OBSERVATIONAL ANALYSIS OF THE INTERACTION BETWEEN A

OBSERVATIONAL ANALYSIS OF THE INTERACTION BETWEEN A

OBSERVATIONAL ANALYSIS OF THE INTERACTION BETWEEN A BAROCLINIC BOUNDARY AND SUPERCELL STORMS ON 27 APRIL 2011 by ADAM THOMAS SHERRER A THESIS Subm...

3MB Sizes 0 Downloads 0 Views

Recommend Documents

Fundamental analysis of the interaction between overburden - SAIMM
Fundamental analysis of the interaction between overburden behaviour and snook stability in coalmines by J.N. van der Me

VirB7 interaction
Jan 30, 2007 - Richard Bayliss*†, Richard Harris*‡, Loic Coutte§, Amy Monier§, Remi Fronzes*, Peter J. Christie§,

GALVANIC INTERACTION BETWEEN CARBON FIBER - A Sagues
Keywords: galvanic corrosio%CFRP, Carbon Fiber Reinforced Plastics, concrete, relative humidity, chloride contamination.

Analytical and Observational Relations Between Landslides Volume
Oded Katz, Asaf Klar, Einat Aharonov,. Borina Kalderon- ... et al.,). The above work presents an analytical examination

A Practice-Based Analysis of Social Interaction in a Massively
Multiplayer Online Games) that investigates how what the players do in the gaming environment can give rise to ... A Pra

Antibody-Antigen Interaction Analysis
Antibody-Antigen Interaction Analysis. Application Note NT0011. Using MST to analyse the binding of Nanobodies and. Nano

A Political and Historic Analysis of the Relationship between the
How the relationship between the United States and Saudi Arabia has influenced U.S. Foreign .... The interest in oil cam

The interaction between price and value
American Dad 990,000. TheTerminal 774,000. Broke Girls 1.68 mln. Guys With Kids 480,000. Bones 1.04 mln. Don't Trust the

Interaction between the invasive macroalga Lophocladia - IMEDEA
bonnel E, Cadoiu G (2001) Espe`ces, peuplements et paysages marins remarquables de la Ciotat, de l'ile Verte a` la calan

A Case of Suspected Drug Interaction Between Topiramate and Co
Oct 10, 2017 - Case description: A 4-year-old boy presented with tuberous sclerosis and symptomatic partial epilepsy of

OBSERVATIONAL ANALYSIS OF THE INTERACTION BETWEEN A BAROCLINIC BOUNDARY AND SUPERCELL STORMS ON 27 APRIL 2011

by

ADAM THOMAS SHERRER

A THESIS

Submitted in partial fulfillment of the requirements for the degree of Master of Science in The Department of Atmospheric Science to The School of Graduate Studies of The University of Alabama in Huntsville

HUNTSVILLE, ALABAMA 2014

ABSTRACT The School of Graduate Studies The University of Alabama in Huntsville

Degree

Master of Science

College/Dept.

Science/Atmospheric Science

Name of Candidate Adam Sherrer Title Observational Analysis of the Interaction Between a Baroclinic Boundary and Supercell Storms on 27 April 2011 A thermal boundary developed during the morning to early afternoon hours on 27 April as a result of rainfall evaporation and shading from reoccurring deep convection. This boundary propagated to the north during the late afternoon to evening hours.

The presence of the boundary produced an area more conducive for the formation of strong violent tornadoes through several processes. These processes included the production of horizontally generated baroclinic vorticity, increased values in stormrelative helicity, and decreasing lifting condensation level heights.

Five supercell storms formed near and/or propagated alongside this boundary. Supercells that interacted with this boundary typically produced significant tornadic damage over long distances. Two of these supercells formed to the south (warm) side of the boundary and produced a tornado prior to crossing to the north (cool) side of the boundary. These two storms exhibited changes in appearance, intensity, and structure. Two other supercells formed well south of the boundary. These two storms remained relatively weak until they interacted with the boundary. These storms then rapidly intensified and produced tornadoes.

iv

Acknowledgements The completion of this thesis would not have been possible without the help and guidance of many people. I would first like to thank Dr. Knupp who has provided me hours of guidance not only with this research and graduate school but throughout my undergraduate study as well. I would also like to thank the other members of my committee Drs. Larry Carey and Donald Perkey for the assistance they have provided as well. I would also like to thank Todd Murphy and Ryan Wade for their continued guidance through this process as well. In addition, many other friends and colleagues helped in the research process. These friends include: Aaron Mayhew, Matthew Saari, Lamont Bain, Stephanie Mullins, Ryan Rogers, Dustin Phillips, and Tony Lyzaa. Finally, I need to thank my family and fiancé. If not for their unwavering support the completion of this process would have seemed impossible. This research was supported by National Science Foundation, Grant ASG1110622.

vi

TABLE OF CONTENTS Page List of Figures.....................................................................................................................ix List of Tables ...................................................................................................................xvi Chapter 1.

INTRODUCTION ..................................................................................................1

2.

BACKGROUND ....................................................................................................4 2.1

3.

4.

5.

Supercell Thunderstorms ……....................................................................4 2.1.1

Environmental Parameters of Supercells ........................................5

2.1.2

Supercell Structure .........................................................................8

2.2

Quasi Linear Convective Systems & Bow Echoes ...................................10

2.3

Thermal Boundaries Roles in Supercell & Tornado Outbreaks ...............12

2.4

Previous Work with Thermal Boundaries and Tornadic Storms ………..15

DATA AND METHODOLGY ............................................................................22 3.1

Radars .......................................................................................................22

3.2

Radar Processing …………………………………….………………......24 3.2.1

Editing …………………………………………………………...24

3.2.2

EVAD Analysis …………………………....................................26

3.3

Surface Data ………………………………………….............................27

3.4

MIPS Data ……………………………………………………………….32

SYNOPTIC & MESOSCALE ENVIRONMENT ………………………………34 4.1

Synoptic Environment …………………………………............……......35

4.2

Mesoscale Environment …………………………………………………40

BOUNDARY DEVELOPMENT & CHARACTERISTICS ……………………44 vii

5.1

Boundary Development ……………………………....…………………44

5.2

Boundary Progression ………………………………………………...…54 5.2.1

Vertical Boundary Development ………………………………..63

5.3 Boundary Characteristics ……………………………………………………67 5.3.1

Temperature and Stability ……………………………………….67

5.3.2 Variation of the Vertical Winds …………………………………...70 5.3.3 6.

Storm-Relative Helicity …………………................................…76

SUPERCELLS INTERACTING WITH THE BOUNDARY….………..………78 6.1

Cullman EF-4 Supercell …………………………………………………79

6.2

Hackleburg EF-5 Supercell ……………………………………………...86

6.3

Smithville EF-5 Supercell ……………………………………….............95

6.4

Jackson County EF-4 Supercell …………............................……………98

6.5

Limestone County Supercell ..……………………………….................102

6.6

Non-Tornadic or Weakly Tornadic Supercells ……………...................105

6.7

6.6.1

Supercells in Central Mississippi ………………...........…….…105

6.6.2

Northwest Alabama Supercell ………………………................105

Summary………………………………………………………………..106

7.

DISCUSSION ………………………….............................................................108

8.

CONCLUSION & FUTURE WORK…………..................................................115 8.1 Conclusions…………………………………………………………………115 8.2 Future Work …………………………................................................……..117

viii

LIST OF FIGURES Page Figure 2.1 Figure 2.1: Tilting of horizontal vortex lines relative to storm motion. Image from Markowski and Richardson 2010, adapted from Davies-Jones 1984. ……………………7 2.2 Conceptual structure of a supercell (Lemon and Doswell 1979) solid lines indicate storm echo. Solid lines with arrows represent storm relative streamlines. The storm’s forward-flank, rear-flank downdrafts are represented as well as the updraft location…....9 2.3 Conceptual model of a QLCS denoting the locations of the pre-squall low, mesohigh, straitiform regions and wake low (Markowski and Richardson 2010, adapted from Johnson and Hamilton 1988). ……………………………………………………..12 2.4 Depiction of the production of baroclinically generated horizontal vorticity (Markowski and Richardson 2010 adapted from Klemp 1987). Solid black lines represent the horizontal vortex lines. The white arrows represent parcel trajectories into the storm. …………………………………………………………………………………………...14 2.5 Bar graphs presenting the frequency of tornadogensis in the presence of a boundary during VORTEX (top) and tornado occurrence relative to distance from boundary in VORTEX (bottom) (Markowski et al. 1998). …………………..…………17 2.6 Figure from Markowski et al. 1998 demonstrating potential storm locations and interactions relative to horizontal vorticity generated by the boundary. ………..……....18 2.7 Coneptual model of a boundary and associatied wind profiles. Wind profiles show backing of the wind at lower levels on cool side of boundary with more unidirectional flow on warm side(Maddox et al. 1980). ..........................................................................19 2.8 Chart showing measured rotational velocity as a function of boundary position (Rasmussen et al. 2000). Relative maximum occurs between 0 and 50 m s-1. …...…….21 3.1 Surface observations and organization type. Locations without a symbol denote other cities mentioned in this study. ……………………………………………….........30 4.1 : Storm Prediction Center’s 12 UTC Convective Outlook for April 27, 2011. A high risk was placed over northern Alabama bordering parts of Mississippi, Tennessee, and Georgia. ……………………………………………………………………………..34 4.2 : 1200 UTC 300 mb observations showing a deep trough with large divergence located over Mississippi, Alabama, and Tennessee. …………………………………….36 4.3 1200 UTC 850 mb observations showing strong southerly low-level winds over the southeast…………….………………………………………………………………..36 ix

4.4 1200 UTC surface observations showing a warm front in extended through Arkansas into Missouri with a cold front in southwestern Arkansas. Southerly flow resides over much of the southeast……..…………………………..……………….…...38 4.5 1200 UTC 300 mb observations displaying the negatively tilted trough axis over Arkansas Mississippi, and Alabama. Strong diffluence existed in the exit region of the Jetstream that enhanced divergence values. ……………………..………………….…..38 4.6 850 hPa analysis at 0000 UTC on 28 April demonstrating strong southerly flow east of the trough axis……………………………………………………………………39 4.7 0000 UTC surface pressure and wind at 0000 UTC on 28 April demonstrating the location of the cold front across Mississippi and Alabama. …………………………….39 4.8 SPC Mesoanalysis graphics for SBCAPE (solid contours J kg-1) and SBCIN (dashed contours with shading, top figure) and 0-1 SRH (m2 s-2). Surface temperatures are not shown due to the presence of cold air over north Alabama which was no accurately diagnosed. ……………….................................................................................................42 4.9. 0.7° PPI from ARMOR at 2000 UTC. White dashed line denotes approximate boundary position at this time. …………………………………………………………..43 5.1 0.5° PPI scan from KGWX at 1400 UTC. Figure shows the eastern portion of the QLCS over northern Mississippi and Alabama. ……………………………………...…48 5.2 0.7 ° PPI scan from KGWX at 1600 UTC showing the fully developed QLCS over northern Alabama.………………………………………………………….…........48 5.3 0.7° PPI scan from ARMOR at 1800 UTC showing scattered convective showers over north Alabama. At 1800 UTC the QLCS was in eastern Alabama.……….……….49 5.4

0 .5° PPI scan from KHTX at 2000 UTC.…………………………………….…49

5.5 Bear Creek, Alabama time series of temperature (red, °C), dew point (green, °C), surface pressure (black, pressure-1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. The location of Bear Creek is shown in Figure 3.1.……………………50 5.6 Muscle Shoals, Alabama time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. The temperature in Muscle Shoals remained cool and nearly saturated through the evening hours. The boundary passed this location just prior to 2300 UTC. The location of Muscle Shoals is shown in Figure 3.1………………………………………50

x

5.7 Russellville, Alabama time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. Dew point was not shown due to bad data. The location of Russellville is shown in Figure 3.1.……………………………………………………………………………………….51 5.8 Hayden, Alabama time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. The location of Hayden is shown in Figure 3.1………………………………………….51 5.9 UAH time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. Power failure occurs at 2215 UTC. This is likely prior to passage of the boundary. The location of UAH is shown in Figure 3.1. ……………………………………………………………,…….52 5.10 Huntsville ASOS time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. The location of KHSV is shown in Figure 3.1. …………………………………...……..52 5.11 Temperature time series from Huntsville (KHSV) (purple), Tuscaloosa (KTCL) (green), Muscle Shoals (KMSL) (black), Anniston (KANT) (red), and Montgomery (KMGM) (blue) AL. Image demonstrates the passage of the early morning (dashed lines 1 and 2) and midday QLCSs (dashed lines 3 and 4) as well as the boundary passage in Huntsville and Muscle Shoals (dashed lines 5 and 6). …………………………………..53 5.12 0.5° ARMOR PPI with θ values for 1800 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location. …………………………………56 5.13 0.5° ARMOR PPI with θ values for 1900 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location. ………………………………....56 5.14 0.5° ARMOR PPI with θ values for 2000 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location. …………………………………57 5.15 0.5° ARMOR PPI with θ values for 2100 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location. …………………………………57

xi

5.16 0.5° ARMOR PPI with θ values for 2200 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location. …………………………………58 5.17 Time series from M3V depicting measured temperature, dewpoint, wind speed, and car speed. Temperature and dewpoint are displayed on the left axis while wind speed and car speed are displayed on the right.………………………………………………...61 5.18

Path displaying path M3V traveled while transecting the boundary. …………...61

5.19 RSA sounding launched at 1700 UTC prior to the passage of the midday QLCS and formation of the boundary. .………………………………………………………...65 5.20

UAH MPR sounding at 2030 UTC showing the fully developed inversion….…65

5.21 UAH MPR sounding at 2115 UTC showing the inversion depth decreasing. The height of the inversion has decreased to approximately 950 mb. The depth of the coldest air has now been reduced to just above the surface….…………………………………..66 5.22 2100 UTC sounding launched from the National Space Science Technology Center on the Campus of UAH. Sounding was launched approximately 25 km into the cool side of the boundary as the Hackleburg storm was located approximately 60 km southwest. ……………………….……………………………………………………….69 5.23 April 2011 00 UTC sounding from Calera, Alabama. Launch location 90 km south of the southernmost extent of the boundary. ……………………………...………69 5.24 20-1000 m EVAD analysis winds at KBMX from 1800-2400 UTC. One barb= 10 kts. KBMX’s location relative to the boundary is depicted in images 5.12-5.16…….…71 5.25 EVAD analysis winds from KGWX from 19-22 UTC over 20-1000 m. Frontal passage occurs prior to 2200 UTC. One barb= 10 kts. KGWX’s location relative to the boundary is depicted in images 5.12-5.16. ……………………………………….…..…71 5.26 EVAD Analysis from KHTX from 18-22 UTC. One barb= 10 kts. KHTX’s location relative to the boundary is depicted in images 5.12-5.16. .……………………73 5.27 ARMOR and 915 MHz Wind Profiler wind profile. Easterly flow dominates the 1900 to 2100 UTC lowest levels. The 0-200 m winds become more southerly at approximately 2030 UTC with the surface winds becoming southerly at 2115 UTC. The lowest 200 m data is supplied by ARMOR radar located at Huntsville International Airport. The Huntsville ASOS is also located there and shows the boundary passage to have occurred between 2130 and 2200 UTC. The 200-1000 m winds are provided by the 915 MHz profiler located on the campus of UAH. One barb=10 kts. ARMOR’s position relative to the boundary is depicted in Figure 5.12-5.16. ………………………………75

xii

5.28 SRH calculation from UAH’s ARMOR and 915 MHz wind profiler. Shows a rapid increase in SRH starting at 2115 UTC just prior to boundary passage. After reaching a maximum in SRH at 2140 UTC SRH values fall off dramatically to values close to 200 m2 s-2. …………………………………………………………………….77 6.1 Six panel display showing the development of the Cullman EF-4 supercell. The Cullman storm is highlighted by the white box in each image. Tornadogenesis occurs at 1942. …………………………………………………………………………..………..82 6.2 Photograph taken of storm at 1922 UTC in northwest Walker County. Image looking northeast. A wall cloud has developed at this time. Image also depicts the relatively high cloud base associated with the warm side of the boundary (refer to Figures 6.5 & 6.6). Photograph by Dr. Timothy Coleman. ..……………………………………82 6.3 Time series from Cullman, Alabama showing temperature (red), dewpoint (green), pressure-1000 mb (black) on left axis. Total rainfall since 0000 UTC (blue) is on right axis. Image shows boundary passage at this location was near 1955 UTC. Cullman received no rainfall from the midday QLCS, but experienced a drop in temperature and shift to northerly winds. The passage of the FFD of the Cullman supercell is near 1945 UTC when 4 mm of rainfall was recorded. An additional 8 mm of rainfall occurred approximately an hour later when a second supercell crossed Cullman. ………..…….83 6.4 ARMOR 0.7° PPI demonstrating the approximate location of the Cullman storm relative to the surface observation. Values of θ for this image were constrained to ± 5 minutes of indicated times. ……………………………………………………….……83 6.5 Picture taken from Cullman AL at approximately 1955 UTC looking northwest. Tornado is approximately 2.5 km northwest of camera. Image shows multiple vortices surrounding the tornado. A higher cloud base is also demonstrated in this image (refer to Figure 6.6). This higher cloud base provides further confidence the storm is located on the warm side of the boundary. Photograph by FOX WBRC in Birmingham. …….….84 6.6 Cullman EF-4 tornado photographed at 2020 UTC looking northwest from Arab, Alabama. Storm is located approximately 6 km northwest. Cloud base surrounding tornado is now closer to ground indicative of a lower LCL height which is produced by the cooler and more humid air on the cool side of the boundary. Photograph by Charles Whisenant. …………………………………………………………………………….84 6.7 Rotational velocity (Vrot) of the Cullman EF-5 storm and approximate position relative to the boundary. Data from KHTX was used to determine values of Vrot. Positive distance denote storm is north of the boundary. A relative maximum of 35 m s-1 was reached when the storm was determined to be nearest the boundary. ………………..85 6.8 0.5° PPI from KGWX at 2001 UTC. Storm is located approximately 20 km north of the radar. Left panel indicates radar reflectivity (Z) while right panel is storm-relative velocity (SRV). …………………………….………………………………………….87 xiii

