USF Libraries
USF Digital Collections

The influence of the El Niño-southern oscillation on cloud-to-ground lightning activity along the Gulf Coast of the Unit...

MISSING IMAGE

Material Information

Title:
The influence of the El Niño-southern oscillation on cloud-to-ground lightning activity along the Gulf Coast of the United States
Physical Description:
Book
Language:
English
Creator:
Lajoie, Mark R
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla.
Publication Date:

Subjects

Subjects / Keywords:
climatology
enso
flash density
La Niña
teleconnection
Dissertations, Academic -- Geography -- Masters -- USF   ( lcsh )
Genre:
government publication (state, provincial, terriorial, dependent)   ( marcgt )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Summary:
ABSTRACT: This study investigates the response of lightning to the El Niño Southern Oscillation (ENSO) in the vicinity of the U.S. Gulf Coast region and nearby adjacent waters of the Gulf of Mexico, for the years 1995-2002. The Gulf Coast region was selected for this study because of its high flash density (Orville and Huffines, 2001) and because it is an area where the ENSO fingerprint is very clearly demonstrated on both temperature and precipitation patterns (CPC, 2002). Additionally, this geographic domain roughly matches the only known study on this topic (Goodman et al., 2000). Winter is the season of greatest response to ENSO (CPC, 2004), and past studies show that summer has the most lightning activity (e.g., Orville and Huffines, 2001). The temporal domain of the study is restricted to 1995 and beyond, as this follows a system-wide upgrade of the National Lightning Detection Network (NLDN) that improved overall flash detection efficiency (Cummins, et. al.1998; Wacker and Orville, 1999). Both qualitative and quantitative methods were employed to explore the lightning data for ENSO teleconnections. Mean flash density maps were constructed for the complete period of record, individual months and the winter and summer seasons. Maps were visually examined for qualitative comparison with past climatologies and the Goodman et al., (2002) ENSO study. Additionally, monthly flash deviations are computed, visualized and correlated with the Niño 3.4 SST anomaly for all months in the study, seeking out variations in both the amount of flash deviation and spatial properties. Abundant literature exists on both ENSO and lightning individually. This study offers an insight into their intersection.
Thesis:
Thesis (M.A.)--University of South Florida, 2004.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Mark R. Lajoie.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 92 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 56137634
notis - AJR7187
usfldc doi - E14-SFE0000363
usfldc handle - e14.363
System ID:
SFS0025057:00001


This item is only available as the following downloads:


Full Text

PAGE 1

The Influence of the El Nio-Southern Oscillation on Cloud-to-Ground Lightning Activity along the Gulf Coast of the United States by Mark R. LaJoie A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Geography College of Arts and Sciences University of South Florida Major Professor: Arlene G. Laing, Ph.D. Steven Reader, Ph.D. Graham A. Tobin, Ph.D. Date of Approval: May 14, 2004 Keywords: climatology, ENSO, flas h density, La Nia, teleconnection Copyright 2003, Mark R. LaJoie

PAGE 2

ACKNOWLEDGEMENTS This research would not have been comp leted but for the assistance of several individuals at key milestones. I am especially indebted to my major advisor, Dr. Arlene Laing, who patiently nudged me along to the finish line over the last several years. Special thanks as well to Dr. Steven Reader and Dr. (Major) Karl D. Pfeiffer. Without their software and computer programming e xpertise I would still be struggling with multiple gigabytes of raw data.

PAGE 3

TABLE OF CONTENTS LIST OF TABLES iii LIST OF FIGURES iv ABSTRACT vii CHAPTER ONE: INTRODUCTION 1 Purpose of the Study 2 Research Problem 4 Research Aim and Objectives 4 Structure of the Thesis 5 CHAPTER TWO: LITERATURE REVIEW 6 The ENSO Cycle 6 El Nio and La Nia 6 Impacts of ENSO on North America 7 Measuring, Quantifying, and Categorizing ENSO Events 11 The Tropical Atmosphere Ocean (TAO)/TRITON Buoy Array 12 The NIO Regions 13 The Oceanic NIO Index (ONI) 14 Lightning 16 The National Lightning Detection Network (NLDN) 17 Lightning Climatology 20 The Development of Lightning Climatologies 20 U.S. Lightning Climatologies 22 Lightning Climatologies for the Southeast U.S. 26 Florida CG Lightning Climatologies 26 Lightning Climatologies for other Areas of the Southeast U.S. 30 Impact of ENSO on Lightning Distributions 33 CHAPTER THREE: METHODOLOGY 35 Geographic Domain 35 Data 36 Description of Data 36 Data Pre-Processing 36 Data Analysis 37 i

PAGE 4

CHAPTER FOUR: RESULTS 39 Lightning Climatology 39 Mean and Annual Lightning Climatology 39 ENSO Correlations 50 Winter 51 December 51 January 53 February 54 Comparison with Goodman et al., (2000), Winter Season 55 Summer 60 June 60 July 61 August 63 CHAPTER FIVE: CONCLUSION 65 Implications 67 Future Research 68 REFERENCES 69 APPENDICIES 75 APPENDIX A: NINO 3.4 TIME SERIES 76 APPENDIX B: PERL SCRIPT 79 APPENDIX C: SPLUS SCRIPT 82 ii

PAGE 5

LIST OF TABLES Table 1: Cold and Warm Episodes based on ONI (Derived from CPC, 2004) 16 Table 2: CG Lightning 1995-2002 (54,943,981 flashes) 44 Table 3: NINO 3.4 VALUES 76 iii

PAGE 6

LIST OF FIGURES Figure 1: El Nio, Normal and La Nia conditions (From NOAA, 2004) 7 Figure 2: ENSO Teleconnections, Warm Episode Winter (From CPC, 2002). 9 Figure 3: ENSO Teleconnections, Warm Episode Summer (From CPC, 2002) 9 Figure 4: ENSO Teleconnections, Cool Episode Winter (From CPC, 2002) 10 Figure 5: ENSO Teleconnections, a Cool Episode Summer (From CPC, 2002) 10 Figure 6: The TAO/Triton Array (From Magnun, 1998) 13 Figure 7: Graphical Depiction of Nio Regions (CPC, 2002) 14 Figure 8: NLDN Sensor Network (from Orville and Huffines 2001) 18 Figure 9: NLDN sample record 19 Figure 10: NLDN Location Errors (from Zajac et al., 2002) 19 Figure 11: NLDN Detection Efficiency (from Zajac et al., 2002) 20 Figure 12: Mean Annual Flash Density (Orville and Huffines, 2001) 23 Figure 13: Mean Measured Annual Flash Density (Zajac and Rutledge, 2001) 24 Figure 14: Vaisala-GAI Lightning Climatology (1996-2000) 25 Figure 16: Mean flash density for northern Florida (from Camp et al., 1998) 29 Figure 17: Warm Season Flash density for Gulf Coast (from Stroupe, 2003) 30 Figure 18: Jul-Aug Flash Density 1986-1994 (from Watson and Holle, 1996) 31 Figure 19: (Top) Steiger and Orville (2003); (Bottom) Steiger et al., (2002) 32 Figure 20: Lightning Days, DJF (Goodman et al., 2000) 34 Figure 21: Study Area 35 Figure 22: Mean 1000mb geopotential heights. Top, 1995-2002; 44 iv

PAGE 7

Figure 23: Average Flashes by Month, 1995-2002 45 Figure 24: Average Flashes by Year, 1995-2002 45 Figure 25: Mean Annual Flash Density, 1995-2002 (54, 943, 981 flashes) 46 Figure 26: Mean Flash Density, 1995 46 Figure 27: Mean Flash Density, 1996 47 Figure 28: Mean Flash Density, 1997 47 Figure 29: Mean Flash Density, 1998 48 Figure 30: Mean Flash Density, 1999 48 Figure 31: Mean Flash Density, 2000 49 Figure 32: Mean Flash Density, 2001 49 Figure 33: Mean Flash Density, 2002 50 Figure 34: December Mean CG Flash Density, 1995-2002 52 Figure 35: December SSTLightning Correlation, 1995-2002 52 Figure 36: January Mean CG Flash Density, 1995-2002 53 Figure 37: January SSTLightning Correlation, 1995-2002 54 Figure 38: February Mean CG Flash Density, 1995-2002 55 Figure 39: February SSTLightning Correlation, 1995-2002 55 Figure 40: DJF Mean Flash Density, 1995-1996 57 Figure 41: DJF Mean Flash Density, 1996-1997 57 Figure 42: DJF Mean Flash Density, 1997-1998 58 Figure 43: DJF Mean Flash Density, 1998-1999 58 Figure 44: DJF Mean Flash Density, 1999-2000 59 Figure 45: DJF Mean Flash Density, 2000-2001 59 Figure 46: DJF Mean Flash Density, 2001-2002 60 v

PAGE 8

Figure 47: June SST Mean CG Flash Density, Jun 1995-'02 61 Figure 48: June SSTLightning Correlation, 1995-2002 61 Figure 49: July Mean CG Flash Density, 1995-2002 62 Figure 50: July SST-Lightning Correlation, 1995-2002 63 Figure 51: August Mean CG Flash Density 1995-2002 64 Figure 52: August SSTLightning Correlation, 1995-2002 64 vi

PAGE 9

The Influence of the El Nio -Southern Oscillation on Cloud-to-Ground Lightning Activity along the Gulf Coast of the United States Mark R. LaJoie ABSTRACT This study investigates the response of lightning to the El Nio Southern Oscillation (ENSO) in the vicinity of the U.S. Gulf Coast region and nearby adjacent waters of the Gulf of Mexico, for the years 1995-2002. The Gulf Coast region was selected for this study because of its high flash density (Orville and Huffines, 2001) and because it is an area where the ENSO fingerprint is very clearly demonstrated on both temperature and precipitation patterns (CPC, 2002). Additionally, this geographic domain roughly matches the only known study on this topic (Goodman et al., 2000). Winter is the season of greatest response to ENSO (CPC, 2004), and past studies show that summer has the most lightning activity (e.g., Orville and Huffines, 2001). The temporal domain of the study is restricted to 1995 and beyond, as this follows a system-wide upgrade of the National Lightning Detection Network (NLDN) that improved overall flash detection efficiency (Cummins, et. al. 1998; Wacker and Orville, 1999). Both qualitative and quantitative methods were employed to explore the lightning data for ENSO teleconnections. Mean flash density maps were constructed for the complete period of record, individual months and the winter and summer seasons. Maps were visually examined for qualitative comparison with past vii

PAGE 10

climatologies and the Goodman et al., (2002) ENSO study. Additionally, monthly flash deviations are computed, visualized and correlated with the Nio 3.4 SST anomaly for all months in the study, seeking out variations in both the amount of flash deviation and spatial properties. Abundant literature exists on both ENSO and lightning individually. This study offers an insight into their intersection. viii

PAGE 11

CHAPTER ONE INTRODUCTION El Nio originally referred to a noticeable warming of the coastal waters off Peru and Ecuador. The warm current usually appeared around Christmastime, and was noticed by local inhabitants for centuries (Philander, 1990). In recent decades, however, researchers have discovered that the equatorial Pacific actually fluctuates between warm (El Nio), neutral and cool (La Nia) states over periods of three to seven years (Trenberth, 1997). These sea surface temperature fluctuations are part of a large-scale, interrelated disruption of typical surface pressure (Southern Oscillation), winds and precipitation patterns that alters global atmospheric circulations and is one of the leading sources of climate disruptions and weather variability across the globe (COMET, 2003). Acknowledging the coupled nature of this ocean-atmosphere system, the entire phenomenon has come to be known as the El Nio-Southern Oscillation, or ENSO (McPhaden, 2002). Beyond investigations of the physical aspects of ENSO, many studies, both formal and informal, have credibly linked environmental hazards with ENSO at a variety of scales. In the United States specifically, ENSO is known to influence temperature and precipitation distributions (Ropelewski and Halpert, 1986; Green et al., 1997; Smith, et al., 1999); cloud cover (Angell and Korshover, 1987); and a wide range of meteorological hazards including tornados (Agee and Zurn-Birkheimer, 1998; Hagemeryer 1998; Schaefer and Tatom, 1998), hurricanes (OBrien, et al., 1996; Pielke, 1999), and even snow pack (Cayan, 1996) and wildfires (Harrison and Meindl, 2001). A study conducted by Harrison and Larkin (1998) on the intense 1

