Florida State University Libraries

Transcription

1 Florida State University Libraries Electronic Theses, Treatises and Dissertations The Graduate School 2003 Warm Season Lightning Distributions over the Northern Gulf of Mexico Coast and Their Relation to the Mesoscale and Synoptic Scale Environments Jessica Raye Stroupe Follow this and additional works at the FSU Digital Library. For more information, please contact

2 THE FLORIDA STATE UNIVERSITY COLLEGE OF ARTS AND SCIENCES WARM SEASON LIGHTNING DISTRIBUTIONS OVER THE NORTHERN GULF OF MEXICO COAST AND THEIR RELATION TO THE MESOSCALE AND SYNOPTIC SCALE ENVIRONMENTS By JESSICA RAYE STROUPE A Thesis submitted to the Department of Meteorology in partial fulfillment of the requirements for the degree of Master of Science Degree Awarded: Fall Semester, 2003

3 The members of the Committee approve the thesis of Jessica R. Stroupe defended on November 10, Henry E. Fuelberg Professor Directing Thesis Guosheng Liu Committee Member Philip Cunningham Committee Member The Office of Graduate Studies has verified and approved the above named committee members. ii

4 ACKNOWLEDGEMENTS First and foremost, I wish to extend my deepest gratitude and appreciation to my major professor, Dr. Henry Fuelberg, for his advice, guidance, and support during my stay at Florida State University (FSU). I also wish to thank my committee members, Drs. Guosheng Liu and Philip Cunningham, for their involvement in my research. I want to express my sincerest thanks and appreciation to Irv Watson of the National Weather Service (NWS) in Tallahassee. He has provided me with a plethora of assistance, advice, and guidance on many aspects of my research. Special thanks also are extended to Dan Smith at the NWS Southern Region Headquarters for obtaining lightning data for this project and to Kent Kuyper at NWS Lake Charles for his helpful comments. This research was funded by the National Oceanic and Atmospheric Administration (NOAA) Cooperative Program for Operational Meteorology, Education and Training (COMET) under grant S and by the NOAA CSTAR program through grant NA03NWS I wish to thank my colleagues in the Fuelberg Lab for making my time spent at FSU memorable. Very special thanks goes to former Fuelberg Lab member Todd Lericos who spent many hours helping me with whatever I needed. Finally, I would like to thank my family. Throughout my lifetime, my parents, Cheryl and Walter Stroupe, and my brother, Matthew Stroupe, have provided me with unconditional love, support, and guidance. I also wish to express my deepest love and appreciation to my best friend, Travis Smith. His unwavering love and support have meant the world to me. They have all helped to make me the person I am today. iii

5 TABLE OF CONTENTS List of Tables...v List of Figures... vi Abbreviations and Acronyms... viii Abstract... ix Chapter 1. INTRODUCTION...1 Chapter 2. DATA AND METHODOLOGY Lightning Data Radiosonde Data...7 Chapter 3. RESULTS Individual Flows Calm Flow Northeast Flow Southeast Flow Southwest Flow Northwest Flow Lightning Maxima in Urban Areas...40 Chapter 4. SUMMARY AND CONCLUSIONS...53 REFERENCES...56 BIOGRAPHICAL SKETCH...59 iv

6 LIST OF TABLES Table 1. The number of days associated with each combination of flow regimes at LCH and LIX...16 Table 2. Statistical parameters for each flow regime in the western portion of the region between 91.5 and 96.0 degrees West. The mean hpa vector wind from LCH was used to categorize flow days. The all days category includes every day, including those for which sounding data were unclassifiable. Thus, it is not the sum of the categories listed above Table 3. As in Table 2, except for the eastern portion of the region between 87.0 and 91.5 degrees West. Days were classified according to the mean hpa vector wind from LIX Table 4. Median 1200 UTC sounding parameters for the five flow regimes. The K-index has units of C; CAPE has units of J kg -1 ; and precipitable water is expressed in inches Table 5. Monthly distribution of days for which LCH sounding data were classifiable...24 Table 6. As in Table 5, except based on classifiable data from LIX Table 7. Statistical parameters for individual study boxes centered over Lake Charles, Baton Rouge, New Orleans, and Biloxi. Each box is approximately 36 km 60 km...43 v

7 LIST OF FIGURES Figure 1. Map of domain extending from N and W. Major cities and geographical features are labeled. Outlines of NWS County Warning Areas (CWA) are shaded....8 Figure 2. Composite lightning flash density maps (flashes km -2 yr -1 ) for all warm season days from , where year corresponds to the warm season from May to September. The upper scale corresponds to (a) the 24-hour composite, while the lower scale is for (b) the nighttime lightning composite from UTC ( CST) Figure 3. Average 1000 hpa heights for the composite case. Contours are in 5 m increments...12 Figure 4. Diurnal distribution of all flashes ( 10 5 ) in the region shown in Fig. 1. Hour 1 UTC denotes flashes between 1:00-1:59 UTC...12 Figure 5. Distribution of days according to their hpa mean vector wind directions for (a) LCH and (b) LIX. Directions are grouped into 5 bins. Flow regimes are labeled at the top of each histogram, with arrows denoting the divisions between regimes Figure 6. Scatter diagram of number of lightning flashes ( 10 4 ) versus hpa mean vector wind direction for (a) LCH and (b) LIX. Flow regimes are labeled and denoted by solid black lines...18 Figure 7. Average 1000 hpa height contours in 5 m increments for the following flow regimes: (a) calm flow, (b) northeast flow, (c) southeast flow, (d) southwest flow, and (e) northwest flow. The flow categories are based on LCH data...25 Figure 8. Lightning flash density maps (flashes km -2 regime-day -1 ) for the five flow regimes: (a) calm flow, (b) northeast flow, (c) southeast flow, (d) southwest flow, and (e) northwest flow. The solid black line indicates the division between the western and eastern components of the domain vi

8 Figure 9. Hourly flash density maps (flashes km -2 regime-day -1 ) for calm (light wind) flow between UTC ( CST). Each map represents a one-hour time period...32 Figure 10. Hourly flash distribution ( 10 4 ) by flow regime for the (a) western portion of the region between 91.5º and 96.0º W and (b) eastern portion between 87.0º and 91.5º W...33 Figure 11. Hourly flash density maps (flashes km -2 regime-day -1 ) for Lake Charles and surrounding areas for southwest flow between UTC ( CST). Each map represents a one-hour time period. The white circle denotes the approximate location of the city s center...44 Figure 12. Hourly flash density maps (flashes km -2 regime-day -1 ) for Baton Rouge and surrounding areas for northwest flow between UTC ( CST). Each map represents a one-hour time period. The white circle denotes the approximate location of the city s center...46 Figure 13. Hourly flash density maps (flashes km -2 regime-day -1 ) for New Orleans and surrounding areas for southwest flow between UTC ( CST). Each map represents a one-hour time period. The white circle denotes the approximate location of the city s center...48 Figure 14. Hourly flash density maps (flashes km -2 regime-day -1 ) for Biloxi and surrounding areas for northwest flow between UTC ( CST). Each map represents a one-hour time period. The white circle denotes the approximate location of the city s center...50 Figure 15. Hourly flash distributions for individual maxima over (a) Lake Charles, (b) Baton Rouge, (c) New Orleans, and (d) Biloxi...51 vii

9 ABBREVIATIONS AND ACRONYMS C CAPE CG COMET CST CWA FSL FSU hpa IMPACT in J ka kg km LCH LIX m MDF m s -1 NCDC NLDN NOAA NWS TOA u g UTC Celsius Convective Available Potential Energy Cloud-to-ground Cooperative Program for Operational Meteorology, Education and Training Central Standard Time County Warning Area Forecast Systems Laboratory Florida State University hectopascal Improved Accuracy from Combined Technology inch Joule kiloampere kilogram kilometer Lake Charles, Louisiana Slidell, Louisiana meter magnetic direction finding meter per second National Climatic Data Center National Lightning Detection Network National Oceanic and Atmospheric Administration National Weather Service time-of-arrival u-component of the geostrophic wind Universal Coordinated Time viii