6.9 0.5° PPIs from KGWX demonstrating the progression from first appearance on radar to tornadogenesis. The storm produced its first damage report within one hour of first appearing on radar……………………………………………………………..…88 6.10 0.5° PPI from KGWX (panels A-D) and 0.7° PPI from ARMOR (panels E&F) demonstrating the Hackleburg’s supercell relative to the boundary’s position from 2001 to 2115 UTC. The boundary’s location is indicated by the white dashed line……….....88 6.11 Time series from Russellville, Alabama showing temperature (red), dew point (green), pressure-1000 mb (black) on left axis. Total rainfall since 0000 UTC (blue) is on right axis.. ……………………………………………………………………………….91 6.12 Photograph taken of the Hackleburg EF-5 at 2130 UTC from 20 km to the northeast. Image shows low cloud base associated with the storm. LCL was estimated to be approximately 100 m at this time. Photograph courtesy of Gary Cosby, Decatur Daily. ……………………………………………………………………………………………91 6.13 Contour of Vrot as of the Hackleburg storm as determined from both ARMOR and KGWX. The right axis demonstrates distance from the radar (solid black line) and the author’s estimation of the mesocyclone distance from the boundary (black line with dots). KGWX data is displayed for 2001 to 2042 UTC. After 2042 UTC KGWX is used to supplement upper level data with ARMOR supplying the 1000-3000 m. Due to extended range from KGWX and a three-tilt scan mode by ARMOR further analysis is not done………………………………………………………………………………………94 6.14 0.5° PPIs from KGWX depicting the evolution from QLCS to supercell that produced the Smithville EF-5 supercell. ………………………………………………...97 6.15 0.5°PPI from KGWX showing approximate boundary position relative to the Smithville EF-5 supercell. The S, L, and H denote the Smithville, Limestone County, and Hackleburg supercells respectively. .……………………………………………………97 6.16 0.5° PPIs from KBMX (A,B,C) & KHTX (D,E,F) depicting the formation and development of the Jackson County EF-4 tornado. The storm remained relatively weak for 2.5 hours until reaching the Jackson County area. Here the storm rapidly intensified and produced a tornado at 2207 UTC. The J, L, C and H denote the Jackson County, Limestone County, Cullman, and Hackleburg supercells respectively ……………….100 6.17 Scottsboro, Alabama measured temperature. A spike in temperature from 20°C to 23°C is shown between 2100 and 2130 UTC corresponding to a period increased solar insolation. This spike was followed by a sharp decrease in temperature that corresponds to the passage of the FFD from the Cullman supercell. Following this decrease a gradual increase is seen from 2100 to 2330 UTC.……………………………………………..100

xiv

6.18 KHTX 0.5° tilt at 2205 UTC showing the Jackson Ef-4 supercell. Surrounding surface theta values shown for 2000 UTC. A temperature gradient exists between north Huntsville and Scottsboro, Alabama. ………………………………………………...101 6.19 0.5° PPI from KHTX depicting the rapid intensification of Vrot between 2134 UTC and 2211 UTC. Uncertainty in boundary position prevents approximate boundary position from the mesocyclone from being displayed. ………………………………...101 6.20 0.5° PPIs from KGWX (A,B,C,D) and 0.7° PPIs from ARMOR (E, F) demonstrating the development of the Limestone County EF-2 supercell. Despite its slow development it developed rapidly upon catching up to the boundary near Limestone County at 2153 UTC. The L, S and H denote the Limestone County, Smithville, and Hackleburg supercells respectively.……………………………………………………104 6.21 0.5 PPI from ARMOR showing the approximate boundary location relative to the Limestone County EF-2 producing supercell at time of tornadogensis. ……………….104 6.22 0.5° PPI from KGWX showing the northern-most supercell that developed from the QLCS in central Mississippi. ………………………………………………………107 6.23 0.5° PPI from KGWX showing the middle supercell (A) from the QLCS and the supercell that developed in northwest Alabama (B). Also pictured are the Hackleburg EF5 supercell, Hackleburg EF-5, and Cullman EF-4 supercells. …………………………107

xv

LIST OF TABLES Page Table 3.1 City, State, Organization and report interval for surface observations. …………29

xvi

CHAPTER 1

INTRODUCTION

The 27 April 2011 tornado outbreak ranks as one of the worst in the history of the United States regardless of the metric used to determine it. On that day 199 tornadoes were spawned. The state of Alabama alone had 63 confirmed tornadoes with 13 being rated EF-3 or greater. Of the 13 EF-3 tornadoes, three were determined to be EF-5 intensity. Of the 316 people were killed by tornadoes on 27 April, 238 deaths occurred in the state of Alabama. An additional 24 fatalities occurred due to other hazardous weather. The total damage loss along the outbreak area was estimated at about 10 billion dollars (Simmons et al. 2012, Knupp et al. 2014). The outbreak was separated into three events over the affected area. First, there was the early-morning quasi-linear convective system (QLCS). This QLCS formed an embedded mesoscale connective vortex (MCV) and was responsible for spawning 76 tornadoes (28 in Alabama), most of which were rated as EF-2 or lesser magnitude and with a total of five reaching EF-3 intensity. The second event was another QLCS that developed in northeastern Mississippi and then progressed across northern Alabama. This QLCS spawned seven tornadoes, the strongest of which rated as an EF-2. The final event

was the supercell storm outbreak that occurred from eastern Mississippi to Georgia and reached as far north and east as Virginia. These supercells produced 29 tornadoes that were rated as EF-3 or higher and accounted for a vast majority of the deaths and damage. While the three events were distinct, the mid-day QLCS and afternoon supercell outbreak are very much related. It will be shown that the mid-day QLCS and subsequent convective showers generated and maintained a cool air mass over the northeastern portion of Mississippi and the northern half of Alabama. This cool air persisted into the late evening hours. These persistent showers also reduced solar insolation over the same region through this portion of the day. The combination of the outflow and limited insolation over north Alabama, coupled with the clear skies and warm air in central and south Alabama, led to the formation of a thermal boundary. The role of thermal boundaries in severe weather has been studied as early as the late 1950’s with the publication of Magor 1959. Since that time multiple studies have been conducted to detail the interactions of these boundaries with supercells and to try and quantify the significance of the boundary in tornadogenesis. These studies however were typically limited to the immediate boundary position and its impacts on a specific storm. In this study the boundary interacted with multiple storms while propagating toward the north. A dense network of observations on the meso-γ scale were collected to show the progression and temperature difference across the boundary. Furthermore, this network of observations provides the ability to relate storm position with the storms intensity and structure as a function of time. The presence of this boundary appeared to be influential to the supercell outbreak. The five tornado producing supercells that formed and/or propagated within a region

2

extending 10 km into the warm side (south) of the boundary to 20 km into the cool side (north) of this boundary include the Cullman EF-4, Hackleburg EF-5, Smithfield EF-5, Jackson County EF-4 and Limestone County EF-2 producing supercells (Knupp et al. 2014). Two more supercells were deemed to have been near the boundary yet they did not produce any tornadoes. The primary objective of this paper is to show how the presence of this boundary impacted the near supercell environment such that storms determined to be within 10 km south to 20 km north of the boundary were able to strengthen. The five supercells that traversed alongside the boundary all produced significant tornado damage (EF-2 or greater) and most produced damage paths of 50 km or greater. Furthermore, four of these storms produced a total of five tornadoes that were classified as EF-4 or greater. The Hackleburg EF-5 supercell was the main beneficiary of this boundary. This supercell produced a tornado that was determined by the NWS to have stayed on the ground for 212 km while producing EF-4 or greater damage for a majority of this path length.

3

CHAPTER 2

BACKGROUND

For the role of the thermal boundary’s influence to be fully explored in this study, it is first helpful to review supercell and outflow basics. Therefore, the following sections will discuss supercell basics as well as the thermodynamic and dynamic processes involved in the formation of a thermal boundary. Finally, the theory of this interaction will be visited using previous case studies. 2.1 Supercell Thunderstorms Supercell thunderstorms are defined as convective storms with a single, quasi steady, rotating updraft that lasts longer than it takes for one air parcel to flow completely through it (Houze 1993, Bluestein 1993, Glickman 2000). The National Oceanic and Atmospheric Association (NOAA) states that, operationally a supercell can be defined as any storm where a mesocyclone has formed such that the rotational velocity (Vrot), defined as Vrot=[(max outbound velocity + max inbound velocity)/2]=20 m s-1. This mesocyclone must also extend to a depth of 2,400 to 3,000 m and must persist for at least 2 volume scans with rotational velocity values of greater than 20 m s-1 (NWS 2009). Essentially, a supercell is a highly organized type of convection. The rotating updraft

is known as a mesocyclone which is defined as a storm scale (3-8 km) cyclonic vortex with large values of vertical vorticity on the order of 10-2 s-1 (Lemon and Doswell 1979, Brooks, et al. 1994, Markowski and Richardson 2010, NWS 2009). Mesocyclone size can vary from 3 to 8 km and can last on the order of minutes to hours depending on the storm from which it originates (Moller et al.1994, Miller et al. 1988). Supercells often meet and surpass the NWS criteria for a severe thunderstorm. Severe thunderstorm criteria include winds in excess of 50 kts and/or hail of 1” in diameter (Galway 1989, Johns and Dowell 1992). Supercells storms are typically found in atmospheric environments that have an abundance of convective available potential energy (CAPE) and or vertical shear of horizontal wind. In most cases these storms produce some type of damage from hail, straightline winds, or tornadoes. This has prompted many studies examining the ranges of environments required to form these storms (Brooks et al. 1994). Studies by Johns et al. (1993), Korotky et al. (1993), Rasmussen (2003), Rasmussen and Blanchard (1998), and Bunkers et al. (2006) have shown that combinations ranging from high CAPE and lowshear to low and CAPE high shear are conducive to supercell environments. In general, these studies all conclude that the preferred environment for supercell storms is most often some combination of high directional wind shear and CAPE. 2.1.1 Environmental Parameters of Supercells An important parameter used when describing the atmosphere’s ability to produce severe weather is instability. Instability is the measure of the atmosphere’s buoyancy. If a parcel of air resides in an environment in which it can be lifted from its current state and then be more buoyant than it was in its original environment, it is said to have positive buoyancy. More commonly this buoyancy is referred to as (CAPE). CAPE is defined as

5

the maximum energy available to an ascending parcel according to parcel theory (GLICKMAN 2000). CAPE is mathematically defined as

θ v′ ( z ) − θ v ( z ) dz θ v (z) LFC EL

CAPE ≡ g



[1]

where g=acceleration due to gravity (m s-2), θ’v is the virtual potential temperature of unmixed air lifted along a saturated adiabatic lapse rate from the parcel’s level of free convection (LFC) to its equilibrium level (EL), and θ v is the virtual potential temperature of the environment. The LFC is defined as the level to which a parcel must be forcibly lifted, by some lifting mechanism, so that it rises freely (Petty 2008, Markowski and Richardson 2010). Wind shear is a second parameter that is important when evaluating the likelihood for severe weather. Wind shear is a measure of both the change in a wind’s direction and magnitude with increasing height. Shear plays two primary roles in the formation and maintenance in severe weather. The first is the tilting of storm updrafts into a storm which allows for the updraft to be located adjacent and not within the downdraft. This in turn allows the updraft to be more efficient in supplying the storm with additional energy. Secondly, wind shear helps in the formation of supercells and more specifically mesocyclones. Kerr and Darkow (1996) and Markowski et al. (1998) showed that high curvature of hodographs at low levels is a good predictor of supercell potential. When an updraft begins to rotate due to the presence of wind shear pressure perturbations develop. These pressure perturbations can then cause the updraft to be more steady state by creating an upwardly directed pressure gradient force allowing a storm to strengthen or remain strong (Rotunno and Klemp 1982, 1985).

6

Storm-relative helicity (SRH), a measure of wind shear, was calculated in this study to demonstrate the potential interaction the boundary had on the supercells. Helicity itself is “a measure of the degree to which the direction of a fluid motion is aligned with the vorticity of the fluid” (Davies-Jones 1984). In this sense helicity is highly correlated to streamwise vorticity which is shown in Figure 2.1.

Figure 2.1: Tilting of horizontal vortex lines relative to storm motion. Image from Markowski and Richardson 2010, adapted from Davies-Jones 1984. Streamwise vorticity is found when the storm motion vector (c) is subtracted from wind velocity vector (v) and the resultant vector is aligned with the horizontal vorticity (ωh). Markowski and Richardson (2010) showed that SRH can be calculated using the equation:  = ∑   −  −   −  −  −  

[2]

where the summation is evaluated from N-1 and n=1. N is the number of measurement levels available, the wind vector v=iu+jv and c=icx +jcy . Like wind shear, SRH has been used to evaluate the probability of supercell formation. Storm environments yielding 7

values greater than 100 m2 s-2 of 0-3 km SRH are typically considered favorable for the development of supercells. In supercell outbreaks it is common to see 0-3 km SRH values greater than 400 m2 s-2. Finally, 0-1 km SRH is often used to help discriminate between environments likely to produce tornadic or nontornadic supercells (Bluestein and Marx 1987, Brooks et al. 1993, Brooks et al. 1994, Markowski and Richardson 2010). 2.1.2 Supercell Structure A supercell storm is made of three principle parts that include: a forward flank downdraft (FFD), rear-flank downdraft (RFD), and a quasi-steady rotating updraft Figure 2.2.About 25% of these storms produce tornadoes. The tornado is formed at the far southwest portion of the storm in the vicinity of the mesocyclone and RFD. Outflow boundaries are produced by the FFD and RFD, and are referred to as the forward flank gustfront and rear flank gustfront respectively. Upper-level winds are generally oriented from west to southwest. This causes the hydrometeors formed within the storm to be advected east to northeast from the updraft. These hydrometeors then evaporate which produces latent cooling of the air surrounding them and results in the air becoming negatively buoyant and descending. This process is what forms the FFD. Often smaller hydrometeors are advected further from the updraft than are larger ones. As radar reflectivity factor is proportional to D6, where D is the diameter of the drop, there will be a strong gradient in reflectivity that extends from the updraft outward (Kumjian and Rhyzkov 2008, 2012, Crowe et al. 2010). The RFD is formed when dry mid- and upperlevel winds impinge upon the backside of an updraft (Markowski 2002a, Grzych et al. 2006, Markowski and Richardson 2010). This leads to latent cooling due to evaporation and melting which produces negative buoyancy. It has also been thought that downward

8

pressure perturbation forces exist that aid in the formation of the RFD (Markowski and Richardson 2010). The updraft is formed when winds (usually southerly) south of the storm converge upon the RFD. The RFD, cooler and denser, serves as a lifting mechanism for the southerly winds that are then brought into the storm. The convergence of the RFD and these inflow winds often lead to a trailing line on the southwest flank of the storm that can be seen on radar as a thin line of showers. This feature is referred to as a flanking line (Markowski 2002a, Markowski 2002b, Markowski and Richardson 2010).

Figure 2.2: Conceptual structure of a supercell (Lemon and Doswell 1979) solid lines indicate storm echo. Solid lines with arrows represent storm relative streamlines. The storm’s forward-flank, rear-flank downdrafts are represented as well as the updraft location. 2.2 Quasi-Linear Convective Systems and Bow Echoes A Quasi-Linear Convective System (QLCS), sometimes referred to as a squall line, is defined as a convective system that is composed of a line of storms that has a characteristically large length to width ratio (Maddox 1980, Bluestein and Marx 1987, 9

Weisman et al. 1988). These storms are often associated with strong winds and sometimes large hail. They are often formed in areas of high CAPE and wind shear. These storms form in similar environments as supercells. However, these storms do not require as much shear as a supercell and are normally more strongly forced by an approaching boundary such as a dry line or cold front (Tripoli and Cotton 1980, Schlesinger 1982). These storms typically occur in environments that have high CAPE and shear although they are capable of forming in environments where meager CAPE and shear is available (Rotunno et al. 1988). The biggest delineator between a QLCS and supercell environment is the orientation of the deep layer shear across the main lifting mechanism. In events where the deep shear is oriented along the forcing mechanism, upscale growth to a QLCS is preferred. When increasing the angle of the shear across the lifting mechanism, the formation of discrete cells becomes favored (Chisholm and Renick 1972, Marwitz 1972 a, b, c, Weisman and Klemp 1982, 1988). On occasion the center of these linear storms will begin to bulge and form a bowlike segment. These features are known as bow echoes (Fujita 1978, Johns et al. 1993, Weisman 1993). Winds found within this bowing segment are typically stronger than the winds found in other parts of the QLCS, thus making bow echoes more dangerous. Bow echoes are a result of the center portion of the storm accelerating ahead of the main line due to the formation of a rear-inflow jet (Weisman 1992, Weisman 1993, Fujita 1978). QLCS storms can further be sub-categorized into trailing stratiform, leading stratiform, and parallel stratiform systems (Parker and Johnson 2000). These archetypes are dictated by the wind shear orientation and variation with height. Trailing stratiform squall lines typically have upper to mid-level shear oriented back to front across the

10

major axis of the system. Leading stratiform squall lines occur when the shear is oriented from rear to front. Finally, parallel stratiform occurs when the deep layer shear oriented along the minor axis of the system (Parker and Johnson 2000). Trapp and Weisman (2003) and Weisman and Trapp (2003) showed when a QLCS enters an environment with high unidirectional shear (20 m s-1 or greater over the 0-6 km layer) and sufficient CAPE (≈2000 J kg-1) the formation of a low-level mesocyclone is possible. These conditions also lead to the formation of bowing segments within the QLCS. These studies indicate that the preferred location for the development of mesovortices to form is on the north side of the bow. This compares well to the midday QLCS that produced seven tornadoes across north Alabama. However, further analysis of this process is outside the scope of this study. Much like a supercell, QLCS’s can be divided in sub-parts that make up the whole system. A trailing stratiform QLCS is typically made up of four parts that are depicted in Figure 2.3. These parts are pre-squall low, convective region, mesohigh, stratiform region, and wake low. The pre-squall low is located in front of the system. The mesohigh is collocated with the main precipitation core within the QLCS. Past the mesohigh lies the stratiform rain area that is also a transition from the higher pressure found in the mesohigh to the lower pressure found in the wake low.

