HPS













Figure 2-7. Definition of horizontal point spacing (HPS).

Using Equations (2-9) and (2-10) and data collection parameter values that are

valid for the current study (H = 750 m, S = 300, t = 200, PRF = 50 kHz, r= 0.019 s, and

y= 0.6 mrad), the HPS and EHS are approximately 0.90 m and 0.42 m, respectively. It

can be seen, therefore, that the point spacing in the horizontal direction is typically better

(smaller) than that in the vertical direction for the system used in this study. The HPS









and EHS can be reduced further with relatively minor modifications to the data collection

parameters. For example, cutting the scan angle in half while keeping all other

parameters the same produces a negative EHS, meaning that the pulses will overlap in the

horizontal dimension, based on the FWHM points of the beam. Although horizontal

spacing cannot be neglected, vertical spacing is currently a greater concern than

horizontal spacing for obstruction detection.

Radiometric Considerations in Obstruction Detection

Figure 2-8 illustrates schematically how radiation emitted from the laser and

reflected from a surface below is detected and used to determine range. As shown in the

figure, reflected radiation that is incident on the receiver optics is passed through a

narrow bandpass filter to remove background radiation (Optech, 1998). Next, a square-

law detector converts the optical signal into a current that is proportional to the incident

optical power (or to the square of the electric field).

The output from the detector is next fed into an amplifier and then into a constant

fraction discriminator (CFD). The purpose of a CFD is to provide accurate triggering

that is nearly independent of the amplitude of the input pulse (see, e.g., Binkley and

Casey, 1988). Digital pulses output by the CFD are then fed into the actual timing

mechanism, known as a Time Interval Meter (TIM) (Optech, 1998). Essentially the same

detection and measurement mechanisms are used for both the transmitted and received

pulses, with the primary difference being that scattered laser light within the system is

captured and used to detect the transmitted pulse (Optech, 1998). The temporal

difference between the corresponding points on the transmitted and received pulses,

combined with the value for the group velocity of the laser light in the atmosphere,

allows ranges to be determined.












Digital
Pulse


Scattered Portion
of Transmitted
Pulse


Radiation


Incident
Laser
Radiation


.I Y Surface
Figure 2-8. Schematic Illustration of the detection and measurement system.

For the purposes of this study, it is important to note that if the signal is below the

detection threshold, the target will not be detected. The first step in estimating the return

signal strength involves calculating, Ea, the irradiance (W.m 2) incident on the receiver.









Derivations of expressions for Ea and the received power, Pr, are contained in Appendix

A. For convenience, equations (A-7) and (A-8) are restated below:

Po-
Ea 2 R4 T (2-11)


where PT is the transmitted power, TATM is the atmospheric transmittance, and cis the

effective target cross section given by Equation (A-6). The power received can then be

computed from:

P,= AETss (2-12)

where Ar is the receiver area and TsYS is the system transmittance, which is limited

primarily by the transmittance of the bandpass filter. Next, the photocurrent generated by

the detector can be computed from:

Iph = 1P, (2-13)

where 91 is the responsivity (A-W 1) of the photodetector.

The exact value of TsYS for the system used in this research was not disclosed by

the manufacturer, but a "typical" value of 0.6 was used in the calculations. The value of

91 was also not disclosed by the manufacturer, so it was not possible to directly calculate

Iph using Equation (2-13). However, the responsivity of the photodetector can be

assumed to be constant over the course of a project. In this study, return signal strength

calculations were performed using Equations (2-11) and (2-12).

Figure 2-9 shows a plot of received power vs. tilt angle. The target in this example

is an antenna that was surveyed by the field crew in 2001. The methods used to obtain

the reflectance of this object are described in Chapter 3. The value of the peak

transmitted power for the system used in this study was obtained from the manufacturer.

Other parameters used in the calculations were based on an actual data collection









configuration used in the study (see Chapter 3). It is interesting to note in Figure 2-9 that

the received power drops off very rapidly with increasing tilt angle. This is because R

increases with t, and Pr decreases as R-4












0.5








0 10 20 30 40 50 60 70 80 90
Tilt Angle (deg)
Figure 2-9. Received power vs. tilt angle based on the following parameters: PT = 11.3
kW (peak, not average); = 0.6 mrad; TATM 07; H 750 m; p 0.318







(reflectance of SPN 452); d= 0.305 m (diameter of SPN 452); Ar = 1.79x10-3
a \
a,










; and T 0060.
From Figures 2-6 and 2-9, we note that a tradeoff exists between the geometric and





radiometric considerations. Specifically, it is desirable to increase the tilt angle as much

as possible to reduce the effective vertical spacing (improving the geometry), but
increasing the tilt angle also has the undesirable effect of reducing the received power








from a target. For example, for a given system we can achieve a zero EVS, but the
Figure 2-9. Received power would only be aboutilt angle thirbased on the follower received with a nadir-viewing11.3
kW (peak, not average); y= 0.6 mrad; TATM = 0.87; H= 750 m; p= 0.318
(reflectance of SPN 452); d= 0.305 m (diameter of SPN 452); Ar = 1.79x10-3
m2; and Tsys =0.60.

From Figures 2-6 and 2-9, we note that a tradeoff exists between the geometric and

radiometric considerations. Specifically, it is desirable to increase the tilt angle as much

as possible to reduce the effective vertical spacing (improving the geometry), but

increasing the tilt angle also has the undesirable effect of reducing the received power

from a target. For example, for a given system we can achieve a zero EVS, but the

received power would only be about one third of the power received with a nadir-viewing

instrument. Likewise, increasing the beam divergence improves the geometry (larger

footprint) but reduces the received signal strength, as seen in Equation (2-11). In






26


configuring an airborne laser-scanning system for obstruction detection, the goal,

therefore, is to choose parameters that optimize the geometry while still enabling a

detectable return signal from targets of interest.














CHAPTER 3
EXPERIMENTS

Airborne Laser Data Collection

Based on the geometric and radiometric considerations discussed in Chapter 2, the

experiment was designed to test fourteen different data collection configurations. These

fourteen configurations consisted of different combinations of tilt angle, beam

divergence, and flying height, as listed in Table 3-1. The survey project areas for the

study consisted of three zones covering the airfield and portions of the runway 10

approach at Gainesville Regional Airport (GNV), as shown in Figure 3-1.

Table 3-1. The 14 data collection configurations used in this study and the predicted
vertical and horizontal point spacing for each. Note: by definition, VPS and
EVS apply only to configurations that employ a tilted sensor.
Config. Tilt Divergence Flying Predicted Predicted Predicted Predicted
# (deg) (Wide/ Height VPS (m) EVS (m) HPS (m) EHS (m)
Narrow) (m)
1 0 N 750 N/A N/A 0.8 0.7
2 0 W 750 N/A N/A 0.8 0.4
3 10 N 750 5.9 5.0 0.9 0.7
4 10 W 750 5.9 3.3 0.9 0.4
5 20 N 750 2.9 2.4 0.9 0.7
6 20 W 1050 2.9 0.9 1.3 0.6
7 20 W 1150 2.9 0.7 1.4 0.6
8 20 W 750 2.9 1.5 0.9 0.4
9 20 W 850 2.9 1.3 1.0 0.5
10 20 W 950 2.9 1.1 1.1 0.5
11 30 N 750 1.8 1.5 1.0 0.8
12 30 W 750 1.8 0.8 1.0 0.5
13 40 N 750 1.2 0.9 1.1 0.9
14 40 W 750 1.2 0.3 1.1 0.5






































Figure 3-1. Survey project areas overlaid on a digital orthophoto and USGS quadrangles.
Zone 3 encompasses the airfield. The GPS reference station, NATASHA, is
located on the UF campus.

Although airborne laser data were collected over a significantly larger area during

the 2001 study, the three zones shown in Figure 3-1 contain 90% of the obstructions

surveyed by the field crew. Because of the large amount of data required for this study, it

was not possible to collect data over the entire runway 10 approach. By limiting the data

collection to these three zones, significant savings in cost, data storage, and processing

time were achieved with minimal impact on the obstruction analysis.

Acquisition of the airborne laser data took place from June 10 through June 15,

2002. The system used in the data collection was an Optech ALTM 2050 mounted in a

Cessna Skymaster. An Ashtech ZXII receiver and 700936 D choke ring antenna located
at NATASHA UF, an NGS Cooperative Base Network (CBN) control station on the UF
Figure 3-1. Survey project areas overlaid on a digital orthophoto and USGS quadrangles.
Zone 3 encompasses the airfield. The GPS reference station, NATASHA, is
located on the UF campus.

Although airborne laser data were collected over a significantly larger area during

the 2001 study, the three zones shown in Figure 3-1 contain 90% of the obstructions

surveyed by the field crew. Because of the large amount of data required for this study, it

was not possible to collect data over the entire runway 10 approach. By limiting the data

collection to these three zones, significant savings in cost, data storage, and processing

time were achieved with minimal impact on the obstruction analysis.

Acquisition of the airborne laser data took place from June 10 through June 15,

2002. The system used in the data collection was an Optech ALTM 2050 mounted in a

Cessna Skymaster. An Ashtech ZXII receiver and 700936 D choke ring antenna located

at NATASHA UF, an NGS Cooperative Base Network (CBN) control station on the UF


~ ~iS~ ''' .9'
~.II ~- '' ~-
,~ii'=..2 L, f- -.... ~r~ ,---,~
,, '. ~r,
: ~;.~:...... ~~ .~~
'''''---L;"' ''
., ':` :
'' : I
1. ., .- 5:









campus, served as the GPS reference station. The average flying speed over ground for

all fourteen configurations was approximately 55 m/s (107 knots), and all configurations

used a scan frequency of 53 Hz, a scan angle of + 150, and a PRF of 50 kHz. To enable

variable tilt angles of 0 to 400, a custom sensor mount (Figure 3-2) was designed and

built by Optech. During the five-day acquisition period, over 378 million laser data

points were collected in the three zones shown in Figure 3-1.















Figure 3-2. Variable-tilt sensor mount designed for this study. The left and right images
show the 200 and 300 tilt angle positions, respectively.

The use of variable tilt angles resulted in a few added complications over typical

airborne laser data collection. Most importantly, the sensor-to-antenna offset vector

("lever arm") components required separate measurements for each tilt angle setting.