PAGE 12

1997-1998 event established that about 90% of the U.S. has at least one statistically significant historical weather association with El Nio. By extension, researchers have, correctly or incorrectly, attempted to draw direct associations between these ENSO spawned hazards and a host of societal impacts, including effects of ENSO on economics (Changnon, 1999) agriculture (Legler et al., 1997) and disease (Diaz and McCabe, 1999). Certain sections of the media have sensationalized ENSO and its apparently hazardous impacts. Indeed, severe and unusual weather events make good copy, and ENSO is an opportune scapegoat (Kumar, 1997). Purpose of the Study Melodrama and sensationalist claims aside, it appears established that ENSO exerts a profound influence on global weather and climate (COMET, 2003). However, although ENSO has been linked to other natural phenomena and hazards, as discussed, little research has been published on the possible relationship between ENSO and lightning. Lightning poses substantial threat to lives, commercial activities and personal property (Curran et al., 2000). According to the National Climatic Data Center (NCDC) publication Storm Data, there were 44 fatalities and 233 injuries nationwide from lightning during 2003 alone (National Weather Service, 2003). Lightning caused $25.6 million dollars in property damage as well. Curran et al., (2000) used Storm Data to summarize lightning casualties and damages from across the United States from 1959-1994. They calculated 3,239 deaths, 9,818 injuries and 19,814 reports of property-damage from lightning strikes across the period of record. Lightning safety experts estimate that there are 10 times as many lightning-related injuries than there are deaths, as it is 2

PAGE 13

generally assumed that injuries are under-reported (Curran et al., 2000). Moreover, a separate study from the Centers for Disease Control determined that over 30% of individuals struck by lightning die, and 74% of lightning strike survivors are left with permanent disabilities (CDC, 1998). These statistics alone demonstrate the hazardousness of lightning. An improved understanding of general lightning distributions could help mitigate this hazard. Geographic distributions of lightning are influenced by a number of factors. Zajac et al., (2002) outlined several important climate controls that influence lightning distributions, paraphrased below. 1. Latitude: solar insolation, which influences low level instability and convection, varies significantly with latitude 2. Continentality: the effect on lightning and thunderstorm caused by land and water surface distributions. Differential heating between land and water masses generates atmospheric circulations such as land, sea and lake breezes 3. Moisture: required for convection. Distance and bearing from moisture sources such as the Gulf of Mexico and Pacific are a prime influence on thunderstorm activity 4. Ocean currents: affect location of synoptic controls such as the subtropical high. Sea surface temperatures affect low-level instability 5. Terrain: can enhance or alter atmospheric circulations, in turn impacting thunderstorm activity and distributions. Zajac et al., (2002) The above factors shape lightning distributions (Zajac et al., 2002), and ENSO is known to influence a broad spectrum of climate and weather conditions. Does the ENSO cycle influence lightning through direct or indirect action on these climate controls, or perhaps other factors? 3

PAGE 14

Research Problem It has been briefly argued that both ENSO and lightning are hazards with significant impact on societies and individuals. The purpose of this study is to determine whether a relationship exists between the two. The rationale for this research topic stems from the plausibility of such a relationship given the established link between lightning activity and various climatic controls, and from the viewpoint that lightning constitutes a significant natural hazard. Only one published study is known to specifically examine the possible ENSO-lightning connection over the U.S. Goodman et al., (2000) examined winter season (December-January-February) lightning data over the Southeast United States and found significant increases in the frequency of lightning days along the Gulf Coast and adjacent waters during the intense 1997-98 ENSO event. While the study is well constructed, a possible shortcoming is that it focuses on thunderstorm days, rather than raw lightning activity. Thunderstorm days are not an ideal index of lightning, since no distinction is drawn between a single stroke of lightning in a day, or a prolonged storm. Such an approach may not be able to reliably indicate a lightning-ENSO link. Research Aim and Objectives The overall aim of the project is to increase knowledge and understanding concerning the influence of the ENSO cycle on lightning activity along the US Gulf Coast. In order to achieve this aim, the project has several research objectives, as follows: 4

PAGE 15

To establish the pattern of lightning activity in the Gulf Coast region for the 8 years from January 1995 to December 2002, using data from the Air Force Combat Climatology Center lightning flash database. To calculate the month-by-month correlations between lightning activity and ENSO for winter and summer seasons using the lightning data and a concurrent 96-month series of sea surface temperature anomaly values from the equatorial Pacific extracted from the Climate Prediction Center (CPC, 2003) database. To determine the significance of the correlations and their temporal and geographic variations for the Gulf Coast region during the warm and cool seasons Structure of the Thesis This initial chapter provided a brief background to the topic, and explained and justified the research problem addressed in the study. Chapter 2 contains a review of the literature relevant to this study, including the characteristics of ENSO, and the climatology of lightning with particular reference to the Gulf Coast region. Chapter 3 describes and explains the methodology used in the study, including the datasets used, the use of ArcMap GIS and the statistical methods utilized. Chapter 4 contains the data analysis and results, including an exposition of lightning distributions, ENSO correlation patterns and the relationships between them. Chapter 5 discusses the results in the context of the findings of previous studies, and assesses the implications of these results for existing and future research. 5

PAGE 16

CHAPTER TWO LITERATURE REVIEW The ENSO Cycle ENSO is a natural phenomenon, a coupled atmospheric-oceanic process caused by recurring redistributions of heat and atmospheric momentum in the Equatorial Pacific (McPhaden, 2002). The oceanic component, El Nio, refers to a periodic warming of sea surface temperatures in the Equatorial Pacific. The Southern Oscillation is the atmospheric component, involving a seesawing of surface pressure across the Equatorial Pacific and occurring in concert with El Nio. It is not certain whether El Nio causes the Southern Oscillation or viceversa but an observation of an anomaly of one heralds the others arrival (McPhaden, 2002). ENSO triggers major shifts in tropical rainfall patterns and deep convection, disrupting atmospheric circulations and climate across the globe. El Nio and La Nia While the term El Nio has historically (and popularly) been used to describe the oceanic component of ENSO, the Equatorial Pacific in fact has been found to phase between a normal and two extreme states, El Nio and La Nia (COMET, 2003). In typical years, the trade winds (tropical easterlies) force the surface waters of the Equatorial Pacific eastward. This promotes an upwelling of deep, cold water from the ocean bottom to surface, instigating atmospheric subsidence and persistent high pressure over the region (Figure 1). The increased atmospheric stability suppresses atmospheric convection and rainfall (COMET, 2003). 6

PAGE 17

During an El Nio, or warm ENSO event, this characteristic pattern collapses. The trade winds fail or even reverse, and cold-water upwelling is capped. The consequence is anomalously high sea surface temperatures, which fuel convection and generate regional low pressure in the eastern tropical Pacific (Figure 1). This instability culminates in increased precipitation over the region (COMET, 2003). Conversely a La Nia, or a cool ENSO event, can be described as an extreme case of the normal, and is really an enhancement of processes that occur during neutral years (COMET, 2003). During La Nia, the trade winds strengthen and cold upwelling is enhanced (Figure 1). The result is a more pronounced regional high pressure and an increase in atmospheric stability over the eastern Tropical Pacific. This serves to even further suppress convection and precipitation across the region (COMET, 2003). Figure 1: El Nio, Normal and La Nia conditions (From NOAA, 2004) Impacts of ENSO on North America ENSO events disrupt global circulations, altering typical patterns of temperature and precipitation worldwide. This link with distant events is referred to as a teleconnection, defined in the Glossary of Meteorology (AMS, 2000) as either a linkage between weather changes occurring in widely separated regions of the globe, or a significant positive or negative correlation in the fluctuations of a field 7

PAGE 18

at widely separated points. most commonly applied to variability on monthly and longer timescales, the name refers to the fact that such correlations suggest that information is propagating between the distant points through the atmosphere. Although global ENSO teleconnections are certainly discussion-worthy, given the regional emphasis of this study and its focus on the potential impact of ENSO on lightning activity in the southeast United States, the discussion here is limited to North American ENSO impacts. The primary ENSO teleconnection with North America is a displacement or disruption of the jet stream (COMET, 2003). ENSO sea surface temperature anomalies (warm or cool) typically commence during late spring or summer, peaking and reaching their maximum areal extent over the tropical Pacific during the Northern Hemisphere winter. The event usually ends by the following summer. ENSO teleconnections (during either warm or cool events) become most evident over North America during this wintertime peak (CPC, 2002). Figure 2 summarizes typical ENSO teleconnections over North America during a warm episode winter. The subtropical jet stream shifts southward, becomes more zonal, and strengthens, displacing the typical storm track from the northern to southern United States and enhancing moisture flow from the Pacific. The consequence is a cooler, wetter, and stormier southern tier of the United States (COMET, 2003). During the Northern Hemisphere summer, warm episodes appear to have negligible effect (Figure 3). Teleconnections are confined primarily to the Caribbean and Southern Hemisphere (CPC, 2002). 8

PAGE 19

Figure 2: ENSO Teleconnections, Warm Episode Winter (From CPC, 2002). Figure 3: ENSO Teleconnections, Warm Episode Summer (From CPC, 2002) La Nia has a reverse effect on North America compared with El Nio, although some variability is evident (COMET, 2003). During a La Nia year, the upper level flow becomes more meridional; the jet stream shifts north (entering the continent over the Pacific Northwest) and becomes more variable in intensity. Much of North America is colder and stormier, with increased precipitation due to the northward shift of the storm track (COMET, 2003). The southern tier of the U.S., however, can expect warmer than normal temperatures, as well as decreased 9

PAGE 20

storminess and precipitation (Figure 4). During the summer, cool episodes appear to have negligible, if any, effect on North America (Figure 5). As with the El Nio summer, teleconnections appear to be confined primarily to the Caribbean and Southern Hemisphere (CPC, 2002). Figure 4: ENSO Teleconnections, Cool Episode Winter (From CPC, 2002) Figure 5: ENSO Teleconnections, a Cool Episode Summer (From CPC, 2002) 10

PAGE 21

While the focus here is upon the most evident winter and summer ENSO teleconnections over North America, some studies have shown that ENSO may also alter precipitation and temperature patterns over North America during other seasons, albeit on a more localized scale. Green et al., (1997) analyzed mean monthly temperature and precipitation totals for 788 North American weather stations for the time period 1947-1986, and classified each season into neutral, cold or warm phases. In addition to detecting the major features of the winter and summer anomalies as discussed above, Green et al., (1997) found enhanced precipitation in regions along the eastern seaboard, including Florida, during springs that followed warm events. Portions of South Texas were found to be cooler than normal. North Texas and northern Alabama were found to be drier during El Nio springs than in neutral years. For cool events, Green et al., (1997) established that the anomalies are sometimes reversed from those associated with warm events, but not everywhere. In spring, warm anomalies are found in northern Florida, Georgia, and South Carolina, and gulf coast regions may exhibit increased precipitation. In summer, the extreme south US is colder than normal, with enhanced precipitation in the southeast, but dry conditions in Texas and Louisiana. Measuring, Quantifying, and Categorizing ENSO Events El Nio and La Nia encompass a wide range of climatic conditions. Philander (1990) states that the amplitude of various El Nios can vary greatly, and offers the useful analogy that the terms El Nio and La Nia are useful in the same way that winter is useful, even though each winter is distinct. Further, he observes that, Due to differences in strength, and the differences in sea surface temperature, 11

PAGE 22

rainfall and pressure, it is best to avoid absolute, strict definitions and to accept the termsas general and qualitative (Philander, 1990). Nevertheless, it is useful for researchers to capture the relative strengths of ENSO events quantitatively so as to compare them with each other and with the normal state. Researchers can accurately monitor ENSO because it impacts so many facets of the ocean-atmosphere system in a conspicuous fashion (Philander, 1990). ENSO is sizable in temporal scope (two to 5-year time scale) and geographic/areal extent (entire Tropical Pacific), and the impacts are global (Hanley et al., 2002). Variables routinely monitored include sea surface temperature (SST) fluctuations, surface pressure shifts, trade wind intensity variations, and many other elements, using direct observations as well as remote sensing techniques (Magnun, 1998). The ENSO index (a measure of environmental conditions during ENSO events) used for this study is derived from direct SST measurements from an array of moored buoys in the equatorial Pacific. Therefore, both the array and the index are described briefly below. The Tropical Atmosphere Ocean (TAO)/TRITON Buoy Array In the past, in situ measurements of ocean surface temperatures were obtained primarily from ship observations. This was not ideal for obtaining quality data, as availability was sporadic at best, being reliant upon the chance proximity of ship traffic and the whims of the crews. Today, real-time data are obtained from a number of sources, including an advanced array of moored ocean buoys stretching across the equatorial Pacific. The Tropical Atmosphere Ocean (TAO)/TRITON buoy array is a multi-national collaboration between United States, Japan, Korea, Taiwan, and France, and 12

