|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 THE RELATIONSHIP BETWEEN THE RADIUS OF GALE FORCE WINDS AND THE RAIN SHIELD OF U NITED STATES LANDFALLING TROPICAL CYCLONES. By ERIN LEIGH BUNTING 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 2009
2 2009 Erin Leigh Bunting
3 To all who provided support and a shoulder to lean on
4 ACKNOWLEDGMENTS When entering gr aduate school everyone tells of the importance of picking the right committee. No wiser words were spoken and from experience I can say a committee is a sounding board, a sho ulder to lean on, and an invaluable source of knowledge. For their guidance and support I thank Dr Corene Matyas, Dr. Jane Southworth, and Dr. Peter Waylen. To Dr. Matyas, I enjoyed the fact that we grew and learned about this process together. I know I am the first of many successful graduate students to come. To my family, one could not as k for a greater support system We laugh together and grow together, I love every moment of it. Finally, to my fellow graduate students I thank you from the bottom of my heart. I came into this a shy, obsessive compulsive student and you taught me that it is alright to let your hair down and have fun. Thank you for taking me o ut of my comfort zone; I am a better person because of you.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 6 LIST OF FIGURES ................................ ................................ ................................ ......................... 7 LIST OF ABBREVIATIONS ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ ................................ ..... 9 CHAPTER 1 TROPICAL CYCLONE LITERATURE REVIEW ................................ ............................... 11 2 THE SPATIAL DISTRIBUTION OF THE TROPICAL CYCLONE RAIN SHIELD IN RELATION TO THE RADIUS OF GALE FORCE WINDS ................................ ............... 18 Introduction ................................ ................................ ................................ ............................. 18 Study Area ................................ ................................ ................................ .............................. 22 Data and Methods ................................ ................................ ................................ ................... 22 Data ................................ ................................ ................................ ................................ .. 23 Methods ................................ ................................ ................................ ........................... 24 Results ................................ ................................ ................................ ................................ ..... 27 Discussion ................................ ................................ ................................ ............................... 32 Conclusions ................................ ................................ ................................ ............................. 33 Fu ture Research ................................ ................................ ................................ ...................... 34 3 THE PLACE OF CLIMATOLOGY IN THE GEOGRPAHIC DISCIPLINE ....................... 46 Spatio Temporal Principles in the Geographic Discipline ................................ ..................... 46 New Methodologies Being Employed by Geographical Climatologists ................................ 48 LIST OF REFERENCES ................................ ................................ ................................ ............... 53 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ......... 57
6 LIST OF TABLES Table page 2 1 National Hurricane Center Operational Forecast Size Variables ................................ ............ 35 2 2 Tropical Cyclones from 1995 2006 Analyzed in Study. ................................ ......................... 36 2 3 Statistical properties of the distributions of the R17 size parameter ................................ ....... 37 2 4 Percentage of Rain Shield Within / Outside of R17 ................................ ................................ 38 2 5 Natural Break Table of Thre e SHIPS Variables That Commonly Affect Tropical Cyclone Size ................................ ................................ ................................ ...................... 39
7 LIST OF FIGURES Figure page 2 1 An example of the outer wind radii, Hurricane Charley (2004) ................................ .............. 41 2 2 Tropical Cyclones Studies fro m 1995 2006. Including Landfall Point, Track, and Analysis Point ................................ ................................ ................................ .................... 42 2 3 Percentage of Rain Shield within and Outside of R17 for All Storms Studied ....................... 43 2 4 Mean Percentage of Rain Shield Occurring Outside of R17 by Category. ............................. 44 2 5 Composition of the Rain Shield Occurring Outside of R17: Tropical Storms ........................ 45 2 6 Composition of the Rain Shield Occurring Outside of R17: Tropical Cyclone Cat 1 5 ......... 45
8 LIST OF ABBREVIATIONS DIR Storm Heading EBT Extended Best Track EYED Eye Diameter GIS Geographic Information System HURDAT Hurricanes Database MCP Minimum Central Pressure NCDC National Climatic Data Center NHC National Hurricane Center NWS National Weather Service POCI Pressure of the Outermost Closed Isobar R17 Radius of Gale Force Winds R26 Radius of Damaging Force Winds R33 Radius of Hurricane Force Winds R CLIPER Rainfall Climatology and Persistence Model ROCI Radius of the Outermost Closed Isobar RVMAX Radius of Maximum Winds SHIPS Statistical Hurricane Intensity Prediction Scheme SPEED Storm Speed TC Tropical Cyclone TRaP Tropical Rainfall Potential Model TS Tropical Storm WSR 88D Weather Surveillance Radar 1988 Doppler
9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Masters of Science THE RELATIONSHIP BETWEEN THE RADIUS OF GALE FORCE W INDS AND THE RAIN SHIELD OF UNITED STATES LANDFALLING TROPICAL CYCLONES. By Erin Leigh Bunting August 2009 Chair: Corene Matyas Major: Geography While TC related deaths can be attributed to strong winds, storm surge, or tornadoes; Rappaport (2000) found that the majority of TC related deaths in the U.S from 1970 1999 resulted from coastal and freshwater flooding caused by heavy rainfall. S ignifican t improvements have been made in forecasts of TC wind field and tracks yet there is much less known on rain shield dynamics. Currently, the National Hurricane Center (NHC) solely takes gale force wind extent into consideration when placing watches and war nings along the U.S. coast; meaning that the extent of the rain fall, one of the most deadly and destructive aspects of the TC, is not even inc luded in storm advisory reports This study analyzes the relationship between TC wind fields and the rain shield within the NHC o perational forecasting procedures, in order to understand how the rainfall extent aligns with the storm size variables calculated at the NHC. Specifically, the research question asks: does the extent of the rain shield align with the radius of gale force winds (R17) during the time at which gale force winds made landfall? Review of the literature and the NHC forecasting policies lead to the hypothesis that increased environmental forcings will lead to more asymmetrical storms and therefore a higher percentage of the rain shield occurring outside of R17. Forty two U S landfalling TCs from 1995 2006 were analyzed. Within the GIS an overlay of the average R17 and composite reflectivity radar
10 images was performed, resulting in area calculation s of rainfall within and outside R17. In tropical storms and weak hurricanes, typically less than 20% of the rain falls within the R17 boundary. Strong hurricanes generally contain more than 70% of the rain shield within the R17 boundary, because the stron g pressure gradient causes the tangential winds to increase in speed, therefore there is a tight wrapping of winds around the center of circulation and an axisymmetrical shape to the rain shield. Analysis of the composition of the rainfall occurring outsid e of R17 shows the majority (85 95%) of i ntense rainfall produced by TCs occurs within the radius of gale force winds for all storm strengths. In other words, heavy rainfall that may lead to flooding will not occur before the onset of gale force winds for most TCs.
11 CHAPTER 1 TROPICAL CYCLONE WIND AND RAINFALL PATTERNS Due to their substantial size and intensity, tropical cyclones (TCs) are the most destructive and costliest of atmospheric systems (Elsner and Kara 1999). A TC is the generic term used to describe a non frontal synoptic scale low pressure system with organ ized convection and cyclonic surfa ce wind circulation. These warm core systems are uniquely composed and thus there is a dynamic wind and precipitation pattern across the system. At the core there is an eye that ranges in diameter from 5 50 km where war m and calm winds are sinking to the ground (Frank 1977). Within the eye there is little deep convection; but surr ounding the eye is the eyewall, which is the most intense aspect of a TC Due to the conservation of angular momentum t he strongest winds and h eaviest rainfall occur within the eyewall, and are associated with vertically developed cumulonimbus clouds (Elsner and Kara 1999; Frank 1977). The other key components to the structure of a TC are the spiral rainbands areas with localized strong winds an d precipitation that wrap around the eye but extend hundreds of kilometers outward. The outer rain bands are composed of vertically developed thunderstorms that trail away from the eye in a spiral fashion (Willoughby 1988). Due mainly to the unique environ mentally dependent composition of TCs there are a wide variety of related hazards. In this study two of the most sever e TC related hazards are studied the winds and the precipitation, to determine if the two make landfall in conjunction with one another or if there is a significant time lag between wind field landfall time and rain fall arrival time Tropical cyclone winds and precipitation are responsi ble for the vast majority of TC related fatalities (Rappaport 2000). Deaths and damage associated with rainfall can be attributed to inland and coastal flooding; whereas wind related destruction can result from microbursts, tornadoes, or intense sustained winds that are naturally associated with TCs. TC rain and wind