Clearly, in determining the offset vector components, it could not be assumed that the z-

axis of the sensor was aligned with the local vertical. To acquire these offsets, a least

squares adjustment of a 3D trilateration was used. Distances were measured from the

bottom of the GPS antenna (specifically, the bottom of the TNC female connector) to

each of the four corners of the top of the ALTM sensor box and also to the screw hole in

the handle. These measurements were repeated for each tilt angle setting. The

coordinates of the box covers and handle in the sensor frame are known from









engineering diagrams provided by Optech. The additional offsets from the screw hole in

the handle to the center of the scan mirror and from the TNC connector on the GPS

antenna to the antenna phase center are also known. Since there are three unknown

sensor-to-antenna offset vector components, more than three measured distances permit a

least squares solution. Custom software was written to perform the least squares

adjustment of the data and output the final offset vector components and their standard

deviations, which averaged approximately 1.5 cm.

Calibration

Careful calibration of an airborne laser-scanning system is essential to obtaining

high positional accuracy. The system used in this study was calibrated by the

manufacturer prior to the data collection in Gainesville. In-flight calibration was

performed on June 15 to refine the calibration parameters. Calibration flights were

performed across (i.e., perpendicular to) runway 6-24 at GNV and also over a large, flat-

roofed building on the UF campus that had been accurately surveyed using GPS. For the

flights over the building, both profile-mode (zero scan angle) and scan mode (40 scan

angle) were used, while the runway flights utilized a scan angle of 200. The data

collected during these flights were used to determine corrections to the pitch, roll, and

scale calibration parameters.

The calibration data were analyzed in Surfer (Golden Software, Inc.) Version 8.00.

The pitch correction value was obtained using the profile-mode data captured over the

field-surveyed building. By comparing the surveyed edges of the building with the

locations at which the corresponding changes in elevation in the laser data occurred, the

amount by which the pitch was over/underreported could be determined. The roll









correction was obtained in a similar manner using the 40 scan angle data and finding

locations at which the edge of the building was captured at the outer edge of the scan.

The mirror angle scale factor was determined by examining elevation profiles from the

200 scan angle data over the runway. The slope along a portion of the runway is

presumed to be constant, so an upward curve ("smile") or downward curve ("frown") in

the elevation profiles indicates a necessary correction to the scale.

Data Processing

The airborne laser data were processed using REALM (TopScan, GmbH, and

Optech, Inc.) Version 3.0.3d. The processing was completed following standard

procedures used by researchers at UF (see Shrestha et al., 1999; Carter et al., 2001) with

three notable exceptions. First, due to the very short baselines (approximately 6 to 11

km), it was not deemed necessary to utilize the Kinematic and Rapid Static (KARS)

software (Mader, 1992) in processing the GPS trajectories; the GPS processing was done

directly in REALM. Second, no filtering or gridding of the data was performed. And

third, a few default processing parameters were changed for the reasons listed below:

* The default setting for the REALM V. 3.0.3d parameter "Max. FL Diff' serves to
eliminate "suspicious" data points by excluding any laser shot for which the
distance between the first and last returns is greater than 100 m. Since many
obstructions have Above Ground Level (AGL) heights of over 100 m, this default
setting cannot be used in obstruction detection.

* The default for the REALM V. 3.0.3d "Min. Intensity" setting is 1. The intensity
value is a unitless digital number that is proportional to the strength of the return
signal, and, hence, to the effective target cross section (see Appendix A). The
"Min. Intensity" parameter is used to exclude laser shots whenever the intensity
value is below the defined threshold. Optech recommends against setting this
parameter below the default value of 1, as a value of zero indicates a potentially
bad range measurement (Tickle, personal correspondence, 2003). In this work,
however, the parameter was set to zero to avoid excluding obstructions with small
effective target cross sections from the output.









* Yet another default in REALM V. 3.0.3d is to output only last return laser data.
While this default setting is suitable for bare-earth terrain mapping, for obstruction
detection, first returns are more important. Therefore, the setting was changed to
output the first return data.

As noted in Chapter 1, an underlying hypothesis in this research is that standard

data collection configurations are not well-suited for obstruction detection. Interestingly,

this statement was found to be equally applicable to the processing software; as described

above, several of the default parameter settings are unsuitable for obstruction detection.

All of the settings can be changed with little difficulty, but, unfortunately, the software

does not have the capability to save the user's settings from one session to the next. In

short, although the "off-the-shelf" processing software was utilized successfully in this

project, it was clearly not designed for this application.

The output laser data points were projected in UTM (WGS84) Zone 17 North.

Elevations were referenced to the WGS84 ellipsoid. In order to compare the airborne

laser data sets against the NGS kinematic GPS runway data and field-surveyed

obstruction data (see Chapter 4), the ellipsoid elevations were converted to orthometric

heights by the analysis software using GEOID99.

Field Spectrometer Data Collection

In addition to the airborne laser data collection, field work was also required to

obtain reflectance measurements for obstructions and other objects in the survey areas.

These reflectance measurements are needed in calculating the return signal strength from

obstructions (see Equations (A-6) through (A-8)). Comparisons of the calculated return

signal strength values with the empirical results allow the minimum detectable return

signal to be estimated (Chapter 4).









The spectrometer used for obtaining reflectance measurements in this study was an

Analytical Spectral Devices, Inc. (ASD) LabSpec Pro. This instrument has a spectral

range of 350 to 2500 nm. The spectral resolution of the LabSpec Pro is 3 nm at 700 nm

and 10 nm at 1400 nm and 2100 nm. The sampling interval is 1.4 nm for the spectral

region 350-1000 nm and 2 nm for the spectral region 1000-2500 nm. Figure 3-3 shows

the ASD LabSpec Pro being used to obtain a spectrum for a guywire of one of the towers

in Survey Zone 1.























Figure 3-3. Obtaining reflectance measurements for a guywire of one of the towers in
Survey Zone 1 using the ASD LabSpec Pro portable spectrometer.

The process of obtaining reflectance spectra involves first acquiring a reference

spectrum with the instrument probe pointed at a calibrated reference panel, typically

made of Spectralon (Labsphere). The reference panel should be highly Lambertian and

have a known reflectance close to unity. Next, data are collected while pointing the

probe at the desired target. Finally, the reflectance values of the target are computed









from the ratio of the measurements made over the target to those made over the reference

panel for each band (Lillesand and Kiefer, 2000).

Data collection with the field spectrometer took place from July 22 to July 24,

2002. During this three-day period, weather conditions ranged from overcast to mostly

sunny. Because of the varying conditions, a reference spectrum was obtained for every

measurement, and each spectrum was acquired within one minute of its corresponding

reference spectrum. Data collection was limited to the hours of 10:00 AM to 2:45 PM

each day, since the sun was used as the light source.

Spectral measurements were acquired for twelve different vertical objects,

including eight field-surveyed obstructions, and also for three horizontal surfaces within

the survey areas. Appendix B contains the reflectance spectra. Water absorption bands

and excessively-noisy regions have been removed from the spectra. Table 3-2

summarizes the reflectance values at 1064 nm for the twelve objects. Table 3-3

summarizes the reflectance values at 1064 nm for the three surfaces.

Table 3-2. Reflectance values at 1064 nm for field-surveyed obstructions and other
objects in the survey areas.
Survey Point # Description /1064
445 Strobe Lighted Tower 0.607
445 Strobe Lighted Tower 0.107
(Guywire)
446 Tower 0.128
449 Pole (wood) 0.309
452 Antenna 0.318
453 Transmission Pole 0.148
454 Flagpole 0.639
456 Pole (wood) 0.332
N/A Pine Tree 0.595
N/A Palm Tree 0.414
N/A Pole (wood) 0.368
N/A Generator (metal painted 0.256
green)










Table 3-3. Reflectance values at 1064 nm for three horizontal surfaces in Survey Zone 1.
Survey Point # Description /1064
N/A Grass 0.567
N/A Concrete 0.402
N/A Asphalt 0.208

Some of the objects listed in Table 3-2 were noticeably more weathered on one side

than on the other sides. In addition, a few of the manmade objects contained both painted

and unpainted sections. In these cases, spectra were acquired for two or more different

parts of the object, and a mean of the reflectance values was taken. In obtaining spectra

for the palm and pine trees, the probe was aimed at the leaves or needles, since the first

laser return is likely to be from the top of a tree, rather than its trunk.

The mean value ofP1064 for the objects in Table 3-2 is 0.352. Twenty-five percent

of the objects in Table 3-2 have P1064 values of less than 0.2, indicating that many

obstructions are relatively poor reflectors of 1064 nm light. The significance of these low

reflectance values is examined in Chapter 4.














CHAPTER 4
DATA ANALYSIS

Preliminary Analysis

Before evaluating how well the field-surveyed obstructions were captured in the

airborne laser data, the fourteen data sets were checked for blunders and elevation biases.

This analysis was performed using an independent data set consisting of 46 check points

positioned by the National Geodetic Survey (NGS) using van-mounted GPS receivers

and kinematic (KGPS) processing techniques. All of the points are located in relatively

flat areas on or around the airfield. Software was written to locate the laser data point

closest in horizontal distance to each point in the independent data set. If the distance to

the closest point was less than the mean laser footprint radius, the laser point was selected

as a match to the NGS check point, and the difference in elevation between the laser

point and the NGS point was computed.