10 ABSTRACT Cloud-to-ground lightning data from the National Lightning Detection Network during the 14-year period are used to create a warm season lightning climatology for the northern Gulf Coast. Days are separated into five flow regimes based on the orientation of the coastline and the low-level flow. Specifically, each day is classified into either a calm (less than 2.5 m s -1 ) or directional flow category based on the mean hpa vector wind data at Lake Charles and Slidell, Louisiana. Flash densities are calculated, and maps are created for daily, hourly, and nocturnal periods. The composite 24-hour and nocturnal flash density maps indicate an east-to-west decrease in lightning over the region. Flash densities for the 24-hour period are greatest over land, with relative maxima located near the major metropolitan areas of Houston, Lake Charles, Baton Rouge, New Orleans, Biloxi, and Mobile. Flash densities during the nocturnal period are largest over the coastal waters. Land breezes, warm and shallow Gulf of Mexico waters, and advection of land-forming convection are responsible for the nighttime offshore convection. Lightning across the northern Gulf Coast depends largely on the prevailing synoptic flow. The low-level flow controls the sea breeze, the dominant forcing mechanism during the warm season. Southwest flow is the most unstable and humid of the five regimes, and it exhibits the most flashes. In this case, sea breeze induced convection is located slightly inland from the coast. Northeast flow, being the driest and most stable of the regimes, exhibits the least amount of lightning. The large-scale flow holds the sea breeze along the coastline. Flash density maxima over urban areas are examined to determine the relationship between lightning and the low-level flow. Analyses reveal that these maxima result from interactions between the prevailing flow and one or more mesoscale circulations. Urban ix

11 influences such as frictional convergence, heat island effects, and air pollution also play a role in the enhancements. Geographic features and local mesoscale circulations affect lightning across the northern Gulf Coast. Geographic features, including lakes, bays, marshes, swamps, and coastline orientations, interact with the low-level flow and mesoscale circulations, such as the sea breeze and lake breezes, to produce complex lightning patterns and distributions. This climatology is useful for operational meteorologists faced with the challenge of forecasting summertime convection and its resulting lightning. x

12 CHAPTER ONE INTRODUCTION Cloud-to-ground (CG) lightning causes injury and death, disrupts human activity and aviation, starts fires, and damages property. This damage includes communications and electrical systems, as well as trees and buildings. An accurate forecast of lightning can help reduce the number of injuries and deaths attributed to lightning, as well as the time and money spent repairing or replacing damaged property. Understanding the mesoscale processes that lead to convective development and its resulting lightning is necessary in order to produce better forecasts. The National Lightning Detection Network (NLDN) (Cummins et al. 1998) was formed to document the occurrence of lightning and assist with lightning research. The NLDN has provided invaluable data to the operational and research meteorological communities since Many studies have used NLDN data to study lightning patterns across the contiguous United States. Analyses by Orville (1991, 1994), Orville and Silver (1997), Huffines and Orville (1999), Orville and Huffines (2001), and Orville et al. (2002) show that Florida, particularly central Florida, has the greatest lightning flash densities in the nation, with the peak occurring during July. The Gulf of Mexico coast from Houston to Alabama exhibits secondary flash density maxima that are described further by Zajac and Rutledge (2001). Regional lightning studies have been conducted for various portions of the United States. Since Florida has the greatest number of lightning flashes, it has been the focus of most investigations. Maier et al. (1984) examined lightning in the vicinity of the Kennedy Space Center and found that lightning peaks occurred approximately three hours after local solar noon. They also noted a weak relationship between the average diurnal distributions of lightning and rainfall. Average monthly flash densities were 1

13 shown to peak during the months of June to August (Hodanish et al. 1997), with maxima occurring in the central portion of the peninsula, particularly just east of Tampa. They attributed the lightning maxima in this region to interactions between the two coastlines, sea breeze circulations and fronts, and the low-level flow, which is determined mainly by the location of the subtropical ridge. Reap (1994) related Florida lightning distributions to synoptic map types determined using sea level pressure forecasts from the Nested Grid Model. He found lightning maxima within coastal sea breeze convergence zones, thus supporting the dependence of lightning on low-level flow patterns. In addition, surface effects, such as lakes and complex coastlines, were shown to alter convergence/divergence patterns and thus affect convection and lightning distributions. Lericos et al. (2002) created a 10-year warm season lightning climatology for the Florida peninsula based on the position of the subtropical ridge axis. The low-level flow was shown to be a primary factor in determining lightning distributions. For example, the greatest median flashes per flow regime day occurred when the ridge axis was located between Tampa and Miami. They also emphasized the role of geographic features in the distribution of lightning. Convex coastlines were shown to enhance convergence, producing increased convection and associated lightning flashes. On the other hand, concave coastlines were associated with divergence, weaker convection, and smaller flash densities. The previously mentioned Florida lightning studies all concentrated on lightning over the peninsula. However, Camp et al. (1998) examined lightning distributions and the factors that influence them in the panhandle region of northern Florida. This area has a more complex coastline and is influenced by only one body of water, the Gulf of Mexico. As in the aforementioned studies, the low-level flow (either perpendicular or parallel to the coastline) and the complexity of the coastline were important factors in determining locations of flash density maxima and minima. They also showed that low-level wind speeds of 2-5 m s -1 were ideal for thunderstorm and lightning formation. The sea breeze plays an important role in coastal mesoscale processes, and many investigators have examined it in great detail. Wexler (1946) gives a complete description of the sea breeze, including discussions on theory and factors influencing the 2

14 sea breeze, its surface and horizontal characteristics, and vertical structure. Simpson (1994) also gives a thorough description of the sea breeze, with discussions of field and laboratory measurements, as well as analytic and numerical modeling. Estoque (1962) modeled the sea breeze phenomenon and found that temperature and large-scale winds played important roles in the development and propagation of the sea breeze. Arritt (1993) used a two-dimensional numerical model to analyze the sea breeze, varying the prevailing wind speeds of onshore and offshore flow. He found that large-scale onshore flow more than 3 m s -1 prohibited thermal circulations. Conversely, calm flow or offshore flow up to 11 m s -1 allowed the formation of thermally induced perturbations and sea breezes. He showed that the strongest thermal circulation and farthest inland penetration of the sea breeze occurred when the u-component of the geostrophic wind (u g ) was offshore in direction and approximately 3-6 m s -1 in magnitude. When u g was offshore and between 6-11 m s -1, the thermal circulation that formed remained entirely over water. The effects of coastlines on the sea breeze phenomenon also have been studied. McPherson (1970) used a three-dimensional numerical model to show that a bay located along a coastline could distort the sea breeze convergence field. During the early stages of a sea breeze circulation, the vertical motion field was maximized on both the eastern and western sides of the bay. However, as time progressed, the vertical motion maximum shifted first to the northwestern corner and then to the northeastern corner of the bay, thus producing a skewed vertical motion field. Visible and infrared satellite imagery was used by Gibson and Vonder Haar (1990) to clarify the influence of geographic features on cloud patterns over a portion of the southeastern United States. Increased cloudiness was noted when the sea breeze enhanced local convergence. Florida s weather during the warm season is predominately governed by the Gulf of Mexico and Atlantic Ocean sea breezes. Florida s sea breezes are a forecasting challenge and of great interest to many researchers. Pielke (1974) used a threedimensional numerical model to simulate sea breezes over south Florida. He demonstrated the importance of the prevailing synoptic flow on sea breeze circulations, as well as the effects of the coastline in enhancing the convergence needed to trigger thunderstorms. In addition, Pielke (1974) showed the importance of Lake Okeechobee 3

15 on the convergence and convection patterns across the south Florida peninsula. Similar results were given by Blanchard and López (1985) who used radar images and thermodynamic conditions to explain the formation and evolution of convection patterns over south Florida. López and Holle (1987) related lightning patterns in Florida to the two sea breezes. They showed that when the synoptic flow was offshore, the convergence associated with the sea breeze developed later in the day but was stronger because of the greater temperature difference between land and water. The sea breeze associated with this case remained stationary along the coastline. The opposite was found true for onshore flow. In this case, the sea breeze circulation developed early, was weaker, and had greater inland penetration. In addition, lightning occurrence was related to the low-level flow through stability and moisture content (López and Holle 1987). Very localized studies have been conducted recently to investigate the relationship between lightning and major cities. Watson and Holle (1996) examined lightning patterns across the southeastern United States, particularly near Atlanta, Georgia, in preparation for the 1996 Summer Olympics. They found a lightning maximum over Atlanta (greater than 3 flashes per km 2 ) which peaked during the late afternoon and early evening hours. Livingston et al. (1996) also studied lightning distributions, as well as synoptic conditions and thermodynamic parameters, in Georgia for the 1996 Summer Olympics. They confirmed the presence of a definite diurnal cycle of lightning with a peak near 2200 UTC and a sea-breeze induced flash density maximum near Savannah, Georgia. Analyses of lightning flashes near 16 major cities in the Midwest were conducted by Westcott (1995). Her statistical analyses suggested that the urban areas did not initiate new convective storms but rather intensified those that had already formed. She showed that the greatest warm season lightning occurred over and downwind of the metropolitan areas. Several factors were listed as possible causes, including increased air pollution, urban size, geographic features, and urban-rural structural differences. While urban areas were shown to affect lightning distributions, no one particular cause was pinpointed as the main reason for this enhancement. Houston, Texas, also has been a region of interest because of its location in the northern Gulf Coast lightning maxima and its poor air quality. Steiger et al. (2002) 4