11

Figure 2.3: Conceptual model of a QLCS denoting the locations of the pre-squall low, mesohigh, straitiform regions and wake low (Markowski and Richardson 2010, adapted from Johnson and Hamilton 1988). 2.3 Thermal Boundaries Roles in Supercell and Tornado Outbreaks Prior to 27 April 2011, there were three major tornado outbreaks that have drawn significant attention in the meteorological community. These events include the 3-4 April 1974 Super Outbreak, the 1994 Palm Sunday Outbreak on 27 March 1994, and the Super Tuesday outbreak on 6 February 2008.. Interestingly, the Palm Sunday outbreak was the only case of these events in which the presence of a thermal boundary was studied. The formation of a thermal boundary can be attributed to several different processes. For example, Businger et al. (1991) and Keeter et al. (1995) have shown that boundaries may be formed by the trapping of cool air by topography by. The most common method for forming a boundary is differential precipitation. As precipitation falls through the atmosphere, the individual raindrops are partially evaporated which

12

produces latent cooling of the atmosphere. The cooler air produced from this process is then pulled to the surface by drag from other falling hydrometeors. The collection of this cold air at the surface is referred to as a cold pool. A by-product of downdrafts is the outflow. Outflows are created when a downdraft reaches the surface and spreads outwardly from the point it impacts (Markowski and Richardson 2010). This outflow may combine with other outflows from other convective showers and result in a larger outflow that covers the area of the two individual outflows. A cold pool is a region of relatively cold air surrounded by warmer air. Cold pools can persist from minutes to over a day. The duration that a cold pool persists is most dependent on the amount of solar insolation received over the area that the cold pool covers. If cloudy conditions persist over the cold pool then the cold pool will tend to last longer. Cold pools can become a focusing mechanism for convective initiation. Often the presence of a cold pool is marked not only by temperature variations across the boundary but by changes in dew point, wind speed, wind direction, and pressure. In many cases the presence of these outflows can be seen on radars and visible satellites as a fine line. Baroclinically generated horizontal vorticity (BGHV) is thought to be an important feature in the process of tornadogenesis (Markowski and Richardson 2010). Although tilting of horizontal vorticity associated with vertical wind shear is thought to be the main source of rotation within a supercell, the addition of baroclinically generated horizontal vorticity can significantly increase the amount of horizontal vorticity. Most importantly, BGHV is internally driven by the supercell, meaning a supercell creates its

13

own vorticity by processes within the FFD (Davies-Jones and Brooks 1993, Davies-Jones et al. 2001, Klemp 1987). This process is demonstrated in Figure 2.4.

Figure 2.4: Depiction of the production of baroclinically generated horizontal vorticity (Markowski and Richardson 2010 adapted from Klemp 1987). Solid black lines represent the horizontal vortex lines. The white arrows represent parcel trajectories into the storm. In this case study, the horizontal temperature gradient is enhanced further due to the cooler air in place from the midday QLCS. Therefore the production of BGHV would theoretically be greater than in cases where only the storm is able to produce this temperature gradient. This effect is compounded when a storm is able to draw upon warmer air south of the boundary. This will be discussed in more detail in a later section. Likewise, if the warm and cool sides of the boundary were reversed, the production of BGHV would likely be mitigated; however, that process is beyond the scope of this study.

14

2.4 Previous Work with Thermal Boundaries and Tornadic Storms It has long been understood that the presence of a thermal boundary or outflow can increase the risk for supercells and tornadoes. Studies such as Magor (1959), Fujita (1963), Doswell (1982), Sanders and Doswell (1995) have mentioned the presence of a bubble high from outflows and their ability to alter the mesoscale environment. Furthermore, studies have shown the interactions of various outflow boundaries with supercells such as Purdom (1976), Maddox et al. (1980) hereafter known as MAD80, Davies et al. (1994), Markowski et al. (1998) hereafter Mark98, Wakimoto and Atkins (1996),Wakimoto et al. (1998), Rasmussen et al. (2000) hereafter Ras2000 and King et al. (2003). Studies by Korotky (1990), Businger et al. (1991), Vescio et al. (1993), and Langmaid et al. (1997) hereafter Lang97 emphasize the importance of shallow boundary features in focusing severe convection in the Southeast. Moreover, the Mad80, Mark98, and Ras2000 papers exclusively details supercell interaction with thermal boundaries. Their results show that storms intensified and had a higher likelihood of becoming tornadic while in close proximity to a boundary. In fact, Mad90 stated “It has long been recognized that the region where a line of thunderstorms intersects such a baroclinic zone is a favored location for the occurrence of severe tornadic storms”. Blanchard (2008) hereafter B08, detailed such an event from the Verification of the Origins of Rotation in Tornadoes Experiment (VORTEX). During this project, a dense network of mobile mesonets was available to measure the near environments of tornadic storms. Specifically, B08 discussed a storm that formed just on the cool side of a stationary front. Using the available mesonets, the interaction between supercell and boundary was measured with high temporal and spatial resolution. The storm featured in

15

the study was moving away from the boundary and had inflow vigorous enough to briefly draw the boundary and warmer air south of the boundary into the storm inflow region. B08 noted that this occurrence was much like the studies of Weaver and Nelson (1982) and Dostalek et al. (2004). Although the storm never produced a tornado, it remained severe and near the boundary for its two-hour lifetime. Mark98 detailed the significance of boundaries in the VORTEX study. Markowski stated that approximately 70% of the tornadoes observed during the project were associated with the presence of a thermal boundary. These results are shown in Figure 2.5. Mark98 also hypothesized that storms moving more parallel to the boundary than across it are more likely to be maintained for longer periods due to their ability to process greater amounts of unstable air. In addition Mark98 postulated that since boundary placement is on the warm side of the gradient tornadogenesis is preferred on the cool side of boundaries due to the additional generation of baroclinic horizontal vorticity. This is demonstrated in Figure 2.6. Finally, Figure 2.5 shows that storms located between 30 km into the cold side of the boundary and less than 10 km into the warm side of the boundary were more likely to produce a tornado during VORTEX. This finding was attributed to the baroclinically generated horizontal vorticity.

16

Figure 2.5: Bar graphs presenting the frequency of tornadogensis in the presence of a boundary during VORTEX (top) and tornado occurrence relative to distance from boundary in VORTEX (bottom) (Markowski et al. 1998). A negative distance in the bottom figure denotes a storm is located on the warm side of the boundary.

17

Figure 2.6: Figure from Markowski et al. 1998 demonstrating potential storm locations and interactions relative to horizontal vorticity generated by the boundary. Lang97 described the impact a thermal boundary had on the 1994 Palm Sunday Outbreak. That outbreak produced 30 tornadoes that stretched from Alabama to the Carolinas. While the outbreak covered a length of over 570 km, it was limited to a width of only 50 km. This 50 km width was controlled by the presence of a quasi-stationary thermal boundary that had formed as a result of heavy precipitation that fell across the southeast on the previous night; however, this case was limited in the density of surface observations available which inhibited the authors from pinpointing the boundary’s location. 18

Mad80 used multiple cases of boundary interactions with supercells in models to determine why boundaries are a preferred location for supercells and tornadoes. These cases include storms interacting with both cool outflow boundaries from previous convection or warm fronts. The scopes of these events range from one supercell interacting with a boundary to multiple storms interacting with a boundary. Mad80 determined that within the baroclinic region of the boundary winds veered more rapidly than those outside of the baroclinic region with a minimal veering of winds occurring on the cool side of the boundary as depicted in Figure 2.7. Mad80 concluded that environments with less than optimal macroscale conditions for supercell formation can be enhanced by a boundary such that miniature outbreaks may occur along these boundaries.

Figure 2.7: Coneptual model of a boundary and associatied wind profiles. Wind profiles show easterly winds at lower levels on the cool side of the boundary (C) with more unidirectional flow on the warm side (B) (Maddox et al. 1980).

19

Ras2000 described an event that occurred on 2 June 1995 near Amarillo Texas. In the early hours of the event, strong thunderstorms produced an outflow boundary in the area. Satellite imagery was able to show the location of the boundary as a fine line of cumulus clouds. This boundary was once again well sampled by the VORTEX fleet of mobile mesonets. These mesonets noted more easterly low-level winds as well as falling temperatures upon crossing the boundary. These easterly winds produce higher values of vertical vorticity as well as increased convergence along the boundary. Soundings were launched from 50 km north, 15 km north, and 15 km south of the boundary during this event. The sounding launched 15 km north of the boundary had the largest 0-2 km shear and greatest SRH. This can be attributed to the easterly low-level winds associated with the cool side of the boundary. The three storms that crossed from the warm to cool side of the boundary produced tornadoes. All three storms exhibited intense rotation within 30 minutes of crossing the boundary and maintained this rotation while remaining within 60 km of the boundary. The intensity of these mesocyclones as a function of distance from the boundary location is demonstrated in Figure 2.8.

20

Figure 2.8: Chart showing measured rotational velocity as a function of boundary position (Rasmussen et al. 2000). Relative maximum occurs between 0 and 50 m s-1. The number of mid-level mesocyclones during the event was found to be split evenly across the boundary; however, of the low-level cyclones formed that day nearly 70% were found to be located on the cool side of the boundary. Rasmussen noted that after crossing the boundary storms had a more easterly direction and attributed this to increased rotation nearer to the surface.

21

CHAPTER 3

DATA AND METHODOLOGY

Data collected for this event includes a wide variety of sources in the form of radars, surface observations, profiling instruments, and upper-air rawinsondes. 3.1 Radars Five radars were used in this study. Three of these were National Weather Service WSR-88D radars located in Hytop, Alabama (KHTX), Columbus, Mississippi (KGWX), and Birmingham, Alabama (KBMX). The other two radars, which were dual polarization, include University of Alabama in Huntsville Advanced Radar for Meteorological and Operational Research (ARMOR) and the Mobile Alabama Xband (MAX). The 88D level II data was downloaded from NOAA’s NCDC website. The level II data includes reflectivity factor, spectral width, and radial velocity. Typically for rapidly evolving storm situations the 88Ds operate in the VCP 12 and VCP 212 modes. These modes are preferred because of their ability to give greater vertical resolution as well as good temporal resolution. Both scan strategies include the following scan elevations (0.5, 0.9, 1.3, 1.8, 2.4, 3.1, 4.0, 5.1, 6.4, 8.0, 10.0, 15.6 and 19.5 degrees. The event occurred after the super resolution upgrade was added to the radars, however it was before the dual-polarimetric upgrade had been completed.

ARMOR was operating in a sector and surveillance scan mode until 2040 UTC. These modes were comprised of a 3-tilt surveillance and a 90 degree sector scan. The sector scans focused on the Cullman EF-4 producing storm and included elevation angles of 0.7, 1.3, 2.0, 3.1, 4.2, 5.2, 6.2, 7.2, 8.2, 9.1, 11.0, 12.5, 14.0, 16.0, 18.4, 21.0 and 25.0 degrees. These sectors varied in azimuth ranges as the storm moved northeast from Cullman into northeast Alabama. After 2040 UTC, ARMOR operated exclusively in a 3-tilt surveillance scan mode with elevation angles of 0.7°, 1.3°, and 2.0°. Once the 3-tilt scan was initialized, the radar remained in this scan sequence until power failure ended data collection at 2231 UTC. NOAA’s method of classifying supercells using radars was used for delineating between supercell and nonsupercell storms. This method states that a supercell must show signs of mesocyclone development. The threshold for this is demonstrated by the formula given in Section 2.1. The vertical extent of this region must be greater than 2.5 km in depth and must persist for at least two volume scans (NOAA 2010). In addition to the method stated above, the more common definition of a supercell was also used. This definition states that a deep mesocyclone extending at least half the depth of the storm must be present and persist for greater than 20 minutes or for the time it takes one parcel of air to pass through the entire updraft.

23

3.2 Radar Processing 3.2.1 Editing The unedited raw data that was downloaded from NCDC had to be corrected for both range and velocity folding. NCAR’s SOLO II and SOLO 3 software packages were used to remove the range folding and dealias the areas of velocity folding. Range folding occurs when the radar receives backscatter beyond its maximum unambiguous range (Rmax) defined by  =

2 ∗ 

where c is equal to the speed of light and PRF is the radar’s pulse repetition frequency. Higher PRF values yield lower values of Rmax. When the radar receives backscatter from targets outside of this range, the backscatter is received after the subsequent radar pulse. This causes the radar to place the target in the corresponding position for the subsequent pulse. Velocity folding occurs when a radar’s Nyquist velocity is less than the measured radial velocity. A radar’s Nyquist velocity (Vmax) is given as

 

∗! .

"

Velocity folding often occurs in severe storm environments. When using lower wavelength radars such as C- or X-band the Nyquist velocity is particularly low. Often a radar’s PRF will be adjusted to optimize data results depending on the type of research being done. For this case ARMOR’s wavelength is 5.3 cm with a PRF of 1200 s-1 yielding a Nyquist velocity of approximately 15 m s-1 while the MAX radar wavelength is 3.15 cm and had a PRF of 1000 s-1 yielding a Nyquist velocity of 7.9

24

m s-1. The WSR 88D radars use a variable PRF at different elevations. Specifically, lower PRFs are used at lower elevations to maximize Rmax. At higher elevation scans, higher values of PRF are used to maximize the value of Vmax. In general, the Vmax is approximately 30 m s-1. When a radar’s Vmax is exceeded the value returned is then said to be “folded”. This means the returned value is equal to V-Vmax. Rotational velocity (Vrot) of a storm was computed using two methods. First the storm motion was set within GR2Analyst software so storm-relative velocity (SRV) was available to be displayed. GR2Analyst is a program designed to display radar data. Storm motion was determined by placing a marker over a constant and identifiable portion of the storm (typically the tornado debris signature if present) and progressing through scans. Once a scan had been advanced GR2Analyst has a feature to compute storm-motion from the markers. This process was repeated four times with the resultant values averaged. Next, GR2Analyst’s vertical slice tool was used to view a vertical section of the mesocyclone. This tool stacks the PPI data from a given volume scan on top of itself so that a vertical view of the storm can be obtained. The tool allows the user to pan through the storm so that maximum inbound and outbound velocities can be found for each height. The program’s automated unfolding algorithm was utilized in this exercise. The values obtained were then converted to units of m s-1. When radial velocities exceeded the program’s ability to unfold the aliased data, SOLO 3 and SOLO II were used. Once the data was unfolded, stormmotion was removed to produce SRV. Using the software’s data widget, maximum inbound and outbound velocities in SRV were extracted for heights from 250 m to 5000 m at 250 m intervals. These SRV values were then averaged to produce Vrot.

25

Possible errors in Vrot are introduced for multiple reasons. First, areas with increased ground clutter or excessive debris can cause false return in the sample. A second source of error arises as a mesocyclones distance from the radar increases. The increase in the radar sample volume could lead to lower values or Vrot.being returned. 3.2.2 EVAD Analysis Extended Velocity-Azimuth Display (EVAD) analyses were produced for each radar during the time of interest using RadX software. The EVAD in the RadX software was designed using the methods explained in Matejka and Srivastava (1991) that expounded on Srivastava et al. (1986). This process includes making horizontal rings (VAD ring) for each range gate for each elevation. Each ring then becomes only a function of azimuth (Matejka and Srivastava 1990). After making each VAD ring, the resultant values are run through a Fourier series. This process yields a wind speed and direction for each corresponding height. Once the EVADS were completed, the resultant data were then checked for quality using the methods described in the Doppler wind profiler section below. Additionally, the RadX code was written so that several parameters could be varied by the user. This allows the user to yield results specific to each case. The EVAD input parameters for this study were as follows: •

5-km radius of influence



minimum first height of 20 m



data output at 20 m height intervals

A 0-1 km wind profile was constructed using the ARMOR EVAD analysis and data from the 915 MHZ profiler. This wind profile demonstrates the easterly flow

26

that was found to be characteristic on the cool side of the boundary. The 915 MHz profiler has a first gate of 196 m whereas the EVAD analysis data extends from 20 to 170 m. This limitation is due to the 5 km radius of influence imposed on the EVAD analysis combined with the 3-tilt mode in which the radar was operating. Finally, ten meter winds from the Huntsville ASOS were inserted into the wind profile at corresponding times. The transition from 10 m ASOS winds to the EVAD winds and finally the profiler winds are smooth throughout, giving higher confidence in both the performance of the EVAD analysis as well as the profiler. 3.3 Surface Data Surface data were the primary data used in this study. The first objective of the study was to acquire all available surface data. The northern halves of Alabama and Mississippi were the primary regions of interest for this study. Data were collected through numerous ground-based sources that were then uploaded to a common website operated by the University of Utah. Forty-eight unique locations were collected from this website. Original sources of these data include the Citizen Weather Observer Program (CWOP), Automated Surface Observing System (ASOS), National Weather Service (NWS), Remote Automated Weather Station (RAWS), and Federal Aeronautics Administration (FAA) groups. Data were acquired from UAH’s surface instrumentation as well as the M3V. Both sources provided rapid surface condition evolution. The temporal resolution of each site varied. Many sites reported with frequency greater than or equal to every 5 minutes, although some reported as infrequently as

27

every 60 minutes. In addition, many of these sites lost power for extended periods. These power failures eliminated an additional five sites from being used in the study. Values of θ and θe were calculated to help determine placement of the boundary at 30 minute intervals. Since not all stations reported at half hour frequencies, on the half hour, a tolerance of plus or minus 10 minutes was placed on each location. All sites listed in Table 3.1 included at least temperature and relative humidity readings. The locations for these stations are demonstrated in Figure 3.1. Locations that did not include an altimeter or pressure reading, a pressure reading was acquired by using nearest neighbor or by averaging methods. Nearest neighbor methods were used if a site was located within 10 km of another. Interpolation was used if a site was located between two other sites both within 30 km. Altimeter readings were converted to pressure once an altimeter reading was obtained for all sites the conversion from altimeter to MSL pressure was calculated using the hypsometric equation: # = #$% &' − Γ' )

28

*  + ,-



.