A non-zero mean difference in elevation between the airborne laser data points and

the NGS points was interpreted by the software as an elevation bias in the airborne laser

data. After removing elevation biases, the RMSE and estimated vertical accuracy at the

95% confidence level were calculated for each of the fourteen airborne laser data sets

(Table 4-1). The calculations were performed in accordance with the National Standard

for Spatial Data Accuracy (Federal Geographic Data Committee, 1998) using the

following equations:










(ZNGS -Z ser)2
RMSE= =1-
E n (4-1)

Accuracy95%cL = 1.96(RMSE) (4-2)

Table 4-1. Results of testing the airborne laser data sets using an independent data set of
NGS kinematic GPS runway points.
Config Parameters (tilt in deg; Day(s) of Mean RMSE, after Accuracy
# wide/narrow divergence; data Difference removing at 95%
flying height in m) collection in Elevation elevation bias CL (m)
(m) (m)
1 Tilt: 0, Div.: N, FH: 750 June 15 -0.06 0.08 0.15
2 Tilt: 0, Div.: W, FH: 750 June 15 -0.04 0.07 0.14
3 Tilt: 10, Div.: N, FH: 750 June 14-15 -0.01 0.05 0.09
4 Tilt: 10, Div.: W, FH: 750 June 14 0.00 0.05 0.10
5 Tilt: 20, Div.: N, FH: 750 June 12 0.09 0.12 0.23
6 Tilt: 20, Div.: W, FH: 1050 June 14 0.20 0.08 0.15
7 Tilt: 20, Div.: W, FH: 1150 June 14 0.12 0.11 0.21
8 Tilt: 20, Div.: W, FH: 750 June 12 0.20 0.12 0.23
9 Tilt: 20, Div.: W, FH: 850 June 13 -0.01 0.09 0.18
10 Tilt: 20, Div.: W, FH: 950 June 14 0.14 0.08 0.15
11 Tilt: 30, Div.: N, FH: 750 June 10-11 0.18 0.22 0.43
12 Tilt: 30, Div.: W, FH: 750 June 10-11 0.27 0.18 0.34
13 Tilt: 40, Div.: N, FH: 750 June 10 0.23 0.12 0.23
14 Tilt: 40, Div.: W, FH: 750 June 10 0.35 0.15 0.29

Several researchers (e.g., Vaughn et al., 1996b; Shrestha et al., 1999; Krabill et al.,

1995) have demonstrated that vertical RMSEs of 5 to 15 cm on terrain are achievable

through airborne laser mapping. Although most of the RMSEs listed in Table 4-1 are

within this range, the RMSEs for the data collected with the 300 tilt angle are notably

poorer. Also noteworthy from Table 4-1 are the mean differences in elevation between

the airborne laser data and the NGS points. These values appear to increase with tilt

angle and become quite large for the data sets collected with 300 and 400 tilt angles. This

correlation of height bias with tilt angle is clearly illustrated in Figure 4-1. Here, the










biases shown were calculated by averaging the biases for the various deployment

modalities of the airborne laser system.


35

30 -

255
20 --

c 15
..l" 2 .3
10,
LU
a, 5

$ 0
S10 20 30 40 5
-5 -5

-10
Tilt Angle (deg)

Figure 4-1. Plot of average elevation bias for each tilt angle setting on the ordinate vs. tilt
angle on the abscissa, based on the values from Table 4-1.

Although careful calibration of an airborne laser-scanning system should reduce

systematic error, elevation biases in airborne laser data are not uncommon. Vaughn et al.

(1996b) reported an elevation bias of 54 cm in their data, and Shrestha et al. (1999)

reported elevation biases that varied from pass to pass and ranged from -10 to +20 cm.

After removing these biases, both groups determined the vertical accuracy of their data to

be 10 cm (1 o) or better.

The apparent correlation between elevation bias and tilt angle shown in Figure 4-1

merits investigation. One possible explanation lies in the relationship between the

attitude parameters (pitch in particular) and the computed elevations of laser points. By

considering the geometry involved, it can be intuited that an error in pitch will have little

effect on the computed elevation of a laser point with a nadir-pointing beam, but as the









tilt angle is increased, the effect on the computed elevation will increase. A mathematical

analysis of this effect involves first expanding Equation (2-1) to give the following

expression for the elevation of a laser point:

Z = ZGPS + sinp(x, 8x + Rcosssint)+ cospsinr(y y + Rsins) (4
+ cospcosr(z1 Rcosscost)

Using Equation (4-3), it is possible to examine how errors in attitude angles

propagate to errors in computed elevations of laser points. Because we are considering

systematic (as opposed to random) errors, the methods applicable to propagation of

systematic errors (see, e.g., Mikhail and Ackermann, 1976; Vanicek and Krakiwsky,

1986) must be applied. Letting Ar, Ap, and Ah be small, known systematic errors in

orientation parameters (which can be either positive or negative), the propagated error in

the elevation of a laser point is given by


AZ = f Ar + Ap + Ah (4-4)
ar Op ah

Removal of systematic errors from airborne laser data has become the focus of

widespread research as the airborne laser scanning community strives towards ever-

increasing positional accuracy (e.g., Filin, 2002; Toth et al., 2002). Unknown systematic

errors in various parameters, including attitude, are likely to exist for any airborne laser

data set. While rigorous treatment of systematic errors is beyond the scope of this

research, we found through numerical curve-fitting techniques that the following errors in

attitude propagate to elevation errors that fit the empirical curve: Ap = 0.0430, Ar =

-0.0400, and Ah = 0. These values were obtained by conducting a systematic search in

2D parameter space to find the (Ap, Ar) pair that would yield the best fit to the empirical










curve. The search interval for both Ap and Ar was [-2o, +20], and the sampling step was

0.001.

Figure 4-2 shows a plot of propagated error in elevation vs. tilt angle using the

errors listed above and the following additional parameter values: R = 750 m (the flying

height for ten of the fourteen configurations), s = 7.50 (half of the maximum

instantaneous scan angle for all configurations), andp = r = 0.50 (typical values). The

blue curve is the experimental data curve shown in Figure 4-1, and the red curve is the

propagated systematic error as a function of tilt angle calculated using Equation (4-4).

Although, as noted above, the red curve was calculated using ad hoc methods, the close

agreement with the experimental data indicates that uncorrected systematic errors in

orientation could account for the observed trend.

30

25 -

20-

15

10

6

0 -

-5[

-10 -
0 5 10 15 20 25 30 35 40
Tilt Angle (deg)
Figure 4-2. Blue curve: average elevation bias vs. tilt angle, based on the data shown in
Table 4-1. Red curve: propagated systematic error in the elevation of a laser
point due to the following systematic errors in orientation: Ap = 0.0430, Ar
-0.0400, Ah = 0.









Errors in GPS (with possible day-to-day variability) could also help explain the

effects seen in Table 4-1. Based on the experience of researchers at UF and NGS, it is

reasonable to expect errors of approximately 4 cm horizontal and 8 cm vertical in the

trajectories, given the lengths of the baselines and the methods used in processing the

GPS data (Sartori, personal correspondence, 2002). In addition, atmospheric effects

should also be considered. For purposes of this study, it was not deemed necessary to

perform a rigorous analysis of atmospheric effects. Nevertheless, increased tilt angle will

clearly magnify any error due to the atmosphere for two reasons: 1) the optical path

length of the laser increases with tilt angle; and 2) refraction increases with increase in

optical path length.

Yet another factor that may have contributed to the trend seen in Figure 4-1 is the

possible reduction in range accuracy with a tilted sensor. The tilted sensor results in an

elongated pulse, which, in turn, leads to a longer rise time for the return pulse. The CFD

may experience difficulty with the longer rise time, leading to reduced range accuracy

(Liadsky, personal correspondence, 2003).

Lastly, height error (bias) may have been introduced when changing the

configuration of the laser system on board the aircraft. In this work, in-flight sensor

calibration was performed for only one configuration: 0 tilt, narrow divergence, and

1200 m flying height. (See Chapter 3 for an overview of the calibration procedures.)

While the importance of a separate calibration for each configuration was recognized,

time constraints permitted only one calibration flight. It is not surprising to find

systematic errors for configurations that are very different from that used in calibration.









Although the main focus of this study is on the detection of airport obstructions,

rather than the absolute vertical accuracy, the results of this analysis provide valuable

information for the implementation of airborne laser mapping technology in airport

obstruction surveying programs. First, to achieve the highest vertical accuracy, in-flight

calibrations should be performed for each configuration of the system to be used in data

collection. Second, with a tilted sensor, additional steps should be taken to reduce or

eliminate errors in attitude angles to the greatest extent possible.

Obstruction Detection Analysis

The primary objective in analyzing the data was to determine how well

obstructions were detected in each of the fourteen airborne laser data sets. The reference

data set used in the obstruction detection analysis consisted of 52 field-surveyed

obstructions. These obstructions were positioned by an NGS survey crew in February

2001, using GPS and conventional survey techniques (Figure 4-3). Many of the objects

surveyed by the field crew did not actually obstruct the FAR Part 77 or ANA surfaces at

GNV, but were selected as being representative of "typical" types of obstructions. The

procedures followed in surveying these objects were identical to those commonly

followed by NGS surveyors in performing FAR Part 77 and ANA obstruction surveys in

accordance with FAA No. 405, Standards for Aeronautical Surveys and Related Products

(U.S. Department of Transportation, 1996). Photographs of many of the obstructions can

be found in Appendix C.

For purposes of this study, "detection" was defined as satisfying the following two

conditions: 1) laser returns were received from the object surveyed by the field crew; and

2) the difference in elevation between the field-surveyed point and the closest laser point

on the object was within a predefined limit. It was not deemed necessary for the laser









point to have hit the same part of the object surveyed by the field crew. For example, if

the survey crew positioned an obstruction light on the west side of the top of a tower

while the closest laser point hit the east side of the tower, this would qualify as a

detection, provided the difference in elevation was within tolerance. Furthermore, for

clusters of trees that are in such close proximity to one another that the branches overlap,

it was not considered necessary (or even possible) to verify that the closest laser point to

the field-surveyed point actually hit the same tree and not its neighbor.





















Figure 4-3. NGS field survey of obstructions at GNV.

Although the NGS data set originally contained 55 obstructions in the project areas,

three of the obstructions were excluded from the analysis based on visual inspection of

the survey area and discussions with airport authorities. Two of the excluded

obstructions, Survey Point Numbers (SPNs) 418 and 438, were trees that were either

burned or blown down between the dates of the NGS field survey and the airborne laser

data collection. The third excluded obstruction was an antenna (SPN 450) that was found

to have been removed. Since the 52 remaining obstructions included seven antennae and









fifteen trees, the absence of these three obstructions was not considered detrimental to the

overall analysis.

Automated Obstruction Detection Analysis

The first step in the analysis entailed writing software to determine the number of

field-surveyed obstructions detected in each airborne laser data set and the RMS

difference in elevation between the field-surveyed points and closest laser points on the

detected obstructions. The software used a search cylinder centered on each field-

surveyed obstruction (Figure 4-4) as the detection criterion. If no laser point was found

within the search cylinder for a particular obstruction, then that obstruction was reported

as "not detected" by the software. If the search cylinder contained one or more laser

points, the laser point in the cylinder and closest in 3D distance to the field-surveyed

point was located and used as the "matching" point. The output of the software included

the number of obstructions detected in each laser data set, the horizontal and vertical

distance from each field-surveyed obstruction to its matching laser point, and the RMS

differences in elevation.