PAGE 23

consists of nearly 70 moored buoys (Magnun, 1998) (Figure 6). The buoys measure a full complement of atmospheric elements including surface winds, sea surface temperature, currents, ambient air temperature, and humidity, and transmit the data back to terrestrial data receivers. Availability of data in real-time from the TAO/Triton array has proven to be an enormous benefit for detecting and predicting climate variations in the Pacific. Figure 6: The TAO/Triton Array (From Magnun, 1998) The NIO Regions Exploiting data from the TAO/TRITON array, satellites, and other sources to full advantage, researchers have created indices based upon sea surface temperature departures in the tropical Pacific in order to monitor the ocean-atmosphere response to the ENSO cycle. According to the International Research Institute (IRI) for Climate Prediction, four key regions are monitored (Figure 7), each for specific ENSO responses, as follows: NIO1+2 (0-10S, 80-90W): Typically first to warm when an El Nio event develops. 13

PAGE 24

NIO3 (5S-5N; 150W-90W): The region of the tropical Pacific with the largest SST variability on El Nio time scales. NIO3.4 (5S-5N; 170W-120W): The region that has large variability on El Nio time scales, and that is closer (than NIO3) to the region in which changes in local sea-surface temperature are important for shifting the large region of rainfall typically located in the far western Pacific. NIO4 (5S-5N: 160E-150W): The region where changes of sea-surface temperature lead to total values around 27.5C, which is thought to be an important threshold in producing rainfall. (IRI, 2003) Figure 7: Graphical Depiction of Nio Regions (CPC, 2002) The NIO 3.4 region is considered to be most suitable for monitoring climate variability on global scales, as SST variability here signals the strongest effect on shifting precipitation patterns from the west to the central Pacific (IRI, 2003). It has also been observed that prognostic numerical weather models exhibit the highest skill when initializing with data from this region (IRI 2003; COMET, 2003). The Oceanic NIO Index (ONI) A confusing assortment of approaches to define and classify ENSO events has been used by various researchers, ranging from purely qualitative to the strict quantitative (Trenberth, 1997). Most frequently, the state and intensity of ENSO has been classified using indices derived from sea surface temperatures or sea level 14

PAGE 25

pressure differences across the equatorial tropical Pacific (Hanley et al., 2002). However, multiple indices are in use, and little consensus exists as to which best captures the state and intensity of ENSO (Hanley et al., 2002). The debate will undoubtedly continue, as new indices are introduced and the established ones undergo continual refinements (Trenberth, 1997). However, in September 2003 NOAA, in collaboration with a team of experts from academia and federal government agencies, announced that an agreement had been reached on a standardized operational index for ENSO, and definitions for both El Nio and La Nia based upon the index (Department of Commerce, 2003). The group decided to use the NIO 3.4 index as a baseline because anomalies in this region of the tropical Pacific exhibit the strongest ENSO signal on global circulation patterns and teleconnections. The index was formally defined as the three-month average of sea surface temperature departures from normal in the NIO 3.4 region of the Pacific (Department of Commerce, 2003). The new index was designated the Oceanic NIO Index (ONI), and formal definitions for El Nio and La Nia were also agreed upon (Department of Commerce, 2003). Based upon ONI, the NOAA operational definitions for El Nio and La Nia are as follows: El Nio: A phenomenon in the equatorial Pacific Ocean characterized by a positive sea surface temperature departure from normal (for the 1971-2000 base period) in the Nio 3.4 region greater than or equal in magnitude to 0.5C, averaged over three consecutive months. La Nia: A phenomenon in the equatorial Pacific Ocean characterized by a negative sea surface temperature departure from normal (for the 1971-2000 base period) in the Nio 3.4 region greater than or equal in magnitude to 0.5C, averaged over three consecutive months. Department of Commerce (2003) 15

PAGE 26

Table 1 depicts warm and cold episodes from 1995-2002 based on ONI. Actual numerical values are at Appendix A. From 1995-2002, based on ONI, there were 23 warm episodes, 36 cool episodes and 37 neutral episodes. Note the extended cold episode running from January 1998 early 2001. Two significant warm episodes occurred from 1997-1998, and another began in the spring of 2002. Table 1: Cold and Warm Episodes based on ONI (Derived from CPC, 2004) Year DJF JFM FMA MAM AMJ MJJ JJA JAS ASO SON OND NDJ 1995 W W W N N N N N C C C C 1996 C C C N N N N N N N N N 1997 N N N N W W W W W W W W 1998 W W W W N N C C C C C C 1999 C C C C C C C C C C C C 2000 C C C C C C N N N C C C 2001 C C N N N N N N N N N N 2002 N N N N W W W W W W W W Lightning For a comprehensive, technical background on the physical mechanisms of lightning, the reader is referred to Uman (2000). However, an abbreviated overview from Uman (2001) is offered in this section, in order to define key terms and facilitate later discussion. Though many theories abound, researches think that the conditions for a lightning discharge are set in a thunderstorm by the clashing of ice particles that induce a charge separation between particles within a cloud. When this occurs, cloud tops tend to develop an overall positive charge, with the base becoming predominantly negatively charged (Uman, 2001). Over time, a strong electrostatic field develops between the earth and the cloud, which causes an electrical discharge (Uman, 2001). 16

PAGE 27

Lightning can be grouped into two primary varieties; cloud-to-ground (CG) and cloud-to-cloud (CC). CG lightning, an electrical discharge between a thunderstorm and the Earths surface, is the most common type, comprising over 90% of all lightning discharges (Uman, 2001). CC lightning consists of all electrical discharges between or within clouds. Both varieties of lightning are commonly referred to as flashes. As this study is concerned with CG lightning, it is discussed here. A CG flash is a composite event; that is to say, it is comprised of several distinct stages (Uman, 2001). In the first stage of a CG discharge, a series of stepped leaders extend from an electrified cloud to the ground (Uman, 2001). As the stepped leader reaches the ground, an upward discharge, or return stroke is induced from the earth back to the source in the cloud. This luminous return stroke is what can be observed by the human eye (Uman, 2001). Uman (2001) defines several other terms associated with lightning, as well. First, modern lightning detection instrumentation is able to sense several physical characteristics of a CG flash, to include whether a flash is negatively or positively charged. This characteristic is known as polarity, and the vast majority of CG flashes have a negative charge (Uman, 2001). CG flashes can also have multiple combinations of stepped leaders and return strokes lowered to the earth in the same or nearby locations during a single event. This is known as multiplicity (Uman, 2001). The National Lightning Detection Network (NLDN) The NLDN detects, measures, and records CG (cloud-to-ground) lightning flashes over the contiguous United States and, with decreasing efficiency, over adjacent waters (Cummins et al., 1998). The network currently comprises 106 17

PAGE 28

sensors strategically dispersed across the U.S., as depicted in Figure 8 (Orville and Huffines, 2001). Of the total, 59 are LPATS (Lightning Positioning and Tracking System) sensors, which locate lightning strikes using a time-of-arrival (TOA) technique. The other 47 are Improved Performance through Combined Technology (IMPACT) sensors, which combine magnetic direction finding (MDF) and TOA techniques (Cummins et al., 1998; Wacker and Orville., 1999). Figure 8: NLDN Sensor Network (from Orville and Huffines 2001) For each CG lightning strike detected by the network, data records include location of the strike, date, time, polarity (positive or negative), signal strength (peak current), multiplicity (strokes) and the number of detectors used to locate each flash (Figure 9). Both the sensors and coverage have been incrementally upgraded over time to improve accuracy, expand coverage and keep pace with evolving technology (Cummins et al., 1998). The most recent, complete upgrade of the network was completed at the end of 1994 (Cummins et al., 1998; Wacker and Orville, 1999). This upgrade increased both the location accuracy of CG flashes and increased the detection efficiency of low peak current CG flashes (Wacker and Orville, 1999). The 18

PAGE 29

overall location accuracy of the network is now 500m (Figure 10). Whereas the NLDN isnt sensitive enough to sense every CG flash, detection efficiency is in the 80 to 90 percent range for flashes with peak currents greater than 5 kA, varying somewhat by region (Figure 11). Figure 9: NLDN sample record Figure 10: NLDN Location Errors (from Zajac et al., 2002) 19

PAGE 30

Figure 11: NLDN Detection Efficiency (from Zajac et al., 2002) Lightning Climatology A descriptive climatology deals with the observed geographic or temporal distribution of meteorological observations over a specified period of time (AMS, 2000). This section reviews the literature concerning the spatial distributions of CG lightning, or lightning climatology over the United States, and more specifically over the Gulf Coast region, prior to investigating the potential relationship between ENSO and CG lightning. For a more extensive, categorized list of lightning climatology studies the reader is referred to Zajac and Rutledge (2001). Additionally, Orville (2002b) compiled a comprehensive list of all known lightning studies (both informal and formal) completed since the 1970s, available on-line. The Development of Lightning Climatologies Even a cursory inspection of the lists compiled by Zajac and Rutledge (2001) and Orville (2002b) reveals that numerous climatologies have examined geographic and temporal CG lightning distributions across the full spectrum of scale including continental, regional and on down to the local and point. In addition to 20

PAGE 31

straightforward studies of CG flash distributions, spatial patterns of physical characteristics have been explored in depth as well to include polarity (positive vs. negative) and peak current, ratios of CC to CG flashes, and more. In fact, most published lightning climatologies explore more than one of these aspects. Lightning climatologies have improved substantially in both quality and sophistication, particularly over the last two decades, with the technological evolution of the NLDN and resulting expansion of the lightning data archive. The NLDN now spans the continent, accurately recording the vast majority of CG lightning flashes (Orville and Huffines, 2002). Satellite-based sensors record lightning of all varieties (CG, CC, etc.) on a global scale as well, although these techniques are not as mature (Christian, et al., 2003). All these data are archived and available to researchers in near real-time. Early CG lightning climatologies were crude by todays standards, constructed primarily through the analysis of surface meteorological observation thunderstorm records, or of single-station, non-networked sensors. Investigators could hand-plot and isopleth basic lightning characteristics such as the number of thunderstorm days. This made for tedious research and limited the scope of studies immensely. For example, Wallace (1975) examined surface observations from 100 weather stations to characterize the diurnal aspects of thunderstorms. MacGorman et al., (1984) conducted a similar study of lightning strike density for the contiguous United States from thunderstorm duration records. Easterling and Robinson (1985) analyzed starting times for thunderstorms at 450 stations over a 25-year period and documented distinct seasonal, spatial and diurnal variations in lightning activity across the U.S. 21

PAGE 32

Based upon these characteristics, they categorized the nation into nine distinct thunderstorm regions. U.S. Lightning Climatologies A series of studies conducted by Texas A & M Professor Richard Orville (in collaboration with others) provides the longest, and most comprehensive NLDN-based lightning climatology available for the continental United States. Orvilles first U.S. lightning climatology using NLDN data was published in 1991, and refined and extended through subsequent years in 1994 (Orville), 1997 (Orville and Silver), 1999 (Orville and Huffines), and 2001 (Orville and Huffines). The most extensive of these studies (Orville and Huffines, 2001) examined the physical characteristics and geographic distribution of more than 216 million CG lightning flashes spanning the continental United States from 1989-1998 (Orville and Huffines, 2001). The analysis was constructed with a spatial resolution of 0.2, which equates approximately to 20x20 km grid cells. The area of individual grid cells varies from 350 km-2 to 425 km-2 according to latitude (Orville and Huffines, 2001). The rationale for this resolution was derived from an earlier study (Reap and Orville, 1990) that established this as the approximate distance at which thunder is audible. Although a mean annual climatology masks year-to-year variations in CG lightning, it is useful nonetheless because it clearly illustrates the overall flash pattern across the U.S. (Orville and Huffines, 2001). For the ten-year period, Orville and Huffines determined that Florida has the greatest mean annual lightning flash densities for the United States. Specifically, regions with more than 9 flashes km-2 year-1 are located in a broad swath across central Florida near the Gulf Coast, extending inland through Orlando to Cape Canaveral and the Atlantic coast (Figure 22