[3]

Table 3.1: City, State, Organization and report interval for surface observations. CITY Anniston

STATE AL

ORGANIZATION NWS

FREQUENCY 5 min

Bankhead

AL

FAA

1 hour

Bear Creek

AL

CWOP

10 min

Belgreen

AL

CWOP

10 min

Birmingham

AL

ASOS

1 min

Brindley Mountain

AL

CWOP

6 min

Columbus Air Force Base

MS

NWS

1 hour

Corinth

MS

CRN

10 min

Courtland

AL

CWOP

5 min

Crossville

AL

CRN

5 min

Cullman

AL

NWS

40 min

Fort Payne

AL

ASOS

1 min

Gadsden

AL

ASOS

1 min

Guntersville

AL

CWOP

5 min

Gurley

AL

CWOP

5 min

Hayden

AL

CWOP

5 min

Huntsville

AL

ASOS

5-min

Jasper

AL

CWOP

10 min

Madison County Airport

AL

FAA

20 min

Margaret

AL

CWOP

4 min

Millport

AL

CWOP

10 min

Monroe

MS

RAWS

1 hour

Moulton

AL

CWOP

20 min

Muscle Shoals

AL

CWOP

20 min

Noxubee

MS

RAWS

1 hour

Okolona

MS

Bureau of Land Management

5 min

Oxford

MS

ASOS

1 min

Russellville

AL

CWOP

10 min

Scottsboro

AL

CRN

5 min

Speake

AL

CWOP

20 min

Tishomingo

MS

RAWS

1 hour

Tombigbee

MS

RAWS

1 hour

Tupelo

MS

CWOP

15 min

Tuscaloosa

AL

CWOP

30 min

Valley Head

AL

CRN

5 min

UAH

AL

MIPS

5 min

Winborn

MS

RAWS

1 hour

29

Figure 3.1: Surface observations and organization type. Locations without a symbol denote other cities mentioned in this study. To calculate potential temperature (θ) the formula θ=T(1000hPa/p)(Rd/cp)

[4]

was used where Rd is the dry air constant (287 J/kg K) and cp is the specific heat of air (1005 J/Kg K). The equivalent potential temperature (Θe) was then calculated using the formula θe = θ*exp(Lw/cpT)

[5]

where L is the latent heat of vaporization, w is the water vapor mixing ratio, cp is the specific heat of water and T is in Kelvin. Furthermore, RH=e/es where e=es(Td) and

7.87∗9:

./ = 611.2 ∗ exp9: ;<=.> Finally, using the approximation q=εe/p=w where ε=0.622. 30

[6]

The placement of the boundary was the author’s subjective analysis using the available surface data. For this study, the boundary location has been defined as the warm side of the gradient (Markowski 1998). The boundary’s placement was determined at 30 minute intervals using the data from the available locations. The data were displayed in the GR2Analyst software by making placefiles for each location. A value for θ was placed at its corresponding location for each 30 minute interval. Sites that did not record an observation within a ±10 minute window for the specified time were skipped. Once all data were displayed an estimate of the approximate boundary’s position was determined by locating the location where the gradient in θ was the largest. Additional adjustments to the boundary’s position were made based upon each locations data set. For example, if the data for a particular location suggests the maximum in temperature occurred prior to the particular time being analyzed the boundary would be moved slightly to the north to correspond. Likewise if the data suggests the maximum in temperature was after the analyzed time the boundary would be moved slightly to the south. Finally, further adjustments were made to the boundary’s position for locations where data were sparse. A constant boundary motion was determined using data from the M3V. This forward motion was then used to adjust the boundary’s position using time to space conversion methods. With the placement of the boundary being subjective potential error in placement does exist. The author is confident that with the comprehensive set of surface observations available the boundary’s determined location is likely within ± 10 km of the actual position of the boundary at each time. Areas that contained more observations would constrain this further to ± 5 km.

31

3.4 MIPS DATA The Doppler wind profiler (DWP) is a vertically pointing radar operating at 915 Mhz. Data from this instrument include signal to noise ratio (SNR), spectral width (SW), and radial velocity (VR). SNR is a return power field expressed as a ratio. SW is a measure of dispersion of velocities within a given radar sample. VR is the component of motion of the target toward or away from the radar. The profiler works by alternating three beam orientations and then combining the measurements. These three beams are comprised of a vertically pointing beam as well as two orthogonal beams oriented at 66° elevation. Typically, the profiler cycles through a four-beam sequence to produce the associated wind field. This sequence alternates two vertically pointing beams with the two off zenith beams. The resultant 2-D wind field is then calculated using trigonometric methods where the vertical motions are taken directly from the vertically pointing beams. Raw data from UAH’s wind profiler were used in this study to produce a wind profile time series. The data were read into IDL where it was then corrected. Each wind value was compared to the others at its height for the prior five minutes and next five minutes. The standard deviation was then calculated for the 11 measurements. Values outside of the two standard deviation range were discarded and replaced by interpolation. In places where extended periods of data were dropped, interpolation was used from the adjacent measurements in the vertical directions. The microwave profiling radiometer (MPR) is a 12-channel passive measurement device capable of constructing vertical profiles of temperature, water vapor, and

32

liquid water content. There is greater vertical resolution at lower levels that provides a more accurate profile of the boundary layer. The MPR is able to provide a complete sounding every minute. These soundings can be viewed sequentially to see the vertical development of the boundary. Soundings produced while rain was present were discarded because of scattering and emission from raindrops. At 2051, UTC the UAH severe weather research team launched a balloon rawinsonde from the National Space Science Technology (NSSTC) building in Huntsville, Alabama. Both IMET software and sounding instrumentation were used to acquire the data from the rawinsonde. Surface conditions were taken locally from UAHuntsville’s surface observation to calibrate the sonde package prior to launch. The rawinsonde experienced an early termination at approximately 550 mb and the GPS system malfunctioned after launch. However, a thorough sampling within the cool air mass was taken. Sounding data from the Redstone Arsenal (RSA) launch at 1700 UTC that morning were used to supplement the data from the UAH rawinsonde so that a full temperature profile could be constructed. This profile was then used to estimate CAPE. Due to the GPS malfunction, no wind data were retrieved from the sounding.

33

CHAPTER 4

SYNOPTIC AND MESOSCALE OVERVIEW

The 27 April 2011 supercell outbreak was a well-forecasted event. The Storm Prediction Center (SPC) placed a large portion of the southeast in a moderate risk for severe weather as early as three days in advance. On the day of the event, a high risk was issued for north Alabama and portions of bordering states. Figure 4.1 shows the 1200 UTC convective outlook issued by the SPC. The SPC outlook was later updated to extend the size of the high-risk area further southwest. This high-risk area was deemed to have a 30 % chance of tornado occurrence within a 20-mile distance from a single point.

Figure 4.1: Storm Prediction Center’s 12 UTC Convective Outlook for April 27, 2011. A high risk was placed over northern Alabama bordering parts of Mississippi, Tennessee, and Georgia.

4.1 Synoptic Environment The SPC was able to place high confidence in their forecast due to the highly favorable synoptic conditions for severe weather that were forecasted to occur over Alabama and adjoining states. While the afternoon to evening supercell outbreak was well forecasted, the two events preceding the supercell outbreak were not anticipated. With the three events related by means of the boundary, it is important to first review the synoptic environment that promoted initiation of these three events. Figures 4.2 and 4.3 show the 1200 UTC 300 and 850 mb analysis while Figure 4.4 shows the 1200 UTC surface observations. Figures 4.5 and 4.6 are the 300 and 850 mb analysis at 0000 UTC on April 28, and Figure 4.6 displays the 0000 UTC surface observations. The synoptic environment on 27 April was one of the most conducive environments for large tornadoes documented in recent decades (Knupp et al. 2014). The 1200 UTC 300 mb chart in Figure 4.2 shows a deep slightly negatively tilted trough with its axis extending down into north Texas. The jet stream maximum winds were approximately 115 kts over Arkansas with significant divergence values located to the east over MS and northeast AL at this time. Figure 4.3 shows a closed circulation at 850 mb extending down into southern Texas with the deepest portion located in Wisconsin. Warm air advection was prominent over the eastern portion of the country with 50 kt winds from the south over Alabama. Additionally, the 1200 UTC surface chart shows a surface low located in western Arkansas with a cold front extending into eastern Texas and a warm front extending from the low into southern portions of Missouri, Illinois, Indiana, and Ohio.

35

Figure 4.2: 1200 UTC 300 mb observations showing a deep trough with large divergence located over Mississippi, Alabama, and Tennessee.

Figure 4.3: 1200 UTC 850 mb observations showing strong southerly low-level winds over the southeast.

36

Figure 4.4 shows the 1200 UTC surface observations. This images shows a surface-based low located over western Arkansas with a cold front located in eastern Texas. The cold front was located over central Arkansas as denoted by the onset of northerly winds. Southerly winds covered the southeastern region and the warm front was located over the Ohio Valley. The 28 April 0000 UTC 300 mb analysis shown in Figure 4.5 shows a more negatively tilted trough with a center axis located over Arkansas and Louisiana. Strong diffluence existed east of the trough axis over Alabama and contributed to the large divergence located over the Mississippi/Arkansas border. A strong 100 kt jet extended from Colorado into Mississippi. Figure 4.6 shows the 0000 UTC 850 mb analysis. This figure shows a sharp trough extended through Mississippi and a 30 m s-1 low-level jet over Mississippi and Alabama. This low-level jet supplied the region with moisture from the Gulf of Mexico and high values of low-level wind shear. The surface cold front had reached the Mississippi-Alabama state line and was connected to a surface low located over northern Kentucky at 0000 UTC as denoted by Figure 4.7.

37

Figure 4.4: 1200 UTC surface observations showing a warm front in extended through Arkansas into Missouri with a cold front in southwestern Arkansas. Southerly flow resides over much of the southeast.

Figure 4.5: 1200 UTC 300 mb observations displaying the negatively tilted trough axis over Arkansas Mississippi, and Alabama. Strong diffluence existed in the exit region of the Jetstream that enhanced divergence values.

38

Figure 4.6: 850 hPa analysis at 0000 UTC on 28 April demonstrating strong southerly flow east of the trough axis.

Figure 4.7: 0000 UTC surface pressure and wind at 0000 UTC on 28 April demonstrating the location of the cold front across Mississippi and Alabama.

39

4.2 Mesoscale Environment The early morning QLCS organized over southeast Missouri and north Louisiana and moved into Alabama at 0900 UTC. Surface observations showed southerly winds with temperatures from Muscle Shoals to Montgomery, Alabama ranging from 18.8 ºC to 23.3ºC. The SPC mesoanalysis indicated SBCAPE values ranged from 1500 J kg-1 at its southern edge in Butler, Alabama. SBCAPE is defined as the total amount of potential energy available to a parcel originating from the surface. SBCAPE values of 500 J kg-1 existed at its northern-most extent in extreme northeast Alabama. The 0-6 km and 0-1 km shear vector magnitudes were approximately 30-35 m s-1 and 25 m s-1respectively. The 0-1 km SRH values were about 750 m2 s-2 over a majority of this region. By 1300 UTC, the QLCS had exited Alabama and entered the state of Georgia. By this time, the temperatures in Muscle Shoals and Montgomery ranged from 16.6 ºC to 21.1 ºC respectively. The midday QLCS formed over north Mississippi near 1130 UTC and entered the state at approximately 1500 UTC. Temperatures in most locations had begun to rebound from the early morning QLCS (Figure 5.11) with the presence of southerly winds and the addition of solar heating. The SPC mesoanalysis indicated that SBCAPE values over the northern half of Alabama were approximately 500 J kg-1 with high values of convective inhibition (CIN) ranging from -150 to -400 J kg-1. However, MUCAPE values of 1000 J kg-1 were also present at this time, where MUCAPE is defined as the total amount of potential energy available to the most unstable parcel. The 0-1 shear vector magnitude was approximately 30 m s-1, and 0-1 km SRH values up to 900 m2 s-2 were diagnosed. The midday QLCS exited northeast Alabama by 1900 UTC when a temperature

40

difference of about 9°C existed between Florence, Alabama (16.6°C) and Birmingham (25.5°C). Following the passage of the midday QLCS, scattered convective showers and thunderstorms covered much of northern Alabama from Cullman to points north. Rainfall evaporation and cloud shading maintained cool surface temperatures over north Alabama. This is particularly true for the northwest parts of Alabama and northeast Mississippi where the convection had persisted longer. This led to cooler surface temperatures that yielded lower SBCAPE values (1000-2000 J kg-1) when compared to areas receiving less rain (greater than 3500 J kg-1). Although north Alabama had lower SBCAPE values, 0-1 SRH values were extremely high over this portion of the state. Values of SRH of greater than 1100 m2 s-2 were diagnosed in the SPC mesoanalysis. Despite the higher values of CIN, multiple supercells had formed by 2000 UTC. Figure 4.8 shows the SBCAPE, CIN and 0-1 km SRH at this time. As the afternoon progressed, the SPC mesoanalysis indicated that the storm environments on both the north and south sides of the boundary were extremely conducive for the development of strong long-lived supercells. The mesoanalysis indicated CAPE values south of the thermal boundary increased to approximately 4500 J kg-1 with limited CIN present. In addition, 0-6 km bulk shear values approached 30 m s-1 with values of 0-1 SRH having reached approximately 400 m2 s-2. On the north side of the boundary, 0-6 km shear was found to be even greater approaching values of 35-40 m s-1 with 0-1 km SRH having reached 1100 m2 s-2 in northeast Alabama. However, many of these values are likely inaccurate due to the inability of the SPC Mesoanalysis to correctly resolve the presence of the cool air mass and thermal boundary near the surface.

41

However, a sounding launched by UAH showed that MUCAPE values of 3500 J kg-1 were available about 35 km into the cool side of the boundary. Figure 4.9 demonstrates the boundary’s approximate position at 2000 UTC. This boundary denotes the separation of the conditions on the warm side and cool side of the boundary. These differences will be discuss further in the next chapter. The Cullman EF-4 supercell had formed and was producing damage at this time.

Figure 4.8: SPC Mesoanalysis graphics for SBCAPE (solid contours J kg-1) and SBCIN (dashed contours with shading, top figure) and 0-1 SRH (m2 s-2). Surface temperatures are not shown due to the presence of cold air over north Alabama which was no accurately diagnosed.

42

Figure 4.9: 0.7° PPI from ARMOR at 2000 UTC. White dashed line denotes approximate boundary position at this time.

43

CHAPTER 5

BOUNDARY DEVELOPMENT, PROGRESSION, AND CHARACTERISTICS

5.1 Boundary Development The midday QLCS entered northwest Alabama at 1330 UTC. Initially the leading edge of this line was oriented west to east. Shortly after entering the state a portion of the line in northern Mississippi developed a bow shape that caused the orientation of the line to become more north south. This change in orientation is demonstrated in Figures 5.1 and 5.2. While this change in orientation occurred large amounts of precipitation continued to fall over the northwest portion of Alabama. This precipitation was accompanied by cool outflow as well. This led to the northwest portion of Alabama as well as northeast Mississippi receiving significant cooling. Figure 5.3 shows the QLCS had advanced into east Alabama by 1800 UTC. Scattered precipitation covered much of northern Alabama to the west of this QLCS. This precipitation produced further cooling over the northern portion of Alabama. Furthermore, the associated cloud cover prevented solar insolation from warming this portion of Alabama, thereby maintaining the thermal contrast across the boundary.

Finally Figure 5.4 illustrates that by 2000 UTC the QLCS had exited the state of Alabama, yet the scattered precipitation was still present over north central Alabama. Portions of northwest Alabama had been receiving precipitation and therefore cool outflow for the 6 hour period whereas areas in east and central Alabama received little to no precipitation. Figure 5.4 also demonstrates that the scattered convection had become slightly more intense at that time and the Cullman supercell had already formed. The variability of precipitation is demonstrated in individual surface stations times series shown in Figures 5.5-5.10, which also depict the variation of surface temperature associated with this precipitation. Figure 5.5 illustrates the surface conditions from Bear Creek, Alabama located approximately 5 km southeast of the midday QLCS. The station recorded no precipitation from the passage of the QLCS nor did it record a significant drop in temperature. This is important as the Hackleburg supercell crossed just north of this location which also marks the southernmost intrusion of the cool air produced by the midday QLCS. When comparing this location to that of Muscle Shoals, Alabama (Figure 5.6) there was a notable difference in both the temperature and precipitation totals. Muscle Shoals, which is located in northwest Alabama received approximately 25 mm of precipitation from the midday QLCS and the subsequent lingering scattered precipitation. A noticeable drop in temperature from 19 to 15°C occurred during this time. The observation at 2153 shows the boundary had passed this location since temperatures had increased to 23°C whereas the previous measurement was 18°C at 2115 UTC. Figure 5.7 illustrates the observations from Russellville, Alabama located between the Bear Creek and Muscle Shoals surface stations. Russellville received

45

approximately30 mm of precipitation that coincided with a drop in surface temperature from 18 to 14°C as the QLCS passed the area. The temperature remained relatively constant until approximately 2000 UTC when the baroclinic zone of the boundary passed as temperatures increased from 17 to 19°C. A maximum temperature of 22°C was reached at approximately 2225 UTC. This maximum temperature marks the boundary passage at this location. Figure 5.8 demonstrates the surface observations from Hayden, Alabama located 50 km south of the southernmost portion of the QLCS. Therefore Hayden received no precipitation or cooling from the midday QLCS allowing the temperature to increase throughout the afternoon. Furthermore, southerly to southwesterly winds persisted at that location during the afternoon. Figure 5.9 shows the data from UAH, which received approximately 30 mm of precipitation from the midday QLCS with associated cooling from 23 to 18°C. The onset of modest warming occurred after 2100 UTC when temperatures increased from 1820°C. A complete profile of the boundary passage was not attained due to power failure at this location prior to boundary passage. Finally, Figure 5.10 depicts the observations from KHSV at the Huntsville International Airport. The drop in temperature was offset from the precipitation at this location due to the orientation of the QLCS. However, a sharp decrease in temperature from 24-18°C was measured. Once again, the onset of warming associated with the passage of the baroclinic region occurred at approximately 2100 UTC, when the temperatures increased from 18 to 24°C. Additionally, the shift in wind occurred at this

46

location after the passage of the midday QLCS when winds became more easterly. The winds shifted to southerly at approximately 2140 UTC. Figure 5.11 shows a composite temperature time series from Anniston (KANT), Muscle Shoals (KMSL), Montgomery (KGMG), Tuscaloosa (KTCL), and Huntsville (KHSV), Alabama. This plot depicts the development of the boundary with the passage of the early-morning and midday QLCSs. A drop in temperature was recorded at all locations except at KMGM with the passage of this QLCS. The recovery in temperature occurs at different times for each location. Interestingly, the temperature at KANT recovered the most quickly and exceeded the temperature at KMGM by 1200 UTC. By 1500 UTC the temperatures at KHSV and KTCL had begun to increase. At that time the temperatures ranged from 30°C at KMGM to 15°C at KMSL. The midday QLCS stopped the increase in temperature at KHSV after1500 UTC. Furthermore, this QLCS did not impact KTCL so temperatures continued to increase until reaching values around 27°C which was near those of KMGM and KANT. By 2100 UTC both KHSV and KMSL had begun to warm with the passage of the baroclinic zone. The temperatures at these locations continued to increase until the boundary passage. At that time the temperature range of these locations was approximately 22°C in Muscle Shoals to 28°C in Montgomery.

47

Figure 5.1: 0.5° PPI scan from KGWX at 1400 UTC. Figure shows the eastern portion of the QLCS over northern Mississippi and Alabama.

Figure 5.2: 0.7 ° PPI scan from KGWX at 1600 UTC showing the fully developed QLCS over northern Alabama.

48

Figure 5.3: 0.7° PPI scan from ARMOR at 1800 UTC showing scattered convective showers over north Alabama. At 1800 UTC the QLCS was in eastern Alabama.