The search cylinder was adopted based on the definition of "detection" and served

two purposes: first, to provide some measure of assurance that the laser point selected by

the software hit the same object surveyed by the field crew; and second, to impose a limit

on the maximum difference in elevation between the field-surveyed point and closest

laser point that would qualify as a detection of that object. The radius of the search

cylinder used in the analysis software was based on three factors: 1) the estimated

horizontal position error in the field-surveyed data; 2) the estimated horizontal position

error in the airborne laser data; and 3) an allowance for the movement of the tops of

obstructions due to wind.










Search Cylinder




Laser Data Points Field-Sreyed
Field-Surveyed
Point


















Figure 4-4. Obstruction detection analysis algorithm. The red laser point is in the search
cylinder and closest in 3D distance to the field-surveyed point, so it is selected
as the "matching" point. If the search cylinder contains no laser points, the
obstruction is reported to be "not detected."

Although no accuracy assessment was performed on the field-surveyed data, the

survey party chief estimated the horizontal and vertical accuracy of the obstruction data

to be no worse than 0.76 m (2.50 ft) and 0.15 m (0.50 ft), respectively, at the 95%

confidence level. These estimates were based on the survey methods and instruments

used and on checks provided by redundant observations during the field survey (Kuper,

personal correspondence, 2002). Tuell (2002) determined the horizontal accuracy of the

airborne laser data collected with Optech ALTM systems during the 2001 study to be

0.15 m at the 1-o level. The corresponding 95%-confidence-level value, 0.26 m, was

used as the estimated horizontal accuracy of the airborne laser data.









Based on weather data from the Southeast Regional Climate Center in Columbia,

South Carolina, wind speeds at GNV during the airborne laser data collection averaged

just 13 km/hr (8 mph), but gusts of up to 37 km/hr (23 mph) were recorded. The wind

conditions during the 2001 field survey are unknown but significantly less important,

since an experienced field surveyor will take steps to minimize the effect of wind on the

surveyed-position of an obstruction. Since many of the trees surrounding the airport are

20 to 25-m pines whose tops tend to sway in even mild breezes, 2 m was selected as a

reasonable (perhaps slightly conservative) allowance for horizontal movement due to

wind. Summing the estimated horizontal position errors and the wind allowance gives a

3-m radius for the search cylinder.

The horizontal accuracy requirement for obstructions specified by FAA No. 405 is

6.1 m (20 feet) or 15.2 m (50 feet), depending on the location of the object within the

Obstruction Identification Surfaces (OIS). These values were not used for the radius of

the search cylinder because they were deemed too large to provide any assurance that the

laser point hit the same object surveyed by the field crew. The height of the search

cylinder, on the other hand, was selected based on the vertical accuracy requirement

specified in FAA No. 405. The vertical accuracy requirement is 0.91 m (3 feet) or 6.1 m

(20 feet), again depending on the location of the object within the OIS. The less-stringent

value, 6.1 m, was used as the maximum difference in elevation (i.e., half the height of the

cylinder) in the analysis software.

Vertical accuracy of 6.1 m is an extremely significant criterion to approach

procedure developers. Specifically, obstruction data that meet this standard can be

designated as vertical code "C," thus satisfying one of the key minimum vertical accuracy









requirements specified in FAA Order 8260.19, Flight Procedures andAirspace (U.S.

Department of Transportation, 1993). Data that do not meet this standard are given a

vertical accuracy code of"D" (50 feet) or worse, which can affect minimum altitudes and

lead to operational restrictions (U.S. Department of Transportation, 1993).

Table 4-2 summarizes the results of the automated analysis. The table has been

sorted based on how well obstructions were detected using each configuration.

Configuration 5 at the top of the table has the highest percent of obstructions detected

(100%) and the lowest RMSE (0.88 m). At the bottom of the table is Configuration 12

with only 63% detection and an RMSE of 2.17 m. The actual output of the analysis

software showing the horizontal, vertical, and 3D distances from each obstruction to its

matching laser point for each of the fourteen data sets is contained in Appendix D.

Table 4-2. Percent of obstructions detected in each airborne laser data set and the RMS
difference in elevation between the field-surveyed points and "matching" laser
points.
Config Parameters (tilt in deg; Percent of RMSE Accuracy
# wide/narrow divergence; flying Obstructions (m) at 95% CL
height in m) Detected (m)
5 Tilt: 20, Div.: N, FH: 750 100 0.88 1.73
1 Tilt: 0, Div.: N, FH: 750 100 1.04 2.04
8 Tilt: 20, Div.: W, FH: 750 100 1.26 2.46
9 Tilt: 20, Div.: W, FH: 850 98 1.81 3.55
13 Tilt: 40, Div.: N, FH: 750 96 1.14 2.23
2 Tilt: 0, Div.: W, FH: 750 96 1.24 2.42
3 Tilt: 10, Div.: N, FH: 750 94 1.23 2.42
4 Tilt: 10, Div.: W, FH: 750 94 1.27 2.49
10 Tilt: 20, Div.: W, FH: 950 87 1.99 3.91
11 Tilt: 30, Div.: N, FH: 750 85 1.83 3.58
14 Tilt: 40, Div.: W, FH: 750 77 2.13 4.17
7 Tilt: 20, Div.: W, FH: 1150 77 2.16 4.22
6 Tilt: 20, Div.: W, FH: 1050 73 2.02 3.97
12 Tilt: 30, Div.: W, FH: 750 63 2.17 4.25









Several interesting observations can be made from the results shown in Table 4-2:

* With constant flying height and tilt angle, narrow divergence was consistently
better than wide divergence for obstruction detection.

* For the 100 and 200 tilt angles, the number of obstructions detected was the same
for the narrow and wide divergence settings, with the only difference being in the
RMSE. For the 300 and 400 tilt angles, the difference in the percent of obstructions
detected between the narrow and wide divergence setting increases to over 20%.

* For the 200 tilt angle and wide divergence, the percentage of obstructions detected
decreases rapidly with flying height.

In Chapter 2, it was shown mathematically that obstruction detection depends on an

interplay between geometry and return signal strength (radiometry). The results and

observations above clearly illustrate the tradeoff that exists between geometry and

radiometry. Based purely on geometric considerations, configurations employing 300 to

400 tilt angles and wide divergence, such as numbers 12 and 14, should have been very

well suited for obstruction detection, whereas configurations with near nadir-pointing

beams and narrow divergence, such as number 1, should have been poor. Instead,

however, configurations 12 and 14 are both near the bottom of the table, while

configuration 1 is near the top. The ability to detect obstructions with a particular

configuration clearly cannot be predicted based on geometry alone.

The importance of radiometric considerations can be understood through

examination of the reflectance data obtained with the field spectrometer (see Table 3-2

and Appendix B). These data indicate that many obstructions are poor reflectors at the

laser wavelength of 1064 nm. Low reflectance at this wavelength combined with small

cross-sectional area leads to very small effective target cross sections (see Equation (A-

6)). Regardless of the number of laser pulses incident on these objects, they will not be

detected unless the irradiance is sufficiently high to result in a detectable return signal.









It is interesting to note in Table 4-2 that three of the four best configurations used a

200 tilt angle. With constant flying height and beam divergence, a nadir-pointing beam

will minimize the range, giving the highest irradiance on a target (see Equation (A-3)),

whereas a large tilt angle will produce better geometry (Equation (2-3) and Figure 2-6).

The results presented in Table 4-2 show that in this study, a tilt angle of 200 provided the

best geometry while still enabling a detectable return signal from the obstructions.

The somewhat anomalous results for configuration 1 merit further investigation.

Although, as noted above, the nadir-pointing beam used in configuration 1 should enable

high return signal strength, the analysis presented in Chapter 2 shows that the obstruction

detection geometry is poor with this configuration. In fact, configuration 1 is very similar

to the configurations used in the 2001 study, which produced relatively poor results.

Nevertheless, this configuration produced the second best results in this study.

A possible explanation for the results achieved with configuration 1 lies in the fact

that fairly substantial overlap between strips (approximately 25%) was used throughout

this study. Apparently, the higher PRF and good strip overlap improved the geometry

enough to enable configuration 1 to outperform the similar configurations used in the

2001 study. Because it is impossible to predict where within the scan pattern an

obstruction will fall, however, increasing the tilt angle is a more reliable method of

improving the detection geometry than increasing the strip overlap. Further tests are

needed to determine whether or not the favorable results achieved with configuration 1

can be replicated.

Despite the steps taken to minimize uncontrolled variables during the study, daily

variations in weather conditions, GPS geometry, and even the performance of the laser









are unavoidable. In at least one instance, these uncontrolled variables may have had a

noticeable effect on the results. Specifically, reports generated by the data processing

software showed that the data collected with the 300 tilt angle (configurations 11 and 12)

on June 10th had a significantly lower percentage of returns than any of the other data

sets collected during this study. It is suspected that the atmospheric conditions were

relatively poor on the evening of June 10th, causing the 300 tilt angle configurations not

to perform as well as they might have under better conditions.

Visual Analysis

The automated obstruction detection analysis was supplemented with a visual

analysis performed using TerraScan Viewer (Terrasolid, Ltd.). A 0.02 km2 subset around

each obstruction of interest was taken from each of the original airborne laser data sets.

Using the TerraScan Viewer, the laser points were displayed in profile mode to determine

how well the obstruction was detected with each configuration. For example, Figure 4-5

shows the data points on an antenna (SPN 452) computed from the laser returns obtained

using configurations 5 and 14. From Table 3-2, the reflectance of this object at 1064 nm

is 0.32. The automated analysis software reported SPN 452 to be "not detected" with

configuration 14. From Figure 4-5 it can be seen that, in fact, with configuration 14, no

detectable laser returns were received from this obstruction.

Figure 4-6 shows the results of a similar analysis performed on a pole (SPN 460).

The automated analysis software reported that this object was detected to within 0.34

meters vertical of the field-surveyed point with configuration 5 and to within 0.69 meters

with configuration 8. The only difference between configurations 5 and 8 is that the

former used narrow divergence and the latter, wide. The obstruction was reported to be









"not detected" with configuration 12. Again, the visual analysis supports and clarifies the

results of the automated analysis. It should be noted that the colors of the laser points in

Figures 4-5 and 4-6 correspond to a classification by elevation range, but the classes were

not standardized between the two figures.
