PAGE 33

12). Correspondingly high values occur between Lake Okeechobee eastward to the Atlantic. Of note, the entire Gulf coast exhibits relatively high CG flash densities when compared with most of the rest of the United States. Close inspection reveals relatively high values near Houston, TX and New Orleans, LA as well. Figure 12: Mean Annual Flash Density (Orville and Huffines, 2001) A more recent study by Orville et al., (2002) incorporates data from the newly integrated Canadian sensor network, extending the coverage and resultant climatology northward across the continent. However, this study covers only the three years (1998-2000) since the Canadian sensors were added, and flash density findings remain consistent with earlier studies in the series for the southeast and Gulf Coast of the United States. Zajac and Rutledge (2001) published a comprehensive climatology using NLDN data for the years 1995-1999 (Figure 13). Though the study covered just five years, it is the most complete national climatology available using data only from the 23

PAGE 34

period following the 1995 NLDN upgrade. Moreover, while the most recent study by Orville and Huffines study limited the maximum value to >9 flashes km-2 year-1, Zajac and Rutledge (2001) extend the maximum plotted value to 14.5 flashes km flashes km-2 year-1. Extending the scale helps pinpoint the locations of high-end CG flash density in Florida and southern Mississippi. However, the next-highest category spans from 5.12 to 10.24. This could be misleading because such a broad class masks local variations in the highly variable, high-flash density southeast U.S. From a spatial distribution perspective, overall results are equivalent to Orville and Huffines (2001). Figure 13: Mean Measured Annual Flash Density (Zajac and Rutledge, 2001) Employees of Vaisala-GAI (owner and operator of the NLDN) created a 5-year (1996-2000) national-scale lightning climatology (Figure 14) available on the National Weather Service Lightning Safety World Wide Web site (Vaisala-GAI, 2002). While the map is not associated with a peer-reviewed paper, it is useful to include here for comparison purposes. Examination reveals localized inconsistencies 24

PAGE 35

in flash density pattern exist when compared to the other studies, but results largely correspond. One disparity is the noticeably reduced flash densities observed along the north Florida coast of the Gulf of Mexico. Flash densities in the 4-8 flashes km-2 year-1 range are analyzed right along the coast. Proceeding north (inland), flash densities increases to the 8-16 flashes km-2 year-1 range, a potential 12 flashes km-2 year-1 difference. It is possible these differences between the other studies could be a function of year-to-year variability in lightning. It is more likely that the majority of the flash densities in this region measure at or around 8 flashes km-2 year-1, right at the break point between classifications. This example brings out the very important fact that selection of classification scheme must be carefully considered. Figure 14: Vaisala-GAI Lightning Climatology (1996-2000) 25

PAGE 36

Lightning Climatologies for the Southeast United States In light of the preceding discussion, Orville and Huffines (2001) study remains, to date, the best nation-wide climatology available for baseline comparison purposes. However, many studies have also been published on the regional scale. Florida, in particular, has been the subject of extensive research, and a number of studies have confirmed that the state has the maximum flash density of anywhere in the United States; in fact, the state is known as the lightning capital of the world (Hodanish et al., 1997). As such, Florida has long been heavily instrumented with lightning sensors, and was fully networked in 1985, four years before the NLDN become operational in 1989 (Hodanish et al., 1997). Florida CG Lightning Climatologies Hodanish et al., (1997) constructed a 10-year (1986-1995) CG flash climatology for Florida using an NLDN database of over 25 million flashes. Monthly flash density maps were produced as well as a mean map for all months. Considerable spatial variation was noted across the state and attributed to the influence of Floridas unique geography (geographic location, peninsular shape and proximity to the Gulf and Atlantic, and numerous land-locked bodies of water) on various combinations of mesoscale processes and seasonal variations at the synoptic-scale. Flash density findings reinforce the studies already discussed. A number of published studies have noted the relationship of warm season Florida CG lightning distributions to prevailing low-level wind patterns. An early study by Lopez and Holle (1987) established this relationship, but is not discussed at length here as it used pre-NLDN data, which was less accurate. Reap (1994) demonstrated the fundamental role of the low-level wind patterns in the initiation of 26

PAGE 37

thunderstorm activity and spatial distribution of CG lightning using synoptic map types of sea level pressure from the meteorological model sea level pressure forecasts. He found and linked significant CG flash density variations with specific sea and lake breeze convergence zones spawned by Floridas complex coastlines. Two additional studies published by Florida State University researchers examined the relationship of the low-level winds to warm season CG flash distributions over the Florida panhandle (Camp et al., 1998) and peninsula (Lericos et al., 2002) for six-year (1989-1994) and 10-year (1989-1998) periods of record, respectively. Although techniques differ to some extent between the two studies, in both lightning data were stratified according to common low-level flow regimes, which are strongly dependent upon the location and orientation of the sub-tropical ridge (i.e., whether the ridge is north, south directly over the state of Florida). Each study day was classified into a flow regime based upon radiosonde-derived low-level (1000-700 mb vector mean) winds from reporting stations within the study area. Individual flashes within the domain were superimposed on 5km x 5km grid cell arrays, and maps of lightning flash density generated for each flow regime for hourly, daily (24-hr), and nocturnal periods. For each study, a composite map for all flow regimes and hours was generated so that general patterns can be discerned (Figure 15 and Figure 16). Both Camp et al., (1998) and Lericos et al., (2002) establish that wind direction and speed, time of day, and coastline complexity all play significant roles in shaping warm season CG lightning distributions over Florida, principally because these factors shape sea and land breezes, the primary mechanism for summertime thunderstorms over the state. Sea and land breezes generate thunderstorm activity 27

PAGE 38

almost daily over Florida, but their strength and penetration extent are heavily dependent upon prevailing low-level winds. For example, Lericos et al., (2002) found that when the large-scale flow is from the southeast, the east coast sea breeze and associated lightning are relatively weak. However, with this same southeast pattern the west coast sea breeze is strong and remains near the coastline, producing more lightning near Tampa. Conversely, when the large-scale flow is from the southwest, little convection occurs along the west coast, but major lightning activity occurs along the east coast. For Florida, Lericos et al., cite four areas of relatively large flash densities near Tampa, Fort Meyers, West Palm Beach and Cape Canaveral (Figure 15). Figure 15: Mean flash density for Florida peninsula (from Lericos et al., 2002) 28

PAGE 39

Figure 16: Mean flash density for northern Florida (from Camp et al., 1998) Stroupe (2003) expanded this work, using the same techniques as Lericos et al., (2002) to analyze lightning data on a finer grid of 2.5km x 2.5 km elements. The geographic coverage of the study area was expanded to encompass the entire northern Gulf Coast, and the temporal domain was extended through 2002 for a 12-year (period of record. Results for all hours and all flow regimes are shown in Figure 16. The finer grid scale brings out more detail, accentuating areas of relatively high and low flash density. The geographic distributions remain similar to Lericos et al., (2000) with maximum CG flash densities located in Florida and tending to diminish progressively from east to west along the Gulf Coast and inland. Stroupe also notes strong maximums near Biloxi, Mississippi and several large metropolitan areas such as Houston, Lake Charles, New Orleans and Mobile (Stroupe, 2003). These areas of relative maxima are attributed to a number of factors to include convergence due to 29

PAGE 40

sea, lake, swamp and river breezes, and convex coastlines (Stroupe, 2003). Also noted are urban heat island effects and possible impacts from pollution. Figure 17: Warm Season Flash density for Gulf Coast (from Stroupe, 2003) Lightning Climatologies for Other Areas of the Southeast U. S. Though lightning climatology literature for the southeast is less extensive outside of Florida, there are a couple other studies that should be mentioned. Watson and Holle (1996) constructed a detailed study of summertime lightning in the southeast U.S. in preparation for the 1996 Summer Olympic Games. Just as in the previously mentioned studies, maximum concentrations of CG lightning were confirmed over the Florida peninsula, with additional local maxima scattered along coastal sections of the Florida panhandle, Georgia, and South Carolina. 30

PAGE 41

Figure 18: Jul-Aug Flash Density 1986-1994 (from Watson and Holle, 1996) Steiger et al., (2002) analyzed CG lightning in the Houston area from 1989-2000 and found 45% more flashes than in the surrounding area (Figure 19). The researchers speculated that this anomaly could be anthropogenic in origin; that is, the urban heat circulations and heavy pollution from Houston could be responsible for the increase. Additionally, the possible effect of salt water due to Houstons proximity to the Gulf of Mexico, and sea breeze enhancements of convection patterns were noted as potential factors. Steiger and Orville (2003) examined 14 years (1989-2002) of CG lightning data for an area centered roughly over the state of Louisiana. Significant enhancements (peak density >7 flashes km-2 year-1) of lightning were found over the western side of the Lake Charles area, and a heavily industrialized region of eastern Louisiana near Baton Rouge (Figure 19). These findings are significant because, unlike the Steiger et al., (2002) study, Steiger and Orville (2003) demonstrated that these areas are small enough that urban heat is insufficient to alter circulations, and the impact of the Gulf is limited (sea breezes and salt water are eliminated as causal 31

PAGE 42

factors). Comparisons of the CG flash density in this area with the locations of airborne particulate matter indicate that pollution is the only likely cause of this lightning anomaly (Steiger et al., 2003). Figure 19: (Top) Steiger and Orville (2003); (Bottom) Steiger et al., (2002) 32

PAGE 43

Impact of ENSO on CG Lightning Distributions One previous study has explored lightning response to ENSO over the Gulf Coast region of the United States. Goodman et al., (2000) examined mean winter season (December, January and February) lightning for a 10-year study period (1989-2002) over the southeast United States (Figure 20). For the strong 1997 warm ENSO event, the group noted a 100-200% increase in lightning flashes, hours and days for a large area centered over the northern Gulf of Mexico and Gulf Coast region (Goodman et al., 2000). Also noted was a doubling in the number of synoptic scale low pressure systems developing within or moving through the northern Gulf basin (Goodman et al., 2000). As discussed previously, warm ENSO events are characterized by a southward shift and intensification of the jet stream over the southern United States (COMET, 2003). Goodman et al., (2000) compared the winter 1997-1998 lightning enhancement with regional climatology data and were able to conclusively attribute it to increased synoptic-scale forcing, directly attributed to ENSO and the stronger than normal upper level jet stream. In addition to NLDN data, the group incorporated total lightning from NASAs Lightning Imaging Sensor (LIS) carried aboard the Tropical Rainfall Measuring Mission (TRMM) satellite (Christian et al., 1999). Total lightning includes all flashes remotely sensed from the TRMM satellite to include intra-cloud and cloud to ground (Christian et al., 1999). As the satellite is only over the Gulf Region 3 to 4 times per day, its data cannot be used exclusively, but it was useful in reinforcing and validating findings from the NLDN dataset (Christian et al., 1999). 33

PAGE 44

For more information on LIS and TRMM, the reader is referred to Christian et al., (1999). Figure 20: Lightning Days, DJF (Goodman et al., 2000) 34

PAGE 45

CHAPTER THREE METHODOLOGY Geographic Domain The study area is the Southeast United States and near-adjacent waters of the Gulf of Mexico. The region is geographically bounded by 33 and 24 north latitude, 79 and 99 west longitude (Figure 21). As discussed in the literature review, multiple studies confirm that this region exhibits the greatest overall CG lightning flash density in the United States, as well as the strongest teleconnections with ENSO with respect to temperature and precipitation. Figure 21: Study Area 35