Figure 5.4: 0 .5° PPI scan from KHTX at 2000 UTC.

49

Figure 5.5: Bear Creek, Alabama time series of temperature (red, °C), dew point (green, °C), surface pressure (black, pressure-1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. The location of Bear Creek is shown in Figure 3.1.

Figure 5.6: Muscle Shoals, Alabama time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. The temperature in Muscle Shoals remained cool and nearly saturated through the evening hours. The boundary passed this location just prior to 2300 UTC. The location of Muscle Shoals is shown in Figure 3.1.

50

Figure 5.7: Russellville, Alabama time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. Dew point was not shown due to bad data. The location of Russellville is shown in Figure 3.1.

Figure 5.8: Hayden, Alabama time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. The location of Hayden is shown in Figure 3.1.

51

Figure 5.9: UAH time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. Power failure occurs at 2215 UTC. This is likely prior to passage of the boundary. The location of UAH is shown in Figure 3.1.

Figure 5.10: Huntsville ASOS time series of temperature (red, °C), dew point (green, °C), pressure (black, -1000 mb), and total rainfall after midnight (blue, mm). Left axis values represent the temperature dew point and pressure, while the right axis defines rainfall. The location of KHSV is shown in Figure 3.1.

52

Figure 5.11: Temperature time series from Huntsville (KHSV) (purple), Tuscaloosa (KTCL) (green), Muscle Shoals (KMSL) (black), Anniston (KANT) (red), and Montgomery (KMGM) (blue) AL. Image demonstrates the passage of the early morning (dashed lines 1 and 2) and midday QLCSs (dashed lines 3 and 4) as well as the boundary passage in Huntsville and Muscle Shoals (dashed lines 5 and 6).

53

5.2 Boundary Progression By 1800 UTC the thermal contrast shown in Figure 5.11 had formed a thermal boundary oriented southwest to northeast south of the Brindley Mountain surface station as shown in Figure 5.12. The boundary moved as far south as Garden City, Alabama or approximately 34.0º latitude by 1900 UTC as shown in Figure 5.13. Horizontal variations in potential temperature (θ) of 7-11 K were typical across the baroclinic zone at this time. Upon reaching approximately 34.0° latitude the boundary retrograded towards the north at approximately 6.6 m s-1. This movement was non-uniform along the boundary axis due to the interaction with various supercells and likely the presence vertical mixing. Surface observations show wind speeds south of the boundary of 2.6-5.2 m s-1. The progression of the boundary was determined using data collected from the M3V as well as tracing the boundary’s movement in surface time series (e.g., Figures 5.5-5.10). Initially the northward progression was forced by southerly winds located south of the boundary. Since the northerly movement of the boundary exceeded the value of the surface winds, it is inferred that the northerly progression was aided by vertical mixing and boundary layer inflows to the supercells located to the boundary’s north. The nonuniform northerly progression of the boundary is depicted in Figures 5.14-5.16. Figure 5.14 shows the boundary at 2000 UTC, after it had begun to propagate toward the north. The Cullman supercell is currently intersecting the boundary at this time. The boundary remained intense at this time with differences in θ of 10 K over an approximate 30-50 km distance. The boundary remained more intense in west Alabama at this time. The difference in temperature was approximately the same, however the width of the baroclinic region was much less.

54

Figure 5.15 shows the boundary has continued propagating toward the north at 2100 UTC. At that time three supercells were currently located within 30 km of the boundary. The boundary had weakened slightly in intensity with differences in θ approximately 7 K over an approximately 30-50 km distance. Finally, Figure 5.16 shows the boundary at 2200 UTC. By this time the boundary had reached the Huntsville area. At that time the Limestone County and Jackson County supercells were interacting with the boundary. The boundary had continued to weaken with differences in θ closer to 4 K over the same 30-50 km distance.

55

Figure 5.12: 0.5° ARMOR PPI with θ values for 1800 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location.

Figure 5.13: 0.5° ARMOR PPI with θ values for 1900 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location.

56

Figure 5.14: 0.5° ARMOR PPI with θ values for 2000 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location.

Figure 5.15: 0.5° ARMOR PPI with θ values for 2100 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location.

57

Figure 5.16: 0.5° ARMOR PPI with θ values for 2200 UTC. White dashed line represents the author’s subjective analysis of the boundary’s location.

58

The M3V made two complete north-south transects through the boundary along I65 as shown in Figures 5.17 and 5.18. The M3V then parked at the I-65/I-565 junction and northern part of the baroclinic region pass over. These figures depict the M3V's measurements and location of the measurements. The first transect occurred while traveling south on I-65 during the 1959-2013 UTC period. The M3V started on the cool side of the boundary at the University of Alabama in Huntsville (point A). At 1959 UTC the initial increase in temperature from 18.4 to 18.5°C was noted in the data. This occurred at point B in Figures 5.17 and 5.18. The M3V continued traveling due south for another 24.5 km before reaching a temperature of 24.5°C at 2013 UTC at point C in Cullman, Alabama. This maximum in temperature along the gradient is by definition the boundary location (Markowski et al. 1995). The 24.5 km distance of increasing temperature was the baroclinic zone. Upon crossing the boundary the M3V turned around in the city of Cullman. While in Cullman, the M3V measured an additional 3°C increase in temperature. This increase can likely be attributed to urban heat island effects. This is supported by the decrease in temperature at 2030 as the M3V left Cullman traveling north on I-65. This is denoted by point D. The M3V crossed back into the cool portion of the boundary at 2043 UTC denoted by point E, where the temperature was measured at 24.6°C. After crossing the boundary and moving through the baroclinic zone, the M3V parked at the I-65/I-565 intersection denoted by point F and sampled the northern portion of the baroclinic region between 2116 and 2132 UTC. Over this period the temperature increased from 18.5°C to 21.1°C. This parking spot was 24.3 km due north of the point where M3V initially measured a temperature increase on the initial southward through transect through the boundary. While static the M3V data shows the increase in

59

temperature beginning at 2116 UTC. Therefore, the leading edge of the baroclinic zone traveled the 24.3 km over the 1:17 minute time frame. From this information it can be deduced that the northward motion of the boundary was 5.27 m s-1. If this forward motion is assumed to be constant (boundary motion likely varied slightly due to the interactions with various supercells and vertical mixing) the north-south cross section of the baroclinic zone would be approximately 20 km wide at this longitude. The rate of temperature change is much less at this time than the previous transects due to the car being static. Additionally, the M3V did not measure the boundary passage. A maximum temperature of 21.9°C was reached at 2212 UTC whereas the maximum temperature at KHSV was measured at 25°C at 2220 UTC. It is possible that some heating was offset by the presence of the FFD from the Hackleburg supercell being located less than 5 km to the north as indicated by ARMOR data. M3V data further supports the boundary’s progression to the north being aided by storm inflows. While the M3V was stationary near the I-65/I-565 junction the Hackleburg tornado passed nine km to the north at 2125 UTC. Despite the tornado being 9 km north of the M3V, the data shows sustained inflow winds of greater than 11 m s-1 leading up to the passage of the storm. A maximum temperature of 21.1°C was achieved at 2132 UTC. A 5-10 m s-1 wind was measured between the period of 2125 UTC and 2132 UTC. This wind was the boundary layer inflow to the Hackleburg storm. This implies that the storm was able to drawn the warmer air located near the boundary. It is this way the storms aided the boundary’s progression to the north. Also, the drawing of this warmer to the north would lead to an isolated portion of the boundary being moved along with it such as Blanchard (2008) presented.

60

Figure 5.17: Time series from M3V depicting measured temperature, dewpoint, wind speed, and car speed. Temperature and dewpoint are displayed on the left axis while wind speed and car speed are displayed on the right.

Figure 5.18: Path displaying path M3V traveled while transecting the boundary. 61

The contribution to forward progression by way of mixing was not directly measured, but inferred from the northern movement of the boundary having exceeded boundary layer flow south of the boundary. The relative contributions of vertical mixing, storm inflow, and southerly winds likely varied from location to location and were dependent on turbulence, vertical and horizontal wind shear, storm intensity, and proximity to the boundary. By 2200 UTC the boundary had moved near the Huntsville area. At this time storms covered most of north Alabama and solar heating was reduced south of the boundary. For these reasons the boundary’s magnitude had lessened. For the duration of the event the portion of the boundary west of Huntsville exhibited a greater north-south gradient than that on the east side. Values of Δθ west of Huntsville were typically 7 K over a 30 km distance, compared to typical differences of 4 K over the same distance in the eastern half of Alabama. At its peak intensity at 2100 UTC the Δθ values were 6.9 K over a 35 km north south distance in west Alabama while simultaneously 5.3 K over that same distance in east Alabama. This is shown in Figure 5.15. The width of the baroclinic zone on the western portion of the boundary varied throughout the event. This is likely a result of multiple storms passing through the region. A higher resolution of surface observations would be needed to state this with more confidence. Interestingly, the Cullman, Hackleburg, and Jackson County storms produced their most significant damage within this baroclinic region. This will be discussed further in Chapter 6.

62

5.2.1 Vertical Boundary Development Before one can understand the impact the boundary had on storms it is first important to understand the effects on the storm environments on the warm and cool sides of the boundary. The cool side of the boundary was well sampled by the MIPS. The wind profiler and MPR both took high-resolution measurements of the environment prior to the boundary formation, and during the passage of the baroclinic region In addition, a balloon sounding was acquired at 2052 UTC prior to the passage of the boundary. Surface measurements from the MIPS (Figure 5.9) demonstrate the cool side of the boundary had a higher relative humidity resulting in lower LCL heights in addition to the cooler temperatures found. Data from the MPR demonstrate the depth of this boundary changed with time as a low-level inversion developed atop the cold pool whose depth decreased with time. The cool outflow from the storms was confined to the lowest 400 m of the atmosphere. The inversion was strengthened by warmer air located to the south of the boundary being forced over the cooler more dense air. The vertical structure and evolution of the boundary are shown below in Figures 5.19-5.21. The initial sounding from RSA at 1700 UTC (Figure 5.19) demonstrates a dry adiabatic layer near the surface prior to the passage of the midday QLCS. This sounding contains modest CAPE values capable of sustaining the QLCS. The 2050 UTC sounding (Figure 5.20) shows the top of the cooler surface-based air had begun to warm slightly with the slope of the inversion having decreased. The top of the inversion extended to approximately 900 mb. By 2115 UTC (Figure 5.21) the coldest air had been eroded away with the surface-based inversion now only extending a few hundred meters from the

63

surface. Additionally, the inversion only extended to 950 mb at that time. SBCAPE values were still meager but CIN values were large. However, MUCAPE values of greater than 2000 J kg-1 were capable of supporting supercell storms. This will be discussed in the next section. The decreasing depth of the inversion at with time at UAH suggests a sloping warm air cool air interface, presumably similar to that shown in the Figure 3.7 schematic. Moreover, a sloped boundary would lead to decreasing amounts of CAPE for parcels located increasingly further north of the boundary where the depth of the cold pool was greater. This partially explains why storms located well into the cool side of the boundary were incapable of maintaining their intensity. This subsequently inhibited the ability of these storms to produce tornadoes despite having higher shear values in these areas.

64

Figure 5.19: RSA sounding launched at 1700 UTC prior to the passage of the midday QLCS and formation of the boundary.

Figure 5.20: UAH MPR sounding at 2030 UTC showing the fully developed inversion.

65

Figure 5.21: UAH MPR sounding at 2115 UTC showing the inversion depth decreasing. The height of the inversion has decreased to approximately 950 mb. The depth of the coldest air has now been reduced to just above the surface.

66

5.3 Boundary Characteristics 5.3.1 Temperature and Stability A sounding launched at 2052 UTC from UAH on the cool side of the boundary (Figure 5.22) demonstrates the significant temperature decrease from the collection of cool outflow from the storms. This sounding was launched approximately 1 h prior to the passage of the Hackleburg storm. With the boundary passage occurring at KHSV at 2227 UTC the boundary’s approximate position relative to the launch site was 36 km to the south. This sounding terminated at 547 mb. Data from the 1700 UTC RSA shown in Figure 5.19 sounding has been used to fill the upper levels for the purposes of this study. A closer look at Figure 5.22 shows a slight discontinuity where the RSA data was added. The existing RSA sounding was taken 4 h earlier and appears to have been warmer at higher levels than the sounding launched by UAH. This implies that the CAPE values given are actually underestimated. Despite the presence of the low-level inversion, areas north of the boundary were able to support strong convection. While SBCAPE at the time of this sounding was only 26 J kg-1 with CIN as much as -591 J kg – 1 the MLCAPE values approached 1735 J kg-1 with only -43 J kg-1of CIN. MLCAPE is defined as the amount of CAPE available to a parcels in the lowest 100 mb of the atmosphere. The MUCAPE value at the time of this launch was 2835 J kg-1 for a parcel origin at 940 mb with negligible CIN present. All tornadic storms on the cool side of the boundary were found to have been located within 15 km of the boundary for a majority of the time they produced tornadic damage. With the boundary being located approximately 25 km south at the time of this

67

sounding it can be assumed that storms residing within the 15 km range from the boundary had higher CAPE values available due to the sloped boundary interface. The 4/28 0000 UTC sounding launched by the NWS in Calera, Alabama (shown as KBMX in Figure 3.1) demonstrates conditions that existed on the warm side of the boundary. This sounding is displayed in Figure 5.23. The combination of no surface inversion along with steep lapse rates produced prodigious amount amounts of CAPE in central Alabama. SBCAPE in this area reached 3458 J kg-1 with MLAPE values reaching 4635 J kg-1. This sounding is only moderately representative of the environments of storms determined to be south of the boundary discussed in this study. This is primarily because Birmingham received more daytime heating than areas in north Alabama. For example, the surface temperature at the time of this sounding was 26°C at KBMX whereas, the maximum temperature in Cullman, Alabama post boundary passage was only 24.5°C. These locations are displayed in Figure 3.1.

68

Figure 5.22: 2100 UTC sounding launched from the National Space Science Technology Center on the Campus of UAH. Sounding was launched approximately 25 km into the cool side of the boundary as the Hackleburg storm was located approximately 60 km southwest.

Figure 5.23: 28, April 2011 00 UTC sounding from Calera, Alabama. Launch location 90 km south of the southernmost extent of the boundary.

69

5.3.2 Variation of the Vertical Wind Profile As discussed in Chapter 2 the low-level wind profile has been found to be an important factor in the formation of supercells. EVAD analyses have been produced from the five radars used in this study to show the differing wind profiles at their respective locations. KBMX was located well south of the boundary and an EVAD analysis from this radar shown in Figure 5.24 demonstrates the associated 0-1 km wind profile. In addition, the 0000 UTC sounding shown in figure 5.23 also demonstrates the profile sampled from KBMX in Calera, Alabama (location shown in Figure 3.1). The KGWX radar situated on the Mississippi-Alabama state line was determined to be nearest the boundary prior to 2000 UTC. Figure 5.25 demonstrates the evolution of the wind profile in Columbus, Mississippi. At 1900 UTC the wind profile is shown to be unidirectional with winds from the south over the lowest 1000 m. However, just after 1900 UTC a slight backing of the lowest levels is apparent which is likely associated with the southernmost intrusion of the boundary in this location (refer to Figure 6.15). After 2000 UTC the profile becomes more unidirectional for the lowest 1000 m. A second shift in the 0-1 km winds occurs at 2100 UTC when these winds shifted to more southwesterly. At 2140 UTC the winds become westerly marking the passage of the cold front.

70

Figure 5.24: 20-1000 m EVAD analysis winds at KBMX from 1800-2400 UTC. A full barb= 10 kts. KBMX’s location relative to the boundary is depicted in images 5.12-5.16.

Figure 5.25: EVAD analysis winds from KGWX from 19-22 UTC over 20-1000 m. Frontal passage occurs prior to 2200 UTC. One barb= 10 kts. KGWX’s location relative to the boundary is depicted in images 5.12-5.16.

71

The Hytop, Alabama (shown as KHTX in Figure 3.1) wind profile, in Figure 5.26, shows westerly low-level winds dominating the area from 18-19 UTC before backing to southerly by 1930 UTC. The westerly winds were a product of the passing of the midday QLCS. Southerly winds then dominate the 0-1 km profile for the remainder of the time shown. The easterly winds did not develop in this region due to the lack of precipitation that fell over this area when compared to other portions of north Alabama. Furthermore, marked wind shifts are apparent prior to 2100 UTC and 2136. These winds shifts are associated with the passage of the Cullman supercell to the south and east and the Jackson county storm to the south and east. The most representative wind profile of the cool side of the boundary is that from Huntsville, Alabama and is shown in Figure 5.27. Due to the 5 km radius of influence imposed on the analysis, ARMOR data was only available for the 20-200 m height interval. Winds above this height were supplied UAH’s 915 MHz wind profiler with 10 m winds being supplied by KHSV. The cool side of the boundary in Huntsville was strong and persisted with time. Also, only light convection covered the area leading to unbiased wind profiles. From 1900 to 2000 UTC the wind profile represents the cool side of the boundary only. At approximately 2000 UTC winds at the 200 m AGL began veering to more southerly. This veering of winds became shallower with time until the 10 m winds at KHSV were measured to be southerly at 2140 UTC. This gradual wind shift with height also depicts the slope like cool air-warm air interface. This process of veering from easterly to southerly matches well with the onset of temperature increase (leading edge of baroclinic zone) at KHSV (Figure 5.10). This is occurred at 2115 when the temperature increased to 19°C.

72

Figure 5.26: EVAD Analysis from KHTX from 18-22 UTC. One barb= 10 kts. KHTX’s location relative to the boundary is depicted in images 5.12-5.16.

73

The easterly to southeasterly winds that comprised the cool side of the boundary were generally light. Surface winds were found to be easterly at many surface stations located on the cool side of the boundary, exceptions at Russellville and Cullman. These areas only experienced a temperature change with the initial passage of the boundary. Winds in these areas remained southerly for the duration of the event. Conversely, the Muscle Shoals surface data show a relatively strong easterly wind that persisted from the passage of the mid-day QLCS until passage of the cold front. This variation indicates that the easterly winds were isolated to the northern portions Alabama where the coolest and deepest air was located. This hypothesis is further supported by data from the Brindley Mountain surface site (Figure 3.1). This site is roughly located at the midway point between Huntsville and Cullman. Winds at this location were found to be variable from 150-190° between the times of 1800 and 2100 UTC.