Figure 4-5.Visual obstruction analysis. The image on the left is a photograph of SPN
452. The middle image shows the data points computed from laser returns
obtained using configuration 5. The image on the right shows that with
configuration 14, no detectable laser returns were received from the object.


Figure 4-6. From left to right: photograph of SPN 460, and data points on this object
based on laser returns obtained using configurations 5, 8, and 12, respectively.
Although it is difficult to discern from the photograph on the far left, the pole
is in front of the tree. The tree was not included in the TerraScan profiles.


rrsxrpep~i~i~









Analysis of Return Signal Strength Calculations

Using Equations (2-11) and (2-12) and the reflectance data given in Table 3-2, the

received power from an obstruction for each configuration can be estimated. By

comparing the calculated return signal strength values against the results of the automated

obstruction detection analysis (Appendix D) it is then possible, in theory, to determine

the value of the minimum detectable return signal. This is somewhat of an inexact

science because of the day-to-day (or even pulse-to-pulse) variability in certain

parameters, such as the output power of the laser and atmospheric transmittance. Further,

the reflectance of an obstruction and its cross-sectional area are typically not constant

over its entire surface. Lastly, it is nearly impossible to predict, for any given laser pulse,

where within the footprint an obstruction will fall.

Based on the above arguments, the return signal strength calculations give, at best,

a sort of estimated average value. It can be countered, however, that with good

geometry, numerous laser pulses (perhaps even hundreds of pulses) will be incident on an

obstruction. Therefore, even an estimated average value of the return signal strength is

useful, under the assumption that at least one pulse will lead to a return signal equal to or

greater than the calculated value.

Table 4-3 shows the calculated received power for SPN 452 (the antenna shown in

Figure 4-5) for each configuration, along with the results of the automated obstruction

detection analysis. Tables 4-4 and 4-5 show the results of similar analyses for SPNs 449

and 454, respectively. SPN 449 is a wood pole, while SPN 454 is a flagpole (see

photographs in Appendix C). The value of the atmospheric transmittance, TATM, used in

the calculations was 0.87.









The data in Tables 4-3 through 4-5 indicate that the minimum received power for

detection is approximately 0.5 [W. However, the following simplifying assumptions

have been made in the return signal strength calculations: 1) all targets are lambertian; 2)

all surfaces are flat; and 3) the power distribution within the footprint is uniform. The

calculated received power values would be smaller if these assumptions had not been

made. Hence, the values shown here should not be interpreted as an accurate portrayal of

the performance characteristics of the Optech system. Nevertheless, the good relative

agreement between the three tables suggests that the analytical methods presented here

may be used to examine the ability to detect targets whose reflectance has been

measured.

Table 4-3. Analysis of return signal strength calculations for SPN 452. Configurations 11
and 12 have been excluded from this analysis because, as noted above, the
atmospheric conditions were poor during the collection of those two data sets.
Configuration Parameters (tilt in deg; Calculated Received Detected
# wide/narrow divergence; flying Power from SPN (Y/N)
height in m) 452 ([LW)
1 Tilt: 0, Div.: N, FH: 750 1.65 Yes
3 Tilt: 10, Div.: N, FH: 750 1.60 Yes
5 Tilt: 20, Div.: N, FH: 750 1.46 Yes
2 Tilt: 0, Div.: W, FH: 750 1.12 Yes
4 Tilt: 10, Div.: W, FH: 750 1.07 Yes
13 Tilt: 40, Div.: N, FH: 750 0.97 Yes
8 Tilt: 20, Div.: W, FH: 750 0.93 Yes
9 Tilt: 20, Div.: W, FH: 850 0.64 Yes
14 Tilt: 40, Div.: W, FH: 750 0.50 No
10 Tilt: 20, Div.: W, FH: 950 0.46 Yes
6 Tilt: 20, Div.: W, FH: 1050 0.34 No
7 Tilt: 20, Div.: W, FH: 1150 0.26 No











Table 4-4. Analysis of return signal strength calculations for SPN 449. Again,
configurations 11 and 12 have been excluded from the analysis for the reasons
mentioned above.
Configuration Parameters (tilt in deg; Calculated Received Detected
# wide/narrow divergence; flying Power from SPN (Y/N)
height in m) 449 (|tW)
1 Tilt: 0, Div.: N, FH: 750 1.61 Yes
3 Tilt: 10, Div.: N, FH: 750 1.56 Yes
5 Tilt: 20, Div.: N, FH: 750 1.42 Yes
2 Tilt: 0, Div.: W, FH: 750 1.04 Yes
4 Tilt: 10, Div.: W, FH: 750 0.99 Yes
13 Tilt: 40, Div.: N, FH: 750 0.94 Yes
8 Tilt: 20, Div.: W, FH: 750 0.86 Yes
9 Tilt: 20, Div.: W, FH: 850 0.59 Yes
14 Tilt: 40, Div.: W, FH: 750 0.47 No
10 Tilt: 20, Div.: W, FH: 950 0.42 No
6 Tilt: 20, Div.: W, FH: 1050 0.31 No
7 Tilt: 20, Div.: W, FH: 1150 0.24 No

Table 4-5. Analysis of return signal strength calculations for SPN 454.
Configuration Parameters (tilt in deg; Calculated Received Detected
# wide/narrow divergence; flying Power from SPN (Y/N)
height in m) 454 (|tW)
1 Tilt: 0, Div.: N, FH: 750 3.32 Yes
3 Tilt: 10, Div.: N, FH: 750 3.22 Yes
5 Tilt: 20, Div.: N, FH: 750 2.93 Yes
13 Tilt: 40, Div.: N, FH: 750 1.43 Yes
2 Tilt: 0, Div.: W, FH: 750 1.06 Yes
4 Tilt: 10, Div.: W, FH: 750 1.01 Yes
8 Tilt: 20, Div.: W, FH: 750 0.88 Yes
9 Tilt: 20, Div.: W, FH: 850 0.61 Yes
14 Tilt: 40, Div.: W, FH: 750 0.48 No
10 Tilt: 20, Div.: W, FH: 950 0.43 No
6 Tilt: 20, Div.: W, FH: 1050 0.32 No
7 Tilt: 20, Div.: W, FH: 1150 0.24 No

Depending on the diameter of the laser footprint and, thus, on the configuration, the

obstructions were sometimes treated as area targets and sometimes as linear targets in the

calculations. In the linear target cases, rather than assuming that the target fell directly in

the center of the laser footprint, the expected value for the extent of the target in the










footprint was used. The expected (average) value for the extent of a linear target in a

circular footprint is obtained by dividing a quadrant of the footprint (i.e., the area under

the curve) by the radius, giving 78.5% of the maximum value.

A potentially more rigorous method of performing the return signal strength

calculations would be to model the interaction of the incident laser radiation with the

target as a convolution, as depicted in Figure 4-7. In this method, for each position of the

kernel (i.e., each laser footprint position) a return signal strength value can be obtained.

One obvious advantage of this method is a more rigorous modeling of both the laser

footprint and the target. Although this method was not used in the return signal strength

calculations in this study due to the obvious complexity of modeling targets in this

manner, it may be worth investigating in future research.



.cv".


Irradiance
incident on target


24 2 Convolve


1 2 1

Kernel representing
irradiance incident


SPt PV P Pl4 P P PV PS
P.I F-- i' C P* FP7 P P Ps

P1i p: P r Pr P P-6 Pr P.

Pn PF1 P P PC' P:4 Pr Pa
PN PC Py PM PIB P V p p
P7I P72 Pn P74 P7J PM P77 P7
Psi Pi P3 P64 PS PS Pg? Pes


Target
-'-of interest


on target Rasterized piece of
world containing target

Figure 4-7. A potentially more rigorous method of performing the return signal strength
computations involving modeling the interaction of the incident laser radiation
with a target as a convolution.














CHAPTER 5
CONCLUSIONS AND RECOMMENDATIONS

The results of this research provide strong indication of the capability to detect and

position airport obstructions requiring 6.1-m vertical accuracy using an airborne laser-

scanning system that has been appropriately configured. Three of the fourteen data

collection configurations tested in this study resulted in 100% of the field-surveyed

obstructions being detected to within predefined tolerances that were established based

on requirements set forth in FAA specifications documents. Particularly encouraging is

the vertical RMSE of 0.88 m achieved with configuration 5.

While these results attest to the potential of airborne laser scanning in airport

obstruction surveying, we may have only begun to scratch the surface in terms of the

ability to detect and accurately position discrete point features. Clearly, improvements

above and beyond those demonstrated in this study are feasible. However, there is a limit

to the improvement that can be achieved through modification of data collection

parameters by the user of the system; additional enhancements will require the

cooperation of the system manufacturer.

Further improvements to geometry will require increasing the pulse repetition

frequency and, even more importantly, the scan frequency. One method of achieving a

higher scan frequency is to utilize a dual-axis scanner. A side benefit of the dual-axis

scanner would be a more regular scan pattern.

Improvements in parameters relating to radiometric considerations are also

possible. Several of the data collection configurations used in this study resulted in a









high density of laser pulses incident on the obstructions, but the obstructions were not

detected because the received signal was below the detection threshold. This suggests

that obstruction detection capability could be further improved by increasing the

sensitivity of the receiver. This could be achieved, at least in theory, through any of the

following methods:

* Lowering the preset signal threshold in the CFD
* Utilizing a photodetector with a higher responsivity
* Increasing the area of the receiving optics
* Cooling the detector
* Modifying the bandpass filter to optimize SNR
* Changing any other parameters necessary to reduce noise in the system

Although the modifications listed above should improve the detection capability,

some of them could have significant drawbacks. For example, reducing the preset signal

threshold would increase the probability of false returns. Only the system manufacturer,

or someone with intimate knowledge of the various system components and performance

characteristics, would be able to evaluate the potential benefits and drawbacks associated

with each. There is little doubt, however, that it is possible to better tune the system for

obstruction detection. While many applications demand the highest possible data

accuracy, for obstruction detection it would be permissible to sacrifice a small amount of

geometric accuracy for increased receiver sensitivity.