PAGE 46

Data Description of Data Two datasets are utilized. The first is the total set of CG lightning flash records (nearly 55 million) detected in the study area by the NLDN over an 8-year (Jan 1995 Dec 2002) period of interest. The Air Force Combat Climatology Center (AFCCC) acquired the raw data from Vaisala-GAI, Inc., owner and operator of the NLDN. AFCCC technicians segregated the flash records into 96, one-month increments, and furnished it in tab-delimited, text file format. Each lightning record contains several measured attributes to include location, date and time of individual flashes. The second dataset is the concurrent 96-month time series of SST anomaly values from the Nio 3.4 region of the equatorial Pacific, downloaded from the Climate Prediction Center (CPC, 2003). The SST anomaly data is included at Appendix A. Researchers have established that the Nio 3.4 region is most representative of the ENSO signal on U.S. climate, as it contains the western half of the equatorial cold tongue (NOAA, 2003). The index is derived from a running, three-month average of SST departures from normal in this region (NOAA, 2003). Data Pre-Processing To prepare the raw data for study, each monthly lightning file was imported into ArcMap GIS as a point feature layer, saved in Environmental Systems Research Inc (ESRI) shapefile format, and projected to a custom Albers Equal Area projection. Albers was selected for the projection because it preserves area, a key property when analyzing flashes per unit area. ESRI shapefiles are comprised of a main file, an 36

PAGE 47

index file and a dBase table, which contains feature attributes with one record per feature (ESRI, 1998). Data Analysis For analysis purposes and usability in ArcMap, flashes were assigned to a fine-resolution grid. This was accomplished by importing the dBase table component of each shape file into the SPLUS statistical package and running a custom script (Appendix B). The script aggregated and binned individual flashes into a grid of 2.5 km x 2.5 km cells (816 X 418; 314,088 total cells), assigned each resultant grid box a corresponding flash count (i.e., if 7 flashes, flash count value of box is 7) and wrote the data back out to 96 monthly ASCII files. Header information specifying the grid corners and geographic extent of the domain was then appended to the files, rendering each readable by ArcMap as monthly raster grids precisely matching the study area. With this processed data, an updated CG lightning climatology was created for the Gulf Coast Region of the United States. Using ArcMap, mean CG flash density maps were constructed for the complete period of record, individual months, and the winter (DJF) seasons. Maps were visually examined for qualitative comparison with past climatologies and the Goodman et al., (2000) ENSO study discussed in the literature review. Following this qualitative review, quantitative methods were employed to explore the lightning data for ENSO teleconnections. Simple Pearsons correlations were computed between concurrent monthly pairings of Nio 3.4 SST and CG lightning flash deviation values from the study area. The correlation procedure is performed initially for the entire domain, then on progressively smaller bin sizes for each month in the period of record. 37

PAGE 48

A perl script was crafted to accomplish the correlations and binning of the 816x418 array of grid cells (Appendix C). The script allows the user to partition the grid into approximately equal PxQ sub-domains. For example, if the user specifies P=2 and Q=2 (or 2x2 sub-domains), the lightning data are aggregated over four bins of dimension 408x209 in the original grid. For each resultant bin, the lightning flash deviation from the monthly mean is calculated, and each 8-month time series of values is compared with the corresponding 8-member series of NIO 3.4 SST anomaly values. Thus SST anomalies are compared to above or below normal monthly lightning values in any given bin. The correlation values computed by the script for each user-specified domain were then written out to monthly text files, and the text files imported into ArcMap and visualized over a base map of the study area. To achieve this, the ArcMap ET GeoWizards tools were used to create a blank grid matching the user specified sub-domains. The blank grid squares were then populated with matching correlation values from the text files. 38

PAGE 49

CHAPTER FOUR RESULTS Lightning Climatology Mean and Annual Lightning Climatology A total of 54,943,981 flashes were analyzed, ranging from a low of just 4,205,344 in 1995 to a high of 8,927,069 in 1997 (Table 1, Figure 22, Figure 23). The average for all years is 6,867,998, making it quite clear that lightning activity is highly variable from year to year. Figures 25-33 depict mean CG flash densities for the entire period of record as well as individual years. While flash activity is comparable in both intensity and pattern from year to year, there are interesting localized variations that become readily apparent upon close inspection. On the mean map (Figure 25) a broad swath with flash densities greater than 9 flashes km-2 year-1 is observed in Florida, stretching from the Tampa Bay region eastward through Orlando, extending to the Atlantic coast near the Cape Canaveral area. An additional zone of enhanced flash densities is found in southeast Florida between Lake Okeechobee and West Palm Beach. These maximums correspond with findings from earlier Florida studies (Reap, 1994; Lericos et al., 2002), and are most likely a function of clashing land and sea breeze circulations. Along the northern Gulf of Mexico, a significant area of elevated flash density extends from the New Orleans area eastward to just past the city of Mobile. An area with enhanced flash densities is observed in the Houston area during most years. This matches the observation by Steiger et al., (2002) with respect to pollution and urban 39

PAGE 50

effects. Interestingly, flash densities drop off past Houston, progressing southward along the Texas Gulf coast. This is observed in all three national studies (Orville and Huffines, 2001; Zajac and Rutledge, 2001; Vaisala-GAI, 2002) discussed in the literature review. In general, enhanced flash densities are found along the Gulf, the result of complex sea and land breeze interactions. In most years, and on the average, flash intensity diminishes with distance from the coast as influence from Gulf sea breeze circulations diminish. As mentioned, lightning activity and flash density patterns vary markedly from year to year. With the lowest total flashes of any year in the study, 1995 (Figure 26) was the least impressive with respect to flash density. For instance, while relatively high flash densities persist along the coasts, these regions are, on the whole, less intense than in subsequent years, and diminished in areal coverage as well. The swath of enhanced flash densities returns across the peninsula during 1996 (Figure 27), as well as in the southeast of Florida. In 1997 (Figure 28), much of the intense activity diminishes in the central part of the peninsula. In both years, an enhanced area varying from 6-9 flashes km-2 year-1 is observed in northeast Louisiana near Shreveport that is not present during other years of the study, or in the mean. The year 1998 (Figure 29) is worthy of mention because a marked increase in flash density is observed over the waters of the northern Gulf of Mexico. While the Gulf does exhibit these intensities in many other years, the pattern is far more expansive during 1998 than in other years. Sea breeze activity is very evident in Florida, as well. 1998 displays most of the maximums and minimums other years in the study havein fact, it could be said to be a prototypical year. 40

PAGE 51

The years 1999 (Figure 30) and 2000 (Figure 31) appear much the same as 1998, with the exception that lightning activity over the Gulf of Mexico becomes much reduced in areal extent. Flash densities over a small area of the northeast Gulf exceed 7 flashes km-2 year-1, but again, the area is not as expansive as that seen during 1998. In 2001 (Figure 32), a large region with flash densities over and above 3 flashes km-2 year-1 penetrates from the north-central Gulf Coast, inland to the northernmost extent of the study area. While this intensity is not remarkable in and of itself, it is more intense and expansive than in any other individual year, or in the mean for this portion of the region. Maximum flash densities in Florida also greatly exceed the average. The year 2002 had more lightning activity than any other year in the study. This is borne out in both the map of the year (Figure 33) and in Table 1 where it is seen that raw lightning flashes approach 9 million in total. An unusual, extremely intense area with flash densities greater than 9 flashes km-2 year-1 is observed off the Texas coast, near Corpus Christi. In fact, Texas coastal lightning activity is considerably more active in 2002 than in any other year. A few, overarching comments must be made about the flash density findings of this study. While the general flash density patterns closely resemble the results of prior studies (e.g. Zajac and Rutledge, 2001; Orville and Huffines, 2001), and from year to year, intensity results arent nearly as impressive for the mean of all years (Figure 25). For instance, the areas with greater than 9 flashes km-2 year-1 observed in Florida and other isolated locations are far less extensive in areal coverage than in any of the earlier studies. There are several plausible explanations for this. 41

PAGE 52

First, CG flash distributions vary from year-to-year, and this study covers just 8 years. This abbreviated study domain is largely unavoidable; the entire period of record with networked data is not much longer (1989-2002), and expanded datasets are expensive and difficult to acquire. Additionally, data from earlier than 1995 would have required complicated correction factors beyond the scope of this paper (Cummins et al., 1998). Nevertheless, anomalously low totals for even one year in a short study can artificially impact overall results. For example flash totals from 1995 (Figure 26), were strikingly low when compared with other years in the study. Conceivably, adding additional years with normal lightning activity could bring mean flash densities up. Of note, flash density findings were consistent with other studies for individual years. For instance, multiple individual grids in some years measured 15-17 flashes km-2 year-1 just north of Tampa, consistent with Zajac and Rutledge (2001). Since the analysis grid was so fine (2.5kmx2.5km) perhaps subtle, year-to-year shifts in flash distributions caused high-density grids to be averaged with lower density grids. Increasing the grid size could help alleviate this problem. Finally, it would perhaps be a stretch to attribute to year-to-year variability of lightning to ENSO based solely on flash density maps, but it bears examination. Examination of the ONI values (Table 1) for individual seasons in the study reveals that this 8-year period was disproportionately affected by cool events and less storminess. During the 8-year period of this study, 23 periods DJF periods were classified as warm episodes (24%) with 36 cool episodes (37%) and 37 neutral episodes (39%). The 1989-1998 period examined by Orville and Huffines (2001) 42

PAGE 53

incorporated 43 warm seasons (36%), 59 neutral periods (49%) and just 18 cool seasons (15%). For individual years, 1995 had by far the lowest amount of lightning; coincidentally (or not) most of the year was under either La Nia or neutral conditions (Table 1). By contrast 2002, with by far the most lightning, was under El Nio conditions for most of the year. 1998, another El Nio year, had slightly more lightning and exhibited enhanced flash densities over the Gulf of Mexico. This theory is further borne out by comparing the mean large scale synoptic circulation during this study period to that of another, for example Zajac and Rutledge (2001) which encompassed a study period of 1995-2002. The Zajac and Rutledge (2001) study is most appropriate for comparison purposes here because the entire temporal domain follows the 1994 NLDN upgrade (Cummins et al., 1998). To accomplish this comparison, maps of the mean 1000mb geopotential heights for the 1995-2002 and 1995-1999 periods were created from the National Center for Environmental Prediction (NCEP) reanalysis data (Figure 22). For both studies, the subtropical ridge extends westward from the Atlantic Ocean to the border of the study area, with the height gradient tightening over eastern Texas. Additionally, troughing is observed extending across the north central section of the domain. For the 1995-1999 period, however, heights display an overall decrease, and the troughing is more pronounced. The decreased heights during the 1995-2002 indicate overall cooler conditions and, almost certainly, increased frontal activity (with attendant increase in lightning) that can be attributed to increased numbers of cool ENSO episodes (Figure 2). 43

PAGE 54

Figure 22: Mean 1000mb geopotential heights. Top, 1995-2002; Bottom 1995-1999 Table 2: CG Lightning 1995-2002 (54,943,981 flashes) Mo. 1995 1996 1997 1998 1999 2000 2001 2002 Avg Jan 203524 90505 254571 454193 477030 67063 39454 92611 209869 Feb 89806 125055 85521 657078 125773 122930 107662 26769 167574 Mar 369473 467506 260169 598464 402669 500257 606848 406322 451464 Apr 434554 426625 941018 223991 373347 336759 467506 511028 464354 May 559349 326419 1095365 291526 1050521 372141 414729 543231 581660 Jun 665956 1478988 1230427 769422 955980 919531 1415519 1015992 1056477 Jul 1537688 1226430 2039592 1575330 1408821 1639936 1193130 2212939 1604233 Aug 1087634 1424000 1036517 1931068 1737535 1547568 1163355 1922176 1481232 Sep 629160 838306 634653 431941 615188 953033 866399 590684 694921 Oct 165888 172187 392069 229590 92947 114412 272313 853456 286608 Nov 156804 143686 206331 109606 16048 176086 414464 261567 185574 Dec 208244 164219 111023 111023 94404 34997 140734 490294 169367 Tot 4205344 6188274 7348764 7272209 7115495 6784713 7102113 8927069 44

PAGE 55

020040060080010001200140016001800JanFebMarAprMayJunJulAugSepOctNovDecThousands Figure 23: Average Flashes by Month, 1995-2002 01000200030004000500060007000800090001000019951996199719981999200020012002Thousands Figure 24: Average Flashes by Year, 1995-2002 45

PAGE 56

Figure 25: Mean Annual Flash Density, 1995-2002 (54, 943, 981 flashes) Figure 26: Mean Flash Density, 1995 46