16 showed a 45% increase in lightning centered over Houston compared to nearby areas. Most of this increase was seen over and downwind of metropolitan Houston and was due to large lightning events. Several theories were presented as possible causes for the lightning enhancement: sea breeze effects, urban heat island effects, urban air pollution and its resulting modification of microphysical processes, and salt water effects (Steiger et al. 2002). None of these could be proven to be the leading cause for the lightning enhancement over the city. Orville et al. (2001) suggested that the large lightning flash densities over Houston s urban area could be due to a combination of enhanced sea breeze convergence from a complex coastline, an urban heat island effect, and the large amount of industrial pollution. Several of the previously mentioned studies have found the northern Gulf Coast to be a hotbed of lightning activity. Others have demonstrated the relationships between lightning, the sea breeze and its associated convection, coastlines, and other geographic features. The landscape of the northern Gulf Coast is populated by numerous lakes, bays, marshes, swamps, and rivers. Its coastline is highly complex, producing a detailed pattern of frictional convergence and divergence based on the low-level flow. The thermally induced sea breeze provides an additional triggering mechanism for convection. Thus, the northern Gulf Coast is an interesting region to study lightning patterns and their relation to mesoscale phenomena and the environment. To the best of the author s knowledge, no lightning studies have been conducted for most of this region. The objective of this study is to create a lightning climatology for the northern Gulf Coast. Data from warm seasons during the 14-year period are analyzed and categorized according to the low-level synoptic flow. Lightning flash densities are calculated every hour to study diurnal variations. Possible causes for the formation and movement of lightning patterns are examined. Finally, special emphasis is placed on lightning maxima located near several metropolitan areas within the region. 5

17 CHAPTER TWO DATA AND METHODOLOGY 2.1 Lightning Data In complete operation since 1989, the NLDN detects and records CG lightning flashes over the continental United States and immediate coastal waters. This network, owned and operated by Vaisala Inc., provides detection data to a variety of commercial, government, educational, and public entities. Specifics concerning network methodology and operations are described by Cummins et al. (1998). One hundred six ground-based sensors across the United States are included in the NLDN. The network utilizes the IMProved Accuracy from Combined Technology (IMPACT) method to detect CG flashes. This new method of flash recording and processing incorporates two types of sensors: time-of-arrival (TOA) sensors and IMPACT sensors, which use both TOA and magnetic direction finder (MDF) methods to accurately determine the location and time of the flash. Even though CG lightning flashes may consist of several return strokes, only the first flash s data are recorded. Included are the flash s time, latitude, longitude, polarity, strength, and multiplicity. The detection efficiency and location accuracy of the NLDN have improved greatly through the years. During the early years of the network s operations, the detection efficiency ranged between 65% and 85%, while the location accuracy was 8 km to 16 km (Cummins et al. 1998). System upgrades in 1995 allowed the detection of a greater number of flashes, thereby improving the network s efficiency and accuracy. Since these upgrades, the NLDN has a detection efficiency of 80%-90% and is accurate to within 0.5 km. Because of the enhanced methods of detection, other forms of lightning may be sensed and recorded. Following the suggestion by Cummins et al. (1998), I removed 6

18 from the data set weak positive flashes with strengths less than 10 ka. The data set also was examined for duplicate lightning flashes. When two or more recorded flashes occurred within 10 km of one another, only the first flash s data were retained, although their multiplicities were combined (Cummins et al. 1998). No corrections were applied to the data to compensate for the variation in detection efficiencies and location accuracies across the study area. This produces an underestimation of flash densities. Data from the months of May to September were used in this study. These warm season months were chosen because of their enhanced convection and minimal synoptic activity. Synoptic-scale forcing typically is weak, and the influence of mid-latitude systems is rare. Instead, mesoscale phenomena, such as sea breezes and lake breezes, interact with their environments, geographic features, and each other to produce complex patterns of convergence and resulting convection. The study domain spanned N and W, encompassing the northern Gulf of Mexico coastline and adjacent waters (Fig. 1). As shown by Cummins et al. (1998), five NLDN sensors are located in this area. The flashes were counted on a 2.5 km 2.5 km grid, corresponding to a array of 6.25 km 2 grid cells. 2.2 Radiosonde Data Radiosonde data were used to categorize each day of the period according to the prevailing low-level flow in the area. The mean vector wind in the 1000 hpa to 700 hpa layer was computed each day using the 1200 UTC sounding. As shown in previous studies (López and Holle 1987; Camp et al. 1998; Lericos et al. 2002), the flow within this layer provides a good indication of sea breeze and thunderstorm movement during the warm season. Two radiosonde sites were chosen to describe the low-level flow in the region Lake Charles (LCH) in the western portion of the domain and Slidell (LIX) in the domain s eastern portion (Fig. 1). Radiosonde data from 1989 to 1999 were available on the Radiosonde Data of North America CD-ROM distributed by the National Climatic Data Center (NCDC) and the Forecast Systems Laboratory (FSL) (FSL and NCDC 1999). Data from the years 2000 to 2002 were obtained directly from FSL s website ( data/fsl-data.html). 7

19 FIG. 1. Map of domain extending from N and W. Major cities and geographical features are labeled. Outlines of NWS County Warning Areas (CWA) are shaded. 8

20 CHAPTER THREE RESULTS To investigate general lightning patterns in the northern Gulf Coast region, all flashes were grouped together and examined as a composite case, without any consideration of wind direction or time of day. Fig. 2a shows a general decrease in lightning from east to west. Enhanced flash densities are seen along the entire Gulf of Mexico coastline (Fig. 2a), suggesting a link to the sea breeze. However, the strongest maximum is in coastal Mississippi, near Biloxi. Its flash density values are greater than 8.0 flashes km -2 yr -1, with year defined as the warm season from May to September. Most of the other maxima along the northern Gulf Coast are located near major metropolitan areas. The areas near Houston, Lake Charles, New Orleans, and Mobile all exhibit flash density maxima of 5-8 flashes km -2 yr -1. City locations and geographic features are shown in Fig. 1. Houston s maximum corresponds to the region of enhanced lightning found by Steiger et al. (2002) and Orville et al. (2001). A weaker maximum of 4-6 flashes km -2 yr -1 is found near Baton Rouge. Each of these maxima may be caused or enhanced by several factors, including convergence due to sea, lake, swamp, and river breezes, convex coastlines, urban heat island effects, and air pollution. These maxima will be discussed in detail later in the manuscript. Reduced flash densities are found over the extreme northern and southern portions of the domain, over the Atchafalaya Basin, and over Lake Pontchartrain (Fig. 2a). Flash densities over coastal Louisiana generally are weaker than those over coastal Mississippi and Alabama. Large lakes in south Florida have been associated with subsidence and diminished lightning (Pielke 1974; Blanchard and López 1985). Similarly, in the current study area, flash densities over Lake Pontchartrain are only 2-4 flashes km -2 yr -1. The Atchafalaya Basin and much of coastal Louisiana is a region of 9

21 a) b) FIG. 2. Composite lightning flash density maps (flashes km -2 yr -1 ) for all warm season days from , where year corresponds to the warm season from May to September. The upper scale corresponds to (a) the 24-hour composite, while the lower scale is for (b) the nighttime lightning composite from UTC ( CST). 10