12 climatologies are com mon in the literatu re. Rodgers (1981) study produced a TC rainfall climatology that looked at the characteristics and patterns of precipitation derived from satellite imagery. In their rainfall climatology Burpee and Black (1989) looked at the spatial extent of rainfall loca ted near the center of circulation of two U.S. landfalling TCs. Examples of wind field climatologies vary in the literature due to the different temporal and spatial scales analyzed. For example, Kimball and Mulekar (2004) examined the pattern of the outer wind radii in storms that originated in the North Atlantic Basin. The wind field variables utilized in Kimball and Mulekar's study are commonly used size parameters in NHC operational forecasts. In this study I am utilizing principles of TC rain and wind climatologies to examine how the TC wind fields and rainfall extent aligned just as the leading edge of the wind field makes landfall Specifically the research question asks: Does the extent of the rain shield coincide with the radius of gale force winds (R17) during the time at which gale force winds start to make landfall? The National Hurricane Center (NHC) is responsible for forecasting TC activity in the North Atlantic and East Pacific ocean basins. Part of the operational forecasting procedure at t he NHC involves warning the public about potential TC landfalls and giv ing the public a timeframe for TC arrival as to prepare their homes or evacuate. Coastal watches and warnings are issued 24 36 hours prior to TC landfall. A TC watch is issued by the NH C when TC conditions are possible in a specific area within the next 36 48 hours (Elsner and Kara 1999). TC warnings, on the othe r hand, indicate that severe TC related weather conditions are expected within 24 hours (Elsner and Kara 1999). The NHC must co nsider three factors when issuing watches and warnings (Sheets 1990). First, sufficient lead time must be provided for protection of life and to prevent property damage. Second, the extent of the warnings must be taken into careful consideration as to avoi d large government expenditures and to avoid complacency among
13 residents in the path of the storm. The final step in the warning process is to communicate the warning to the possibly effected communities in a manner designed to generate a timely response. As part of the forecasting parameters the NHC issues advisory reports that address the position, bearing, and other TC related hazards information. Climate research that utilizes the NHC data is common within the literature due to its reporting of numerou s TC hazards variables. model of TCs and Fran klin et al (2007) investigated positional data. Included in the storm advisor y report are several variables of storm speed, motion, and directionality. As seen in the excerpt below from Hurricane Katrina, direction and storm speed are two position variables reported in the advisory reports. In the case of Katrina, the storm was mov ing at 10 mph in a west northwest direction. In terms of information of TC related hazards there is a wide variety of information available in the advisory reports, including: maximum sustained wind speeds, minimum central pressure, tornado potential, esti mated storm surge, and estimated rainfall amounts. From the advisory report below we can see that Hurricane Katrina had a minimum central press ure of 935 mb and maximum sustained winds of 145 mph. The advisory report also gives general insight into the spatial extent of TC related winds. In terms of Hurricane Katrina, see below, the hurricane force winds extended 85 miles from the center of circulation wherea s the tropical storm force winds extend 185 miles from the center. Th ere is no note of the gale force wind / rain shield alignment within the NHC advisory reports. Total rainfall estimates are seen in the advisory reports and there is some regional defini tion to the rainfall maximum, Hurricane Katrina had an estimated rainfall total of 5 10 inches and the advisory report mention s how the rainfall was projected to fall along the G ulf C oast, but the size
14 of the rain shield and vulnerability of coastal region s are not addressed within the storm advisories. NORTHWEST NEAR 10 MPH. A GRADUAL TURN TOWARD THE NORTHWEST IS EXPECTED LATER TODAY. MAXIMUM SUSTAINED WINDS ARE NEAR 145 MPH WITH HIGHER GUSTS. KATRINA IS A CATEGORY FOUR HURRICANE ON THE SAFFIR SIMPSON SCALE. SOME STRENGTHENING IS FORECAST DURING THE NEXT 24 HOURS. HURRICANE FORCE WINDS EXTEND OUTWARD UP TO 85 MILES FROM THE CENTER...AND TROPICAL STORM FORCE WINDS EXTEND OUTWARD UP TO 185 MILES. COASTAL STORM SURGE FLOODING OF 15 TO 20 FEET ABOVE NORMAL TIDE LEVELS...LOCALLY AS HIGH AS 25 FEET ALONG WITH LARGE AND DANGEROUS BATTERING WAVES...CAN BE EXPECTED NEAR AND TO THE EAST OF WHERE THE CENTER MAKES LANDFALL. RAINFALL TOTALS OF 5 TO 10 INCHES...WITH ISOLATED MA XIMUM AMOUNTS OF 15 INCHES...ARE POSSIBLE ALONG THE PATH OF KATRINA ACROSS THE GULF COAST AND THE SOUTHEASTERN UNITED STATES. THE HURRICANE IS STILL EXPECTED TO PRODUCE ADDITIONAL RAINFALL AMOUNTS OF 2 TO 4 INCHES OVER EXTREME WESTERN CUBA...AND 1 TO 3 IN CHES OF RAINFALL IS EXPECTED OVER THE YUCATAN PENINSULA. ISOLATED TORNADOES WILL BE POSSIBLE BEGINNING SUNDAY EVENING OVER SOUTHERN PORTIONS OF LOUISIANA...MISSISSIPPI...AND ALABAMA...AND OVER THE FLORIDA PANHANDLE. REPEATING THE 4 AM CDT POSITION...25.4 N... 87.4 W. MOVEMENT TOWARD...WEST NORTHWEST NEAR 10 MPH. MAXIMUM SUSTAINED The spatial extent of TC winds and the uncertainty associated (Sheets 1990) with TC operational forecasts have led to w a tches and warnings being placed along broad sweeping portions of the coastline. The radius of gale force winds (R17), the radius of damaging force winds (R26), and the radius of hurricane force winds (R33) are TC size parameters that illustrate the spatial distribution of strong TC wind s and are collectively referred to as the outer wind radii (Kimball and Mulekar 2004). Each of these wind field variables is used in a different manner to understand the dynamics of the storm. In particular, the R17 radius is used to place watches and warnings along coastal areas. R17 is a measure of the spatial extent of 17.5 m/s winds (Kimball
15 and Mulekar 2004). Even though the R17 wind speeds are barely tropical storm strength, they are ideal for watches and warnings placement because it is one of the larger size variables and therefore covers a large proportion of the storm. In this study, the R17 variable is used to define the extent of the TC related winds because of its use in the N HC operational forecasting procedures. If a relationship exists between the wind field and rain shield, it may be possible to forecast the time of rainfall arrival within the scope of operational forecasting watches and warnings. Rainfall forecasts are no t as precise as those of winds, thus the NHC does not incorporate them into the operational forecasting procedure; but with TCs contributing 10 17% of global precipitation (Roth 2008), there is a need to better understand the distribution of rainfall asso ciated with TCs. Currently, the NHC employs the Hydrometeorolog ical Prediction Center (HPC) model for TC related precipitation forecasting The HPC gridded rainfall models provide estimates of the amount of precipitation that will occur as a result of a T C landfall, and the generalized region over which this rainfall is expected to occur. Instead of incorporating modeled rainfall estimates into the watches and warnings the NHC issues storm advisories with estimated rainfall amounts. These modeled rainfall estimates have no spatial or temporal relationship with the NHC operational forecasting framework; therefore gaining a sense of the relationship between the landfall time of rainfall and winds is difficult in the scope of TC watches and warnings. In this s tudy, we are not looking to improve upon rainfall models such as the baseline Rainfall Climatology and Persistence Model (R CLIPER). Instead we are looking to define the rain shield within the spatial parameters of operational forecasting variables, specif ically R17. Rappaport (2000) found that 82% of TC related deaths in the U.S. from 1970 1999 were attributed to water. This figure can be broken down further to show that rainfall and associated
16 flooding cause d 59% of TC related deaths; w hereas, storm su rge cause d 23% of TC related deaths. There is the potential for flood indu cing rainfall across all parts of TCs. But the core, and particularly the eyewall, is associated with the heaviest rainfall within a TC (Willoughby 1988) TCs can produce heavy rainf all throughout the storm, it is therefore important to classify the spatial extent of rainfall to improve rainfall forecasts and climatologies. The rain shield, or nearly continuous area of rain that typically become heavier as one approaches the eye (Sen n and Hiser 1959), is used to define the extent of the precipitation in this study. The size of the rain shield is highly variable due to a variety of environmental factors including: upper level wind shear, interaction with land, relative humidity, and th e intensity of the TC. The rain shield is composed of both stratiform and convective precipitation, which leads to different levels of cloud development and rainfall rates The outer edge of the rain shield is defined, in this study, through the use of radar reflec tivity returns which are commonly used in forecasting to illustrate the spatial extent of incoming precipitation (Matyas 2007; Matyas 2008). The radar sensor detects precipitation by measuring the strength of the electroma gnetic signal reflecte d back to it after passing through the low or middle levels of a storm. Radar reflectivity measures are converted to rainfall rates through the use of the Z R algorithm. Researchers employ the climatological Z R relationship to convert reflectivity values (unit decibels Z) to rainfall rates (R) (Marshall et al. 1948 ; Jorgensen and Willis 1982). Storms of tropical nature generally have higher rainfall rates compared to non tropical storms (Rosenfield 1993). Rosenfield (1993) updated the Z R relationship to i nclude a separate scale for storms of tropical origin. The updated algorithm (Z=250R 1.2 ) is used by the National Weather Service (NWS) and the NHC ( Rosenfield 1993 ). The tropical Z R relationship illustrates that exponentially more precipitation fall s from tropical systems than othe r storms. For
17 example, at 55 dB Z the climatological Z R estimates rainfall rates at 0.144 m mh 1 (5.68 in/hr) whereas the tropical Z R estimates rainfall at 0.385 m mh 1 (15.14 in/hr) In this study, the rain shield is defined as re gions with returning radar reflect ivity values greater than 20 dB Z (decibels). Values under 20 dB Z are generally considered no data ground noise, or light non accumulating precipitation whereas values at or greater than 20 dB Z mark the initial level of accumu lating rainfall. O ther climatol ogists have employed this threshold to define the edge of the rain shield (Jorgensen 1984) A gap in the literature and the need for a rainfall climatology that can be included in the NHC ope rational forecasts led to the following research question. Does the extent of the rain shield align with the radius of gale force winds (R17) during the ti me at which gale force winds mak e landfall? To form a hypothesis for this research question, I drew f rom four different areas of TC research: the NHC watches a n d warnings policies, literature on rain shields, st orm dynamics, and wind fields, i t is hypothesized that i ncreased environmental forcings lead to more asymmetrical storms and therefore a higher pe rcentage of the rain shield outside of R17. If there is a correlation between the rain shield and the R17 wind field, then the watches and warnings put in place by the NHC are sufficient to forecast TC precipitation arrival.