PAGE 57

Figure 27: Mean Flash Density, 1996 Figure 28: Mean Flash Density, 1997 47

PAGE 58

Figure 29: Mean Flash Density, 1998 Figure 30: Mean Flash Density, 1999 48

PAGE 59

Figure 31: Mean Flash Density, 2000 Figure 32: Mean Flash Density, 2001 49

PAGE 60

Figure 33: Mean Flash Density, 2002 ENSO Correlations In addition to the updated climatology, simple Pearsons correlations were computed between concurrent monthly pairings (summer and winter months) of Nio 3.4 SST and CG lightning flash deviation values from the study area. An initial bin encompassing the entire spatial domain (1x1) was run first to establish a baseline. It correlated poorly, which is not at all surprising; the gross bin size likely masked any detail. However, just as this large bin size smoothed local variation, it is assumed very small bin size selections could prove to be too noisy to be useful as well. Therefore, progressively smaller bin sizes of 10X5, 20x10, 40x20 and 80x40 (roughly 204x209 km, 102x105km, 51x52km, and 25x26km bins respectively) were explored to establish an optimal bin size that would reveal positive and negative correlations on scales that would be useful for climate applications. 50

PAGE 61

Bin size selections are rather arbitrary, but this was the most elegant way to subset the data. It is important to note that the full spectrum of computed bin sizes matched up well in both spatial and intensity variability for all months, therefore only the finest, 80x40 bin size, is discussed here (Figures 35, 37, 39, 48, 50 and 52). Additionally, the average flash density maps for each DJF period are displayed and discussed in this section together with the ENSO correlation results (Figures 34, 36, 38, 47, 49 and 51). This is merely for convenience. As the actual correlations are computed as pair-wise, eight-member series, the locations of or concurrences between areas of positive or negative correlations and elevated or diminished lightning activity have no bearing on whether or not an ENSO-lightning link exists. Winter December Past studies have demonstrated that lightning activity is at a minimum during the month of December (i.e., Hodanish et al., 1997). This is particularly true in central and South Florida, where flash densities average 0-0.1 flashes km-2 month-1 for the month (Fig 34). During December, most of the lightning activity is associated with frontal passages, and sea breeze convection is virtually non-existent. Lightning maxima are oriented from SSW-ENE in bands across the study area, which is indicative of these frontal passages. The area of maximum lightning for this period is located in Texas, and just off the Texas and Louisiana coasts. The maximum SST-lightning correlations (Figure 35) occur in a SSW-ENE banding pattern as well, but mostly over the eastern Gulf, central Florida, and along the Atlantic coast of Florida. Some lightning enhancement is coincident with the 51

PAGE 62

maximum correlation areas in the eastern Gulf, but overall the areas of maximum lightning do not match up with the areas of maximum SST correlations for December. Figure 34: December Mean CG Flash Density, 1995-2002 Figure 35: December SST-Lightning Correlation, 1995-2002 52

PAGE 63

January An overall lightning increase is noted during the month of January, particularly in the north-central Gulf of Mexico, east Texas and Louisiana (Figure 36). As in December, lightning activity is oriented in a banded structure, indicating the prevalence of frontal activity. Positive correlations shift in maximum to the western Gulf, Texas and the central Gulf (Figure 37). Unlike December, however, the maximum lightning regions are highly coincident with positive SST correlations. Again, this is interesting but not necessarily indicative of an SST-lightning link during the month. Figure 36: January Mean CG Flash Density, 1995-2002 53

PAGE 64

Figure 37: January SST-Lightning Correlation, 1995-2002 February Lightning activity shows a slight decrease and shift to the eastern Gulf in February (Fig 38). Positive correlations, however, exhibit a marked increase over the previous two months, with moderate to high values dominating all oceanic regions, especially (Fig 39). The maximum correlations are scattered in a zonal pattern, with clustering over southern Florida, Texas and isolated scattering over the Gulf. Correlations in the extreme southern Gulf should be discounted, as this is beyond the range of the network. However, even after discarding these data points, significant correlations are evident. When compared with the raw flash density for the month, the areas of highest correlation appear to correspond with the areas of lowest flash densities. 54

PAGE 65

Figure 38: February Mean CG Flash Density, 1995-2002 Mean CG Flash Density Figure 39: February SST-Lightning Correlation, 1995-2002 Comparison with Goodman et al., (2000), Winter Season For comparison purposes with the Goodman et al., (2000) study, average flash density maps were prepared for all December-January-February (DJF) periods during 55

PAGE 66

the study (Figures 40-46). Overall, year-to-year lightning distributions were relatively equivalent, and findings for this study largely validate those of Goodman et al., (2000). The two study periods overlapped for four winters (1995-1996, 1996-1997, 1997-1998 and 1998-1999), with a single common DJF period (1997-1998) classified as a warm event according to ONI (Table 1). Lightning activity during this one, shared warm season was considerably more intense and extensive than in any other study year. Enhanced flash values are observed surging across much of the northern Gulf of Mexico region, comparable to Goodman et al., (2000). The higher resolution of the analysis in this study reveals details that were not evident in that single study of this teleconnection. Figure 41 shows an envelope of peak lightning from northeastern Texas to the eastern Gulf of Mexico. Within that envelope, streaks of lightning are aligned along northeast to southwest lines. The larger envelope is indicative of a southerly shift in the mid-latitude cyclone track, as illustrated in Figure 2. Four DJF periods in this study (1995-1996, 1998-1999, 1999-2000 and 2000-2001) were ONI-classified cool events. The 1995-1996 (Figure 40) and 1998-1999 (Figure 43) DJF periods were relatively unremarkable, although lightning is a bit less extensive than the prior warm event year. The 1999-2000 (Figure 44) and 2000-2001 (Figure 45) winter seasons were not covered by Goodman et al., (2000), but a significant decrease in lightning activity is noted across the entire area. When compared to neutral or warm event seasons, the 2000-2001 DJF period exhibited the lowest flash densities of any in the study (Figure 45). The 2001-2002 DJF period marked a return to neutral ENSO conditions, with a slight increase in lightning activity observed over the previous years DJF (Figure 44). Notable about 56

PAGE 67

this period was that the lightning was primarily concentrated in the western half of the study area, with low flash density over the Florida peninsula when compared to other years. As the Goodman et al., (2000) depicts lightning days, and this study raw lightning activity there are some differences noted, especially with respect to areal coverage and locations of maximums. However, some of this difference likely results from scale and classification choices. Figure 40: DJF Mean Flash Density, 1995-1996 Figure 41: DJF Mean Flash Density, 1996-1997 57

PAGE 68

Figure 42: DJF Mean Flash Density, 1997-1998 Figure 43: DJF Mean Flash Density, 1998-1999 58

PAGE 69

Figure 44: DJF Mean Flash Density, 1999-2000 Figure 45: DJF Mean Flash Density, 2000-2001 59

PAGE 70

Figure 46: DJF Mean Flash Density, 2001-2002 Summer June Very high flash densities are observed over Florida during the month of June, with more moderate lightning across the entire northern Gulf coast to just past Houston (Figure 47). Isolated areas of positive correlation are scattered across much of East Texas and the western Gulf (Fig 48). Elsewhere, there exists little response to equatorial Pacific SST changes, and enhanced lightning areas are not coincident with areas of positive correlations. 60

PAGE 71

Figure 47: June SST Mean CG Flash Density, Jun 1995-'02 Figure 48: June SST-Lightning Correlation, 1995-2002 July July is, on average, the month of maximum lightning and flash densities, and this is readily apparent in Figure 49. The usual pattern of high flash densities is 61

PAGE 72

evident across Florida. Maximums of greater than 14 flashes km-2 month-1 extend well inland on the northern Gulf coast as well, with additional, more localized areas of enhanced lightning scattered throughout most of the eastern two thirds of the study area. The region of lowest flash density was observed in eastern Texas, past Houston and extending southward along the coast. A large cluster of positive correlations is centered over Louisiana and southern Mississippi while most of the oceanic regions show little response (Figure 50). Based upon this analysis, the maximum flash density over Florida during July is not strongly influenced by large scale circulation but rather modulated by local sea-land breeze effects. In contrast, the region of weak flash densities over Texas appears to be very sensitive to the changes in the large-scale circulation patterns that result from changes in the equatorial Pacific SSTs. Figure 49: July Mean CG Flash Density, 1995-2002 62

PAGE 73

Figure 50: July SST-Lightning Correlation, 1995-2002 August CG flash densities remain high overall during the month of August (Fig 51). A barely perceptible decrease can be seen over northern Gulf coast and oceanic areas, but interestingly, Florida flash densities appear to increase, particularly in the southern half of the state. This increase from average is likely a simple year-to-year variation, as climatologically, 8 years is not enough to establish a long-term pattern. However, accurate lightning data does not extend much farther back in time, so it is assumed some of these variations will be smoothed as more lightning data comes available to add to the climatologies. August is the only month in which negative correlations are dominant across the domain (Fig 52). While these negative correlations do not appear to be associated with a significant change in lightning flash density between July and August, it is possible to speculate that the negative correlations are due to a decrease in convective precipitation across the domain. El Nio summers are noted by a precipitation decrease across the domain. El Nio summers are noted by a decrease in precipitation 63

PAGE 74

across Florida and the Caribbean (Figure 3). It should be noted that the months of June, July and August have comparatively small areas of positive correlation compared with the winter months. During the warm season, large portions of the domain are not strongly correlated to the ENSO anomalies. As the warm season progresses, the correlations become mostly negative or neutral in August. Figure 51: August Mean CG Flash Density 1995-2002 Figure 52: August SST-Lightning Correlation, 1995-2002 64

PAGE 75

CHAPTER FIVE CONCLUSION An 8-year dataset (1995-2002) of cloud-to-ground (CG) lightning flashes was examined to determine if the ENSO cycle has an influence on lightning activity along the Gulf Coast region of the United States. Flash density maps were constructed for the complete period of record, individual months and all winter (DJF) seasons. Maps were first visually examined for qualitative comparison with past lightning climatologies and the Goodman et al., (2000) ENSO-lightning study discussed in the literature review. Following this qualitative review, simple Pearsons correlations were computed between concurrent monthly pairings of Nino 3.4 SST and CG lightning flash deviation values from the study area to determine if a relationship exists between the two. Results for the updated climatology were, overall, as expected. Variations were noted but CG spatial distributions are consistent with the studies cited in the literature review (e.g., Orville and Huffines, 2001; Zajac and Rutledge, 2001; Vaisala, 2002). A notable departure was found, however, in the mean annual flash density findings (Figure 24). A marked decrease in flash intensity was noted for the mean. There are several plausible explanations to explain this. First, it nearly goes without mentioning that lightning activity is, by its very nature, highly erratic. And, from a climatological perspective, an eight-year period of record is too short to be reliable. This abbreviated temporal domain is largely unavoidable as the entire period of record with networked lightning data is not much 65

PAGE 76

longer (1989-2002), and expanded datasets are expensive and difficult to acquire. Additionally, data from earlier than 1995 would have required complicated correction factors beyond the scope of this paper (Cummins et al., 1998). Nevertheless, close inspection of the data revealed that this study included a disproportionately large number of cool ENSO periods when compared to the others reviewed. This was borne out through analysis of seasonal Operational NINO Index values, and comparisons with the other studies. Additionally, an examination of the mean synoptic circulation for the period showed that 1995-2002 likely experienced increased frontal passages, a primary ENSO teleconnection. This indicates, at least indirectly that lightning activity shows a net decrease during ENSO years. Results from the analysis of winter season lightning data validated the findings of Goodman et al., (2000), further substantiating the role of the ENSO cycle in winter season lightning fluctuations. This is especially evident for the one common warm event winter (1997-1998) between the two studies where a marked increase in lightning activity was noted. Lightning decreases were observed during cool event winters, as well. Results from further qualitative analysis of the data appears to validate the results of Goodman et al., (2000), further substantiating the role of the ENSO cycle in winter season lightning fluctuations. This is especially evident for the one common warm event winter (1997-1998) between the two studies where a marked increase in lightning activity was noted. Lightning decreases were observed during cool event winters, as well. Correlation findings indicate an ENSO lightning link, at least of winter months. Some high correlation values existed in June and July over the study area, 66