22 swamps, marshes, lakes, and small rivers, also exhibiting relatively small flash densities. This is consistent with the findings of Hodanish et al. (1997), who noted that wetlands are associated with small amounts of lightning due to less differential heating and weaker thermal circulations. The northern portion of the study region (Fig. 2a) has small flash densities because of its greater distance from the Gulf of Mexico. Thus, the sea breeze typically does not propagate that far inland, and moisture supplies are more limited. Over the southern portion of the domain (the open waters of the Gulf of Mexico), the number of flashes is thought to be low because of three factors. First, the dominant forcing mechanism, the sea breeze, is a daytime inland phenomenon. Second, convection is usually weaker over water than over land, resulting in a smaller number of flashes (Orville and Henderson 1986). Third, the NLDN sensors are located only over land, causing detection efficiency and location accuracy to decrease with increasing distance from the coastline (Cummins et al. 1998). To further understand the synoptic environment of the region, reanalysis data were obtained from the National Center for Environmental Prediction (Kalnay et al. 1996). Specifically, 1000 hpa geopotential height data from each warm season day during the 14-year period were averaged to create one composite chart (Fig. 3). The subtropical ridge axis extends westward from central Florida into Louisiana. A trough is located over southern Alabama, with a relatively strong height gradient evident over southeast Texas and southwest Louisiana. The diurnal distribution of lightning flashes for the composite (all days, all flows) case is shown in Fig. 4. Hourly values range from approximately during the nighttime to near 1.4 million at the afternoon peak of 2000 UTC (1500 CST). A smaller, secondary peak occurs during the early morning at 1200 UTC (0700 CST). It is thought to be attributed to the land breeze, which is associated with early morning offshore convection. The smallest number of flashes occurs at 0500 UTC (0000 CST) when convection is minimal. To investigate the small number of flashes during the nighttime, flash densities were calculated and plotted for the period UTC ( CST) (Fig. 2b). Flash densities for this nighttime period are much smaller than those of the 24-hour 11

23 FIG. 3. Average 1000 hpa heights for the composite case. Contours are in 5 m increments. 16 Number of Lightning Flashes ( x 10 5 ) Hour of the Day (UTC) FIG. 4. Diurnal distribution of all flashes ( 10 5 ) in the region shown in Fig. 1. Hour 1 UTC denotes flashes between 1:00-1:59 UTC. 12

24 composite case (Fig. 2a), indicating that the majority of flashes during the 24-hour period is due to daytime convection. However, the decrease in lightning from east to west is still evident during the night. Enhanced nighttime flash densities stretch from just offshore of Galveston Bay (southeast of Houston) to the eastern edge of the domain (Fig. 2b). The area of maximum flash densities is offshore of coastal Mississippi. The reason for this maximum is uncertain. However, it may be due to a merger of land breezes from the Mississippi and Louisiana coasts. Specifically, the southward moving land breeze from coastal Mississippi may be colliding with an eastward moving land breeze from Louisiana, leading to enhanced convergence. As suggested by Lericos et al. (2002), enhanced offshore nighttime flash densities also can be due to the warm, shallow waters in that region. Conversely, weaker offshore flash densities occur just off the coastline southwest of New Orleans. This relative minimum may be due to the convex coastline in the area, which promotes localized divergence. 3.1 Individual Flows The low-level synoptic flow plays a significant role in the formation and evolution of the sea breeze and its resulting convection and lightning (e.g., Arritt 1993). Therefore, it is important to classify individual days according to their flow. For each day during the 14-year period, the mean vector wind in the 1000 to 700 hpa layer was calculated from the 1200 UTC radiosonde releases at Lake Charles and Slidell, Louisiana (LCH and LIX). Based on these mean vector winds, each day was placed into a particular flow regime. Of the possible 2142 days during the 14-year period, radiosonde data were available for 2069 days from LCH and 2108 days from LIX. Two thousand forty-one days had radiosonde data from both locations. To examine the distribution of mean vector winds, directional values were grouped into 5 bins and plotted in a histogram. Fig. 5 shows the histograms for LCH (the western part of the domain) and LIX (the eastern portion). The distribution from LCH (Fig. 5a) exhibits unimodal behavior, with the peak number of days having southerly flow. The 190 bin has 89 days, while the number of days in the 205 bin totals 13

25 FIG. 5. Distribution of days according to their hpa mean vector wind directions for (a) LCH and (b) LIX. Directions are grouped into 5 bins. Flow regimes are labeled at the top of each histogram, with arrows denoting the divisions between regimes. 14

26 90. Flow with a northerly component occurs infrequently. Only six days register a mean hpa direction in the 0 to 5 range. Mean wind directions from LIX are somewhat different than those from LCH. The LIX distribution (Fig. 5b) is more uniform, and the peak is skewed slightly to the right of the LCH peak. The greatest number of days is 70, occurring when the mean wind is from the southwest (225º bin). Days with southerly winds are the second most common. Northeast flow days are the least frequent of the four categories. The low-level flow in a coastal region can be described by two directional components: parallel and perpendicular to the coastline. The flow also can be light and variable. Thus, I included a calm flow category to account for days when mean vector wind speeds were less than 2.5 m s -1. All directions were included in this regime. Unlike Florida s relatively straight coastlines, the coastlines in Louisiana and Mississippi are more complex with varying orientations. Using different coastline orientations and thus different flow regimes for each portion of the region would have been impractical. Therefore, an average coastline orientation of 86 was assumed across the region, and days with mean vector wind speeds greater than 2.5 m s -1 were categorized into four quadrants or flow regimes based on this orientation. These flow regimes have equal directional ranges (90 ) and are denoted northeast ( ), southeast ( ), southwest ( ), and northwest ( ). The distribution of days within each regime is shown in Fig. 5. It was unclear whether sounding data from LCH or LIX should be used to classify days according to their low-level flow. To investigate this aspect, hpa vector mean wind directions from the 2041 days having data at both LCH and LIX were compared. Table 1 lists the number of days in each possible combination of flow regimes at these two sites. Flow regimes for LIX are listed vertically, and those from LCH are read horizontally. The southwest flow regime contains the greatest number of days at both LIX and LCH (Table 1), while the northeast quadrant has the smallest number of days. With the exception of calm flow, approximately 66% of days at both locations fall into the same flow regime. Of the remaining days, a large portion has mean wind directions in an adjacent quadrant. There are few days when the low-level flow at the two stations is in 15

27 TABLE 1. The number of days associated with each combination of flow regimes at LCH and LIX. LIX / LCH Calm Northeast Southeast Southwest Northwest Total Calm Northeast Southeast Southwest Northwest Total

28 opposite quadrants. This suggests that, even though the low-level synoptic flow at LCH and LIX often is very similar, there is a significant number of days when the flow is at least somewhat different. These different days indicated that mean vector winds from both radiosonde sites should be used to classify days within flow regimes. To further explore differences that might exist between mean wind directions at LCH and LIX, the number of flashes across each half of the region was summed for each day and related to the mean low-level wind direction for that day using scatter diagrams (Fig. 6). To quantify the flash characteristics of each regime, I also calculated various statistical parameters using daily lightning and radiosonde data (Tables 2 and 3). All of the study days also were combined in a separate category ( all days ) to examine lightning characteristics as a whole, regardless of the mean vector wind. This category is analogous to the composite case discussed earlier (Fig. 2). The scatter diagram in Fig. 6a and statistical quantities in Table 2 are based on mean hpa vector wind data from LCH and lightning data from the western half of the domain ( W). Radiosonde data were available for 2069 days out of a possible Of these days with data, 2045 days, or 98.8%, were classified into one of the five flow regimes. The remaining days were not classified because of missing wind data or problems with the sounding, but they were included in the all days category. Fig. 6a and Table 2 show that the number of days and lightning flashes over the western half of the domain vary widely between flow regimes. The southwest category has the largest number of days (840) and flashes ( ). In fact, the maximum number of flashes in one day (44 846) occurs on a day with low-level southwesterly flow. Conversely, the northeast and northwest flow regimes have the smallest number of days and flashes (approximately 225 and , respectively). More detailed discussions about each flow regime are provided in later sections. Mean hpa vector wind data from LIX and lightning data from the domain s eastern half ( W) are compiled in Fig. 6b and Table 3. Two thousand ninety-five days, or 99.4% of the possible 2108 days, were classified into either the calm, northeast, northwest, southeast, or southwest flow regimes. Although the eastern and western domains have the same areas, considerably more flashes ( ) occur in the eastern region, compared to in the western portion. Within the 17

29 FIG. 6. Scatter diagram of number of lightning flashes ( 10 4 ) versus hpa mean vector wind direction for (a) LCH and (b) LIX. Flow regimes are labeled and denoted by solid black lines. 18