18 CHAPTER 2 THE SPATI AL DISTRIBUTION OF T HE TROPICAL CYCLONE RAIN SHIELD IN RELATION TO THE RADI US OF GALE FORCE WINDS Introduction Due to their substantial size and intensity, tropical cyclones (TCs) are the most destructive and costliest of the atmospheric systems (Elsner and Kara 1999). While TC related deaths can be attributed to strong winds, storm surge, or tornadoes; Rappaport (2000) found that the majority of TC related deaths in the U.S from 1970 1999 resulted from coastal and freshwater flooding caused by heavy rain fall. This is exemplified in Tropical Storm (TS) Amelia (1978), which produced 1220 mm (48 inches) of rainfall and caused the deaths of 30 people in southeast Texas ( Lawrence 1979 ); and Hurricane Floyd (1999), which produced 610mm (24 inches) of rainfall a nd caused 57 deaths across Florida and the Carolinas ( Atallah and Bosart 2003 ). While significant improvements have been made in forecasts of TC wind field and tracks (Franklin et al. 2003; Aberson 2001) much less is known about rain shield dynamics. Curr ently, the National Hurricane Center (NHC) solely considers gale force wind extent when placing watches and warnings along the U.S. coast; meaning that the extent of the rain fall, one of the most deadly and destructiv e aspects of the TC, is not considered when forecasting TC landfall location and conditions This study analyzes the relationship between the TC wind fields and the rain shield within the framework of the NHC operational forecasting procedures, in order to understand how the rainfall extent al igns with the storm size variables calculated at the NHC. Currently, t he NHC does not include localized rainfall maxima into the o perational forecasts. S torm advisories are issued every six hours with generalized precipitation totals but they do not inclu de information specific to when rainfall will co mmence in a given region. As storm summaries do not define the spatial extent of the leading edge of the precipitation, residents of coastal areas cannot ga u ge the arrival time of precipitation when looking at the TC
19 watch and warning maps. The generalized precipitation totals, from the storm summaries, are based on gridded model estimates of rainfall (Marchok et al 2006) Baseline gridded estimates are produc ed using the Rainfall Climatology and Persistence model (R CLIPER) (Marks 2002) or the Tropical Rainfall Potential model (TRaP) (Kidder et al. 2005). These rainfall models are not compatible with forecasts of storm size due to their use of standardized gri ds, therefore the modeled measurements of TC rainfall do not overlay with TC watches and warnings. When forecasting the track and intensity of TCs the NHC consults with national and state go vernment agencies (Sheets 1990) and takes into consideration atmos pheric dynamic and storm size variables derived from both objective and subjective climate / meteorologi cal models. Based on the models the NHC issues watches and warnings along the U.S. coastline based on the extent of gale force winds. The operational wind forecast procedure at the NHC involves estimati on of six TC size parameters: eye diameter (EYED), radi us of maximum winds (RVMAX), radius of gale force winds (17 m/s) (R17), radius of damaging force winds (26 m/s) (R26), radius of hurricane for ce wind s (33 m/s) (R33) and mean radius of the outer most closed isobar (ROCI) (Table 2 1) (Kimball and Mulekar, 2004; Pennington et al., 2000). Variables R17, R26 and R33 are TC size variables that illustrate the spatial distribution of strong winds and are coll ectively referred to as the outer wind radii (Kimball and Mulekar, 2004). The outer wind radii are distance measurements from the center of circulation to the edge of 17 ms 1 26 ms 1 and 33 ms 1 wind. Figure 2 1 depicts the relationship between the outer wind radii for Hurricane Charley (2004). Hurricane Charley (2004) had a R17 value of 125.5 km. TC size is typically measured by forecasters as the extent of R17 or ROCI (Merill 1984). To protect coastal residents, TC watches are put in place 36 hours pri or to landfall; TC warnings are issued 24 hours prior to TC landfall
20 when the forecast of the storm is more precise (Sheets 1990). The NHC places the watches and warnings based on the extent of gale force winds (R17) in order to forewarn as many residents within the largest wind swath associated with the storm (Merrill, 1984; Kimball and Mulekar, 2004). Rainfall forecasts are not as precise as those of wind and surge, because of the complex environment al forcings that can influence them ; but with TCs cont ributing 10 17% of global rainfall, there is a need to understand the distribution of rainfall associated with a TC (Roth 2008). In this study the spatial extent of the precipitatio n, or rain shield, is defined using radar reflectivity da ta. The rain shiel d, or nearly continuous area of rain that typically become s heavier as one approaches the eye, is composed of both stratiform and convective precipitation and is therefore associated with varying rainfall rates (Senn and Hiser 1959). A multitude of atmosph eric variables affect the shape and elongation of the rain shield includ ing shear, relative humidity, t rough interaction, and sea surface temperatures. Upper level vertical wind shear (850 200 mb) weakens the TC by interfering with the organized deep conve ction that occurs around the core and outer bands of the storm (Corboserio and Molinari 2003) Storm intensity also affects the extent of the rain shield. As a TC intensifies the minimum central pressure decreases leading to a steep pressure gradient from the ROCI to the eye (Thorpe 1985; 1986). A steep pressure gradient generates rapid, inward spiraling banding features and therefore a tighter wrapping of winds around the eye (Thorpe 1985; 1986; Willoughby et al 1984). This spatial pattern was seen in the secondary landfall of Hurricane Katrina (2005); the category three storm had a pressure difference from the outermost closed isobar (POCI) to the minimum central pressure (MCP) of 103mb and a rain shield that wa s axisymmetrical. Relative humidity, a measure of cloud saturation, can also influence the rain shield by gauging if entrainment of dry
21 air into the system is occuring Frank (1977) found that strong convective regions of TCs have much higher mean relative humidity values than the surrounding tropical atmosphere. Higher relative humidity values imply that entrainment does not hamper deep convection (Frank 1977). This m ean s that rain shields expand with increase relative humidity, and decrease with low relat ive humidity value because dry air entrainment inhibits the growth of the storm. This study looks at the relationship between the leading edge of a TC rain shield (as defined by radar reflectivity data) and the edge of R17. Specifically the research q uestion asks: To what degree does the rain shield align with the radius of gale force winds (R17) during the time at which gale force winds made landfall? If the two coincide the n the lead time for wind conditions can also be used as the lead time for prec ipitation. Previous research determined the spatial extent of TC precipitation by overlaying various sized grids on rainfall data and determining the quantity within each cell (Rao and MacArthur 1994; Rodger et al. 1994). In t his study takes a different ap proach in that rainfall quantities are not calculated. Instead, the percentages of rainfall within and outside of R17 are calculated in order to understand how the edge of R17 relates to the edge of the rain shield. Although rainfall climatologies have ana lyzed the pattern and type of precipitation within TCs (Frank 1977; Dvorak 1975; Corobosiero and Molinari 2002), and case studies of individual TCs have determined the impacts of TC related precipitation across the southeast U.S (Knight and Davis 2007), no ne of these have specifically addressed the differential pattern of the wind fields and rain shield. Specifically, quantifying how well the wind field (R17) contains the rain shield in U.S. landfalling TCs L iterature and th e NHC forecasting policies sugge st that increased environmental forcings (i.e. strong upper level wind shear, trough interaction, low relative humidity, or any other atmospheric variables that can
22 influence storm dynamics) lead to more asymmetrical storms and therefore a higher percentag e of the rain shield occurring outside of R17. Study Area From 1995 to 2006, fifty four TCs m ade landfall in the US. Of the 54 TCs, 30 were tropical storms with maximum sustained wind speeds less than 33 m/s and 24 were hurricanes with maximum sustained wind speeds greater than 33 m/s. This study analyzes 42 of the 54 (Table 2 2). As seen in figure 2 2 these systems made landfall across various parts of the southeast U.S, and were of varying degrees in strength. To be included, storms must have: 1) made landfall in the US, 2) gale force winds data for all four quadrants at the time R17 made landfall, and 3) long range composite reflectivity imagery available from the National Clima te Data Cente r (NCDC) Storms that made multiple landfalls in the U S were taken into special consideration I n order to include multiple landfalls into the study the TC needed to have sufficient time over warm ocean waters to reintensify. Arbitrarily, a minimum of six hours had to elapse between the first and second landfalls to be included in the study. The addition of multiple TCs landfall brought the number of storms analyzed up to 42. Data and Methods Geogra phic Information Systems (GIS ) enables spat ial analysis and statistics to be performed on TC wind fields and rain shield. Composite radar reflectivity imagery from the National Climatic Data Center (NCDC) and quadrant averaged R17 wind field data from the Extended Best Track (EBT) dataset are used to define the spatial extent of the TC related winds and precipitation within a GIS. Whereas the Statistical Hurricane Intensity Prediction Scheme (SHIPS) dataset is used to draw conclusions as to why the rain shield shape is highly variable. These dataset s are derived from multiple sources including: surface observations, aircraft
23 reconnaissance data, and satellite images (Moyer et al. 2007). Analysis of climatological TC datasets within a GIS allows for the spatial relationship of two of the most destruct ive TC related hazards to be analyzed and quantified. Data R adar reflectivity data (from WSR 88D stations) w ere obtained from the National Climatic Data Center (NCDC) for the site closest to the R17 landfall location (Klazura and Imy 1993), and the Java NEXRAD tool w ere used to convert the data in to a format utilized with a GIS (Ansari and Del Greco 2005). Level III long range composite reflectivity data were utilized to represent the rain shield as the ir spatial exte nt of 460 km is sufficient to depict the rain shield. Composite reflectivity depicts the maximum reflectivity found on any constant elevation angle (Klazura and Imy 1993). The radar reflectivity data depicts the amount of energy returned to the radar s sen sor in decibels of Z (dB Z ), with a range of values from 5 d B Z to 75 dB Z (Klazura and Imy 1993). This range of values represents light to heavy rainfall. Values below 20 dB Z generally represent ground clutter or a false radar signal. Because the portion of the atmosphere being sensed at the 460km distance is 16.8 km above the ground, only the tops of clouds at this distance from the radar site c an be detected This means that the deep convection present at 2 4 km from the ground will not be sensed at long di (1993) tropical Z R relationship is used to convert observed radar R relationship is used to convert radar reflec tivity date to rainfall rates for non tropical systems. Radar values of 20 dB Z correspond with 0.51 mmh 1 rainfall, the minimum rainfall detectable from radar reflectivity returns (Marshall et al 1947; 1948; Rosenfield 1993). Therefore, this study defines the rain shield as areas with radar reflectivity returns greater than or equal to 20 dB Z