PAGE 77

but it was difficult to attribute these to ENSO. This is not surprising; ENSO influences are known to be at a minimum during the summer for the study area. This is because the primary teleconnection of ENSO with the southeast United States is a displacement of the jet stream and storm track. The primary mechanism for thunderstorm activity in the study region is air mass thunderstorms and convection sparked by mesoscale processes such as land and sea breeze interactions. Frontal activity is virtually non-existent at these latitudes during summer months. It was a different story for the winter, however. High correlation values were noted over large swaths of the study area during each month. In January and February in particular, these areas of high correlation were coincident with areas of enhanced flash density. These enhanced CG flash regions and high correlation values and patterns are indicative of the southerly shift in the mid-latitude storm track that is known to be influenced by ENSO. Implications Lightning is hazardous to both societies and individuals. ENSO is known to alter climate and weather patterns across the globe; it is a hazard in its own right. Any study that contributes to a better understanding of either, or their intersection, could potentially save lives and dollars. Teleconnections are difficult to prove, as they involve, by definition, links between distant, apparently unrelated events. It is even more difficult to substantiate a teleconnection between two phenomena that are as highly variable and as little understood as lightning and ENSO. Though no smoking gun was found here, comparisons with past studies and analysis of new data in this thesis furnish compelling evidence that the ENSO cycle impacts lightning activity through its influence on the storm track, particularly for the winter season. 67

PAGE 78

Future Research Much work is available for the way ahead, even with the data and methods available here. The perl script used to compute the correlations for this study is scalable; virtually any grid size can be computed with the code, visualized and examined for ESNO teleconnections. The user is limited only by computing power and the original 2.5X2.5 km grid size. The perl script also computed one to 11 month lags, which have yet to be processed and examined. It is expected analysis of the lags will yield promising results, as precipitation anomalies for the Caribbean and Central America have been found to have a lagged response to El Nio. In another area, the ENSO cycle may impact distributions of other physical characteristics of lightning. Ratios of positive to negative polarity lightning, multiplicity, and diurnal trends can easily be correlated with ENSO anomalies. Perhaps there are more nocturnal lightning flashes during ENSO warm and cool events, for example. Finally, although this study (and the discussion) was limited to ENSO impacts in the Gulf Coast region of the U.S., there are well documented ENSO teleconnections with much of the rest of the U.S. Given more data, the same methods and procedures could be applied to expand the study across the U.S. 68

PAGE 79

References Agee, E., and S. Zurn-Birkhimer. 1998. Variations in USA tornado occurrences during El Nio and La Nia. Preprints, 19th Conference on Severe Local Storms, Minneapolis, MN. American Meteorological Society, 287-290. American Meteorological Society (AMS). 2000. Glossary of Meteorology. 2nd ed. Boston: AMS, 2000. ISBN 1878220349. Angell, J.K., and J. Korshover. 1987. Variability in United States cloudiness and its relation to El Nio. Journal of Climate and Applied Meteorology, v26 n5, pp 580-584. Camp, J. P., A. I. Watson, and H. E. Fuelberg. 1998. The diurnal distribution of lightning over north Florida and its relation to the prevailing low-level flow. Weather and Forecasting, 13, 729. Cayan, D. R.. 1996. Interannual Climate variability and snow pack in the western United States. Journal of Climatology, 9(5), 928-948. Centers for Disease Control and Prevention. 1998 Lightning-associated deaths United States, 1980-1995. Morbidity and Mortality Weekly Report, May 22; 47(19):391-4. Changnon, S.A. 1999. Impacts of 1997-98 El Nio-generated weather in the United States. Bulletin of the American Meteorological Society, 80, 1819-1827. Christian, H. J., R. J. Blakeslee, S. J. Goodman, D. A. Mach, M. F. Stewart, D. E. Buechler, W. J. Koshak, J. M. Hall, W. L. Boeck, K. T. Driscoll, and D. J. Bocippio, "The Lightning Imaging Sensor," Proceedings of the 11th International Conference on Atmospheric Electricity, Guntersville, Alabama, June 7-11, 1999, pp. 746-749. Climate Prediction Center. 2002. El Nio/La Nia Home. Accessed at http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/lanina 01 May 04. Climate Prediction Center. 2003. Monthly atmospheric and SST indices. Accessed at ftp://ftp.ncep.noaa.gov/pub/cpc/wd52dg/data/indices/sstoi.indices 01 May 04. Climate Prediction Center. 2004. Cold and warm episodes by season. Accessed at http://www.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ensoye ars.htm 10 May 04 Cooperative Program for Operational Meteorology (COMET). 2003. The El Nio 69

PAGE 80

Southern Oscillation (ENSO) Cycle. By Dr. Vernon E. Kousky. Accessed at http://www.meted.ucar.edu/climate/enso/index.htm 10 May 04. Cummins, K. L., M. J. Murphy, E. A. Bardo, W. L. Hiscox, R. B. Pyle, and A. E. Pifer. 1998. A combined TOA/MDF technology upgrade of the U.S. National Lightning Detection Network. Journal of Geophysical Research, 103 (D8), 9035. Curran, Brian E., Ronald L. Holle, Ral E. Lpez. 2000. Lightning casualties and damages in the United States from 1959 to 1994. Journal of Climate 13: 3448-3464. Department of Commerce (DOC). 2003. NOAA gets U.S. consensus for El Nio/La Nia Index, Definitions. National Oceanic and Atmospheric Administration (NOAA) Press Release NOAA03-119. 30 September. Diaz, H.F., and G.J. McCabe. 1999. A possible connection between the 1878 Yellow Fever Epidemic in the Southern United States and the 1877-78 El Nio Episode. Bulletin of the American Meteorological Society, 80(1), 21-28. Easterling, D.R., and P. J. Robinson. 1985. The diurnal variation of thunderstorm activity in the United States. Journal of Climate and Applied Meteorology, 24, 1048-1058. Environmental Systems Research Inc. 1998. ESRI shapefile technical description. ESRI whitepaper. Accessed at http://www.esri.com/library/whitepapers/pdfs/shapefile.pdf 02 Feb 04. Goodman, S.J., D. E. Buechler, K. Knupp, K. T. Driscoll, and E. W. McCaul. 2000. The 1997 El Nio event and related wintertime lightning variations in the southeastern United States. Geophysical Research Letters, 27, 541. Green, P. M., D. M. Legler, C. J. Miranda V, and J. J. O'Brien. 1997. The North American climate patterns associated with the El Nio-Southern Oscillation: Center for Ocean-Atmosphere Prediction Studies, 1997. Accessed at http://www.coaps.fsu.edu/lib/booklet/ 30 Apr 04. Hagemeyer, B.C. 1998. Significant extratropical tornado occurrences in florida during strong El Nio and strong La Nia events. Preprints, 19th Conference on Severe Local Storms, Minneapolis, MN. American Meteorological Society, 412-415. Hanley, Deborah E., Mark A. Bourassa, James J. O'Brien, Shawn R. Smith, Elizabeth R. Spade. 2003. A Quantitative Evaluation of ENSO Indices. Journal of Climate, 16: 1249-1258. Harrison, D.E., and N.K. Larkin. 1998. Seasonal U.S. temperature and precipitation 70

PAGE 81

anomalies associated with El Nio: Historical results and comparison with 1997. Geophysical Research Letters, 25: 3959. Harrison, M., and C.F. Meindl. 2001. A statistical relationship between El NioSouthern Oscillation and Florida wildfire occurrence. Physical Geography, 22: 187-203. Hodanish, Stephen, David Sharp, Waylon Collins, Charles Paxton, Richard E. Orville. 1997. A 10-yr monthly lightning climatology of Florida: 1986-95. Weather and Forecasting, 12, 439-448. International Research Institute for Climate Prediction (IRI). 2003. ENSO Web. Accessed at http://iri.columbia.edu/climate/ENSO/index.html 30 Apr 04. Kalnay, E. and Coauthors. 1996. The NCEP/NCAR reanalysis 40-year project. Bulletin of the American Meteorological Society, 77, 437-471. Kumar, Arun. 1997. Further thoughts on ENSO. From Review of the causes and consequences of cold events: a La Nia summit. Summary report. Accessed at http://www.esig.ucar.edu/laNia/report/kumar.html 30 Nov 03. Lericos, T. P., H. E. Fuelberg, A. I. Watson, and R. L. Holle. 2002. Warm season lightning distributions over the Florida peninsula as related to synoptic patterns. Weather and Forecasting, 17, 83-98. Legler, David M., J. Bryant Kelly, and James J. O'Brien. 1997. Impact of ENSOrelated climate anomalies on crop yields in the US. Accessed at http://www.coaps.fsu.edu/~legler/Legler98-1/Legler98-1.htm 20 Apr 04. Lopez, R. E., and R. L. Holle. 1987. The distribution of lightning as a function of low-level wind flow in central Florida. NOAA Technological Memo. ERL ESG-28, National Severe Storms Laboratory, Norman, OK, 43 pp. MacGorman, D. R., M. W. Maier, and W. D. Rust. 1984. Lightning strike density for the contiguous United States from thunderstorm duration records. Rep. NUREG/CR-3759. U.S. Nuclear Regulatory Commission, Washington, DC, 44 pp. [Available from National Severe Storms Laboratory, 1313 Halley Circle, Norman, OK 73069.] Magnun, L.J., D.C. McClurg, L.D. Stratton, N.N. Soreide, M.J. McPhaden. 1998. The Tropical Atmosphere Ocean (TAO) Array World Wide Web Site. Accessed at http://www.pmel.noaa.gov/tao/proj_over/pubs/argos.html 10 May 04. McPhaden, M.J. 2002. El Nio and La Nia: Causes and Global Consequences In: Encyclopedia of Global Environmental Change, Vol 1, John Wiley and Sons, LTD., Chichester, UK, 353-370. 71

PAGE 82

National Climatic Data Center, 2002. El Nio/La Nia. Accessed at http://lwf.ncdc.noaa.gov/oa/climate/elNio/elNio.html 01 May 04. National Weather Service, 2003. National hazard statistics. Accessed at http://www.nws.noaa.gov/om/hazstats.shtml 03 May 04. National Oceanic and Atmospheric Administration (NOAA). 2002. NOAA El Nio Page. Accessed at http://www.elNino.noaa.gov/ 01 May 04. O'Brien, James J., Todd S. Richards, Alan C. Davis. 1996. The effect of El Nio on U. S. landfalling hurricanes. Bulletin of the American Meteorological Society, 77, 4, 773-774. Orville, R. E. 1991. Lightning ground flash density in the contiguous United States 1989. Monthly Weather Review, 119, 573-577. ________. 1994. Cloud-to-ground lightning flash characteristics in the contiguous United States: 1989-1991. Journal of Geophysical Research, 99, 10 833-10 841. ________ and A. C. Silver. 1997. Lightning ground flash density in the contiguous United States: 1992-95. Monthly Weather Review, 125, 631-638. ________ and G. R. Huffines. 1999. Lightning ground flash measurements over the contiguous United States: 1995-1997. Monthly Weather Review, 127, No. 11, 2693-2703. ________ and G. R. Huffines. 2001. Cloud-to-ground lightning in the United States: NLDN results in the first decade, 1989-98. Monthly Weather Review, 129, 1179-1193. ________, G. R. Huffines, W. R. Burrows, R. L. Holle, and K. L. Cummins. 2002. The North American Lightning Detection Network (NALDN)first results: 1998-2002. Monthly Weather Review, 130, 2098-2109. ________. 2002b. Alphabetical bibliography on real-time lightning detection networks. Accessed http://www.cira.colostate.edu/ramm/visit/ltg_biblio_holle.pdf 08 May 04. Philander, S. G. H. 1990. El Nio, La Nia and the Southern Oscillation. Academic Press. San Diego, CA, 289 pp. Pielke, R.A., JR. and C. N. Landsea. 1999. La Nia, El Nio, and Atlantic hurricane damages in the United States. Bulletin of the American Meteorological Society 80, 2027-2033. 72