30 19

31 20

32 eastern half of the study area, the southwest regime again has the largest number of days and flashes (719 and , respectively) (Fig. 6b and Table 3). The northwest category also has days that produce large amounts of lightning. In fact, the second greatest maximum flashes per day (48 111) occurs on a northwest flow day. Clearly, the northeast flow classification ranks last among the regimes (245 days and flashes). Thus, in addition to being the least frequent of all days, northeast flow is characterized by small amounts of lightning (Fig. 6b). Because of the differences between coastlines and geographic features, mean vector winds (Fig. 5), and lightning distributions between LCH and LIX (Fig. 6), I used data from both radiosonde sites to classify each day into a flow regime. The mean vector winds from soundings at LCH are considered representative of conditions across the western portion of the region ( W). Conversely, sounding data from LIX represent conditions in the eastern portion of the area ( W). The directional ranges for each flow regime are the same as described previously (356-86, etc). Finally, because of the large variation in the number of flow days and flashes between the flow regimes (Fig. 6, Tables 2 and 3), flash densities were normalized using the procedure described by Lericos et al. (2002). That is, flash counts for each grid cell were divided by the area of the cell and the number of days in the flow regime. This established the units of flash density as flashes km -2 regime-day -1, allowing easy comparisons between flow regimes. Each flow regime is discussed separately in the following sections. 3.2 Calm Flow The calm (light wind) flow regime contains all days when the hpa mean vector wind speed is less than 2.5 m s -1, regardless of wind direction. Three hundred three (352) days were classified as calm flow days in the western (eastern) portions of the domain (Tables 2 and 3). Of these days, 85% (91%) experience lightning in the western (eastern) portion. The large percentages of lightning days are supported by sounding parameters in Table 4. For each flow regime, median values of K-index, Convective Available Potential Energy (CAPE), and precipitable water were calculated from each day s

33 TABLE 4. Median 1200 UTC sounding parameters for the five flow regimes. The K-index has units of ºC; CAPE has units of J kg -1 ; and precipitable water is expressed in inches. K-index CAPE Precipitable Water LCH Calm Northeast Southeast Southwest Northwest LIX Calm Northeast Southeast Southwest Northwest

34 UTC sounding. Both K-index and CAPE are measures of atmospheric stability. The K- index incorporates low and mid-level temperature and moisture, with larger values representing greater thunderstorm potential. CAPE is a measure of the available buoyant energy needed to vertically accelerate air parcels; large values indicate greater instability. It was calculated using each afternoon s convective temperature as the surface temperature. Precipitable water is a measure of atmospheric moisture available for precipitation. The results show that, in both halves of the domain, calm (light wind) flow is associated with large values of K-index, CAPE, and precipitable water. Thus, convection is favored because of moist, unstable atmospheric conditions. In the western portion of the study area (Table 2), calm flow days have the greatest mean flashes per day (3742) and the greatest median flashes per day (1424) and per lightning day (2059) of all five regimes. The single day maximum of flashes ranks second among the flow regimes. Calm flow days in the west are most common during July and least common in May and June (Table 5), when mid-latitude synoptic systems still may increase low-level wind speeds. Across the eastern section, calm flow days occur most often in August and least often during May (Table 6). Calm flow in the east exhibits the second greatest mean and median flashes per day (4472 and 1889, respectively) and per lightning day (2424) (Table 3). These statistics reveal that calm (light wind) days are frequent and are large lightning producers. The warm season months of May to September are characterized by minimal synoptic activity. This is evident in the mean 1000 hpa geopotential height analysis where height data from calm flow days were averaged to create a composite analysis for that regime (Fig. 7). The figure shows patterns based on the LCH regimes; however, very similar analyses were obtained for the LIX flow categories (not shown). In Fig. 7a, high pressure associated with the subtropical ridge dominates the southeastern United States during calm flow days. The ridge axis extends across Florida into east Texas. The weak height gradient over the Gulf area is indicative of the light winds over Louisiana and Mississippi. The 24-hour composite flash density map for calm flow (Fig. 8a) reveals an active lightning pattern similar to the all days composite shown in Fig. 2a. Maxima are located in the same areas, with most near major metropolitan areas. This similarity 23

35 TABLE 5. Monthly distribution of days for which LCH sounding data were classifiable. Calm Northeast Southeast Southwest Northwest May June July August September Total TABLE 6. As in Table 5, except based on classifiable data from LIX. Calm Northeast Southeast Southwest Northwest May June July August September Total

36 FIG. 7. Average 1000 hpa height contours in 5 m increments for the following flow regimes: (a) calm flow, (b) northeast flow, (c) southeast flow, (d) southwest flow, and (e) northwest flow. The flow categories are based on LCH data. 25

37 FIG. 7 continued. 26

38 FIG. 7 continued. 27

39 FIG. 8. Lightning flash density maps (flashes km -2 regime-day -1 ) for the five flow regimes: (a) calm flow, (b) northeast flow, (c) southeast flow, (d) southwest flow, and (e) northwest flow. The solid black line indicates the division between the western and eastern components of the domain. 28

40 FIG. 8 continued. 29

41 FIG. 8 continued. 30

42 occurs because only mean vector wind speed is considered when classifying days as calm; therefore, all wind directions are included. These variations in wind direction cause flash density patterns to resemble the all days composite case. The density values are smaller because they were normalized per regime day rather than per entire warm season. There are several regions of enhanced and diminished flash densities across the region during calm (light wind) flow (Fig. 8a). Flash density minima are found over large bodies of water and in regions of marshes and swamps. Thus, Galveston Bay (southeast of Houston), the Atchafalaya Basin, Lake Pontchartrain, and Mobile Bay (adjacent to Mobile) all exhibit flash densities between flashes km -2 regimeday -1. On the other hand, flash density patterns are rather disorganized at locations well inland from the coast; only several small maxima of flashes km -2 regime-day -1 are evident. Hourly flash density patterns (not shown) do not reveal an organized progression in this area. Thus, subtle forcing mechanisms may be responsible for the convection. Conversely, enhanced flash densities in the coastal areas are organized in a broken line that parallels the coastline. Hourly flash density maps indicate that this line of enhanced densities is caused by the sea breeze penetrating inland only slightly. Largescale winds less than 2.5 m s -1 are too weak to keep the circulation offshore or move it farther inland. The strongest density maximum within the study area (greater than 0.08 flashes km -2 regime-day -1 ) is located in coastal Mississippi (Fig. 8a). A closer view of this maximum is shown in the series of hourly maps in Fig. 9. Convection develops west of Mobile Bay, near the border of Mississippi and Alabama, at 1700 UTC (1200 CST) (Fig. 9b). This area is favored for convection because of enhanced convergence due to flow interaction with the bay. At 1800 UTC (1300 CST, Fig. 9c), the area of greatest flashes is farther west and centered around a small bay. This is another region where convergence is enhanced because of the coastline s shape. However, by 1900 UTC (1400 CST, Fig. 9d), convection is widespread and indicative of the larger-scale sea breeze. Finally, by 2000 UTC (1500 CST), the flash density peaks, with a swath of large values extending from Slidell to Mobile (Fig. 9e). This peak corresponds to the 31

43 FIG. 9. Hourly flash density maps (flashes km -2 regime-day -1 ) for calm (light wind) flow between UTC ( CST). Each map represents a one-hour time period. 32

44 Number of Flashes a) (x 10 4 ) Calm Northeast Southeast Southwest Northwest West Hour of the Day (UTC) Number of Flashes b) (x 10 4 ) Calm Northeast Southeast Southwest Northwest East Hour of the Day (UTC) FIG. 10. Hourly flash distribution ( 10 4 ) by flow regime for the (a) western portion of the region between 91.5º and 96.0º W and (b) eastern portion between 87.0º and 91.5º W. 33