24 The EBT dataset is an extension of the best track dataset (Neumann et al. 1999); compiled from various publications and represents a detailed post season analysis of North Atlantic basin originating TC intensities and tracks at six hourly intervals (Moyer et al. 2007 ). The newer version of the best track dataset includes TC size parameters (Kimball and Mulekar 2004). The EBT dataset of Pennington et al. (2000) was com piled from advisories issued by the NHC from 1988 to present (Moyer and Powell 2006). The NHC maintains position and intensity data for all Atlantic tropical cyclones since 1851; this dataset is referred to as HURDAT (Neumann et al. 1999). Storm heading ( DIR) and storm speed (SPEED) were extracted from this dataset to assist with th e interpolation of R17 data. When variables DIR or SPEED were not available in the HURDAT dataset, they were calculated by extracting the positional data from the NHC storm advisories and then calculating the storm speed and bearing using the Laake (2003) Excel function The SHIPS dataset, which is comprised of TC environmental data from 1982 2006 (DeMaria et al 2005), is used to determine why certain tropical cyclones have a high percentage of rainfall outside of R17. Included in this dataset are climatological, atmospheric environment, ocean, and storm properties variables. From this dataset 4 variables were extracted: storm category (a measure of intensity), generalized 850 200 mb shear (SHRG), 700 500 mb relative humidity, and the climatological sea surface temperature (CSST). While many environmental and atmospheric variables can impact the rain shield shape these four are commonly employed (Corbosiero and Molinari 2002). Methods The EBT and SHIPS data must be interpolated to R17 landfall time: (1) to avoid land surface interaction issues, (2) to fall in line with NHC watches and warning policies, and (3) to ga uge whether the winds and rain make landfall at the same time. To do this, first we must
25 determine the time of R17 landfall. T he 6 hourly position data were interpolate d to hourly positions beginning 24 hours prior to the landfall of the center of circulation At each hour, the distance between the circulation center and landfall location was calculated. The quadrant average d R17 radius was then subtracted from the distance calculation, and the time when the value was nearest 0 is utilized as R 17 landfall time. Table 2 3, illustrates the analysis time for each storm co mpared with the center of circulation landfall time. Note that R17 landfall occurred for most storm 12 18 hours prior to the center of circulation landfall. To quantify the spatial extent of the rain shield and compare it to the extent of R17, the rada r reflectivity data and R17 data were entered into a GIS. After the radar reflectivity data were converted into GIS format the rain shield must be defined. All radar refl ectivity values less than 20 dB Z were exported out of the composite reflectivity imag e ; l eaving radar returns with the greatest potential to produce accumulating rainfall, and thus the rain shield is defined as radar reflectivity region s greater than or equal to 20dB Z The center of circulation was then identified within the rain shield us ing the HURDAT location data ; this was done in order to produce the wind field buffer from the eye outward Next, within the GIS the R1 7 wind field extent with modeled and overlaid a top the rain shield using a buffer. Using the X Tools Pro 5.1 extension of GIS several area calculations were completed, including: the total area of the rain shield, the area of the rain shield occurring within the R17 wind field, and the area of the rain shield occurring outside of the R17 extent. The total area of the rain sh ield was calculated for two reasons: (1) to gauge the total storm rain shield extent and (2) so that the portion of the rainfall within and outside of R17 can later be calculated. Rain shield size s varied greatly thus simple area totals are insufficient to understand the wind/rain relationship in every storm. To convert the rain shield area totals to percentages the
26 attribute tables of each storm were exported to excel. The percentage of rainfall within and outside of R17 was then calculated. By using t he percentage area calculation s, issues of storm size are alleviated With a spatial extent of 460 km the radar reflectivity data do not always cover the back two quadrants of the storm Due to the lack of complete spatial coverage of the radar images the percentage of rainfall outside of R17 is used for the analysis as we are mostly concerned with the rain that is making landfall prior to the gale force wind landfall The percentage of the rain shield outside R17 is justifiable for analysis because resea rch has shown that the majority of rainfall occurs in the leading edge of the storm (Elsberry 2002). Since we are concerned with the arrival time of the precipitation and not the duration, the percentage of the rain shield area outside of R17 is a tenable calculation. Lastly, the type of precipitation occurring outside of R17 is defined This was done in order to determine whether those areas where the rain shield and wind field did not align were composed of light or heavy precipitation. This is importa nt because the leading edge of precipitation is often heavy and if it does not align with the R17 wind field which is used to place coastal watches and warnings, than coastal residents may not be prepared for potentially flooding precipitation. The precip itation was classified based on the radar reflectivity value. Radar reflectivity values r anging from 20 35 dB Z were extracted from the GIS layer and classified as light precipitation. The light precipitation layer consists of accumulating rainfall with rai nfall rates ranging from 0.51 mmh 1 to 8.38 mmh 1 The heavy precipitation layer consists of all ra dar polygons greater than 35 dB Z ; these radar reflectivity values are associated with intense rainfall rates. The area and percentage area of light and heavy rainfall areas was then calculated, in order to understand the composition of the rainfall occurring outside of R17.
27 Results The distribution of R17 is skewed to the right and has a large spread from the smallest to the largest R17 wind field radius. The mean R17 is 199.5 km with a standard deviation of 88.57 km. Kimball and Mulekar (2004) calculated mean R17 figure of 222.3 km and a standard deviation of 104.3 km respectively The differences are largely due to the different number of years analyzed between the two studies. Overall, the 42 TCs in this study had a range of R17 wind radii of 308.98 km. Table 2 3 lists the statistical properties of the R17 size parameter for all TCs analyzed. The smallest R17 wind field occurred during tropical storm Bonnie (2004). This TC had a R17 radius of 71.77 km, which was caused largely by strong southwesterly shear that weakened the storm 24 36 hours prior to landfall (Avila 2005). Tropical storm Isidore (2003) had the largest R17 radius of 380.74 R17 wind field is roughly one third the size of the largest and most intense TC on record, Supertyphoon Tip (1100 km) (Kimball and Mulekar 2004). A ll TCs have some portion of their rain shield located outside of R17 (Table 2 4) T he average TC had 57% of the rain field located outside of R17 This figure gives the general connotation that the wind field and rain shield do not coincide very well; but upon closer exam ination there are many TCs with rain shields well contained within the R17 wind field. Ther e are several extreme cases on both ends of the spectrum of note. Firstly, TC Helene (2000) had more than 99% of its rain shield outside of R17. This system became a tropical storm in the southeast Gulf of Mexico under marginal environmental condition s bu t the vertical wind shear increased causing the system to become asymmetrical with most of its deep convection, winds, eye (Franklin et al. 2001) TC Helene (2000) also had a relatively small gale for ce wind field of only 84 km, which can be attributed to the vast majority of the precipitation occurring outside of R17. On the other end of the spectrum
28 is Hurricane Wilma (2005) which had only 2.3% of the rain shield outside of R17 due largely t o the int ensity, the steep pressure gradient, and the large R17 wind field of 324.7km (Pasch et al. 2005). Hurricane Wilma (2005) had one of the largest R17 wind fields of the storms in this study S everal environmental forcings were examined (Table 2 5) ; v ertica l wind shear, relative humidity, sea surface temperature, and intensity play a large part in the composition of the wind and r ain shield and therefore are most likely to cause rain shield asymmetries. Table 2 5 reports these environmental variables by nat ural breaks in the data based on the percentage of the rain shield occurring outside of R17. The generalized wind shear value, extracted from the SHIPS dataset, indicates that TCs in the lower (25%) quartile had relatively low wind shear values that range from 6.98 16.91 ms 1 Whereas in the upper quartile vertical wind shear appears to have affected the shape of the TCs, these TCs (Tropical Storm Cindy 2005; Tropical Storm Bill 2003; Tropical Storm Gabrielle 2001; the first landfall of Tropical Storm Ern est 2006; Tropical Storm Bonnie 2004; Tropical Storm Helene 2000) had generalized vertical wind shear values ranging from 18.60 22.07 ms 1 ; within the upper quartile there appears to be one outlier, in terms of the shear values, Tropical Storm Barry (2001) had a low vertical wind shear value of 5.88ms 1 but 82.13% of the rain shield was outside of the R17 wind field. In terms of relative humidity and sea surface temperature there did not appear to be a direct link with the percentage of the rain shield outs ide of R17. In the upper quartile, several of the storms (Tropical Storm Cindy 2005; Tropical Storm Barry; Tropical Storm Bill 2003) had high relative humidity percentage but a consistent trend across all of the 42 storms was not observed. The final varia ble extracted from the SHIPS dataset is intensity. As seen in table 2 5 all of the storms in the upper two quartiles were weak TCs, Saffir Simpson category 1 or tropical