74

Figure 5.27: ARMOR and 915 MHz Wind Profiler wind profile. Easterly flow dominates the 1900 to 2100 UTC lowest levels. The 0-200 m winds become more southerly at approximately 2030 UTC with the surface winds becoming southerly at 2115 UTC. The lowest 200 m data is supplied by ARMOR radar located at Huntsville International Airport. The Huntsville ASOS is also located there and shows the boundary passage to have occurred between 2130 and 2200 UTC. The 200-1000 m winds are provided by the 915 MHz profiler located on the campus of UAH. One barb=10 kts. ARMOR’s position relative to the boundary is depicted in Figure 5.12-5.16.

75

5.3.3 Storm-Relative Helicity SRH has been shown to be a good predictor of both storm type and supercell duration with higher values of SRH being more favorable for supercell development. With SRH being a measure of the amount of helicity available to be tilted by a storm’s updraft the value of SRH is dependent on both directional and speed shear as well as storm motion. With the existence of a thermal boundary varying values of SRH across the boundary would be expected. Figure 5.28 illustrates the SRH values determined by the KHSV-ARMOR-MIPS wind profiles. This time series shows the varying values of SRH that were available prior, during, and after boundary passage. While the synoptic environment was more than sufficient enough to produce large supercells the figure demonstrates a local enhancement coincident with the passage of the baroclinic region of the boundary. There are two portions of the time series where 0-1 SRH increases markedly. First, an increase from 400 m2 s-2 to greater than 600 m2 s-2 occurred from 2045 UTC to 2105 UTC. Secondly, the 0-1 SRH values tripled from 200 m2 s-2 to 600 m2 s-2 between 2115 and 2145 UTC. Both maximum values occur during the passage of the baroclinic region. The first maximum in SRH occurred after the winds began to shift to southerly. The second maximum in SRH corresponds to the time the 0-200 m AGL winds became southerly and increased in intensity. This allowed for increased values in SRH despite the lower vertical directional shear. As mentioned above a SRH value of 200 m2 s-2 is more than adequate to produce supercells but the enhancement to 600 m2 s-2 helps to explain why the baroclinic region and points south were the preferred location for supercells.

76

Figure 5.28: SRH calculation for the 0-1 km AGL layer from UAH’s ARMOR and 915 MHz wind profiler. A rapid increase in SRH started at 2115 UTC just prior to boundary passage. After reaching a maximum in SRH at 2140 UTC SRH values fall off dramatically to values close to 200 m2 s-2.

77

CHAPTER 6

SUPERCELLS INTERACTIONS WITH THE BOUNDARY

As stated in Chapter 2, Markowski (1998) determined that approximately 70% of tornadic storms formed in the presence of a thermal boundary during the 1995 VORTEX1 project. More specifically, a region ranging from 10 km south of the boundary to as much as 30 km north of the boundary seemed to be the area most likely to produce tornadic storms as shown in Figure 2.5. During the 27 April outbreak this range was found to be even further constrained from 10 km south to 20 km north. The position of this boundary was determined using surface measurements across western Mississippi and north Alabama. The boundary was shallow in nature and exhibited thermodynamic changes over relatively large distances of approximately 10 km such that it did not produce a fine line on radar. Also, the scattered cloud cover prevented the boundary from being detected by GOES visible satellite like in Markowski et al. (1998). All supercells that were determined to have traveled within this range produced tornadoes. Furthermore, all but one of these tornadic storms produced tornadoes that were classified as strong (EF-3 or greater). The following sections provide details of these interactions.

6.1 Cullman EF-4 Supercell The Cullman EF-4 tornadic supercell was the first storm to interact with the boundary and produce a tornado. This storm formed southwest of Tuscaloosa, Alabama on the Mississippi-Alabama state line at 1713 UTC and strengthened while traveling northeast at 27 m s-1. Figure 6.1 shows the development of the storm from its first appearance on radar until tornadogenesis. Photographic documentation of the storm’s visual appearance is shown in Figure 6.2. This photograph documents the formation of a wall cloud in northwest Walker County as well as the relatively high cloud base that was typical of storms forming south of the boundary on 27 April. Upon reaching Crane Hill, Alabama (Figure 3.1) at 1942 UTC, the storm produced its first tornadic damage. The tornado quickly strengthened to EF-4 intensity upon reaching the city of Cullman 13 minutes later. Data from the Cullman surface station show an onset of warming from a temperature of 15.7°C at 1855 UTC to a maximum in temperature (boundary passage) of 24.5°C at 1955 UTC. At this time, the storm was located near the city of Cullman 6.5 km southwest of the surface station. Despite the course 20-minute temporal sampling at this station it can be stated with reasonable confidence that the mesocyclone was located within a 10 km distance south of the boundary as demonstrated in Figures 6.3 and 6.4. Figure 6.3 shows data from the Cullman surface station while Figure 6.4 shows the author’s subjective analysis of the boundary’s position relative to the storm Figure 6.4 demonstrates that while Cullman received no rain from the midday QLCS surface temperatures dropped 5°C which coincided with winds shifting to northerly. Figure 6.5 indicates a relatively high cloud base associated with the storm while in the city of

79

Cullman. Upon crossing to the cool side of the boundary the surrounding cloud base lowered. This tornado remained on the ground for approximately 55 minutes over a 41 km distance. The storm was located on the warm side of the boundary during its formation. The storm crossed from the warm side of the boundary to the cool side prior to passing the town of Arab, Alabama at 2020 UTC. This is demonstrated in both surface measurements (Figure 5.14 and surrounding data) as well as photographs taken of the storm (Figures 6.5 and 6.6). Figure 6.4 shows the storms relative position to the boundary and Figure 6.6 shows the visibly lower cloud base associated with the cooler air north of the boundary as seen from Arab, Alabama at 2020 UTC. After passing from the warm to cool side of the boundary, the surrounding cloud base lowered significantly as shown by Figures 6.5 and 6.6. These lower cloud bases can be associated with lower LCL heights found on the cool side of the boundary. The lower LCL heights can be attributed to the cooler temperatures combined with the maintained high dew points on the cool side of the boundary. LCL heights have been estimated using the equation LCL (km)= (T-Td)/8 (Lawrence 2005) (where T and Td are in °C) to be 200 m (18.7 °C, 17.0°C) on the cool side of the boundary ,whereas the warm side had an LCL height of approximately 700 m (24.8°C,19.3°C). Figure 6.7 demonstrates the tornado’s rotational velocity intensity as a function of the tornado’s position relative to the boundary. The image depicts an increase in Vrot to approximately 40 m s-1 at 2010 UTC, at approximately the same time the storm was determined to have moved within 5 km south of the boundary. Furthermore, the depth of the higher Vrot also increased to approximately 2000 m.

80

The storm produced a second tornado approximately 10 km southeast of Scottsboro, Alabama at 2101 UTC. This tornado also produced EF-4 damage over a 60 km path and formed less than 10 km north of the boundary. This was determined using data from the Scottsboro station where the temperature increased from 19.9 °C at 2050 UTC to 23.2°C at 2120 UTC. The storm continued traveling northeast into the state of Georgia. Despite being poorly sampled, it is evident the temperature difference across the boundary was reduced.

81

Figure 6.1: Six panel display showing the development of the Cullman EF-4 supercell. The Cullman storm is highlighted by the white box in each image. Tornadogenesis occurred at 1942.

Figure 6.2: Photograph taken of storm at 1922 UTC in northwest Walker County. Image looking northeast. A wall cloud has developed at this time. Image also depicts the relatively high cloud base associated with the warm side of the boundary (refer to Figures 6.5 & 6.6). Photograph by Dr. Timothy Coleman.

82

Figure 6.3: Time series from Cullman, Alabama showing temperature (red), dewpoint (green), pressure-1000 mb (black) on left axis. Total rainfall since 0000 UTC (blue) is on right axis. Image shows boundary passage at this location was near 1955 UTC.

Figure 6.4: ARMOR 0.7° PPI demonstrating the approximate location of the Cullman storm relative to the surface observation. Values of θ for this image were constrained to ± 5 minutes of indicated times.

83

Figure 6.5: Picture taken from Cullman AL at approximately 1955 UTC looking northwest. Tornado is approximately 2.5 km northwest of camera. Image shows multiple vortices surrounding the tornado. A higher cloud base is also demonstrated in this image (refer to Figure 6.6). This higher cloud base provides further confidence the storm is located on the warm side of the boundary. Photograph by FOX WBRC in Birmingham.

Figure 6.6: Cullman EF-4 tornado photographed at 2020 UTC looking northwest from Arab, Alabama. Storm is located approximately 6 km northwest. Cloud base surrounding tornado is now closer to ground indicative of a lower LCL height which is produced by the cooler and more humid air on the cool side of the boundary. Photograph by Charles Whisenant.

84

Figure 6.7: Rotational velocity (Vrot) of the Cullman EF-5 storm and approximate position relative to the boundary. Data from KHTX was used to determine values of Vrot. Positive distance denotes storm is north of the boundary.

85

6.2 Hackleburg EF-5 Supercell The first echo of the EF-5 Hackleburg supercell appeared on radar at 1915 UTC 40 km southwest of KGWX. The cell moved NNE until merging with a pre-existing storm at which time the storm began traveling northeast at 23 m s-1. Despite extensive ground clutter evidence of rotation was apparent in both cells at the time of the merger as demonstrated in Figure 6.8. Rotation from the secondary storm persisted until 2028 UTC. The main circulation produced what would become the Hackleburg tornado at 2014 UTC northwest of Hamilton, Alabama (Figure 3.1). Perhaps one of the most impressive features of the Hackleburg storm was its ability to produce a tornado within an hour of its first echo. This progression is documented in Figure 6.9. Furthermore, this storm produced EF-4 damage for a majority of the 212 km path length. This may be attributed to the close proximity of the storm to the boundary. Figure 6.10 demonstrates the storm’s proximity to the boundary from 2000 UTC till 2115 UTC.

86

Figure 6.8: 0.5° PPI from KGWX at 2001 UTC. Storm is located approximately 20 km north of the radar. Left panel indicates radar reflectivity (Z) while right panel is stormrelative velocity (SRV).

87

Figure 6.9: 0.5° PPIs from KGWX demonstrating the progression from first appearance on radar to tornadogenesis. The storm produced its first damage report within one hour of first appearing on radar.

Figure 6.10: 0.5° PPI from KGWX (panels A-D) and 0.7° PPI from ARMOR (panels E&F) demonstrating the Hackleburg’s supercell relative to the boundary’s position from 2001 to 2115 UTC. The boundary’s location is indicated by the white dashed line. 88

By 2015 UTC Figure 6.10 (B) shows a locally warmer region located northeast of the storm that extends northward to Russellville, Alabama. The surface station in Russellville shows the onset of warming temperatures occurred at 2005 UTC. The 2000 UTC observation measured value of θ of 290 K whereas the 2005 UTC observation showed a 2 K increase. By 2025 UTC a maximum value of θ (295.8 K) was recorded when the FFD of this storm was overhead. This maximum value marks the passage of the boundary at this location Figure 6.10 demonstrates the storms position relative to the boundary between 2000 UTC and 2100 UTC. Despite the evaporative cooling of rain within the FFD, surface temperatures values remained nearly constant until 2042 UTC as demonstrated in Figure 6.11. At that time, radar data shows the FFD had passed Russellville with surface reports that indicated θ values had decreased to 293 K. This drop in temperature occurred prior to the passage of the dissipating secondary mesocyclone as well as the passage of the RFD. The drop in temperature associated with the FFD was slightly delayed (likely due to the five minute update). However, it is likely the rotation within the storm also caused the RFD to pull cool air from north of the boundary southward. The mesocyclone likely intersected the boundary at 2042 UTC (Figure 6.10 D). This uncertainty is due to lack of temporal and spatial reports in this area. In addition, the decrease in temperature found in Russellville after the passage of the storm introduces further uncertainty of the boundary’s position. This cooling raises the question of whether the boundary itself was advected to the south or if the cooling was a localized effect and the boundary position was unaffected. It is the author’s opinion that the supercell’s lowlevel circulation influenced the boundary and pushed it to the south. Furthermore, due to

89

the RFD forcing cool air southward, as indicated by data from Russellville, the production of cool outflow from within the storm, and inflow to the storm drawing the warm air northward, a precise location of the boundary would be difficult to place. Moreover, these processes would likely lead to the baroclinic region becoming discontinuous, wavy, and difficult to replicate. Figure 6.10 (F) shows the storm had completely crossed into the cool side of the boundary by 2115 UTC. Data obtained from the Moulton, Alabama surface station further supports this statement. This site reached a maximum temperature at 2110 UTC of 295.2 K with the previous measurement of 293.2 K coming only seven minutes earlier at 2103 UTC. At 2115 UTC the main circulation was centered approximately 20 km north of the boundary. Further evidence of this storm crossing to the cool side of the boundary is demonstrated in Figure 6.12. This photograph was taken near Tanner, Alabama (Figure 3.1) at approximately 2130 UTC looking southwest at the tornado as it was producing strong EF-4 damage. The photograph depicts the low cloud base (estimated at 100 m) that was associated with the cool humid air north of the boundary.

90

Figure 6.11: Time series from Russellville, Alabama showing temperature (red), dew point (green), pressure-1000 mb (black) on left axis. Total rainfall since 0000 UTC (blue) is on right axis.

Figure 6.12: Photograph taken of the Hackleburg EF-5 at 2120 UTC from 20 km to the southwest. Image shows low cloud base associated with the storm. LCL was estimated to be approximately 100 m at this time. Photograph courtesy of Gary Cosby, Decatur Daily.

91

The storm remained steady state till 2144 UTC when the circulation weakened on radar. The NWS storm report mentions a brief pause in damage, but concluded the break was not long enough to classify it as two tornadoes. The circulation reappeared on radar again at 2207 UTC while the storm was crossing the Alabama-Tennessee state line. Data from the Madison County Airport AWOS station show that a maximum temperature was reached at 2235 UTC when a maximum value of θ of 292.5 K occurred. The onset of this temperature increase occurred at 2135 UTC when the temperature increased from 288.5 K to 289.5 K. This more gradual increase in temperature implies the temperature gradient had decreased at this time. Furthermore, temperatures near Huntsville at this time were closer to 296 K meaning the boundary was still located to the south of Madison County Airport. Furthermore, with θ maxing out at only 292.5 it indicates the boundary had stalled south of this location. The boundary likely remained in this region while being further diminished by the numerous supercells over the region during this period. Additionally, the constant interaction with the boundary by downdrafts and boundary layer inflows of the storms would lead to a decrease in the gradient. The increased width of the baroclinic zone (Figures 5.15 and 5.16) gives high confidence that the storm despite being 20-30 km north of away from the boundary, resided within the baroclinic region at the time of this second tornadogenesis. The NWS storm survey states that the tornado again produced damage in Huntland, Tennessee at 2207 UTC. This damage path was 17.6 km long with a maximum width of 400 m. This last portion of the tornado was determined to have caused EF-3 damage. The exact position of the boundary relative to the storm is unknown due to lack of surface observations in this area. However, as stated above the boundary was located

92

south of the Madison County Airport with the baroclinic region extending northward from there. This problem is compounded further due to ARMOR’s severely attenuated data and a communication failure occurring at KHTX at 2216 UTC. The combination of these factors makes further analysis of this storm futile. The National Weather Service determined this storm produced its first EF-5 rated damage at Hackleburg, AL at approximately 2025 UTC. The tornado maintained this strength over a 13 km distance into Phil Campbell (2035 UTC) and Oak Grove (2045 UTC). The tornado was determined by the NWS to have reached maximum intensity of 94 m s -1 with a path width of 1.9 km in the town of Oak Grove, Alabama. It is here the tornado path width was determined to be slightly over 1.9 km and the damage was described as being “slightly more intense”. The available data indicate that during the period that the storm produced EF-5 damage, the tornado was located within ±5 km of the boundary. Furthermore, there is high confidence that the tornado crossed the boundary within 10 minutes of producing this EF-5 damage. This range is determined by using the available data in combination with the forward speed calculated by the M3V. Despite extensive ground clutter evidence of rotation was apparent in both cells at the time of the merger as demonstrated in Figure 6.8. Figure 6.13 relates the mesocyclone’s rotational velocity to the boundary’s position by contouring Vrot and plotting the estimated distance from the boundary. This image depicts the rapid intensification in Vrot occurring when the tornado was located approximately 5 km south of the boundary. This intense rotation persisted through the town of Oak Grove before it weakened as the storm moved greater than five miles north of the boundary.

93

Figure 6.13: Contour of Vrot as of the Hackleburg storm as determined from both ARMOR and KGWX. The right axis demonstrates distance from the radar (solid black line) and the author’s estimation of the mesocyclone distance from the boundary (black line with dots). KGWX data is displayed for 2001 to 2042 UTC. After 2042 UTC KGWX is used to supplement upper level data with ARMOR supplying the 1000-3000 m. Due to extended range from KGWX and a three-tilt scan mode by ARMOR further analysis is not done. 94

6.3 Smithville Supercell The storm that produced the Smithville EF-5 formed from a QLCS that was moving eastward over central Mississippi. This QLCS produced three supercells. This in and of itself is interesting because the common progression of a storm event is to evolve from discrete to linear, not linear to discrete. This transition from linear to supercell occurred in a region where the boundary was poorly documented. The evolution from linear to discrete is shown in Figure 6.14. The southernmost portion of this QLCS produced the Smithville EF-5 tornado and was the strongest of the three storms. The storm began showing signs of rotation by 1915 UTC while traveling northeast at 27 m s-1 and became a strong supercell by 1933 UTC. This storm produced a tornado at 1952 UTC in Bellefontaine, Mississippi (Figure 3.1) which quickly strengthened to EF-3 intensity. The data available show a weak thermal gradient between observation sites to the east and northwest with Δθ values of less than 3 K. This indicates the storm was north of the boundary at this time. The largest gradient in θ appeared to be located just north of the storm’s FFD. The tornado from this storm remained on the ground for 47 minutes until it dissipated at 2039 UTC. The same weak gradient in θ was consistent around the storm at this time as demonstrated by Figure 6.15. θ values of 298 K 24 km to the storm’s southeast marked the approximate boundary location. Theta values northwest of the storm were approaching 296 K. Moreover, a much larger difference in θ can be seen between these sites and those in northeast Mississippi where values where only 292 K. Once again the exact strength of the gradient is unknown due to a lack of observation sites in this region of Mississippi. The weaker gradient in this region is expected

95

considering the midday QLCS only impacted the northeastern portion of Mississippi. Furthermore, the existence of the boundary in this region is confirmed by the Tishomingo, Mississippi (Figure 3.1) data that show a change from 288.2 K to 292.6 K between the 2014 UTC observation and 2114 UTC observation. At 2042 UTC the storm produced a second tornado. This tornado quickly produced EF-5 damage in Smithville, Mississippi and remained on the ground for 59 km with a maximum path width of 1.2 km. The progression of the storm and tornado are shown in Figure 6.15. At approximately 2100 UTC the storm crossed into the state of Alabama. At this time the boundary was likely located north of the storm. This is indicated by data from the Tupelo, Mississippi surface station where a 6 K increase occurred between 1857 and 2020 UTC. Unfortunately, there was only one other observation at 1953 UTC during this period. However, the station reported a θ value of 295.4 K at 1953 UTC before reaching 296.1 K at 2020 UTC leading the author to believe the boundary passage was closer to 2020 UTC than the initial 1857 reporting time. By 2120 UTC the tornado dissipated in west Alabama. By this time the supercell had crossed to the warm side of the boundary and was located approximately 10 km south of the boundary’s position. The supercell then began merging with a supercell trailing the Hackleburg supercell (Section 6.5) and the distance between the supercell and the boundary was increasing. Furthermore, the strength of the boundary had been greatly diminished in west Alabama as demonstrated in Figures 5.14 and 5.15. The mesocyclone from the Smithville supercell persisted till 2200 UTC, but never produced any more damage.