The possible use of array detectors for further improving obstruction detection

capability also merits investigation. Degnan (2002) proposes a spaceborne system

employing a highly-pixellated (e.g., 10x10) array detector and a dual wedge optical

scanner. Although the intended application of this proposed system is the mapping of

planets, such as Mars, from orbiting spacecraft, the use of array detectors could also

prove beneficial in airport obstruction detection from airborne platforms. Potential









advantages include increased spatial resolution and, depending on the type of detectors

used, increased sensitivity.

Although this study has focused on detection of airport obstructions, other

important research topics involve extraction and classification of obstructions in airborne

laser data sets. In completing an obstruction survey through airborne laser scanning,

detection of obstructions is only the first step. The obstruction data must then be subset

from the larger data set and classified. For example, it is important to know whether an

object that penetrates the FAA survey surfaces is a manmade feature, such as a pole, or a

natural feature, such as a tree. Most modem airborne laser-scanning systems output

"intensity" data in addition to the range measurements. These data consist of digital

numbers proportional to the generated photocurrent in the receiver. The photocurrent is,

in turn, proportional to the received optical power and, hence, to the reflectance of the

target at the laser wavelength. This intensity data combined with aerial photography, if

available, will undoubtedly prove beneficial in classification. Nevertheless, the problem

of classification is nontrivial and will require continued research.

Lastly, necessary enhancements to the processing software should not be ignored.

In this study, it was found that just as commercial airborne laser mapping systems have

been optimized for bare-earth terrain mapping, so too have data processing software

packages. For airborne laser-scanning systems to be used in production airport

obstruction surveying programs, software that is better tailored to this application would

be extremely beneficial. Existing software could be used with the simple addition of a

specialized airport obstruction processing module. This module would disable functions

that serve to eliminate "suspect" last returns or interpolate or filter the data in any way.









In addition, the module would include tools to allow better visualization of obstructions,

such as in profile mode.

Through further experimentation and possible incorporation of the ideas mentioned

above, it is likely that airborne laser scanning will continue to become an increasingly

effective technology for airport obstruction surveying. The density of laser points

incident on the vertical faces of obstructions will continue to increase, as will the

probability of detecting obstructions with small effective target cross sections. Improved

software will give data processors greater ability to capture, detect, and visualize

obstructions in the laser data set. It should be kept in mind, however, that the goal of

these efforts is to supplement, rather than replace, existing technologies. It is unrealistic

to expect, for example, that airborne laser scanning could eliminate the need for field

surveys or entirely replace photogrammetric procedures. However, expanding the

number of viable airport obstruction surveying technologies will give survey planners

increased options and allow the survey methods to be better tailored to specific project

requirements. The implementation of airborne laser scanning could represent the next

major step in the technological evolution of airport obstruction surveying.















APPENDIX A
DERIVATION OF RANGE EQUATION









The following is a derivation of an equation for the received signal power,

commonly referred to as the "range equation" or "laser radar range equation." The range

equation can be found in various forms in numerous journal articles, including the

following: Wyman (1968), Jelalian (1992), and Baltsavias (1999a). To start, it is noted

that the irradiance incident on the receiver, Ea, measured in W- m2, can be expressed as

P
E, = TATM (A-l)


where Prefl is the reflected power from a target, Qs is scattering solid angle of the target, R

is the range, and TATM is the atmospheric transmittance. Prefl is given by

Pref = pE,,Aar,, (A-2)

where p is the target reflectance, Ear is the irradiance on the target, and Ar is the target

area. Next, Eta can be expressed as

P
Et P--2 T
Etar ~ 2 1ATM
(A-3)

where PT is the transmitted power, and Qt is the solid angle into which the transmitted

power is radiated (i.e., the solid angle subtended by the laser footprint), which is given by

AF
SR2 (A-4)

In Equation (A-4), AF is the area of the footprint, given approximately by


AF -R2 2
4 (A-5)

where yis the beam divergence in radians. Finally, adapting the definition of effective

target cross section, o, from Jelalian (1992):


= pAtar (A-6)









and combining equations (A-i) through (A-6) gives the following expression for Ea:

PT, C
EF -2R4-T
7 2 2R4 T (A-7)

Using Equation (A-7), the received power can then be calculated from

PA, 2A
P = A, Ea Tsys 2 24 T2TM 7TSYS (A-8)


where A, is the receiver area and Tsys is the system transmittance, which is limited

primarily by the transmittance of the bandpass filter. By considering the size and

position of the target in relation to the size and position of the laser footprint, three

different types of targets can be defined: 1) an "area target" fills the entire footprint; 2) a

"linear target" extends the entire length of the footprint but has a width that is small in

comparison with the footprint diameter; and 3) a "point target" has an area much smaller

than that of the footprint. Jelalian (1992) gives the following expressions for ofor area,

linear and point targets, respectively:

Sarea = pR2y2 (A-9)

Crnear = 4pRyd (A-10)

pomnt = 4pA,, (A-11)

In Equation (A-10), dis the diameter of the linear target. The targets are assumed

to be Lambertian (Q, = 71). The significance of equations (A-9) through (A-1 1) is that Ea

is inversely proportional to R2 for an area target, R3 for a linear target, and R4 for a point

target.

In the special case that the target is lambertian and fills the entire footprint, and the

system transmittance can be neglected, Equation (A-8) reduces to






63


P, A,P 2
P, = R2 ATM
2 (A-12)

which is identical to the range equation given in Baltsavias (1999a).















APPENDIX B
REFLECTANCE SPECTRA FOR OBSTRUCTIONS AND OTHER OBJECTS
WITHIN THE SURVEY AREAS











SPN 445: Strobe Lighted Tower


0.8


0.6


'; 0.4


0.2


400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)

SPN 445: Strobe Lighted Tower (Guywire)



0.8


C 0.6
0
'- 0.4


0.2


400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)

SPN 446: Tower



0.8


C 0.6
Ca

t 0.4


0.2 -


0 I I L L
400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)
Figure B-1. Reflectance spectra for SPN 445 strobe lighted tower (top), a guywire for
the strobe lighted tower (middle), and SPN 446 tower (bottom).







66



SPN 449: Pole (wood)
III I I


400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)

SPN 452: Antenna
I L L I I I L L L














400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)

SPN 453: Transmission Pole


400 600 800 1000 1200 1400 1600
Wavelength (nm)


1800


Figure B-2. Reflectance spectra for SPN 449 pole (top), SPN
and SPN 453 transmission pole (bottom).


2000 2200 2400


452 antenna (middle),


I I


0.8


0.6


0.4


0.2


u


C 0.6


S 0.4


0.2


0




1


0.8

CD
O 0.6
C,
S 0.4


0.2


0


f~l -I-------1^


I I I I I


11'Nr~











SPN 454: Flagpole


1


0.8


O 0.6


S0.4
nc-


0.2


0




1


0.8


S0.6

a,
t' 0.4
ar

0.2


0




1


0.8


C 0.6
0

'b 0.4
0r

0.2


0


SPN 456: Pole (wood)


400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)


Pine Tree


400 600 800 1000 1200 1400 1600 1800
Wavelength (nm)


2000 2200 2400


Figure B-3. Reflectance spectra for SPN 454 flagpole (top), SPN 456 pole (middle),
and a pine tree (bottom).


400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)












Palm Tree


1


0.8


S0.6 -


t 0.4
a:

0.2


0
400




10


0.8 -


U4
400 600


800 1000 1200 1400 1600 1800
Wavelength (nm)


2000 2200 2400


Generator (metal painted green)
III I I I I


400


600 800 1000 1200 1400 1600 1800 2000 2200 2400


Wavelength (nm)

Figure B-4. Reflectance spectra for a palm tree (top), a pole (middle), and a generator
(bottom).


600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)

Pole (wood)
II I I I


C 0.6
a0

5 0.4
Q-2


I II


a,
= 0.6


| 0.4


0.2


nI


I I I I I I I I


ill


0.2


I-


~yn~


I I I







69



Grass


CD
0.6
0







0





1


0.8

a,
C 0.6


'3 0.4


0.2


0


O 0.6
0

'| 0.4
Mr


400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)

Concrete


400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)

Asphalt


0.2


rI


u
400


600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavelength (nm)


Figure B-5. Reflectance spectra for grass (top), concrete (middle), and asphalt (bottom).


I















APPENDIX C
PHOTOGRAPHS OF FIELD-SURVEYED OBSTRUCTIONS















SPN 414 TREE


SPN 415 TREE
..............


Figure C-1. Photographs of SPN 414 tree (top), and SPN 415 tree (bottom).





71





72





SPN 4 IS TREE
SPN -I1 TREE
/


Figure C-2. Photographs of
pole (bottom).


N 1s tree (top), ana


N 4. 1 ODStructlOn ilgnt on






73






SPN 445 ANT ON
STROBE LTD TWR
























STROBE LTD T\\-R

















Figure C-3. Photographs of SPN 446 tower, SPN 445 antenna on strobe lighted tower
(top), and SPN 448 antenna on strobe lighted tower (bottom).


















































Figure C-4. Photographs of SPN 449 pole, SPN 453 transmission pole (top), and SPN
452 antenna (bottom).








SPN 454 FLGPL I
/


|SPN 45, POLEI


JSPN 455 SIGN
,


SPN 45"
TRNISN POLE
/


Figure C-5. Photographs of SPN 454 flagpole (top), SPN 456 pole, SPN 455 sign,
and SPN 457 transmission pole (bottom).






























ISPN 459 POLE


Figure C-6. Photographs of SPN 457 transmission pole (top), and SPN 459 pole
(bottom).






77






SPN 460 POLE
/P ,6 O


Figure C-7. Photograph of SPN 460


pole.