29 storms. Figure 2 4 which depicts the mean percentage of the rain shield outside of R1 7 by Saffir Simpson category, and indicates a possible relationship between the percentage of the rain shield outside of R17 and storm intensity. Tropical storms and weak hurricanes (category 1) on average had 73.13% and 66% of the rainfall outside of R17 (Figure 2 4 ) This corresponds to the hypothesis that increased environmental forcings will cause more of the rain shield to be outside of R17 because weaker TCs are weak due to various environmental conditions. TSs, in this study, varied greatly in the pe rcentage of the rain shield falling outside of R17; with TS Erin (1995) having only 15.44% of the rain shield outside of R17 to TS Helene (2000) which had 99.2% of the rain shield outside of R17. The mean percentage value associat ed with TS was 77.6%. Stor ms on or around the mean include: TS Josephine (1996) with 77.31%, the first landfall of Irene (1999) at 77.66%, and Charley (1998) with 77.95% of the rainfall outside of the wind measure. The five category 1 storms in this study had a range of percentage from 60.04% to 72.21%. Examples of Category 1 storms in this study include: Bertha (1996) with a percentage value of 61.35%, Earl (1998) with 64.11%, and Gordon (2000) with 71.75% of the rainfall outside of R17. As seen in figure 2 4, the mean percentage of rainfall occurring outside of R17 decreases with an increase in storm intensity The five category 2 hurricanes included in this study had a mean percentage value of 29.77%. This includes hurricanes such as Frances (2004) with 14.85%, and the 2 landfal ls of Georges with 29.38% and 34.47%. There is a marked decline in the mean percentages outside of R17 from category 1 hurricanes to category 2; meaning that more intense category 2 storms contain more rainfall within the R17 boundary. As stated in the hyp othesis, TCs that are strongly influenced by various environmental forcings will have more of the rain shield outside of R17. Intense storms are defined as hurricanes of category three or greater
30 (Simpson 1974). The spatial calculation of intense TCs falls in line with our hypothesis because these systems are either not easily influence d by forcings due to the stage of maturity or are contained within the ideal environmental. With a mean percentage outside of R17 of 25.71% category 3 storms are even better contained within the wind field. The four category 3 hurricanes in this study: Fran (1996), Bonnie (1998), Jeanne (2004), and Rita (2005) had 49.89%, 10.12%, 19.04%, and 23.78% o f their rainfall outside of R17 While the range of percentages appears to be large, the mean of the values illustrates the point that the wind field aligns well with the rain shield in more intense storms. Category 4 hurricanes had on average 36.99% of the rainfall outside of R17 The five category 4 hurricanes in this study inclu de: Opal (1995) at 39.05%, Ivan (2004) at10.55% outside, and the first landfall of Charley (2004) with 46 .65% of rainfall outside of R17. Through the percentage outside of R17 is higher for category 4 storms than categories 2 and 3 these storms are still b etter contained with R17 as compared to category 1 and tropical storms The higher mean and median values for category 4 hurricanes, as compared with other intense hurricanes, can be explained by further analysis of Hurricane Bret (1999). While Bret made l andfall as a strong category 4 hurricane its formation was greatly hindered due to high vertical wind shear caused by an upper level trough (Lawrence et al. 2001). Even though the trough dissipated and Bret intensified rapidly the asymmetrical shape remain ed, and thus a larger portion of rainfall was outside of R17 (Lawrence et al. 2001). The composition of the rain shield falling outside of R17 is also related to storm intensity. Figures 2 5 and 2 6 depict the composition of the rain shield outside of R17 for all storms in this study. From these figures we can see that the vast majority of the precipitation occurring outside of R17 is l ight precipitation (Equal or less than 35 dB Z ). In weaker TCs (TS and Category 1) more of the rain shield outside of R17 was comprised of heavy precipitation when compared to
31 more intense TCs where most of the heavy rainfall was contained within R17. This can be attributed to the fact that more of the TS rain shield falls outside of R17 and the asymmetrical nature of these systems. Figure 2 5 shows the composition of the rain shield outside of R17 for tropical storms From this figure we can see that the vast majority was light precipitation ( 90%), whereas only 10 % of the r ainfall was intense (greater than 35 dB Z ). Likewise, for category 1 TCs the average amount of intense precipitation occurring outside of R17 was 13.43%. With a percentage of intense rainfall outside of R17 at 21.7% the second landfall of Danny (1997) had the greatest amount of intense precipitation outsid e of R17 of the entire category 1 TCs. This was c aused by a wedging of the hurricane between a high pressure system and a mid troposphere trough that disrupted the TCs winds and rain fields (Rappaport et al. 1999) The percentages of intense rainfall occur ring outside of R17 drops quite substantially from weaker storms to intense storms, as seen in Figure 2 6 The average amount of intense rainfall outside of R17 for category 2 storms was 9.34%. Of these storms Frances (2004) and the first landfall of Georg es (1998) had roughly the same percentages of intense rainfall outside of R17, approximately 5%. The other three category 2 TCs had a rain shield that was composed of approximately 12% intense rainfall outside of R17. Five category three storms were analy zed in this study, and these storms had an average of 4.53% of intense rainfall outside of R17. The four category 3 storms in the study, which represent 10% of the overall storms, include: Hurricanes Fran (1996), Bonnie (1998), Jeanne (2004), and Rita (200 5).Of these storms Hurricane Bonnie (1998) had the highest percentage of intense rainfall outside of R17 with 6.5%. This system moved slowly and had ample time to strengthen over the Atlantic Ocean. T he weakening of steering currents brought it on shore n ear Cape Fear, North Carolin a ; but the compact shape of the rain shield is thought to be caused by the trough interaction just prior to landfall ( Pasch et al
32 2001) Included in the category 4 storms are Hurricane Opal (1995), Hurricane Bret (1999), the fir st landfall of Hurricane Charley (2004), Hurricane Ivan (2004), and Hurricane Dennis (2005). The lone category 5 storm was one of the deadliest TC s in the Atlantic Basin h istory. When Hurricane Katrina (2005) made landfall near the Mississippi / Louisian a border it was an intense category 3 storm, but at the time of R17 landfall it had yet to weaken and was a category 5 storm. The percentage of intense rainfall occurring outside of R17 for these storms i s 9.9%, higher than expected considering the trend seen in category 3 and 4 storms. This deviation from the trend is largely due to the first landfall of Hurricane Charley (2004). Outside of R17 the rain shield of Hurricane Charley (2004) was composed of roughly 23.4% intense rainfall; this is largely to t he high sea surface temperature s (29.1 C) and ideal atmospheric conditions (generalized vertical wind shear of 6.55 ms 1 ) that produced deep convection many kilometers away from the center of circulation (National Hurricane Center; Matyas 2009). Other cate gory 4 TCs fell in line with the trend previously seen; Hurricane Ivan (2004) and Hurricane Dennis (2005) both had a rain shield composition outside of R17 of only 2% intense precipitation. D iscussion The R17 variable, a wind field of minimal tropical st orm strength, is the primary size variable utilized by the NHC when coastal watches and warnings are issued (Kimball and Mulekar 2004; Sheets 1990). Therefore, the greatest winds and storm hazar ds, including precipitation, might be expected to oc cur within this area The fact that 6 0% of TC related deaths result from precipitation and the associated inlan d and coastal flooding, indicates a need to understand how sufficient the wind size variable (R17) is in gauging the e xtent of other TC related hazards. This is especially true considering the fact that TC related flooding has been documented hundreds of kilometers away from the center of circulation, in areas that may not have been in the TC warning zone and therefore not prepared for this h azard
33 Various environmental forcings cause the elongation and asymmetry of the rain shield, particularly the pressure gradient (intensity), vertical wind shear, relative humidity, and sea surface temperatures. P reliminarily analysis show s that the rain s hield and wind fields are not well aligned for weak TCs (tropical storms and category 1 TCs); meaning that operational forecasts would not be sufficient for forecasting flood inducing precipitation. On the other hand, more intense TCs (category 2 storms or greater) were fairly well contained within the R17 wind field; due largely to the intense pressure gradient that results in a tight wrapping of winds around the eye and a general compactness of the storm. This means that watches and warnings issued by the NHC are sufficient for forecasting the precipitation arrival. Given the destructive aspect of these intense systems these findings are good news for coastal residents who would have ample time to prepare / evacuate before TC related winds and heavy rainf all make landfall. A secondary component of this study looked at the composition of the rainfall occurring outside of R17. Areas classified as light rainfall have little chan ce of producing intense rainfall and flooding. Heavy precipitation, on the other hand, is capable of producing a rainfall rate of 21.59mmh 1 or higher. These heavy precipitation regions have the greatest potential for flooding to occur. The spatial analysis shows that the vast majority of the rainfall occurring outside of R17 is light precipitation; with more intense storms having the lowest percentage of intense precipitation outside of R17 when compared with weaker systems. The radar composition analysis gives great insight into NHC watches and warnings and how well such forecasts ar e at defining the onset of winds and precipitation While the rain shield is not well contained within the R17 field the rainfall typically occurring outside of R17 is light precipitation Conclusions In TS and category 1 H urricanes, a large portion (on ave rage around 70%) of the rain shield is located outside of R17
34 Intense storms, which are either mature or contained within an ideal environment, on average have approximately 30% of the rain shield outside R17. This means that the R17 wind field and rain shield for the most part coincide, therefore the NHC operational forecasts of regions likely to experience tropical storm force winds are also sufficient to forecasts regions likely to experience TC related rainfall. Of the rain shield occurring outside of R17 the composition is mostly light precipitation that is not likely to produce flooding The NHC watches and warnings are sufficient for forecasting the landfal l of intense (potentially flood inducing) precipitation in most landfalling TCs Exceptions occur when there is a weak asymmetrical TS with an elongated rain shield, or when upper level vertical wind shear values are high and therefore the rain shield is elongated. Future Research This project lays the foreground for a larger rainfall climatol ogy project. Future research will examine the radius of the outermost closed isobar (ROCI), the radius of damaging force winds (R26), and the radius of hurri cane force winds (R33) and the spatial variation / relationship with the rain shield (Table 2.1) This type of research will require a similar methodology as what was used in this study; it will differ in that we will compare the edge of the rain shield to the edge of the other size variables such a ROCI Other future research goals include: calculatin g the landfall time of the TC rain shield, which would closely relate to the timing of TC watches and warnings; and correlating the effects of vertical wind shear, storm motion, and other parameters to the rainfall distribution.