45 maximum in cloudiness and deep convection found by Gibson and Vonder Haar (1990) in their study of cloud frequencies over this region. It is clear that complex and sometimes subtle forcing mechanisms are important in producing the flash patterns that are observed in the region (Figs. 2, 8a, and 9). One important mesocale forcing mechanism is the large-scale sea breeze, and a second is enhanced convergence/divergence patterns due to the shape of the coastline and to lakes. Since the maximum in Fig. 9 is located near Biloxi/Gulfport, it also may be influenced by urban factors, including air pollution and heat island effects (e.g., Orville et al. 2001; Steiger et al. 2002; Westcott 1995). Additional, weaker maxima (Fig. 8a) are noted in the vicinity of other metropolitan areas (i.e., Lake Charles and New Orleans). These will be discussed in further detail later in the manuscript. Hourly flash distributions for calm (light wind) flow (Fig. 10) are similar in shape and magnitude for the western and eastern portions. The western half s peak of ~ (Fig. 10a) occurs at 2200 UTC (1700 CST), one hour later than the eastern half s peak of ~ flashes (Fig. 10b) at 2100 UTC (1600 CST). Both distributions also exhibit smaller, secondary peaks between UTC ( CST). Early morning offshore convection is the main cause for these small peaks in flash count. 3.3 Northeast Flow Days when the hpa mean vector wind direction is between 356º-86º are classified in the northeast flow category. This regime has the fewest number of days (218 and 234) and flashes ( and ) for both halves of the domain (Tables 2 and 3). In the western segment, only 70% of the 218 flow days have lightning. The mean, median, and maximum flashes per day also are the smallest of all five regimes. Within the eastern sector, a greater percentage of days (82% of the 245 flow days) produce lightning (Table 3). It is interesting that the northeast regime exhibits the greatest number of flashes on a single day (51 327). However, this maximum is due to an individual event rather than a general characteristic of the regime. Convection is less likely during northeast flow because the atmosphere is relatively dry and stable. In fact, values of the K-index, CAPE, and precipitable water for both halves of the domain are the smallest of all flow categories (Table 4). 34

46 Strong high pressure over Illinois and low pressure over the Yucatan peninsula govern the northeast flow regime (Fig. 7b). A ridge axis extends southwestward into southeast Texas, while an inverted trough, possibly associated with frontal boundaries, extends northeastward from the western Gulf of Mexico along the East Coast. The flow along the high pressure s southern edge and around the trough produce large-scale northeasterly winds over Louisiana and Mississippi. The subtropical ridge does not influence the domain. Northeast flow occurs most frequently during September (Tables 5 and 6), a transitional month. Post cold frontal high-pressure systems are more common in this month than during the traditional summer months, bringing drier, more stable air into the region (Table 4). Thus, June and July experience the smallest number of northeast flow days. Magnitudes and patterns of flash density differ considerably over the western and eastern domains during northeast flow (Fig. 8b), with values in the western half being much smaller. The reason for this diminished activity is unclear. The largest densities ( flashes km -2 regime-day -1 ) are in a broken line stretching from Houston to Lake Charles. They are associated with the sea breeze but are farther inland and weaker than expected. The coastal portions of southwest Louisiana consist of marshes, swamps, and lakes. Because of these geographic features, the thermal circulation of the sea breeze may be farther inland where there are stronger temperature gradients between land and water. Flash densities are weaker because of the weaker thermal circulation. There is little lightning activity over the coastal waters. Flash densities across the eastern half of the domain are greater in magnitude and areal coverage (Fig. 8b). Two areas of large values are evident: along the Mississippi coastline and along the Louisiana coast south of New Orleans. Thunderstorms remain along the coasts because the advance of the northward moving sea breeze is limited by the large-scale northeasterly winds. Enhanced flash densities also are found over the coastal waters south and southwest of New Orleans. Hourly flash density maps (not shown) indicate that this lightning is caused by early morning offshore convection. Northeast flow produces a flash density minimum over and southwest of Lake Pontchartrain. The large-scale flow advects the cooler and more stable air into this 35

47 region, producing a minimum known as a lake shadow. Reduced densities also are noted over the Atchafalaya Basin. Hourly flash counts for northeast flow rank last among the five regimes (Fig. 10). Lightning activity in the western section peaks at 2200 UTC (1700 CST) (Fig. 10a), while it reaches a maximum one hour earlier at 2100 UTC (1600 CST) in the eastern portion (Fig. 10b). Most of the remaining counts are less than Southeast Flow The southeast flow regime contains days whose mean vector wind directions are between 86º-176º. The number of flow days and total number of flashes during the 14- year period are similar for both halves of the domain (Tables 2 and 3), with the number of flow days for each half ranking second among the five regimes. Specifically, the western portion has 450 southeast flow days and flashes, compared to 421 days and flashes for the eastern portion. Of the southeasterly flow days, 89% (92%) produce some lightning in the western (eastern) portion of the region (Tables 2 and 3). However, in the eastern sector, southeast flow days exhibit the smallest mean and maximum number of flashes per day and the second smallest median flashes per day (Table 3). On the other hand, the median and maximum flashes day -1 in the western section rank third out of the five flow categories (Table 2). Southeast flow is moister and more unstable than northeast flow. Values of K-index and precipitable water are relatively large (Table 4), but the CAPE values are somewhat smaller than those of the other categories, especially at LIX, indicating less energy for convection. To summarize, there is a large percentage of lightning days, but those days do not produce much lightning. The 1000 hpa analysis (Fig. 7c) shows a closed high-pressure region over North Carolina. The subtropical ridge axis extends from the Atlantic Ocean across the Carolinas and Tennessee to northern Louisiana, placing the entire domain under the influence of large-scale southeasterly flow. Moreover, height contours are tightly packed across Louisiana compared to the rest of the Southeast. Monthly frequencies of southeasterly flow vary greatly between the western and eastern halves of the domain. In the eastern portion (Table 6), southeast flow occurs 36

48 most often during September. Conversely, in the western portion (Table 5), August has the most southeast flow days. This flow occurs least frequently in both halves during July. The flash density map for southeast flow (Fig. 8c) indicates little lightning across the region. The sea breeze is weak and advances onshore quickly, often without producing significant convection. However, a weak, broken line of enhanced densities stretches from Houston to Mobile, slightly inland from the coast. This line appears to result from the weak sea breeze. An area of enhanced flash densities (only 0.03 and 0.06 flashes km -2 regime-day -1 ) also stretches from Baton Rouge northward into Mississippi. This area may result from collisions of individual outflow boundaries, interactions between geographic features and the low-level flow, or urban influences (i.e., heat island effects, industrial pollution, etc.). As with northeast flow, relative flash density minima are noted over Lake Pontchartrain and the area immediately northwest of it (the lake shadow) and over the Atchafalaya Basin. In the western portion of the region, the peak number of hourly flashes ranks third greatest in magnitude (Fig. 10a). This peak of ~ flashes occurs at 2100 UTC (1600 CST), while most of the remaining totals are less than Although the eastern portion has greater flash densities than the western half (Fig. 8c), its diurnal distribution is similar to that of the western portion. However, the afternoon peak of ~ occurs one hour earlier (Fig. 10b). 3.5 Southwest Flow Days with southwest flow are characterized by mean hpa wind directions between 176º-266º. The 1000 hpa synoptic chart (Fig. 7d) shows the subtropical ridge axis extending from the Atlantic Ocean to southern Louisiana, with a broad low-pressure system over western Texas and Oklahoma. These two features produce a tight height gradient and southwesterly winds across the northern Gulf Coast. May, June, and July exhibit the largest numbers of southwest flow days, while September has a definite minimum (Tables 5 and 6). Southwest flow is the most unstable, most frequent, and most active of all categories. In both halves of the study area, the number of days (840 and 719) and 37

49 flashes ( and ) far exceed those of the other categories (Tables 2 and 3). Eighty-five percent (90%) of the days have lightning in the western (eastern) portion. Sounding parameters explain the propensity for convection during southwest flow (Table 4). The large values of CAPE in both halves of the area are the greatest of all regimes, representing unstable conditions. The K-index and precipitable water also are relatively large, signifying abundant deep-layer moisture. Because of the atmosphere s moisture and instability, conditions across the region are conducive for convection and its resulting lightning. The flash density pattern in Fig. 8d exemplifies a typical warm season sea breeze pattern. As in the composite case (Fig. 2a), there is an east-to-west decrease in densities. Maxima also are located near major metropolitan areas. In the western half of the region, areas of enhanced flash densities ( flashes km -2 regime-day -1 ) are over Houston and Lake Charles, with smaller values ( flashes km -2 regime-day -1 ) in between. As previously discussed, the swampy geographic features in the nearcoast areas cause weaker sea breeze circulations that take longer to form and evolve, producing less convection and lightning. Also, compared to some of the other flow regimes, the southwest category has relatively little offshore activity. Flash densities across the eastern portion of the region are greater than farther west (Fig. 8d). Maxima of flashes km -2 regime-day -1 are over New Orleans, in coastal Mississippi, and in a north-south line extending northward from Mobile. The Mississippi and Alabama maxima are consistent with findings of Medlin and Croft (1998) who noted that southwesterly low-level flow produced the greatest radar echo frequencies. The thermal circulation between the warm, shallow water and the land along the sharply-defined coastline forms, strengthens, and progresses inland during the morning. Convection typically begins at the time of maximum heating and instability. The convection is enhanced by interactions between the sea breeze and local circulations (i.e., bay breezes) and topography. Medlin and Croft (1998) also showed a maximum of convection on the north and northwest sides of Mobile Bay during southwesterly flow. They stated that convergence in this region is enhanced by interactions between the lowlevel flow and locally higher elevations. This convective maximum corresponds to the maximum in flash densities seen in Fig. 8d. 38