96

Figure 6.14: 0.5° PPIs from KGWX depicting the evolution from QLCS to supercell that produced the Smithville EF-5 supercell.

Figure 6.15: 0.5°PPI from KGWX showing approximate boundary position relative to the Smithville EF-5 supercell. The S, L, and H denote the Smithville, Limestone County, and Hackleburg supercells respectively.

97

6.4 Jackson County EF-4 Supercell The Jackson County EF-4 producing supercell formed near Tuscaloosa, Alabama at 1930 UTC. The storm slowly strengthened while traveling northeast at 28 m s-1, and developed a mesocyclone1.75 hours later near Cullman, Alabama by 2047 UTC. The storm continued to intensify until reaching the Scottsboro area at 2152 UTC. At this time the interaction of reflectivity segments coincided with rapid intensification. The lowlevel inversion located north of the boundary supplied a ducting mechanism through which these reflectivity segments could propagate (Murphy et al. 2013). At 2207 UTC the storm produced its first tornadic damage northeast of the town of Hollywood, Alabama (Figure 3.1) and continued producing EF-4 damage over a path length of approximately 50 km. Figure 6.16 illustrates the development of this storm through tornadogenesis. With the exception of Scottsboro, surface observations are scarce for the region where this storm produced its tornadic damage. Furthermore, the boundary was not well defined in this region at this point in the day. However, the tornado produced its first damage at 2205, UTC roughly 15 km northeast of the Scottsboro surface station. Between 2140 and 2230 UTC this station recorded a continued increase in θ from 294.0 K to 297.4 K (Figure 6.17). This warming occurred after the FFD of this storm passed over the station at 2130 UTC, producing a 1.8 K drop in θ. Prior to this drop the temperature had risen from 292.4 K at 2055 UTC to 295.7 K at 2120 UTC. This process leaves doubt as to where the actual boundary is located. However, using the constant forward motion mentioned earlier in conjunction with the data depicted in Figure 6.17, it can be shown the storm was located in a baroclinic region during the time of

98

tornadogenesis. This is also indicated by the secondary increase in temperature at 2200 UTC. Furthermore, the surface stations located along this latitude were also showing increases in temperature at this time. Values of θ for these stations as well as the boundary location are demonstrated in Figure 6.18. Figure 6.19 further demonstrates the rapid intensification of this storm in Vrot between 2134 UTC and 2111 UTC. Due to power failure at KHTX past this time further analysis could not be completed. This image depicts moderately low Vrot of less than 20 m s-1 until approximately 2152 UTC when the values increased to 30 m s-1. Furthermore, the depth of the increased Vrot increased with the increased in magnitude.

99

Figure 6.16: 0.5° PPIs from KBMX (A,B,C) & KHTX (D,E,F) depicting the formation and development of the Jackson County EF-4 tornado. The storm remained relatively weak for 2.5 hours until reaching the Jackson County area. Here the storm rapidly intensified and produced a tornado at 2207 UTC. The J, L, C and H denote the Jackson County, Limestone County, Cullman, and Hackleburg supercells respectively.

Figure 6.17: Scottsboro, Alabama measured temperature. A spike in temperature from 20°C to 23°C is shown between 2100 and 2130 UTC corresponding to a period increased solar insolation. This spike was followed by a sharp decrease in temperature that corresponds to the passage of the FFD from the Cullman supercell. Following this decrease a gradual increase is seen from 2100 to 2330 UTC. 100

Figure 6.18: KHTX 0.5° tilt at 2205 UTC showing the Jackson Ef-4 supercell. Surrounding surface theta values shown for 2000 UTC. A temperature gradient exists between north Huntsville and Scottsboro, Alabama.

Figure 6.19: 0.5° PPI from KHTX depicting the rapid intensification of Vrot between 2134 UTC and 2211 UTC. Uncertainty in boundary position prevents approximate boundary position from the mesocyclone from being displayed. 101

6.5 Limestone County Supercell The last storm that interacted with the boundary and produced a significant tornado first appeared on radar at 1930 UTC southwest of KGWX. Like the Hackleburg storm this storm initially moved north-northeast from 210°. Once the storm passed north of KGWX at 2020 UTC it also took on a northeast trajectory moving from 245° at 24 m s –1 and began trailing the Hackleburg supercell. The Limestone County supercell storm began merging with both the Smithville and Hackleburg storms as early as 2105 UTC when it was located between the cities of Hamilton and Russellville. The storm exhibited relatively weak rotation for 1.5 hours. During the majority of this time the storm was approximately 20-30 km south of the boundary. However, as the storm progressed to northeast the distance between the boundary and the storm decreased. This is likely due to a slowing of the eastern portion of the boundary by the convection ahead of this storm. This storm rapidly developed a hook echo and mesocyclone after crossing the Tennessee River at 2145 UTC. This is interesting considering the storm showed no signs of further intensification after forming 2.25 hours earlier. This development is demonstrated in Figure 6.20. The storm’s first tornadic damage occurred in southeast Limestone County 4 km northwest of Belle Mina, Alabama (Figure 3.1) at 2153 UTC. This storm produced spots of high-end EF-2 damage over of its 24 km path length. At the time of tornadogensis the mesocyclone was likely ± 5 km from the boundary. The 2153 UTC observation from KHSV (Figure 5.10), that was located 30 km due east of the storm, shows the boundary had passed having reached a maximum in θ of 296.8 K. Uncertainty in the boundary’s position is introduced with the previous measurement of θ (290.3 K) coming 39 minutes earlier at 2114 UTC. However, the boundary did not pass

102

UAH until approximately 2157 UTC giving further confidence the boundary was located closer to the storm at this time. The tornado continued producing EF-1 damage for an additional 9.5 km after crossing into Madison County at 2202 UTC before it dissipated at 2207 UTC. At this time, the storm was located 8 km due north of the UAH surface instrumentation that showed the boundary had passed only ten minutes earlier. Assuming the 6.6 m s -1 boundary speed the storm was likely only 5 km north of the boundary. Furthermore, the UAH (Figure 5.9) data shows strong sustained southerly winds indicating the storm’s boundary layer inflow was still ingesting warm air from south of the boundary. Figure 6.21 demonstrates the storms position relative to the estimated boundary position at the time of tornadogenesis.

103

Figure 6.20: 0.5° PPIs from KGWX (A,B,C,D) and 0.7° PPIs from ARMOR (E, F) demonstrating the development of the Limestone County EF-2 supercell. Despite its slow development it developed rapidly upon catching up to the boundary near Limestone County at 2153 UTC. The L, S and H denote the Limestone County, Smithville, and Hackleburg supercells respectively.

Figure 6.21: 0.5 PPI from ARMOR showing the approximate boundary location relative to the Limestone County EF-2 producing supercell at time of tornadogensis. 104

6.6 Non-Tornadic or Weakly Tornadic Storms One of the most impressive features of the day was the efficiency with which supercells were able to produce tornadoes. These storms not only produced tornadoes, but most produced significant tornadoes (EF2 or greater) with wide path widths and long tracks. However, two supercell storms in northwest Alabama and northeast Mississippi did not produce a tornado, while an additional supercell produced only a short-tracked EF-0 rated tornado. 6.6.1 Supercells in Central Mississippi The first non-tornadic supercell formed from a line of severe storms oriented north-south over central Mississippi at 1830 UTC. This line broke into three individual storms that all became supercells by 1930 UTC. The northern-most storm, identified as A in Figure 6.22, did not produce a tornado. The middle storm, shown in Figure 6.23, produced only a brief EF-0 tornado with a 1.6 km path length. The boundary was less defined in this part of Mississippi. Despite a lack of surface observations in Mississippi, the available data suggest a weak thermal boundary or gradient existed in north central Mississippi. The northern-most storm developed approximately 80 km north of the boundary, while the middle storm was closer to the boundary, but on the cool side as well. The last storm, the southernmost of the three, produced the Smithville EF-5 supercell described in section 6.3. 6.6.2 Northwest Alabama Supercell The second storm that did not produce a tornado developed in northwest Alabama and identified as B in Figure 6.23. This storm began as a cluster of ordinary convection that merged and organized into a supercell at 1950 UTC. This storm exhibited high Vrot 105

values of 33 m s-1 at 1100 m. This storm had a more northerly storm motion when compared to the other supercell storms that developed that afternoon. Most storms that formed in northern Alabama traveled with a path oriented at 240°, whereas this storm had a path oriented closer to 230°. The more easterly progression was common in the other storms after developing a stronger mesocyclone. Initially, this storm formed approximately 30 km on the north side of the boundary but the more northerly track steered this storm into cooler air which likely reduced the CAPE available to the storm. 6.7 Summary In summary, only three supercells that formed north of Cullman, Alabama produced minor to no damage during the evening hours. The two supercells that did not produce a tornado formed and remained at least 60 km into cool side of the boundary while the third supercell only produced a short-track EF-0 tornado approximately 30 km north of the boundary. An EF-2 tornado was produced by a supercell that formed well south of the boundary and remained at a constant distance for a majority of its existence. Upon catching up to the boundary, the storm rapidly intensified and produced a tornado. Four of the five supercells that were determined to have interacted with the boundary produced tornadoes of EF-4 or greater intensity. The storm motion of these supercell’s caused them to cross the boundary at a relatively small angle when compared to Rasmussen et al. (1995). Not only did these storms produce violent tornadoes (EF-3 or greater) they also produced damage over a distance of greater than 25 km.

106

Figure 6.22: 0.5° PPI from KGWX showing the northern-most supercell that developed from the QLCS in central Mississippi.

Figure 6.23: 0.5° PPI from KGWX showing the middle supercell (A) from the QLCS and the supercell that developed in northwest Alabama (B). Also pictured are the Hackleburg EF-5 supercell, Hackleburg EF-5, and Cullman EF-4 supercells.

107

CHAPTER 7

DISCUSSION

The 27 April 2011 super outbreak occurred over a relatively small area when compared to other major outbreaks. The region impacted by the thermal boundary was further limited to northern Alabama and northeast Mississippi. A comprehensive study of the thermal boundary’s movement and characteristics was achieved with the combination of multiple surface measurements from various sites, a network of radars, and the UAH MIPS. The thermal boundary was initially formed by the cool outflow produced by the morning MCS over central to northern AL, and strongly reinforced by the midday QLCS that impacted the northern portion of Alabama and Mississippi. Once the QLCS exited the state of Alabama, scattered convective rain and extensive clouds lingered over northern Alabama. This continued rain along with overcast conditions reinforced the cool air mass allowing the boundary to persist and strengthen. By 2000 UTC the difference in θ between Birmingham and Muscle Shoals was 12 K. The warm and cool air masses were separated by a baroclinic region that was approximately 30 km wide in the north-south direction. The boundary location is defined by the warm side of this baroclinic region. This study showed that storms located from 20 km into this baroclinic region to 10 km south of the boundary were more likely to

produce stronger mesocyclones, particularly in the lower-levels .These stronger mesocyclones then produced long-lived tornadoes. This is much like the results shown in Markowski (1998) that showed during the VORTEX experiment storms located from 10 km on the warm side of the boundary to 30 km on the cool side of the boundary were most likely to produce tornadoes. Furthermore, three supercells on 27 April formed greater than 30 km into the cool air. Two of these supercells did not produce a tornado and their mid-level mesocyclones dissipated within 1.5 hours of moving over the cool air mass. The third supercell produced an EF-0 tornado with a 1 km path length. After producing the tornado this storm also lost all supercellular characteristics within an hour over the cool air mass. Conversely, multiple supercells developed well south of the boundary’s influence and still produced strong tornadoes. These supercells however, did form in a weak baroclinic region formed by the early-morning QLCS. The most notable of these storms were the Cordova EF-4 and Tuscaloosa EF-4 producing supercells. These supercells, along with the Cullman supercell, were deemed to be two of the most prolific supercells of the entire event (Knupp et al. 2014). This implies that: 1. The synoptic environment was conducive for the development for severe weather without the presence of the thermal boundary. However, supercells that formed south of the boundary were found to have formed in the presence of a second weaker baroclinic zone. 2. The three supercells that formed well into the cool portion of the boundary were likely inhibited by the cool air such that they were unable to produce tornadoes. Furthermore, northeast Mississippi and northwest Alabama (region where coolest air resided) received little to no damage after the passage of the midday QLCS.

109

Five tornado producing-supercells that included the Hackleburg EF5, Smithville EF5, Cullman County EF-4, Jackson County EF-4, and Limestone County EF-2 were determined to have interacted with the boundary. Of these supercells, time height sections of rotational velocity were presented for the Hackleburg, Cullman, and Jackson County storms. In each case rotational velocity was shown to rapidly increase in magnitude over the lowest 2-3 km AGL after each storm entered the baroclinic region. The Hackleburg storm formed very near the boundary. This storm began producing damage south of the boundary before crossing to the cool side of the boundary. The storm then remained within 10-20 km north of the boundary for the duration of its life cycle. This storm produced a single tornado with a path length of 212 km. Additionally, the Hackleburg storm produced tornadic damage within 1 hour of its first appearance. This is more impressive when considering this storm formed over a layer of cool stable air. Moreover, the top of this cool air was marked with a strong capping inversion. Despite this, the storm rapidly intensified from ordinary convection to a supercell in approximately 30 minutes. This rapid intensification occurred as the storm approached the boundary. Conversely, the Jackson County supercell appeared on radar nearly 2.5 hours prior to producing a tornado. A majority of this time the storm was benign in nature as it moved over the more homogenous air mass well south of the boundary. However, upon approaching the boundary the storm’s mesocyclone rapidly intensified and produced a tornado. The same effect was observed with the Limestone County storm. This storm appeared on radar 2.2 hours prior to tornadogenesis. Once again, upon intersecting the boundary the storm’s mesocyclone quickly intensified and produced an EF-2 tornado.

110

The rapid intensification and process of tornadogenesis can be attributed to multiple features found within the baroclinic zone. Markowski et al. (1998) and Rasmussen et al. (2000), both hypothesized that thermal boundaries are a preferred location for tornadogenesis due to hypothesized enhancements in horizontal vorticity produced by the production of baroclinically generated horizontal vorticity (BGHV). Similarly, Maddox et al (1980) hypothesized that the existence of a thermal boundary increases vertical vorticity via horizontal shears (producing vertical vorticity) within the boundary zone. In this study, this additional vertical vorticity was shown to exist on the cool side of the boundary where low-level winds within the boundary were found to be more easterly. Additional horizontal vorticity was likely supplied by the production BGHV. This vorticity was generated by solenoidal effects and was then readily available to be tilted and stretched by a storm’s updraft (Ras2000). Furthermore, after completing EVAD analyses a time series of wind profiles was created. These wind profiles were then used to calculate 0-1 SRH. Values of 0-1 SRH in Huntsville, Alabama increased from 400 m2 s-2 to 600 m2 s-2 as the boundary approached and passed. Such a narrow corridor of high SRH has not been demonstrated in previous studies. Despite the presence of a low-level inversion that was formed as a result of the cool surface-based air, an elevated unstable layer existed that was capable of supporting approximately 3000 J kg-1 of CAPE. This value was significantly lower than the 4000 J kg-1 of CAPE diagnosed at BMX. This differs from results presented by Ras2000 who compared soundings taken from 50 km into the cool side, 15 km into the cool side, and 15 km into the warm side, of the boundary the boundary. In that study a sounding launched 15 km north of the boundary also showed CAPE with a value of 1462 J kg-1.

111

However, the reduction in CAPE was less significant when compared to that on the warm side of the boundary. This can be attributed to the relative magnitudes of the boundaries in both studies. Temperature differences of 12 K were found to exist in this study between central and northwest Alabama whereas the temperature difference in Ras2000 was closer to 5 K. In the current study, the combination of the elevated CAPE combined with the likely additional sources of vorticity produced a more hospitable environment for the formation of strong long-lived tornadoes. The duration and extent of the interaction between each storm and the boundary varied. For example the trajectories of the Hackleburg, Cullman, and Limestone County storms had a small angle of incidence (less than 30°) relative to the boundary. This allowed the storms to remain in close proximity to the boundary for an extended period of time. Furthermore, upon crossing the boundary the increased mesocyclone strength caused the storms to develop a more easterly motion that also allowed for extended interaction periods. The Jackson County storm however, had an angle of incidence greater 30°. This storm was still able to produce EF-4 damage, yet produced damage over a shorter distance (48 km) than the Hackleburg and Cullman tornadoes which had path lengths of 212 and 73 km respectively (further details of this storm were unavailable due to power loss at KHTX). Interactions between the storms and the boundary were two-way in nature. Multiple surface observation sites showed a sharp increase in temperature that was followed by a quick drop in temperature. Moreover, evidence was shown that supports that the boundary’s forward progression was locally aided by the boundary-layer inflows to storms. This process briefly drew the boundary’s position to the north before it receded

112

after the storm’s passage. Blanchard (2008) documented a similar behavior of a nontornadic supercell storm during VORTEX-1 using mobile Mesonet observations of a non-tornadic supercell. The occurrence of five tornadoes of EF-4 or greater intensity (two from Cullman supercell) in a single event would at best be described as rare. The thermal boundary presented in this study was shown to have interacted with the storms that produced these tornadoes. Despite the high number of significant tornadoes on 27 April the boundary still played a significant role in the total amount of damage produced. For example, the Damage Potential Index (DPI), which is a measure of the area a tornado produces damage multiplied by its EF rating (Thompson and Vescio 1998, Doswell et al. 2006), for the 27 April super outbreak reached 21980. Storms associated with the thermal boundary accounted for 1650 of this total. For further perspective the second highest DPI recorded for an outbreak is 7330 occurring on 24 April 2010. It should be noted that in 1994 DPI was calculated using maximum path width where it had previously been mean path width. The findings of this study directly contradict the results of Nowotarski et al. (2011) who found that numerically simulated supercells over stable inversion layers, similar to that measured in the current study (Figure 5.22) generally have lower vorticity near the surface (Knupp et al. 2014). The studies of Ras2000 and Markowski et al. (1998) also demonstrate that the presence of thermal boundary is likely to lead to tornadogenesis. This study documents the intensification of five supercells that were located near or above a cool stable layer. Two of these supercells had existed for more than two hours over the warm unstable air before producing a tornado upon entering the

113

boundary baroclinic region. An additional storm remained ~10 km on the cool side of the boundary during much of its lifetime. This study also demonstrates the formation and evolution of this boundary. A sounding launched in advance of the midday QLCS demonstrated a convective boundary layer. A sounding was then launched after the development of the boundary. This sounding in addition with the thermodynamic profiles produced by UAH’s MPR demonstrates the evolution of the cool air near the surface. Ras2000 was the only previous study to demonstrate the structure of the air mass on the cool side of the boundary. The stability of the air mass was significantly less (i.e., conditionally unstable) than in the current study. The greater temperature difference in this study can be attributed to the convective showers and cloud cover after the passage of the QLCS.