35 Table 2 1 National Hurricane Center Operational Forecast Size Variables Abbreviation Meaning Definition Wind Measure (ms 1 ) EYED Eye Diameter A measure of the calmest part of the storm. Surrounding the eye is the eye wall, which is the most violent part of a hurricane NA RVMAX Radius of Maximum Winds A measure of the distance from the center of hurricane circulation to the band of strongest winds NA R17 Radius of Gale Force Winds A wind field measure slightly above tropical storm force (39.15 mph). Wind measure used as the baseline for TC advisories. 17.5 R26 Radius of Damaging Force Winds Tropical Storm force wind measure at or above 57.49mph. 25.7 R33 Radius of Hurricane Force Winds A measure of the extent of hurricane winds (74mph) 32.9 ROCI Radius of the Outermost Isobar A measure from the center of circulation to the outermost closed isobar. NA Source: Kimball and Mulekar 2004
36 Table 2 2 Tropical Cyclones from 1995 2006 Analyzed in Study Name Year Category at Eye Landfall Category at R17 Landfall Hours from R17 Landfall to Eye Landfall Erin L1 1995 1 1 7.25 Erin L3 1995 1 Tropical Storm 8 Opal 1995 3 4 9.5 Bertha 1996 2 1 10.75 Fran 1996 3 3 15.25 Josephine 1996 Tropical Storm Tropical Storm 3.5 Danny L2 1997 1 1 18.75 Bonnie 1998 3 3 21 Charley 1998 Tropical Storm Tropical Storm 10.75 Earl 1998 1 1 7 Frances 1998 Tropical Storm Tropical Storm 17 Georges L1 1998 2 2 9.5 Georges L2 1998 2 2 20.75 Mitch 1998 Tropical Storm Tropical Storm 8 Bret 1999 3 4 10.75 Dennis 1999 Tropical Storm Tropical Storm 11.5 Floyd 1999 2 2 11 Irene L1 1999 1 Tropical Storm 10.25 Gordon 2000 Tropical Storm 1 15.5 Helene 2000 Tropical Storm Tropical Storm 14.5 Barry 2001 Tropical Storm Tropical Storm 7 Gabrielle 2001 Tropical Storm Tropical Storm 5.5 Eduardo 2002 Tropical Storm Tropical Storm 7.25 Isidore 2002 Tropical Storm Tropical Storm 19.25 Bill 2003 Tropical Storm Tropical Storm 9 Bonnie 2004 Tropical Storm Tropical Storm 2 Charley L1 2004 4 4 3.75 Charley L3 2004 1 1 2.5 Frances 2004 2 2 15 Gaston 2004 1 Tropical Storm 6.5 Ivan 2004 3 4 19 Jeanne 2004 3 3 11.5 Cindy L1 2005 1 Tropical Storm 8.75 Cindy L2 2005 Tropical Storm Tropical Storm 5 Dennis 2005 3 4 9.75 Katrina L1 2005 3 4 9 Katrina L3 2005 3 5 19.75 Rita 2005 3 3 16 Wilma 2005 3 2 10 Alberto 2006 Tropical Storm Tropical Storm 8 Ernesto L1 2006 Tropical Storm Tropical Storm 5 Ernesto L3 2006 Tropical Storm Tropical Storm 5
37 Table 2 3. Statistical properties of the distributions of the R17 size parameter R17 (km) Quartiles Maximum 380.74 90% 324.75 75% 266.22 Median 172.5 25% 129.64 10% 84.5 Minimum 71.77 Moments Mean 199.5 Standard Deviation 88.57 No. of Records 42
38 Table 2 4 Percentage of Rain Shield Within / Outside of R17 Name Year Category (R17) Percentage Area Outside 1 Wilma 2005 2 2.27 2 Bonnie 1998 3 10.12 3 Ivan 2004 4 10.55 4 Frances 2004 2 14.85 5 Erin L3 1995 Tropical Storm 15.44 6 Isidore 2002 Tropical Storm 17.42 7 Jeanne 2004 3 19.04 8 Rita 2005 3 23.78 9 Georges L1 1998 2 29.38 10 Katrina L2 2005 5 29.69 11 Georges L2 1998 2 34.47 12 Opal 1995 4 39.05 13 Dennis 2005 4 41.75 14 Charley L1 2004 4 46.65 15 Bret 1999 4 47 16 Fran 1996 3 49.89 17 Frances 1998 Tropical Storm 52.73 18 Erin L1 1995 1 60.04 19 Dennis 1999 Tropical Storm 61.29 20 Bertha 1996 1 61.35 21 Earl 1998 1 64.11 22 Danny L2 1997 1 66.55 23 Floyd 1999 2 67.87 24 Gordon 2000 1 71.75 25 Charley L3 2004 1 72.21 26 Mitch 1998 Tropical Storm 72.52 27 Ernesto L3 2006 Tropical Storm 74.08 28 Josephine 1996 Tropical Storm 75.00 29 Gaston 2004 Tropical Storm 76.69 30 Katrina L1 2005 Tropical Storm 76.89 31 Cindy L2 2005 Tropical Storm 77.03 32 Irene L1 1999 Tropical Storm 77.66 33 Charley 1998 Tropical Storm 77.95 34 Cindy L1 2005 Tropical Storm 78.49 35 Alberto 2006 Tropical Storm 79.07 36 Barry 2001 Tropical Storm 82.13 37 Bill 2003 Tropical Storm 84.22 38 Gabrielle 2001 Tropical Storm 84.35 39 Ernesto L1 2006 Tropical Storm 84.42 40 Edouard 2002 Tropical Storm 91.98 41 Bonnie 2004 Tropical Storm 94.88 42 Helene 2000 Tropical Storm 99.21
39 Table 2 5. Natural Break Table of Three SHIPS Variables That Commonly Affect Tropical Cyclone Size Name Year Intensity R17 (km) Percentage Area Outside Shear (ms 1 ) Relative Humidity (%) CSST (C) Wilma 2005 2 324.748 2.27 12.18 0.56 27.77 Bonnie 1998 3 282.82 10.12 11.78 0.61 29.5 Ivan 2004 4 367.18 10.55 7.29 0.56 28.6 Frances 2004 2 285.32 14.85 13.28 0.69 28.97 Erin L3 1995 TS 149.7 15.44 6.98 0.57 29.5 Isidore 2002 TS 380.74 17.42 14.87 0.6 28.41 Jeanne 2004 3 246.55 19.04 13.43 0.39 28.4 Rita 2005 3 282.43 23.78 16.91 0.49 28.16 Georges L1 1998 2 208.54 29.38 9.47 0.49 29 Katrina L2 2005 5 320.6 29.69 10.93 0.71 29.1 Georges L2 1998 2 214.72 34.47 9.91 0.47 27.74 Opal 1995 4 349.18 39.05 22.50 0.49 27.56 Dennis 2005 4 274.58 41.75 16.28 0.66 28.83 Charley L1 2004 4 125.498 46.65 19.53 0.48 29.67 Bret 1999 4 143.34 47.00 8.70 0.51 29.3 Fran 1996 3 216.02 49.89 11.71 0.57 ~ Frances 1998 TS 337.22 52.73 13.28 0.69 28.97 Erin L1 1995 1 194.46 60.04 7.50 0.52 29.13 Dennis 1999 TS 180.57 61.29 7.69 0.45 27.14 Bertha 1996 1 266.225 61.35 14.14 0.58 27.59 Earl 1998 1 148.55 64.11 23.24 0.44 28.92 Danny L2 1997 1 137.1 66.55 5.25 0.47 30.5 Floyd 1999 2 325.26 67.87 23.55 0.49 28.13 Gordon 2000 1 209.314 71.75 18.29 0.42 28.5 Charley L3 2004 1 118.065 72.21 ~ ~ 28.63 Mitch 1998 TS 263.91 72.52 ~ ~ 26.5 Ernesto L3 2006 TS 144.65 74.08 18.26 0.44 ~ Josephine 1996 TS 132.53 75.00 23.15 0.62 27.7 Gaston 2004 TS 84.495 76.69 15.24 0.57 28.3 Katrina L1 2005 TS 98.4 76.89 8.12 0.68 29.13 Cindy L2 2005 TS 129.64 77.03 15.06 0.5 ~ Irene L1 1999 TS 153.94 77.66 8.70 0.51 29.3 Charley 1998 TS 223.98 77.95 6.55 0.4 29.1
40 Table 2 5 Continued. Name Year Intensity R17 (km) Percentage Area Outside Shear (ms 1 ) Relative Humidity (%) CSST (C) Cindy L1 2005 TS 126.35 78.49 19.59 0.61 28.73 Alberto 2006 TS 143.148 79.07 ~ 28.5 Barry 2001 TS 138.13 82.13 5.88 0.66 28.41 Bill 2003 TS 164.365 84.22 18.60 0.69 28.33 Gabrielle 2001 TS 162.05 84.35 20.42 0.52 28.83 Ernesto L1 2006 TS 94.915 84.42 19.96 ~ ~ Edouard 2002 TS 72.3 91.98 21.20 0.57 ~ Bonnie 2004 TS 71.765 94.88 22.07 0.39 29.5 Helene 2000 TS 83.97 99.21 21.61 ~ ~
41 Figure 2 1. An example of the outer wind radii, Hurricane Charley (2004)
42 Figure 2 2. Tropical Cyclones Studies from 1995 2006. Including Landfall Point, Track, and Analysis Point
43 Figure 2 3. Percentage of Rain Shield within and Outside of R17 for All Storms Studied Storms in the study
44 Figure 2 4. Mean Percentage of Rain Shield Occurring Outside of R17 by Category. n=42
45 Figure 2 5. Composition of the Rain Shield Occurring Outside of R17: Tropical Storms Figure 2 6. Composition of the Rain Shield Occurring Outside of R17: Tropical Cyclone Cat 1 5 0 10 20 30 40 50 60 70 80 90 100 Composition of Radar Reflectivity Outside R17 Tropical Storms Percentage Area Outside R17 Less than 40 dBZ Percentage Area Outside R17 Greater Than 40 dBZ 0 10 20 30 40 50 60 70 80 90 100 Composition of Radar Reflectivity Outside R17 Hurricanes Percentage Area Outside R17 Less Than 40 dBZ Percentage Area Outside R17 Greater Than 40 dBZ
46 CHAPTER 3 THE PLACE OF CLIMATOLOGY IN THE GEOGR AP HIC DISCIPLINE Spatio Temporal Principles in the Geographic Discipline As a spatial discipline, geography seeks to understand patterns on the landscape and the processes which create them from both a human and environmental perspective (Holt Jensen 1999). The spatial and temporal principles of geography make it unique from ot her disciplines and allow research to be conducted at various scales and extents (Matthews and Herbert 2004). This chapter focuses on the varying effects of scale on climate research and the impacts of my research on geography and atmospheric sciences. The intent of the first section is to define the role of spatial and temporal theories within climate research, particularly my own research. The intent of the second section of this chapter is to gage what was original about my research, and how such rainfal l climatologies contribute to the academic community. Scale is about size, whether it be relative or absolute. The spatial aspect of geography refers to the distribution and relationship of features across space. Individual geographers tend to emphasize differing aspects of the spatial continuum. For example, physical geographers study the distribution of climatic patterns, vegetation, soils, landforms, and a multitude of other ctions on research are alleviated and therefore research conducted within the geographic discipline can examine various environmental / human variables are multiple scales (Carleton 1999). This change in spatial scale is exemplified in contemporary climate change research, as regional climate models (RegCM) incorporate multiple environmental, oceanic, and atmospheric boundary layers to model how variously scaled atmospheric events will impact a region. Geography is a discipline of diversity, under whose spa tial umbrella we study and analyze processes, systems, behaviors,
47 and countless other phenomena that have a spatial expression. The spatial interests in patterns, distribution, and interactions unite geographers across all subfields. Within the geographi c subfield of climatology research is conducted at various spatial scales. In my study, syn optic scale TCs are analyzed to examine the patterns of their associated winds and precipitation. Examples of varying spatial scales in the subfield of climatology are abundant including: global scale events like the trades winds synoptic scale events which include tropical cyclone and midd le latitude cyclones, mesoscale events like the tornad oes and thunderstorms, and micro scale events like microbursts or dust dev ils. The spatial scale of research for both meteorology and climatology range from the global scale to the micro scale (Thompson and Perry 1997; Barry and Carleton 2001; Carleton 1999). Climatologists approach extreme weather research by looking at the gen eral patterns of multiple storms over multiple years, by studying storms of various size and over a long period of time general conclusions can be reached on the atmospheric physics and dynamics involved on the storms life cycle. Space and time are inevit ably connected. This relationship can be seen simply as a frame of reference within which mechanisms become processes that cause change in differing phe nomena (Richards et al 2004). All geographic phenomena vary as a function of time. Consequently, one mus t consider the rate at which geographic phenomena change. Dependent on the geographic subfield the temporal scale of research will vary. It is vital to examine and understand the temporal scale of research being undertaken by a geographer, because it can mean the difference between finding correlations between events or not. Temporal resolution is particularly interesting within the context of aerial photography, satellite imagery, and other remotely sensed data. Most applications within geography do not require an extremely fine temporal resolution. In terms of temporal resolution meteorologists
48 and climatologists use some of the most fine scale data available. Through the use of hourly or sub hourly radar or other remotely sensed images climatologists and meteorologists can examine alterations to the life cycle and dynamics of events such as tropical cyclones or middle latitude cyclones. Climatology and meteorology have long been associated with one another due to their common studies o f atmospheric d ynamics. The major difference in the two fields being that meteorology is very fine scale temporally, and climatology deals with the spatial and temporal generalization of weather so that longer term trends can be recognized. By examining variously scaled atmospheric ev ents, climatologists can focus on the trends and driving forces associated with weather patterns. Whereas, meteorology focuses on the day to day or event moment to moment changes in weather systems or patterns. As studies on the long term t rends of weather have been used to implement forecasting procedures, continued climate research helps fine tune such pro cedures. While weather extremes exist and can pose a challenge to forecasters (i.e. Hurricane Katrina), the general patterns of weather are better for predictive models as they are the conditions that occur on most days New Methodologies Being Employed by Geographical Climatologists In 2006, the NCDC released software which transforms WSR 88D radar data into GIS shapefile format (Ansar i and Del Greco 2005). Radar has assisted weather predictions for over forty years, but its use in research, particularly in the geographic discipline, is much more recent (Krajewski and Smith 2002; Xie e t al 2005). Radar data, which are effective at depic ting storm size and rainfall, allow for many new and different types of climate research to be conducted. NEXRAD radar products have been used to analyze the statistical characterization of extreme rainfall frequency, for validation of satellite remote sen sing data, and to examine the dynamics of rainfall distribution (Krajewski and Smith 2002; Habib and Krajewski 2002; Xie et al. 2005).