APPENDIX D
OUTPUT OF AUTOMATED OBSTRUCTION DETECTION ANALYSIS SOFTWARE









Configuration 1


Table D-1. Tilt: 0; Div: N; FH:750
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
410 ROD ON OL ASOS!SENSOR 1.2328 0.8823 0.8610
[GNV]
436 OBST# 436 TREE 0.6160 0.5282 0.3170
437 OBST# 437 TREE 0.3563 0.1586 0.3190
439 OBST# 439 FENCE 0.4524 0.4524 -0.0010
440 OBST# 440 ANT ON HGR 0.6055 0.2659 0.5440
444 ANT ON STROBE LTD 0.6229 0.1159 0.6120
TWR!444
445 ANT ON STROBE LTD 0.7897 0.2314 0.7550
TWR!445
446 OBST# 446 TWR 0.1283 0.1209 -0.0430
447 ROD ON STROBE LTD 0.9752 0.9195 0.3250
TWR!447
448 ANT ON STROBE LTD 5.0592 2.0324 4.6330
TWR!448
449 OBST# 449 POLE 0.2608 0.2563 0.0480
451 OBST# 451 BLDG 0.8668 0.8668 -0.0030
452 OBST# 452 ANT 0.3562 0.3354 0.1200
453 OBST# 453 TRMSN POLE 0.2430 0.2384 0.0470
454 OBST# 454 FLGPL 0.2069 0.1436 -0.1490
455 OBST# 455 SIGN 0.9851 0.7070 0.6860
456 OBST# 456 POLE 0.9750 0.9470 0.2320
457 OBST# 457 TRMSN POLE 0.2118 0.2105 0.0230
458 OBST# 458 BLDG 0.3248 0.2875 0.1510
459 OBST# 459 POLE 0.2281 0.2045 0.1010
460 OBST# 460 POLE 1.0108 0.8408 0.5610
461 OBST# 461 ROD ON OL 1.1850 0.5257 -1.0620
AMOM
462 OL VORTAC!462 [GNV] 0.2879 0.2851 0.0400
"NCM"
463 OBST# 463 LT POLE 0.7598 0.1086 0.7520
464 OBST# 464 LT POLE 0.2711 0.2708 -0.0130
465 OBST# 465 FENCE 0.4637 0.3949 0.2430
466 OBST# 466 FENCE 0.3325 0.3069 0.1280
467 OBST# 467 TREE 0.4545 0.2971 0.3440
468 OBST# 468 TREE 0.2478 0.2297 0.0930
469 OBST# 469 TREE 2.8884 2.0830 2.0010
470 OBST# 470 TREE 0.7550 0.0871 0.7500
471 OBST# 471 TREE 0.4776 0.0497 0.4750
472 OBST# 472 TREE 0.4138 0.2912 0.2940









Table D-1-Continued
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
473 OBST# 473 TREE 0.1700 0.1697 0.0090
474 OBST# 474 TREE 0.3912 0.1329 0.3680
475 OBST# 475 HGR 0.3616 0.3563 -0.0620
476 OBST# 476 SIGN 1.0931 0.1869 1.0770
477 OBST# 477 FENCE 0.7263 0.6391 0.3450
478 OBST# 478 FLGPL 1.1430 0.7848 0.8310
479 OBST# 479 TREE 0.7573 0.3032 0.6940
480 OBST# 480 ANT ON BLDG 2.4871 0.5566 2.4240
302 ANT ON OL ATCT!ATCT 4.6637 2.7009 3.8020
FLOOR164
306 OL ON LTD WSK 0.5995 0.5846 0.1330
25 ROD ON OL APBN!APBN 1.0998 0.3240 1.0510
412 OL ON LOC!(28) 0.4656 0.4656 0.0040
415 TREE 0.2852 0.2517 0.1340
423 TREE 0.6122 0.5364 -0.2950
430 BLDG 0.1711 0.1711 0.0000
431 OL ON POLE 1.8813 1.8813 0.0070
402 ROD ON OL GS!(28) 0.8102 0.3993 0.7050
414 TREE 0.4778 0.1100 0.4650
425 TREE 0.4921 0.4865 -0.0740

RMSE: 1.04065
Accuracy: 2.03968
Percent Detected: 100.
Search Radius: 3.
LIDAR data file:
C:\lChris\University of Florida\New AccuracyTest\TiltO DivN Fh750.alf


Configuration 2

Table D-2. Tilt: 0; Div: W; FH: 750
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
410 ROD ON OL ASOS!SENSOR 1.2247 0.3455 1.1750
[GNV]
436 OBST# 436 TREE 0.2734 0.1252 0.2430
437 OBST# 437 TREE 0.4934 0.1860 0.4570
439 OBST# 439 FENCE 0.6672 0.3533 0.5660
440 OBST# 440 ANT ON HGR 5.2558 1.0090 5.1580
444 ANT ON STROBE LTD 3.0710 2.4787 1.8130
TWR!444










Table D-2-Continued
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
445 ANT ON STROBE LTD 4.1782 0.5015 4.1480
TWR!445
446 OBST# 446 TWR 0.7792 0.6001 0.4970
447 ROD ON STROBE LTD 0.8820 0.4154 0.7780
TWR!447
448 ANT ON STROBE LTD 0.9346 0.8788 -0.3180
TWR!448
449 OBST# 449 POLE 0.6199 0.5664 0.2520
451 OBST# 451 BLDG 0.4753 0.4739 -0.0360
452 OBST# 452 ANT 0.4710 0.4541 0.1250
453 OBST# 453 TRMSN POLE 0.4136 0.3998 0.1060
454 OBST# 454 FLGPL 0.3527 0.3413 -0.0890
455 OBST# 455 SIGN 0.3905 0.3887 -0.0370
456 OBST# 456 POLE 0.3595 0.3527 0.0700
457 OBST# 457 TRMSN POLE 0.2098 0.1981 -0.0690
458 OBST# 458 BLDG 0.2669 0.2647 0.0340
459 OBST# 459 POLE 0.4305 0.2548 0.3470
460 OBST# 460 POLE 0.5633 0.4960 0.2670
461 OBST# 461 ROD ON OL 1.2677 0.8150 -0.9710
AMOM
462 OL VORTAC!462 [GNV] 0.2436 0.2433 0.0110
"NCM"
463 OBST# 463 LT POLE 0.4518 0.4464 0.0700
464 OBST# 464 LT POLE 0.2907 0.2885 -0.0360
465 OBST# 465 FENCE 2.1580 0.4754 2.1050
466 OBST# 466 FENCE 2.0950 0.4854 2.0380
467 OBST# 467 TREE 0.5921 0.1529 0.5720
468 OBST# 468 TREE 0.6561 0.5885 0.2900
469 OBST# 469 TREE 3.1639 1.9871 2.4620
470 OBST# 470 TREE 0.8718 0.2959 0.8200
471 OBST# 471 TREE 0.6156 0.2974 0.5390
472 OBST# 472 TREE 0.2945 0.2942 0.0120
473 OBST# 473 TREE 0.2388 0.2319 0.0570
474 OBST# 474 TREE 0.5225 0.2786 0.4420
475 OBST# 475 HGR 0.1443 0.1000 -0.1040
476 OBST# 476 SIGN 1.1316 0.1392 1.1230
477 OBST# 477 FENCE 2.0282 0.1145 2.0250
478 OBST#478FLGPL 0.3542 0.2417 0.2590
479 OBST# 479 TREE 1.0034 0.2819 0.9630
480 OBST# 480 ANT ON BLDG ND ND ND









Table D-2-Continued
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
302 ANT ON OL ATCT!ATCT ND ND ND
FLOOR164
306 OL ON LTD WSK 0.7204 0.6439 0.3230
25 ROD ON OL APBN!APBN 1.1052 0.0963 1.1010
412 OL ON LOC!(28) 0.1876 0.1715 0.0760
415 TREE 0.7163 0.1267 0.7050
423 TREE 0.4329 0.2991 -0.3130
430 BLDG 0.0873 0.0830 0.0270
431 OL ON POLE 1.1387 1.1322 0.1220
402 ROD ON OL GS!(28) 0.7321 0.2950 0.6700
414 TREE 0.4697 0.4649 -0.0670
425 TREE 0.4524 0.3084 0.3310

RMSE: 1.23575
Accuracy: 2.42208
Percent Detected: 96.15
Search Radius: 3.
LIDAR data file:
C:\lChris\University of Florida\New AccuracyTest\TiltO DivW Fh750.alf

Configuration 3

Table D-3. Tilt: 10; Div: N; FH: 750
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
410 ROD ON OL ASOS!SENSOR 1.1040 0.5417 0.9620
[GNV]
436 OBST# 436 TREE 0.5105 0.4229 0.2860
437 OBST# 437 TREE 0.6040 0.5781 0.1750
439 OBST# 439 FENCE 0.1095 0.1007 0.0430
440 OBST# 440 ANT ON HGR 5.6905 1.2025 5.5620
444 ANT ON STROBE LTD 2.7029 1.9863 1.8330
TWR!444
445 ANT ON STROBE LTD 0.3955 0.3112 0.2440
TWR!445
446 OBST# 446 TWR 0.6567 0.5810 0.3060
447 ROD ON STROBE LTD ND ND ND
TWR!447
448 ANT ON STROBE LTD ND ND ND
_TWR!448_
449 OBST# 449 POLE 1.4932 0.6689 1.3350
451 OBST# 451 BLDG 0.3812 0.3608 -0.1230









Table D-3-Continued
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
452 OBST# 452 ANT 0.3050 0.2302 0.2000
453 OBST# 453 TRMSN POLE 0.6971 0.6507 -0.2500
454 OBST# 454 FLGPL 1.9684 0.7760 1.8090
455 OBST# 455 SIGN 0.4176 0.4117 0.0700
456 OBST# 456 POLE 0.4989 0.4201 0.2690
457 OBST# 457 TRMSN POLE 0.3086 0.3005 0.0700
458 OBST# 458 BLDG 0.3189 0.2759 0.1600
459 OBST# 459 POLE 0.5197 0.5179 -0.0430
460 OBST# 460 POLE 0.5767 0.5582 0.1450
461 OBST# 461 ROD ON OL 1.5547 0.6780 -1.3990
AMOM
462 OLVORTAC!462 [GNV] 0.4395 0.4306 0.0880
"NCM"
463 OBST# 463 LT POLE 0.6831 0.6300 0.2640
464 OBST# 464 LT POLE 0.4883 0.4879 0.0180
465 OBST# 465 FENCE 0.4544 0.3726 0.2600
466 OBST# 466 FENCE 0.3149 0.2441 0.1990
467 OBST# 467 TREE 0.5216 0.4710 0.2240
468 OBST# 468 TREE 0.4369 0.2401 0.3650
469 OBST# 469 TREE 3.5618 1.5864 3.1890
470 OBST# 470 TREE 1.0985 0.9451 0.5600
471 OBST# 471 TREE 1.1680 1.0357 0.5400
472 OBST# 472 TREE 0.1022 0.0918 0.0450
473 OBST# 473 TREE 0.3937 0.3317 0.2120
474 OBST# 474 TREE 0.5500 0.1703 0.5230
475 OBST# 475 HGR 0.2818 0.2816 -0.0110
476 OBST# 476 SIGN 0.7289 0.5342 0.4960
477 OBST# 477 FENCE 2.1520 0.4879 2.0960
478 OBST#478FLGPL 0.2563 0.2147 -0.1400
479 OBST# 479 TREE 0.8971 0.3124 0.8410
480 OBST# 480 ANT ON BLDG ND ND ND
302 ANT ON OL ATCT!ATCT 4.5412 2.9329 3.4670
FLOOR164
306 OL ON LTD WSK 0.9616 0.9113 0.3070
25 ROD ON OL APBN!APBN 1.1466 0.8629 0.7550
412 OL ON LOC!(28) 0.2434 0.2420 0.0260
415 TREE 0.8733 0.5832 0.6500
423 TREE 0.9269 0.8912 0.2550
430 BLDG 0.3583 0.3559 0.0420
431 OL ON POLE 1.7611 1.5335 0.8660
402 ROD ON OL GS!(28) 0.9555 0.3951 0.8700