50 Hourly flash counts during southwest flow are the largest of the five regimes (Fig. 10). The distribution for the western half peaks at 2000 UTC (1500 CST) with flashes (Fig. 10a), with a secondary maximum between UTC ( CST). Analyses of hourly flash density maps (not shown) reveal that this early morning lightning is associated with near-shore convection. As with the previous regimes, the distribution for the eastern portion peaks one hour earlier than the western distribution. Thus, its peak of flashes occurs at 1900 UTC (1400 CST) (Fig. 10b). 3.6 Northwest Flow Days with hpa mean vector wind directions between 266º-356º are classified as having northwest flow. Northwest flow is a prodigious producer of lightning. However, the statistics vary considerably between the two halves of the region. The eastern portion (Table 3) has approximately 150% more days and double the number of flashes compared to the western portion (Table 2). Specifically, there are only 234 days with flashes in the western portion, with 80% of these producing lightning. On the other hand, the eastern section has 358 days and flashes (Table 3), with a larger percentage (85%) of days producing lightning. Furthermore, of all the directional flows, northwest flow has the largest mean flashes per day (3702 for the western portion and 4995 for the eastern portion). These means increase on days with lightning. The western half has a mean number of flashes per lightning day of 4608 (Table 2), while the eastern section has a mean of 5882 (Table 3). This prolific amount of lightning also is indicated by the daily flash counts in Fig. 6. Twelve days in the western portion (Fig. 6a) and eleven in the eastern half (Fig. 6b) have greater than flashes per day. Northwesterly flow over the study area typically has continental origins. The 1000 hpa analysis indicates (Fig. 7e) that this regime is not governed by the subtropical ridge. Instead, it is influenced by mid-latitude synoptic systems, with a trough of low pressure, most likely associated with individual wave cyclones, located over Alabama and Georgia. Northwest flow is most common during May and least common during July (western portion) and September (eastern portion) (Tables 5 and 6). Due to its continental origin, northwest flow is not as moist and unstable as its 39

51 southerly counterpart (southwest flow) (Table 4). However, it also is not as dry and stable as its easterly counterpart (northeast flow). Values of K-index and CAPE are similar for both halves of the study area (approximately 28º C and 1750 J kg -1, respectively). However, precipitable water is somewhat larger in the eastern half (1.70 in. versus 1.59 in.). The flash density map for northwest flow (Fig. 8e) shows very active regions of lightning. Sea breeze induced maxima are confined to the coastline by the large-scale flow. Values are largest in the eastern portion of the region. Densities greater than 0.08 flashes km -2 regime-day -1 extend along the coast from Biloxi to Mobile. In addition, there is considerable lightning farther offshore. Hourly maps (not shown) reveal that these offshore areas of enhanced flashes are associated with nighttime and morning convection. Relative minima of flash density are noted over Lake Pontchartrain and its shadow region, as well as over the Atchafalaya Basin. Hourly flash counts for two halves of the domain differ from one another (Fig. 10). Specifically, all of the previously discussed distributions peaked during the middle of the afternoon and exhibited an overnight minimum. However, the northwest flow distribution for the western portion of the area does not follow this pattern. Instead, it is fairly uniform, varying only by flashes between its peak and minimum (Fig. 10a). The peak flash count of occurs at 0000 UTC (1700 CST) and the minimum of at 1700 UTC (1200 CST), a time when flash counts normally are increasing to the afternoon peak. Conversely, the northwest distribution in the eastern section (Fig. 10b) resembles the other distributions. The peak number of flashes ( ) occurs at 2000 UTC (1500 CST), while a secondary peak (69 709) occurs at 1400 UTC (0900 CST). This smaller peak corresponds to the offshore lightning activity discussed above. 3.7 Lightning Maxima in Urban Areas As discussed in earlier sections (e.g., Fig. 2a), flash density maxima are located near the major urban areas of Houston, Lake Charles, Baton Rouge, New Orleans, Biloxi, and Mobile. Since these areas are centers of population and industry, it is important to understand the mesoscale processes that produce their enhanced lightning. 40

52 Understanding and better forecasting of lightning are goals of operational meteorologists along the Gulf Coast. Researchers began to investigate relationships between lightning and cities during the 1990s. Westcott (1995) documented lightning maxima around major midwestern cities, finding increases in lightning frequency over and downwind of the cities. She noted that the larger numbers of flashes were from convection that was more intense rather than from additional convective cells. Although she examined pollution and its effects on cloud microphysics, structural characteristics and urban heat island effects, and topography as possible causes for the city enhancements, no single factor was determined to be primarily responsible for the increased lightning. Watson and Holle (1996) and Livingston et al. (1996) studied flash distributions at several Georgia cities in preparation for the 1996 Summer Olympics. Both groups found maxima over the larger population centers and emphasized the distinct diurnal progression of lightning in these areas. Houston also has been the focus of lightning research in recent years (Orville et al. 2001; Steiger et al. 2002), where enhancements of 45% were noted over the city. They attributed the localized enhancements to several factors, including sea breeze convergence, frictional lift, urban heat island circulations, lack of available moisture, and increased levels of pollution particles. The relative importance of each influence and the degree of interaction between them were not established. This section investigates several urban flash maxima within the NWS Lake Charles and Slidell CWAs--Lake Charles, Baton Rouge, New Orleans, and Biloxi. City locations were shown in Fig. 1. Statistical parameters, flash densities, and diurnal distributions were examined for each urban area. Study boxes of approximately 36 km 60 km were superimposed on each city, and lightning data within each were compiled and analyzed separately. Flow regime classifications are the same as described earlier, with each day categorized according to the vector mean hpa wind at the nearest radiosonde site. Houston and Mobile are not emphasized because of the previous research done in those areas (Orville et al. 2001; Steiger et al. 2002; Gibson and Vonder Haar 1990; Medlin and Croft 1998). Lake Charles is an area of interest because several flows exhibit flash density maxima in its vicinity. The enhancement is best defined for the composite case (Fig. 2a) 41

53 but also is evident for the calm and southwest classifications (Figs. 8a and 8c). Specifically, during the 14-year period, flashes are recorded in the Lake Charles study box, with the most flashes (54 826) occurring during southwest flow (Table 7). This regime also has the largest percentage of lightning days of all five categories (50%). Calm flow also is a major lightning producer in Lake Charles ( flashes). The calm (light wind) category has the largest mean number of flashes per day (88) and per lightning day (180), as well as the largest median number of flashes per lightning day (73). Hourly flash densities for Lake Charles and the surrounding areas during southwest flow are shown in Fig. 11 for the period UTC ( CST). The white circle denotes the approximate city center of Lake Charles. Note that the area in the individual panels of Fig. 11 (as well as in Figs. 12, 13, and 14) does not correspond to the area of the study box. The larger region provides a frame of reference and better view of the hourly progression of flash densities in the area. Flash densities caused by the inland penetration of the sea breeze first appear in the coastal parishes of Louisiana at 1500 UTC (1000 CST) (not shown). However, there is no well-defined line of enhanced flash densities associated with the sea breeze. Rather, the areas of enhanced densities are discontinuous. Furthermore, the northward progression of sea breeze induced densities is not obvious. Additionally, flash densities associated with southwest flow are weaker near Lake Charles than near other regions (e.g., Biloxi) because of the geography of coastal Louisiana (Fig. 8d). As previously mentioned, the swampy nature of the region tempers thermal contrasts, weakening the forcing required for focused convection. Flash densities near Lake Charles begin to increase at 1700 UTC (1200 CST) (Fig. 11b), reaching a peak of flashes km -2 regime-day -1 at 1800 UTC (1200 CST) (Fig. 11c). Even though these flash densities are small, they are the greatest in the area. The localized enhancement along the poorly defined sea breeze may be due to the industry and pollution associated with the city. Lake Charles and its suburbs have many petrochemical plants, and the literature (e.g., Orville et al. 2001; Steiger et al. 2002) suggests that lightning enhancements occur within cities having major industry. By 2000 UTC (1500 CST) (Fig. 11e), the enhanced flash densities have shifted somewhat 42