49 The inability of traditionally use d climate data, like rain gauges, to adequately depict the spatial distribution of rain fall volume is why climatologists and meteorologists have looked to radar as an alternative tool to obtain spatially accurate rainfall data. The spatial distribution of radar over a large area, in my study 460 km 2 provides continuous data across the radar radar gives researchers a way to see between the rain gauges, and to interpret the type of rainfall (stratiform or convective) occurring in a region. Since 2006, more atmospheric research is utilizing radar data within a GIS to model the rainfall patterns What makes my research unique from other studies that used radar and GIS is the integration of NHC operational forecasting size variables. These forecasting procedures helped to define the extent of the wind field therefor e we can examine how the rain shield interacts with previous defined forecasting size variables Previous studies examining storm size including Kimball and Mulekar ( 2004 ) study utilized statistics and grid overlays to understand the relationship between storm size and storm development. In my study the GIS model was built to examine the relationship between the rain shield and wind field just as R17 win ds started to make landfall. The research methods developed for this study allow for other size variables, including ROCI and R 2 6, to be studied in a similar way; thus giving us a better understanding of the complex rain / wind relationship. My research on the rain shield and wind field interaction showed that in weak TCs the vast majority of the ra in shield fell outside of R17; whereas in stronger storms (category 3 or greater) the R17 boundary better contained the rain shield. Intense storms are well contained within R17 because the strong pressure gradient causes the tangential winds to increase i n speed, therefore there is a tight wrapping of winds around the center of circulation and an axisymmetrical shape to the rain shield These conclusions were reached by calculating the area
50 of the whole rain shield within and outside of R17. Based on these calculations the current NHC operational wind forecasts are insufficient for forecasting the landfall of TC related precipitation. But analysis of the composition of the rainfall occurring outside of R17 shows that watches and warnings are not necessarily insufficient for forecasting TC rainfall The majority of i ntense rainfall produced by TCs occurs within the radius of gale force winds for all storm strengths. In other words, heavy rainfall that may lead to flooding will not occur before the onset of g ale force winds for most TCs. This mean s that NHC wind forecasts are sufficient for forecasting the arrival time of intense precipitation. Hewitt (1997) argued that disasters are quintessentially geographic phenomena due to the scope and involvement of m ultiple environmental factors. Within the geographic discipline one of the largest areas of research involves natural hazards. This area of study looks to understand the climatological, hydrological, and geomorphological trends of naturally occurring extre mes (Matthews and Herbert 2004). Research on natural hazards involves but is not limited to: monitoring hazards, developing management practices for both the short and long terms, and understanding how humans are agents of change (Alexander 2004). While ex tensive research has been done on natural hazards, particularly tropical cyclone s there is still much unknown about these events This study contributes, in a small manner, to the geographic discipline due to its methodologies, technologies employed, and links to scale. Within the scope of my research there were multiple dataset s employed which are rarely used in the geographic discipline. As stated above, bringing radar images into a GIS is a relatively new process but one that is quickly growing in popu larity. Other datasets such as the EBT, HURDAT, and SHIPS are traditionally used for statistical atmospheric research; but in this study t hese datasets are visualized and studied within a GIS which is a relatively new
51 development to the geographic discipli ne. The use of the aforementioned datasets and radar reflectivity data were necessary for this study because we wanted to produce a rainfall climatology that aligns with the NHC ope rational forecasting guidelines This study contributes to geography a new perspective on datasets, a better understanding of the spatial relationship between TC wind fields and rain shield, and new methodologies for modeling the spatial relationship of TC size variables. Traditionally, climatological research has involved the u se of grids and statistics to quantify the relationship among climate variables. As a result of being associated with geography, climate research methodologies are expanding and coming up to date with geographic technologies and spatial statistics. GIS, re mote sensing, and other geographic technologies are relatively new to climatology but are potentially very powerful. One of the f i r st publications of GIS application in climate research was provided by Shipley and Graffman (1999); they used GIS to model cl imate data collected by the Air Force (Shipley et al. 2000). The implications of GIS in climate research vary from new modeling techniques to new analysis methods that involve spatial statistics. Climate data can be displayed in a GIS in a variety of ways; lightning strikes as point vector data, radar reflectivity data as vector or raster layers, an d isolines layers of pressure or temperature are common examples. Direct meteorological observations or indirect observations provide different thematic layers o f information that can be useful parameters for describing the varying state of the atmosphere. GIS and spatial statistics allow us to build models and calculate the neighborhood statistics, variance, and density on datasets in a manor rarely done before. There is so much not yet know n about tropical cyclones that geographic technologies could shine new light on. This baseline study exemplifies the use of GIS and radar to analyze
52 atmospheric hazards. From this project, multiple new research questions were developed, all of which include radar data and GIS methodologies. Many feel that the integration of radar into a GIS would greatly impact the NHC and NWS forecasting policies (Shipley et al. 2000; Schultz and Reeves 1999), which could lead to more precise forecasting procedures and more time for residents to prepare for severe weather. The use of datase ts such as the Hurricane Database (HURDAT; Jarvinen et al 1984 ), Extended Best Track (EBT; DeMaria et al. 2005 ), and the Statistical Hurricane Intensity Pred iction Scheme (SHIPS; DeMaria et al 1999 ) are crucial to climatological studies because they contain an abundance of data on TC size, atmospheric conditions, and environmental variables. By applying these datasets in a geographic framework, new models can be created, comparison of storm variables can be done with more accuracy, and spatial statistics can be applied to the datasets. Future research will examine the radius of the outermost closed isobar (ROCI), the radius of damaging force winds (R26), and th e radius of hurri cane force winds (R33) and their spatial variation / relationship with the rain shield. To take this study one step further, a comparison of the rain shield composition will be done to study the convective and stratiform precipitation patt ern within and outside of R17. It is anticipated that this study will lay the framework for a larger research project which will produce a rainfall climatology that examines all the size variables utilized by the NH C when forecasting for synoptic scale eve nts such as tropical cyclones.
53 LIST OF REFERENCES Aberson, S. D. 2001: The Ensemble of Tropical Cyclone Track Forecasting Models in the North Atlantic Basin (1976 2000) Bull. American Meteorological Society 82, 1895 1904. Alexander, D. E. 2004: Natur al Hazards on an Unquiet Earth. Unifying Geography: Common Heritage, Shared Future by J. A. Matthews and D. T. Herbert, 266 282. New York: Routledge, Taylor, and Francis Group Ansari, S., and S. Del Greco. 2005: GIS Tools for V isualization and Analysis of NEXRAD Radar (WSR 88D) Archived Data at the National Climatic Data Center. 21st International Conference on Interactive Information Processing Systems (IIPS) for Meteorology, Oceanography, and Hydrology. Atallah, E. H., and L. F. Bosart. 2003: The Extratropical Transition and Precipitation Distribu tion of Hurricane Floyd (1999). Monthly Weather Review 131 1063 1081. Barry, R. G., and A. M. Carleton. 2001: Synoptic and Dynamic Climatology. New York: Routledge. Burpee, R. W ., an d M. L. Black. 1989: Temporal and Spatial Variations of Rainfall Near the Ce nters of Two Tropical Cyclones. Monthly Weather Review 117 2204 2218. Carleton, A. M. 1999: Methodology in Climatology Annals of the Association of American Geographers 89 713 735. Corbosiero, K. L., and J. Molinari. 2002: The Relationship between Storm Motion, Vertical Wind Shear, and Convective Asymmetries in Tropical Cyclones. Journal of the Atmospheric Sciences 60 3 66 376. Demaria, K. L., and J. Kaplan. 1999: An Update d Statistical Hurricane Intensity Predication Scheme (SHIPS) for the Atlantic and Eastern North Pacific Basins. Weather and Forecasting 14 326 337. DeMaria, M, M. Mainelli, L. K. Shay, J. A. Knaff, and J. Kaplan. 2005: Further Improvements to the Statist ical Hurricane Intensity Prediction Scheme (SHIPS). Weather and Forecasting 20 531 543. Dvorak, V. F. 1975: Tropical Cyclone Intensity Analysis and Forecasting from Satellite Imagery. Monthly Weather Review 103 420 430. Elsberry, R. L. 2002: Predicting Hurricane Landfall Precipitation: Optimistic and Pessimistic Views from the Symposium on Precipitation Extremes. Bulletin of the American Meteorlogical Society 83 1333 1340. Elsner, J. B., and A. B. Kara. 1999: Hurricanes of the North Atlatic: Climate and Society. US: Oxford University Press. Frank, W. M. 1977: The Structure and Energetics of the Tropical Cyclone I. Storm Structure. Monthly Weather Review 105, 1119 1135. H abib, E., and W. F. Krajewski. 2002: Uncertainty Analysis of the TRMM Gr ound Validation Radar Rainfall Products: Applications to the TEFLUN B Field Campaign. Journal of Applied Meteorlogy 41 558 572.