PAGE 83

Reap, R. M., 1994: Analysis and prediction of lightning strike distributions associated with synoptic map types over Florida. Monthly Weather Review, 122, 1698-1715. Ropelewski, C.F., and M.S. Halpert. 1986. North American precipitation and temperature patterns associated with the El Nio Southern Oscillation (ENSO). Monthly Weather Review, 114, 3352 2362. Schaefer, J.T., and F.B. Tatom. 1998. The relationship between El Nio, La Nia, and United States tornadoes. Preprints, 19th Conference on Severe Local Storms, Minneapolis, MN. American Meteorological Society, 416-419. Smith, Shawn R., David M. Legler, Mylene J. Remigio, and James J. OBrien. 1999. A Comparison of 1997-98 U.S. temperature and precipitation anomalies to historical ENSO warm phases. Journal of Climate 12(12), 3507-3515. Spivey, Ed Jr. 1998. Shocking News: El Nio Causes Hair Loss! Sojourners Magazine, January-February 1998 (Vol. 27, No. 1, pp. 66). Accessed at http://www.sojo.net/archives/magazine/index.cfm/mode/printer_friendly/actio n/so journers/issue/soj9801/article/980157.html 30 Nov 04. Steiger S. M. and R. E. Orville. 2003. Cloud-to-ground lightning enhancement over southern Louisiana. Geophysical Research Letters, 29, 10.1029/2003GL017923. Steiger, S.M, R. E. Orville, and G. R. Huffines. 2002. Cloud-to-ground lightning characteristics over Houston, Texas: 1989-2000. Journal of Geophysical Research, 107, D11, 10.1029/2001JD001142. Stroupe, Jessica. 2003. Northern Gulf coast lightning climatology. Accessed at http://bertha.met.fsu.edu/~jstroupe/gulfcoast.html 02 May 04. Trenberth, Kevin E. 1997. The definition of El Nio. Bulletin of the American Meteorological Society, 78, 2771-2777. Uman, M. A. 2000. The Lightning Discharge. Mineola NY: Dover Publications, Inc. Vaisala-GAI. 2002. Five year lightning strikes map. Created by Vaisala-GAI, Inc. for National Weather Service Lightning Safety World Wide Web Site. Accessed at http://www.lightningsafety.noaa.gov/lightning_map.htm 08 May 04. Wacker, R.S., and R. E. Orville. 1999. Changes in measured lightning flash count and return stroke peak current after the 1994 U.S. National Lightning Detection Network upgrade: 1, Observations; 2, Theory. Journal of Geophysical Research, 104, 2151-2162. 73

PAGE 84

Wallace, JM. 1975. Diurnal variations in precipitation and thunderstorm frequency over the conterminous United States. Monthly Weather Review, Boston, MA. Vol. 103, no. 5, pp. 406-419. Watson, A.I., and R.L. Holle. 1996. An eight-year lightning climatology of the southeast United States prepared for the 1996 Summer Olympics. Bulletin of the American Meteorological Society, 77, 883-890. Zajac, B. A., J. F. Weaver, and D. E. Bikos. 2002. Lightning meteorology I: Electrification and lightning activity in typical thunderstorms. Virtual Institute for Satellite Integration Training. Accessed at http://www.cira.colostate.edu/ramm/visit/ltgmet1.html 01 May 04. Zajac, B. A., and S. A. Rutledge. 2001. Cloud-to-ground lightning activity in the contiguous United States from 1995 to 1999. Monthly Weather Review, 129, 999. 74

PAGE 85

APPENDICES 75

PAGE 86

APPENDIX A NINO 3.4 Time Series Table 3: NINO 3.4 VALUES YEAR MONTH SST ANOM 1995 1 27.55 1.04 1995 2 27.45 0.76 1995 3 27.63 0.48 1995 4 27.93 0.25 1995 5 27.73 -0.04 1995 6 27.59 0.09 1995 7 27.01 -0.07 1995 8 26.33 -0.38 1995 9 25.96 -0.68 1995 10 25.67 -0.93 1995 11 25.66 -0.86 1995 12 25.57 -0.91 1996 1 25.74 -0.77 1996 2 25.85 -0.85 1996 3 26.62 -0.52 1996 4 27.36 -0.32 1996 5 27.37 -0.39 1996 6 27.32 -0.17 1996 7 27.09 0.01 1996 8 26.56 -0.14 1996 9 26.35 -0.3 1996 10 26.24 -0.36 1996 11 26.19 -0.32 1996 12 26.02 -0.45 1997 1 25.96 -0.55 1997 2 26.36 -0.33 1997 3 27.03 -0.11 1997 4 28.03 0.34 1997 5 28.6 0.84 1997 6 28.94 1.45 1997 7 28.92 1.85 1997 8 28.84 2.14 1997 9 28.93 2.29 1997 10 29.23 2.64 1997 11 29.32 2.8 1997 12 29.26 2.78 1998 1 29.1 2.59 1998 2 28.86 2.17 1998 3 28.67 1.53 1998 4 28.56 0.87 1998 5 28.47 0.71 1998 6 26.72 -0.78 76

PAGE 87

1998 7 25.94 -1.14 1998 8 25.49 -1.22 1998 9 25.61 -1.04 1998 10 25.34 -1.26 1998 11 25.18 -1.33 1998 12 24.79 -1.69 1999 1 24.9 -1.61 1999 2 25.41 -1.28 1999 3 26.25 -0.89 1999 4 26.84 -0.84 1999 5 26.97 -0.79 1999 6 26.6 -0.9 1999 7 26.35 -0.73 1999 8 25.59 -1.12 1999 9 25.71 -0.94 1999 10 25.64 -0.96 1999 11 25.12 -1.39 1999 12 24.9 -1.57 2000 1 24.65 -1.86 2000 2 25.19 -1.51 2000 3 26.08 -1.06 2000 4 27.01 -0.67 2000 5 27.12 -0.64 2000 6 27.03 -0.46 2000 7 26.72 -0.36 2000 8 26.45 -0.25 2000 9 26.21 -0.43 2000 10 25.96 -0.63 2000 11 25.78 -0.74 2000 12 25.59 -0.88 2001 1 25.74 -0.77 2001 2 26.11 -0.58 2001 3 26.84 -0.3 2001 4 27.52 -0.16 2001 5 27.6 -0.16 2001 6 27.68 0.19 2001 7 27.32 0.24 2001 8 26.87 0.17 2001 9 26.55 -0.09 2001 10 26.59 0 2001 11 26.45 -0.07 2001 12 26.17 -0.3 2002 1 26.5 -0.02 2002 2 26.95 0.25 2002 3 27.32 0.17 2002 4 27.94 0.26 2002 5 28.15 0.39 77

PAGE 88

2002 6 28.43 0.94 2002 7 27.98 0.9 2002 8 27.79 1.08 2002 9 27.83 1.19 2002 10 28.05 1.46 2002 11 28.27 1.75 2002 12 28.09 1.62 78

PAGE 89

APPENDIX B PERL SCRIPT #!/usr/bin/perl #Written by Karl D. Pfeiffer MAIN: { for ($lag = 0; $lag < 33; $lag++) { my $outfile = sprintf("%02d-month-lag.txt",$lag); if ( -e $outfile ) { print "File $outfile already exists .. exiting\n"; } else { open(OUTFILE,">$outfile"); my @sst = get_sst(); my $x; my $y; for ($x = 0; $x < 418; $x++) { for ($y = 0; $y < 816; $y++) { my $datafile = sprintf("grids/timeseries/%03d/%03d/point",$x,$y); open(DATAFILE,$datafile) or die "Could not open $datafile:$!\n"; my @data = ; close(DATAFILE); my @lightning; for ( @data ) { my ($y,$m,$v) = split / /; push @lightning,$v; } my $lg; for ( $lg = 0; $lg < $lag; $lg++) { shift @lightning; } my $r = correlation(\@sst,\@lightning); my $r = correlation(\@sst,\@lightning); printf(OUTFILE "%03d %03d %12.7f\n",$x,$y,$r); printf("%s %03d %03d %12.7f\n",timestamp(),$x,$y,$r); } } 79

PAGE 90

close(OUTFILE); } ## end else } ## end for lag } ## end MAIN sub correlation { my $X = shift @_; my $Y = shift @_; my $n = @$X; my $SUM_X = 0; my $SUM_Y = 0; my $SUM_X2 = 0; my $SUM_Y2 = 0; my $SUM_XY = 0; my $i; for ( $i = 0; $i <= $n ; $i++ ) { $SUM_X = $SUM_X + $X->[$i]; $SUM_Y = $SUM_Y + $Y->[$i]; $SUM_XY = $SUM_XY + ($X->[$i] $Y->[$i]); $SUM_X2 = $SUM_X2 + $X->[$i]*$X->[$i]; $SUM_Y2 = $SUM_Y2 + $Y->[$i]*$Y->[$i]; } my $r; if ( $SUM_X == 0 || $SUM_Y == 0 ) { $r = 0; } else { $r = ($SUM_XY ($SUM_X $SUM_Y)/$n)/ sqrt(($SUM_X2 $SUM_X*$SUM_X/$n)*($SUM_Y2 $SUM_Y*$SUM_Y/$n)); } return $r; } sub get_sst { my @sst; open(SST,"sst.txt"); my @data = ; close(SST); foreach ( @data ) { s/\s+/ /g; 80

PAGE 91

my ($y,$m,$v) = split / /; push @sst,$v; } return @sst; } ## end get_sst() sub timestamp { my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime(time); my $stamp = sprintf("%04d/%02d/%02d %02d:%02d:%02d ", 1900+$year, $mon+1, $mday, $hour, $min, $sec); return $stamp; } 81

PAGE 92

82 APPENDIX C SPLUS SCRIPT ************Script run for each of the 96 months in the study********** guiImportData(FileName = "H:\\CompletedXY\\1995\\sr\\XYgulf_9501_lgt.dbf", FileTypes = "dBASE file (dbf)", TargetDataFrame = "d9501", TargetStartCol = "", TargetInsertOverwrite = "Create new data set", NameRowAuto = "Auto", NameColAuto = "Auto", StartCol = 1, EndCol = "", StartRow = 1, EndRow = "", PageNumberAuto = "Auto", StringsAsFactors = T, SortFactorLevels = T, LabelsAsNumbers = F, CenturyCutoffYear = 1930, KeepOrDropList = "", SeparateDelimiters = T, ASCIIDateInFormat = "M/d/yyyy", ASCIITimeInFormat = "h:mm:ss tt", ASCIIDecimalPoint = "Period (.)", ASCIIThousandsSeparator = "None") x1

xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200409Ka 4500
controlfield tag 006 m d s
007 cr mn
008 040712s2004 flua sbm s000|0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0000363
035
(OCoLC)56137634
9
AJR7187
b SE
040
FHM
c FHM
090
G116
1 100
Lajoie, Mark R.
4 245
The influence of the El Nio-southern oscillation on cloud-to-ground lightning activity along the Gulf Coast of the United States
h [electronic resource] /
by Mark R. Lajoie.
260
[Tampa, Fla.] :
University of South Florida,
2004.
502
Thesis (M.A.)--University of South Florida, 2004.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 92 pages.
520
ABSTRACT: This study investigates the response of lightning to the El Nio Southern Oscillation (ENSO) in the vicinity of the U.S. Gulf Coast region and nearby adjacent waters of the Gulf of Mexico, for the years 1995-2002. The Gulf Coast region was selected for this study because of its high flash density (Orville and Huffines, 2001) and because it is an area where the ENSO fingerprint is very clearly demonstrated on both temperature and precipitation patterns (CPC, 2002). Additionally, this geographic domain roughly matches the only known study on this topic (Goodman et al., 2000). Winter is the season of greatest response to ENSO (CPC, 2004), and past studies show that summer has the most lightning activity (e.g., Orville and Huffines, 2001). The temporal domain of the study is restricted to 1995 and beyond, as this follows a system-wide upgrade of the National Lightning Detection Network (NLDN) that improved overall flash detection efficiency (Cummins, et. al.1998; Wacker and Orville, 1999). Both qualitative and quantitative methods were employed to explore the lightning data for ENSO teleconnections. Mean flash density maps were constructed for the complete period of record, individual months and the winter and summer seasons. Maps were visually examined for qualitative comparison with past climatologies and the Goodman et al., (2002) ENSO study. Additionally, monthly flash deviations are computed, visualized and correlated with the Nio 3.4 SST anomaly for all months in the study, seeking out variations in both the amount of flash deviation and spatial properties. Abundant literature exists on both ENSO and lightning individually. This study offers an insight into their intersection.
590
Adviser: Laing, Arlene G.
653
climatology.
enso.
flash density.
La Nia.
teleconnection.
0 690
Dissertations, Academic
z USF
x Geography
Masters.
773
t USF Electronic Theses and Dissertations.
949
FTS
SFERS
ETD
G116 (ONLINE)
nkt 7/20/04
856
u http://digital.lib.usf.edu/?e14.363