54 TABLE 7. Statistical parameters for individual study boxes centered over Lake Charles, Baton Rouge, New Orleans, and Biloxi. Each box is approximately 36 km 60 km. Flow No. No. flow Mean % ltng Mean fls Median fls Regime flashes days fls day -1 days ltng day -1 ltng day -1 LAKE CHARLES Calm NE SE SW NW All days BATON ROUGE Calm NE SE SW NW All days NEW ORLEANS Calm NE SE SW NW All days BILOXI Calm NE SE SW NW All days

55 FIG. 11. Hourly flash density maps (flashes km -2 regime-day -1 ) for Lake Charles and surrounding areas for southwest flow between UTC ( CST). Each map represents a one-hour time period. The white circle denotes the approximate location of the city s center. 44

56 northward, with the only remaining enhancement extending northeast from Houston. The flash density maximum over the Lake Charles study box is depicted in the hourly counts of Fig. 15a. At 1800 UTC (1300 CST), values for southwest flow peak at 5434 flashes, the largest number of flashes per hour of all five regimes. Much weaker peaks for southwest and northwest flow occur several hours later during the afternoon. In summary, the Lake Charles flash maximum appears to result from interactions between the low-level flow, sea breeze induced convection, and urban processes. The second urban area being studied, Baton Rouge, exhibits flash density maxima in the composite case (Fig. 2a), during calm flow (Fig. 8a), and with northwest flow (Fig. 8e). The location of Baton Rouge is somewhat different from the other three cities being examined since it is farthest inland and is not adjacent to any large bodies of water. The only significant geographic feature is the Mississippi River, which flows on the western side of the city. Like Lake Charles, Baton Rouge is a major industrial area, with many petrochemical plants located along the Mississippi River. The composite flash count for Baton Rouge ( ) is similar to that of Lake Charles (Table 7). Southwest and northwest flow days have the largest numbers of flashes ( and , respectively). Although southwest flow has the greater number of flashes, northwest flow yields the largest means (73 flashes per day and 196 flashes per lightning day) and largest median flashes per lightning day (79). Thus, even though northwest flow does not occur frequently, a significant amount of lightning occurs on those days. This is consistent with the scatter diagram for the eastern half of the region (Fig. 6b). Hourly flash densities for northwest flow are given in Fig. 12. At 2000 UTC (1500 CST) (Fig. 12a), densities are greatest on the northwest side of Lake Pontchartrain. Convergence in this region is enhanced by the interaction of lake breezes from Lake Pontchartrain and Lake Maurepas (just west of Lake Pontchartrain) and the large-scale northwesterly flow. Flash densities over Baton Rouge at this time are minimal. By 2100 UTC (1800 CST), flash densities increase around Lake Pontchartrain and in Baton Rouge (Fig. 12b). Flash densities over Baton Rouge, especially on its southeastern side, now have magnitudes of flashes km -2 regime-day -1. The area of enhanced flash densities strengthens and increases, and by 2300 UTC (1900 CST), a maximum of 45

57 FIG. 12. Hourly flash density maps (flashes km -2 regime-day -1 ) for Baton Rouge and surrounding areas for northwest flow between UTC ( CST). Each map represents a one-hour time period. The white circle denotes the approximate location of the city s center. 46

58 flashes km -2 regime-day -1 is located over the city (Fig. 12d). Enhanced convergence from outflow boundaries of convection near Lakes Maurepas and Pontchartrain, as well as urban enhancements, may be responsible for these large flash densities. During the next hour, densities decrease dramatically over the area, and by 0000 UTC (1900 CST), values over Baton Rouge and surrounding areas are only ~ flashes km -2 regime-day -1 (Fig. 12e). The 2300 UTC peak (5689 flashes) and subsequent decrease in lightning for northwest flow also are illustrated in Fig. 15b. The peak for southwest flow occurs earlier, at 2000 UTC (1500 CST). To summarize, lake and urban processes appear to largely influence and enhance lightning in the Baton Rouge area. The sea breeze does not appear to be a significant factor. New Orleans, located along the southern edge of Lake Pontchartrain, is the largest in area and population of the four cities being examined. The Mississippi River runs through the city, and the Gulf of Mexico is in close proximity. Maxima are evident in several of the previously discussed flash density maps, including the composite case (Fig. 2a), calm flow (Fig. 8a), southwest flow (Fig. 8d), and northwest flow (Fig. 8e). Southwest flow is the most common and the most prodigious lightning producer for the New Orleans study box, with 719 days and total flashes. The mean number of flashes per day is 68, the largest of all five flows. Its median flashes per lightning day ranks second behind northwest flow. Enhanced flash densities during southwest flow first appear over New Orleans at 1600 UTC (1100 CST) (Fig. 13a). The densities then increase in New Orleans and the other sides of Lake Pontchartrain (Figs. 13b and 13c) between UTC ( CST). These areas of enhancement probably are due to locally increased convergence from the lake breeze and its interaction with the southwest flow. The sea breeze likely is a factor as well, although it is poorly defined in the hourly maps. By 0900 UTC (1400 CST) (Fig. 13d), values are maximized over New Orleans ( flashes km -2 regime-day -1 ). Three flows appear to anchor this maximum over the city: the sea breeze, the lake breeze, and the large-scale flow. Even though there is the potential for major urban lightning enhancement, the total flash count is similar to those of the smaller cities described earlier. This lack of a major increase may be due to New Orleans proximity to Lake Pontchartrain, which gives the atmosphere a relatively 47

59 FIG. 13. Hourly flash density maps (flashes km -2 regime-day -1 ) for New Orleans and surrounding areas for southwest flow between UTC ( CST). Each map represents a one-hour time period. The white circle denotes the approximate location of the city s center. 48

60 maritime (stable) flavor. This is consistent with the results of Westcott (1995) who noted weaker urban lightning enhancements associated with cities adjacent to major lakes. Hourly flash counts during southwest flow (Fig. 15c) begin to increase at 1300 UTC (0800 CST) to the afternoon peak of 7226 flashes at 1900 UTC (1400 CST). Each of the other flow regimes exhibits a weaker peak near this same time. Hence, lightning patterns and distributions in the New Orleans area are driven by sea and lake breeze circulations and the prevailing low-level flow. Flash counts (and thus flash density values) are larger in coastal Mississippi and Alabama than any other part of the study area because of the stronger nature of the forcing in the eastern region. The coastline is sharply defined, producing strong thermal circulations and strong convergence zones compared to areas farther west. Thus, Biloxi has the most lightning ( flashes) of all four cities (Table 7). Both southwest and northwest flow produce substantial lightning. Southwest flow has the most days (719) and flashes (58 988), with 56% of these days having lightning. However, northwest flow, though less frequent, tends to produce more lightning. Fifty percent of the 358 northwest flow days produce a total of flashes. The mean flashes per day (119), mean flashes per lightning day (239), and median flashes per lightning day (93) are the largest of all five flows. Calm flow also is an active producer of lightning, with means and the median ranking second of five flows. Large-scale northwesterly flow confines the sea breeze circulation along the coastline. Since Biloxi is a coastal city, it experiences large amounts of lightning during this scenario. This is evident in the hourly flash density maps given in Fig. 14. The inland penetration of the sea breeze is first evident at 1800 UTC (1300 CST) (Fig. 14b). Flash densities associated with the sea breeze related convection quickly intensify, reaching peak values of flashes km -2 regime-day -1 at 1900 UTC (1400 CST) (Fig. 14c). From UTC ( CST) (Figs. 14d and 14e), values across coastal Mississippi and Alabama still are quite large. However, the maximum now is located east of Biloxi. Flash counts in the Biloxi study box for northwest flow (Fig. 15d) show a peak of 8130 flashes at 1900 UTC (1400 CST). Since Biloxi is a city with growing industry and population, localized urban effects are becoming increasingly important, in addition to influences by sea breeze circulations and the large-scale flow. 49

61 FIG. 14. Hourly flash density maps (flashes km -2 regime-day -1 ) for Biloxi and surrounding areas for northwest flow between UTC ( CST). Each map represents a one-hour time period. The white circle denotes the approximate location of the city s center. 50