Table D-3-Continued
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
414 TREE 0.3709 0.3443 0.1380
425 TREE 0.6098 0.4334 0.4290

RMSE: 1.23274
Accuracy: 2.41618
Percent Detected: 94.23
Search Radius: 3.
LIDAR data file:
C:\lChris\University of Florida\New AccuracyTest\TiltlO DivN Fh750.alf


Configuration 4

Table D-4. Tilt: 10; Div: W; FH: 750
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
410 ROD ON OL ASOS!SENSOR 1.4240 0.2658 1.3990
[GNV]
436 OBST# 436 TREE 0.3808 0.2281 0.3050
437 OBST# 437 TREE 0.4292 0.1792 0.3900
439 OBST# 439 FENCE 0.4839 0.2351 0.4230
440 OBST# 440 ANT ON HGR 5.7306 1.5677 5.5120
444 ANT ON STROBE LTD 2.6042 1.8733 1.8090
TWR!444
445 ANT ON STROBE LTD 3.9322 0.7131 3.8670
TWR!445
446 OBST# 446 TWR 0.4808 0.1185 0.4660
447 ROD ON STROBE LTD ND ND ND
TWR!447
448 ANT ON STROBE LTD 0.5122 0.4936 -0.1370
TWR!448
449 OBST# 449 POLE 0.3567 0.3561 0.0200
451 OBST# 451 BLDG 0.1087 0.1087 0.0020
452 OBST# 452 ANT 0.3851 0.3378 0.1850
453 OBST# 453 TRMSN POLE 0.1687 0.1646 -0.0370
454 OBST# 454 FLGPL 0.4268 0.3651 -0.2210
455 OBST# 455 SIGN 0.4655 0.3958 0.2450
456 OBST# 456 POLE 0.2723 0.1376 0.2350
457 OBST# 457 TRMSN POLE 0.2206 0.2180 -0.0340
458 OBST# 458 BLDG 0.2848 0.2474 0.1410
459 OBST# 459 POLE 0.3931 0.2211 0.3250
460 OBST# 460 POLE 0.5908 0.4311 0.4040









Table D-4-Continued
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
461 OBST# 461 ROD ON OL 1.5205 1.2753 -0.8280
AMOM
462 OL VORTAC!462 [GNV] 0.4436 0.4428 0.0270
"NCM"_
463 OBST# 463 LT POLE 0.4003 0.3989 0.0330
464 OBST# 464 LT POLE 0.4193 0.4163 -0.0500
465 OBST# 465 FENCE 1.9764 0.2461 1.9610
466 OBST# 466 FENCE 2.0964 0.1759 2.0890
467 OBST# 467 TREE 0.5538 0.2745 0.4810
468 OBST# 468 TREE 0.4668 0.1085 0.4540
469 OBST# 469 TREE 2.6613 0.9781 2.4750
470 OBST# 470 TREE 0.7176 0.4026 0.5940
471 OBST# 471 TREE 0.5720 0.4222 0.3860
472 OBST# 472 TREE 0.2171 0.1900 0.1050
473 OBST# 473 TREE 0.4238 0.4195 0.0600
474 OBST# 474 TREE 0.6157 0.1088 0.6060
475 OBST# 475 HGR 0.2662 0.2483 -0.0960
476 OBST# 476 SIGN 1.1811 0.3163 1.1380
477 OBST# 477 FENCE 2.1547 0.5161 2.0920
478 OBST# 478 FLGPL 0.6021 0.3248 0.5070
479 OBST# 479 TREE 0.9774 0.1670 0.9630
480 OBST# 480 ANT ON BLDG ND ND ND
302 ANT ON OL ATCT!ATCT ND ND ND
FLOOR164
306 OL ON LTD WSK 1.6682 0.6917 1.5180
25 ROD ON OL APBN!APBN 1.0391 0.3125 0.9910
412 OL ON LOC!(28) 0.2251 0.2080 -0.0860
415 TREE 0.3916 0.3163 0.2310
423 TREE 0.5273 0.5206 0.0840
430 BLDG 0.2182 0.2146 0.0390
431 OL ON POLE 1.2977 1.2662 0.2840
402 ROD ON OL GS!(28) 0.8698 0.3366 0.8020
414 TREE 0.0364 0.0232 0.0280
425 TREE 0.5761 0.3345 0.4690

RMSE: 1.27157
Accuracy: 2.49227
Percent Detected: 94.23
Search Radius: 3.
LIDAR data file:
C:\1Chris\University of Florida\New AccuracyTest\Tilt 10DivW Fh750.alf









Configuration 5


Table D-5. Tilt: 20; Div: N; FH: 750
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
410 ROD ON OL ASOS!SENSOR 1.4770 0.4645 1.4020
[GNV]
436 OBST# 436 TREE 0.2998 0.1561 0.2560
437 OBST# 437 TREE 0.3465 0.3084 0.1580
439 OBST# 439 FENCE 0.8246 0.7516 0.3390
440 OBST# 440 ANT ON HGR 1.1372 0.8141 0.7940
444 ANT ON STROBE LTD 2.1425 0.1665 2.1360
TWR!444
445 ANT ON STROBE LTD 2.3963 0.7251 2.2840
TWR!445
446 OBST# 446 TWR 0.1925 0.1916 0.0180
447 ROD ON STROBE LTD 0.6692 0.2838 0.6060
TWR!447
448 ANT ON STROBE LTD 0.6050 0.5348 -0.2830
TWR!448
449 OBST# 449 POLE 1.0022 0.6659 0.7490
451 OBST# 451 BLDG 0.5406 0.5002 -0.2050
452 OBST# 452 ANT 0.4515 0.4488 0.0500
453 OBST# 453 TRMSN POLE 0.1002 0.0707 -0.0710
454 OBST# 454 FLGPL 0.3412 0.2574 -0.2240
455 OBST# 455 SIGN 0.6802 0.6736 0.0940
456 OBST# 456 POLE 0.8395 0.8281 0.1380
457 OBST# 457 TRMSN POLE 0.5876 0.5758 -0.1170
458 OBST# 458 BLDG 0.5294 0.5285 0.0310
459 OBST# 459 POLE 0.7662 0.7295 0.2340
460 OBST# 460 POLE 0.9116 0.8446 0.3430
461 OBST# 461 ROD ON OL 1.0323 0.4967 -0.9050
AMOM
462 OL VORTAC!462 [GNV] 0.2815 0.2752 0.0590
"NCM"
463 OBST# 463 LT POLE 0.8736 0.8734 0.0190
464 OBST# 464 LT POLE 0.2213 0.1589 0.1540
465 OBST# 465 FENCE 0.5452 0.5380 0.0880
466 OBST# 466 FENCE 0.3493 0.3149 0.1510
467 OBST# 467 TREE 0.8078 0.3778 0.7140
468 OBST# 468 TREE 0.5312 0.2638 0.4610
469 OBST# 469 TREE 2.7547 2.1485 1.7240
470 OBST# 470 TREE 0.8340 0.3808 0.7420
471 OBST# 471 TREE 0.7738 0.6037 0.4840









Table D-5-Continued
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
472 OBST# 472 TREE 0.5147 0.3312 0.3940
473 OBST# 473 TREE 0.1867 0.0771 0.1700
474 OBST# 474 TREE 0.6312 0.4866 0.4020
475 OBST# 475 HGR 0.6480 0.6480 0.0000
476 OBST# 476 SIGN 1.2029 0.1372 1.1950
477 OBST# 477 FENCE 0.8258 0.6720 0.4800
478 OBST# 478 FLGPL 0.4588 0.3687 0.2730
479 OBST# 479 TREE 0.9437 0.8010 0.4990
480 OBST# 480 ANT ON BLDG 1.1148 0.8460 0.7260
302 ANT ON OL ATCT!ATCT 4.9239 2.9598 3.9350
FLOOR164
306 OL ON LTD WSK 0.4589 0.3718 0.2690
25 ROD ON OL APBN!APBN 1.2111 0.5001 1.1030
412 OL ON LOC!(28) 0.8603 0.8425 -0.1740
415 TREE 0.4401 0.4302 -0.0930
423 TREE 1.1912 0.9351 0.7380
430 BLDG 0.1622 0.1458 -0.0710
431 OL ON POLE 1.4131 1.3294 0.4790
402 ROD ON OL GS!(28) 0.9245 0.4948 0.7810
414 TREE 0.3673 0.3400 0.1390
425 TREE 0.6229 0.5864 -0.2100

RMSE: 0.881212
Accuracy: 1.72718
Percent Detected: 100.
Search Radius: 3.
LIDAR data file:
C:\lChris\University of Florida\New AccuracyTest\Tilt20 DivN Fh750.alf

Configuration 6

Table D-6. Tilt: 20; Div: W; FH: 1050
SPN Description Dist to Closest LIDAR 2D Dist Delta Elev
Pt (m) (m) (m)
410 ROD ON OL ASOS!SENSOR ND ND ND
[GNV]
436 OBST# 436 TREE 0.3413 0.0790 0.3320
437 OBST# 437 TREE 0.5860 0.2414 0.5340
439 OBST# 439 FENCE 0.6164 0.5565 0.2650
440 OBST# 440 ANT ON HGR 5.6915 0.3621 5.6800
444 ANT ON STROBE LTD 5.6582 1.2995 5.5070
TWR!444