54 Hewitt, K. 1997: Regions at Risk: A Geographical Introduction to Disasters. Longman: Harlow Holt Jensen, A., and B. Fullertor. 1999: Geography, History and Concepts: A Student's Guide. New York: SAGE Jorgensen, D. P. 1984: Mesoscale and Convective Scale Characteristics of Mature Hurricanes. Part I: General Ob servations by Resarch Aircraft. Journal of Atmospheric Science 41 1268 1285. Jorgen sen, D.P., P.T. Willis. 1982: A Z R Relationship for Hurricanes Journal of Applied Meteorology 21 356 366 Kidder, S. Q., S.J. Kusselson, J.A. Knaffm R.R. Ferr aro, R.J. Kuligowski, M. Turk. 2005: The T ropical Rainfall Potential (TraP) Technique. Part I: Descript ion and Examples. Weather and Forecasting 20 456 464. Kimb all, S. K., and M. S. Mulekar. 2004: A 15 Year Climatology of North Atlantic Tropical Cyc lones. Part I: Size Parameters. Journal of Climate 17 3555 3575. Klazura, G. E., and D. A. Imy. 1993: A Description of the Initial Set of Analysis Products Available from the NEXRAD WSR 88D System. Bulletin of the American Meteorological Society 74 1293 1311. Knight, D., and R. Davis. 2007: Cli matology of Tropical Cyclone Rainfall in the Southeastern United States. Phyisical Geography. 55 126 147. Kraj ewski, W. F., and J. A. Smith. 2002: Radar Hydrology: Rainfall Estimation. Advances in Water Resources 25 1387 1394. Lawrence, M. B. 1979: Atl antic Hurricane Season of 1978. Monthly Weather Review 107 477 491. Lawrence, M. B., L. A. Avila, J. L. Beven, J. L. Franklin, J. L. Guiney, and R. J. Pasch. 2001: Atl antic Hurricane Season of 1999. Monthly Weather Review 129 3057 3084. Marks, F. D ., G Kappler, and M. DeMaria. 2002: Development of a Tropical Cyclone Rainfall Climatology an d Persistence (R CLIPER) Model. 25th Conference on Hurricanes and Tropical Meteorology. San Diego, CA: American Meteorological Society, 327 328. Marsha ll, J. S., and W. M.K. Palmer. 1948: The Distr ibution of Raindrops with Size. Journal of the Atmospheric Sciences 101 165 166. Marshall, J. S., R. C. Langille, and W. M.K. Palmer. 1947: Me asurement of Rainfall by Radar. Journal of the Atmospheric Science s 1 22 186 192. Matthews, J. A., and D. T Herbert. 2004: Unifying Geography: Common Heritage, Shared Future. New York: Rout ledge, Taylor and Francis Group. Matyas, C. J. 2009: A Spatial Analysis of Radar Reflectivity Regions w ithin Hurricane Charley (200 4). Journal of Applied Meteorology and Climatology 48 130 142. Matyas, C. J. 2007: Quantifying the Shapes of U.S. Landfalling Tropical Cyclone Rain Shields. The Professional Geographer 59 158 172.
55 Matyas, C. J. 2008: Shape Measure of Rain Shields as Indictors of Changing Environmental Conditions in a Landfalling Tropical Storm. Meteorological Applications 59 259 271. Merrill, R. T. 1984: A Comparison of Lar ge and Small Tropical Cyclones. Monthly Weather Review 112 1408 1418. Moyer, A. C., J. L. Evans, and M. Powell. 2006: Comparison of Observed Gale Radius Statistics. Meteorology and Atmospheric Physics 97 41 55. Neumann, C. J., B. R. Jarvinen, C. J. McAdie, and G. R. Hammer. 1999: Tropical Cyclones of the North A tlantic Ocean, 1871 1998. National Oceanic and Atmospheric Administration 293 206. Pasch, R. J., E. S. Blake, H. D. Cobb, and D. P. Roberts. 2005: Hurricane Wilma (October 15 26, 2005). Tropical Cyclone Report, Washington D.C. National Oceanic and Atmosp heric Association (NOAA ), U.S. Department of Commerce. Pennington, J. M. DeMaria, and K. Williams. 2000: Development of a 10 year Atlantic Basin Tropical Cyclone Wind Structure Climatology. Dataset available at: http://www.bios.edu/rpi/private/pubs/2000pre/demaria/demaria.html Ra o, G. V., and P. D. MacArthur. 1994: The SSM / I Estimated Rainfall Amounts of Tropical Cyclones and Their Potential in Predicting the Cyclone Intensity Changes. Monthly Weather Review 1 12 1568 1574. Rappaport, E. N. 2000: Loss of Life in the United States Associated with Recent Atlantic Tropical Cyclones. Bulletin of the American Meteorlogical Society 81 2065 2073. Richards, K., M. Bithell, and M. Bravo. 2004: Space, Time, and Science In Unifying Geography: Common Hertiage, Shared Future by J. A. Matthews and D. T. Herbert, 327 352. New York: Routle dge, Taylor, and Francis Group. Ro dgers, E. B., and R. F. Adler. 1981: Tropical Cyclone Rainfall Characteristics as Determined from Satel lit e Passive Microwave Radiometer. Monthly Weather Review 109 506 521. Rodgers, E. B., S. W. Chang, and H. F. Pierce. 1994: A Satellite Observational and Numerical Study of Precipitation Characteristics in Western No rth Atlantic Tropical Cyclones. Journa l of Applied Meteorlogy 33 129 139. Rosenfield, D., D. B. Wolff, and D. Atlas. 1993: General Probability Matched Relations between Ra dar Reflectivity and Rain Rate. Journal of Applied Meteorology 32 50 72. Roth, D. Tropical Cyclone Rainfall. 2008: NOAA Hydrometeorological Prediction Center. Camp Springs, MD. Sch ultz, J. R., and R. W. Reeves. 1999: GIS @ NWS Forum for the Discussion of Applications and Requirements for Geographic Information Systems (GIS) a t the National Weather Service. Silver Springs, MD: 30 Jun 1 Jul Senn, H. V., and H. W. Hiser. 1959: On the Origin of Hurricane Spiral Rain Bands. Journal of Atmospheric Sciences 16 419 426.
56 Sheets, R. C. 1990: The National Hurricane Center Past, Present, and Future. Weather and Forecasting 5 185 232. Shipley, S. T., I. A. Graffman, and J. K. Ingram. 2000: GIS Applicat ions in Climate and Meteorology. ESRI User Conference. Paper 159. Simpson, R. H., H.S. Saffir. 1974: The Hurricane Disast er Potential Scale. Weatherwise 27 169 186. Skaggs, R. H. 2004: Climatology in American Geography. Annals of the Association of American Geographers 94 446 457. Thompson, R. D., and A. Perry. 1997: Applied Climatology. New York: Routledge. Thorpe, A. J. 1985: Diagnosis of Balances Vortex Structure Using Potential Vorticity. Journal of Atmospheric Sciences 42 397 406. Thorpe, A. J. 1986: Synoptic Scale Distur bances with Circular Symmetry. Monthly Weather Review 114 1384 1389. Willoughby, H.E. 1988: The Dynamics of the Tropical Cyclone Core. Aust Meteorological Magazine 36 183 191. Willoughby, H. E., F. D. Marks, and R. J. Feinberg. 1984: Stationary and Moving Convective Bands in Hurricanes. Journal of the Atmospheric Sciences 41 3189 3211. Xie, H., X. Zhou, E. R. Vivoni, J. M. H. Hendricks, and E. E. Small. 2005: GIS Based NEXRAD Stage III Precipitation Database: Automated Approaches for Data Processing and Vizualization. Computers and Geosciences 31 65 76.
57 BIOGRAPHICAL SKETCH Erin Leigh Bunting is originally from Ga inesville, Flori da. After completing a Bachelor of Science from the University of Florida in the Spring of 2006 she applied to the graduate program in the D epartment of Geography. After several years and assistance from committee member completed. With a Master of Science in g eography, Erin applied to the doctoral program at UF. At the doctoral level she plans to examine land use patterns and the effect of clima te change across parts of Florida and Mexico
THE RELATIONSHIP BETWEEN THE RADIUS OF GALE FORCE WINDS AND THE RAIN SHIELD OF U.S. LANDFALLING TROPICAL CYCLONES. Erin Leigh Bunting 352 3920494 Geography Department Dr. Corene Matyas Masters of Science May 2009 Tropical Cyclones are naturally occurring weather phenomena that affect the entire state of Florida. This study focuses on the relationship between the wind fields and rain fields of landfalling tropical cyclones. With a high percentage of tropical cyclone deaths related to inland and coastal flooding it is vital to understand their relationship in conjunction with the wind fields. This is a preliminary study that will hopefully lead to a rainfall climatology that can be utilized in conjunction with the Nat forecasts.