114

CHAPTER EIGHT

CONCLUSIONS AND FUTURE WORK

8.1 Conclusion The 27, April 2011 Super Outbreak was one of the worst severe weather outbreaks in the history of the United States. Due to the violent nature of the storms the event killed hundreds of people despite being forecasted days in advance. A thermal boundary formed from the extended precipitation produced over northern Mississippi and Alabama. This thermal boundary played a key role in the formation of multiple violent tornadoes over this same region. This thermal boundary consisted of a cool side where potential temperatures at the surface were roughly equal to 290 K and a warm side that saw potential temperatures approaching 300 K. These values were separated by a 20-30 km wide baroclinic region. This baroclinic region did not appear as fine lines on radar or narrow lines of cumulus clouds often seen in visible satellite images as in previous studies. The cooler surface temperatures, north of the boundary, produced a stable surface based layer. Surface winds on the cool side of this boundary were found to be easterly at multiple locations while those on the warm side were southerly. While not directly measured in this study, the presence of this thermal gradient likely generated horizontal vorticity. The easterly winds found on the cool side of the boundary helped provide added SRH and shear for

the storms to ingest and further strengthen their mesocyclones. The region of increasing temperatures was found to be the most common place for a storm to produce its most significant damage. This zone extended approximately 20 km into the cool side of the boundary and 10 km into the warm side. These results are much like those found from VORTEX-1 shown in Markowski (1998), who determined that storms forming or traveling beyond this 20 km into the cool side of the boundary were unable to sustain supercellular characteristics or produce tornadoes. Furthermore, the coolest temperatures resulting from these midday storms were found to be in northwestern Alabama and northeastern Mississippi Multiple surface stations were used to show how the storms interacted with this boundary region. Through these measurements rapid temperature increases or falls as well as wind and pressure changes were noted with the approaching and passing of each storm. The boundary was shown to advance to the north at a rate that was faster than the southerly winds south of the boundary. This implies that the boundary advancement was aided by some combination of both vertical mixing and storm inflow. Two supercells formed just south of the boundary and crossed to the warm side during times they were producing tornadic damage. After crossing the boundary these storms remained near the boundary’s location for a majority of the time they produced damage. Between the two storms a combined damage length of just less than 350 km was produced with a majority of this damage being classified as either EF-4 or EF-5. Two supercells formed well south of the boundary and approached it with time. These supercells had relatively weak mesocyclones for approximately two hours. Once

116

these storms intersected the boundary the mesocyclones rapidly intensified and a significant tornado was produced. The results of this study have shown that the presence of a thermal boundary can be influential on the formation and intensity of supercells and their tornadoes. This knowledge can lead forecasters to be more aware of these boundaries during severe weather events. This could lead to advanced lead time for each storm and potentially save lives. 8.2 Future Work Spatial and temporal surface measurements of this boundary varied with geographic location and time. Multiple surface stations that would have provided useful data to this study were incapacitated on this day due to power failures from proceeding storms. Furthermore, other potentially important areas were void of data due to no surface instrumentation being in place. Given the large gradients in temperature along with the sometimes poor data resolution, modeling of this event would be difficult and likely inaccurate. However, given the significance of this boundary on this event the study of future thermal boundary interactions would be valuable. With the expanding private sector and the increase of privately owned surface stations more surface stations are becoming available for researchers. These additional sites will continue to make studies involving thermal boundaries more complete. With the increased data, numerical simulation of an event with a baroclinic boundary will become more likely. Furthermore, this data combined with some well-placed instruments such as Doppler wind lidars, additionaly Doppler radars, and other methods of measuring wind profiles, will allow future research to quantify the relative importance of baroclinically generated vorticity.

117

References Blanchard, D O., 2008: Interactions between a Supercell and a Quasi-Stationary Frontal Boundary. Mon. Wea. Rev., 136, 5199–5210. Bluestein, H. B., G. T. Marx, M. H. Jain, 1987: Formation of Mesoscale Lines of Precipitation: Nonsevere Squall Lines in Oklahoma during the Spring. Mon. Wea. Rev., 115, 2719–2727. Brooks, H.E., C. A. Doswell, and R. P. Davies-Jones, 1993: Environmental helicity and the Evolution of low-level mesocyclones, The Tornado: its Structure, Dynamics, Prediction, and Hazards, Geophys. Monogr., No 79, Amer. Geophys Union, 97104. Brooks, H. E., C. A. Doswell, J. Cooper, 1994: On the Environments of Tornadic and Nontornadic Mesocyclones. Wea. Forecasting,9, 606–618. Bunkers, M. R. Hjelmfelt, and P. L. Smith, 2006: An observational examination of longlived supercells. Part I: Characteristics, evolution, and demise. Wea. Forecasting, 21, 673–688. Businger, S., W. H. Bauman, and G. F. Watson, 1991: The development of the Piedmont front and associated severe weather on 13 March 1986. Mon. Wea. Rev., 119, 2224-2251. Chisholm, A. J., and J. H. Renick, 1972: The kinematics of multicell and supercell Alberta Hailstorm. Albertahail studies, Research Council of Alberta Hail Studies, Rep. 72-2, 24-31, 53 pp. Davies, J. M., C. A. Doswell, D. W. Burgess, J. F. Weaver, 1994: Some Noteworthy Aspects of the Hesston, Kansas, Tornado Family of 13 March 1990. Bull. Amer. Meteor. Soc.,75, 1007–1017. Davies-Jones, R., 1984: Streamwise Vorticity: The Origin of Updraft Rotation in Supercell Storms. J. Atmos. Sci., 41, 2991–3006. Davies-Jones, R.P., and H. Brooks, 1993; Mesocyclogenesis from a theoretical perspective. Geophysical Monograph, The tornado, Its structure, dynamics, and hazards. 79, 105-117. Davies-Jones, R. P., R. J. Trapp, and H. B. Bluestein, 2001: Tornadoes and Tornadic Storms. Meteor. Monogr., No. 50, Amer. Meteor. Soc., 167-221. Dostalek, J. F., John F. Weaver, G. L. Phillips, 2004: Aspects of a Tornadic Left-Moving Thunderstorm of 25 May 1999. Wea. Forecasting,19, 614–626.

118

Doswell, C.A. III, 1982: The operational meteorology of convective weather. Vol. I: Operational mesoanalysis. NOAA Tech, Memo. NWS NSSFC-5 [NTIS Accession No. PB83-162321, 158 pp. Doswell, C. A., R. Edwards, R. L. Thompson, J. A. Hart, K. C. Crosbie, 2006: A Simple and Flexible Method for Ranking Severe Weather Events.Wea. Forecasting, 21, 939–951. Fujita, T. T.,1963: Analytical mesometeorology: A review. Meteor. Monogr., 5, No. 27, 77-125. Fujita, T. T. 1978, Manual of downburst identification for project NIMROD. Te Mesometeorol. Res. Pap. No. 117, Dep. Geophys. Sci. University of Chicago Galway, J. G., 1989: The Evolution of Severe Thunderstorm Criteria within the Weather Service. Wea. Forecasting, 4, 585–592. Grzych, M. L., B. D. Lee, C. A. Finley, 2007: Thermodynamic Analysis of Supercell Rear-Flank Downdrafts from Project ANSWERS. Mon. Wea. Rev., 135, 240–246 Johns, R. H., C. A. Doswell, 1992: Severe Local Storms Forecasting. Wea. Forecasting, 7, 588–612. Johns, J. H., J. M. Davies, P. W. Leftwich,. (1993): Some Wind and Instability Parameters Associated With Strong and Violent Tornadoes: 2. Variations in the Combinations of Wind And Instability Parameters Johnson, R. H., P. J. Hamilton, 1988: The Relationship of Surface Pressure Features to the Precipitation and Airflow Structure of an Intense Midlatitude Squall Line. Mon. Wea. Rev., 116, 1444–1473 King, P. W. S., M. J. Leduc, D. M. L. Sills, N. R. Donaldson, D. R. Hudak, P. Joe, B. P. Murphy, 2003: Lake Breezes in Southern Ontario and Their Relation to Tornado Climatology. Wea. Forecasting, 18, 795–807 Klemp, J. B., 1987: Dynamics of tornadic thunderstorms. Annu. Rev. Fluid Mech., 19, 369-402. Knupp, K. R.,T. A. Murphy, T. A. Coleman, R. A. Wade, S. A. Mullins, C. J. Schultz, E. V. Schultz, L. D. Carey, A. T. Sherrer, E. W. McCaul, B. Carcione, S. Latimer, A. Kula, K. Laws, P. T. Marsh, K. Klockow, 2013: Meteorological Overview of the Devastating 27 April 2011 Tornado Outbreak. Bull. of the Amer. Met. Soc. Korotky, W. D., 1990: The Raleigh tornado of November 28, 1988: The evolution of a tornadic environment. Preprints, 16th Conf. Severe Local Storms (Kananaskis Park, Alberta), Amer. Meteor. Soc., 532-537. 119

Korotky W., R. W. Przybylinski, J. A. Hart, (1993) The Plainfield, Illinois, Tornado of August 28, 1990: the Evolution Of Synoptic and Mesoscale Environments. The Tornado: Its Structure, Dynamics, Prediction and Hazards, Geophys. Monogr.No. 79, Amer. Geophy. Union, 611-624. Keeter, K. K., S. Businger, L. G. Lee, J. S. Waldstreicher, 1995: Winter weather forecasting throughout the eastern United States. Part III: The effects of topography and the variability of winter weather in the Carolinas and Virginia. Wea. Forecasting, 10, 42-60. Kerr, B. W., G. L. Darkow, 1996: Storm-Relative Winds and Helicity in the Tornadic Thunderstorm Environment. Wea. Forecasting, 11, 489–505. Langmaid, A. H., A. J. Riordan, 1998: Surface Mesoscale Processes during the 1994 Palm Sunday Tornado Outbreak. Mon. Wea. Rev., 126, 2117–2132. Lawrence, M. G., 2005: The Relationship between Relative Humidity and the Dewpoint Temperature in Moist Air: A Simple Conversion and Applications. Bull. Amer. Meteor. Soc., 86, 225–233. Lemon, L. R., C. A. Doswell, 1979: Severe Thunderstorm Evolution and Mesocyclone Structure as Related to Tornadogenesis. Mon. Wea. Rev., 107, 1184–1197. Maddox, R. A., L. R. Hoxit, C. F. Chappell, 1980: A Study of Tornadic Thunderstorm Interactions with Thermal Boundaries. Mon. Wea. Rev., 108, 322–336. Maddox, R., 1980: Mesoscale convective complexes, Bull. Amer. Meteor. Soc. 61, 13741387. Magor, B. W., 1959: Mesoanalysis: Some operational analysis techniques utilized in tornado forecasting. Bull Amer. Meteor. Soc., 40, 499-511 . Markowski, P. M., J. M. Straka, E. N. Rasmussen, D. O. Blanchard, 1998: Variability of Storm-Relative Helicity during VORTEX. Mon. Wea. Rev., 126, 2959–2971. Markowski, P. M., E. N. Rasmussen, J. M. Straka, 1998: The Occurrence of Tornadoes in Supercells Interacting with Boundaries during VORTEX-95. Wea. Forecasting, 13, 852–859. Markowski, P. M., 2002: Hook Echoes and Rear-Flank Downdrafts: A Review. Mon. Wea. Rev., 130, 852–876. Markowski, P. M., J. M. Straka, E. N. Rasmussen, 2002: Direct Surface Thermodynamic Observations within the Rear-Flank Downdrafts of Nontornadic and Tornadic Supercells. Mon. Wea. Rev., 130, 1692–1721. 120

Markowski, P., and Y. Richardson, 2010: Mesoscale Meteorology in Midlatitudes. Oxford: Wiley-Blackwell, 407 pp. Matejka, T., R. C. Srivastava, 1991: An Improved Version of the Extended VelocityAzimuth Display Analysis of Single-Doppler Radar Data. J. Atmos. Oceanic Technol., 8, 453–466. Marwitz, J. D., 1972: The Structure and Motion of Severe Hailstorms. Part I: Supercell Storms. J. Appl. Meteor., 11, 166–179. Marwitz, J. D., 1972: The Structure and Motion of Severe Hailstorms. Part II: Multi-Cell Storms. J. Appl. Meteor., 11, 180–188. Marwitz, J. D., 1972: The Structure and Motion of Severe Hailstorms. Part III: Severely Sheared Storms. J. Appl. Meteor., 11, 189–201. Moller, A. R., C. A. Doswell, M. P. Foster, G. R. Woodall, 1994: The Operational Recognition of Supercell Thunderstorm Environments and Storm Structures. Wea. Forecasting, 9, 327–347. Miller, L. J., J. D. Tuttle, and C. A. Knight, 1988: Airflow and hail growth in a severe northern High Plains supercell. J. Atmos. Sci., 45, 736-762. Parker, M. D., S. A. Rutledge, 2000:Organizational modes of midlattitude mesoscale convective systems. J. Atmos. Sci. 61, 545-567. Petty, G., 2008: A First Course in Atmospheric Thermodynamics. Sundog Publishing, 337 pp. Purdom, J. F. W., 1976: Some Uses of High-Resolution GOES Imagery in the Mesoscale Forecasting of Convection and Its Behavior. Mon. Wea. Rev., 104, 1474–1483. Rasmussen, E. N., D. O. Blanchard, 1998: A Baseline Climatology of Sounding-Derived Supercell and Tornado Forecast Parameters. Wea. Forecasting, 13, 1148–1164. Rasmussen, E. N., S. Richardson, J. M. Straka, P. M. Markowski, D. O. Blanchard, 2000: The Association of Significant Tornadoes with a Baroclinic Boundary on 2 June 1995. Mon. Wea. Rev., 128, 174–191. Rasmussen, E. N., 2003: Refined Supercell and Tornado Forecast Parameters. Wea. Forecasting, 18, 530–535. Rotunno, R., J. B. Klemp, 1982: The Influence of the Shear-Induced Pressure Gradient on Thunderstorm Motion. Mon. Wea. Rev., 110, 136–151.

121

Rotunno, R., Jo. Klemp, 1985: On the Rotation and Propagation of Simulated Supercell Thunderstorms. J. Atmos. Sci., 42, 271–292. Sanders, F., C. A. Doswell, 1995: A Case for Detailed Surface Analysis. Bull. Amer. Meteor. Soc., 76, 505–521. Simmons, K.M., D. Sutter, and R. Pielke, 2012: Blown away: monetary and human impacts of the 2011 U.S. tornadoes. Extreme events and insurance: 2011 annus horribilis (C Courbage and W.R. Stahel, eds.), pp. 107-120. Schlesinger, R. E. 1982: Effects of mesoscale lifting, precipitation and boundary-layer shear on severe storm dynamics in a three-dimensional numerical modeling study. Preprints 12th Conf. Severe Local Storms, San Antonio, Amer Meteor. Soc., 536541 Trapp, R. J., M. L. Weisman, 2003: Low-Level Mesovortices within Squall Lines and Bow Echoes. Part II: Their Genesis and Implications.Mon. Wea. Rev., 131, 2804– 2823. Tripoli, G. J., W. R. Cotton, 1980: A Numerical Investigation of Several Factors Contributing to the Observed Variable Intensity of Deep Convection over South Florida. J. Appl. Meteor., 19, 1037–1063. Vescio, M. D., Keeter. K. K., Dial, G., Badgett, P., 1993: A low-top weak-reflectivity severe weather episode along a thermal/moisture boundary in eastern North Carolina. Preprints, 17th Conference on Severe Local Storms, St Louis, Amer. Meteor. Soc., 628-632 Wakimoto, R. M., N. T. Atkins, 1996: Observations on the Origins of Rotation: The Newcastle Tornado during VORTEX 94. Mon. Wea. Rev., 124, 384–407. Wakimoto, R. M., C. Liu, 1998: The Garden City, Kansas, Storm during VORTEX 95. Part II: The Wall Cloud and Tornado. Mon. Wea. Rev., 126, 393–408. Weaver, J. F., S. P. Nelson, 1982: Multiscale Aspects of Thunderstorm Gust Fronts and Their Effects on Subsequent Storm Development.Mon. Wea. Rev., 110, 707–718. Weisman, M. L., J. B. Klemp, 1982: The Dependence of Numerically Simulated Convective Storms on Vertical Wind Shear and Buoyancy. Mon. Wea. Rev., 110, 504–520. Weisman, M. L., J. B. Klemp, R. Rotunno, 1988: Structure and Evolution of Numerically Simulated Squall Lines. J. Atmos. Sci., 45, 1990–2013.

122

Weisman, M. L., 1992: The Role of Convectively Generated Rear-Inflow Jets in the Evolution of Long-Lived Mesoconvective Systems. J. Atmos. Sci., 49, 1826– 1847. Weisman, M. L., 1993: The Genesis of Severe, Long-Lived Bow Echoes. J. Atmos. Sci., 50, 645–670. Weisman, M. L., R. J. Trapp, 2003: Low-Level Mesovortices within Squall Lines and Bow Echoes. Part I: Overview and Dependence on Environmental Shear. Mon. Wea. Rev., 131, 2779–2803.

123