1. A Comprehensive Study of Cool Season Tornadoes in the Southeast United States Alicia C. Wasula July 13, 2005

2. Committee Members Lance BosartLance Bosart John MolinariJohn Molinari Dan KeyserDan Keyser Chris ThorncroftChris Thorncroft Russ SchneiderRuss Schneider Morris WeismanMorris Weisman

3. The Problem Nocturnal peak in tornado occurrence during cool season along Gulf coastNocturnal peak in tornado occurrence during cool season along Gulf coast Difficulty warning public increased injury and loss of life (e.g., Fike 1993)Difficulty warning public increased injury and loss of life (e.g., Fike 1993) Difficult to discriminate between true threats and null events in real timeDifficult to discriminate between true threats and null events in real time

4. Goals Provide evidence that this phenomenon existsProvide evidence that this phenomenon exists Provide possible physical explanations via:Provide possible physical explanations via: –Climatological analysis of low-level winds –Composite analysis of southeast US tornado episodes –Case study: 22-23 February 1998 Relate results to tornadogenesis mechanismsRelate results to tornadogenesis mechanisms

5. Outline Historical BackgroundHistorical Background Review of low-level mesocyclone-genesis mechanismsReview of low-level mesocyclone-genesis mechanisms ClimatologyClimatology Composite resultsComposite results Case Study: 22-23 February 1998Case Study: 22-23 February 1998 ConclusionsConclusions Future WorkFuture Work

6. Background Cool season nocturnal tornado peak exists along Gulf coast (e.g. Knupp and Garinger 1993, Brooks et al. 2003)Cool season nocturnal tornado peak exists along Gulf coast (e.g. Knupp and Garinger 1993, Brooks et al. 2003) –Peak exists during early morning hours ~ 1200 UTC Can occur in low CAPE/high shear environments Return Flow air trajectory effects PBL thermodynamics

7. Background Cool season nocturnal tornado peak exists along Gulf coast (e.g. Knupp and Garinger 1993, Brooks et al. 2003) Can occur in low CAPE/high shear environmentsCan occur in low CAPE/high shear environments Return Flow air trajectory effects PBL thermodynamics

8. 0-2 km positive mean shear (x 10 -3 s -1 ) Convective Available Potential Energy (CAPE, J/kg) Johns et al. 1993 CAPE vs. Shear for Cold Season Tornado Cases

9. Background Cool season nocturnal tornado peak exists along Gulf coast (e.g. Knupp and Garinger 1993, Brooks et al. 2003) Can occur in low CAPE/high shear environments Return Flow air trajectory effects PBL thermodynamicsReturn Flow air trajectory effects PBL thermodynamics

10. Types of Return Flow (Crisp and Lewis 1992)

11. Previous Composite Study: Peninsular Florida Tornado Environments (Hagemeyer and Schmocker 1991) 1) Thermodynamics Dry Season Wet Season w All non-TC tornadoes ± 2 h of 00/12Z 1980-1988 Dry season w = ~5°C w = ~5°C Wet season w = ~1°C w = ~1°C

12. Previous Composite Study: Peninsular Florida Tornado Environments (Hagemeyer and Schmocker 1991) Dry Season Wet Season 2) Shear u v All non-TC tornadoes ± 2 h of 00/12Z 1980-1988

13. Question #1: How does buoyancy/shear relationship effect convective mode?

14. High, Dry High, Moist Mod. Low 0102030405060 OrdinaryMulticellSupercell Relationship of Buoyancy/Shear to Convective Mode CAPE (J/kg) Shear (m/s)

15. Question #2: Given that shear is sufficient for supercells, how does hodograph shape determine supercell type?

16. Splitting Storm in Sheared Flow Klemp (1987)

18. Straight Hodograph Left & right mover equally dominant Left mover faster Source: A Convective Storm Matrix http://meted.ucar.edu/convectn/csmatrix/

19. Clockwise ¼ turn Hodograph Right mover dominant Left mover dissipates Source: A Convective Storm Matrix http://meted.ucar.edu/convectn/csmatrix/

20. Counterclockwise ¼ turn Hodograph Left mover dominant Right mover dissipates Source: A Convective Storm Matrix http://meted.ucar.edu/convectn/csmatrix/

21. Question #3:How does near- surface hodograph shape effect low-level mesocyclone generation?

22. Source of low-level rotation: baroclinic generation of horizontal vorticity in FFD stretched in updraft FFD Source: A Convective Storm Matrix http://meted.ucar.edu/convectn/csmatrix/

23. Streamwise Vorticity Maximized when vorticity vector and velocity vector are parallelMaximized when vorticity vector and velocity vector are parallel 0 when vorticity vector and velocity vector are perpendicular0 when vorticity vector and velocity vector are perpendicular

24. Wicker (1996) HorizontalVorticityVector ( H ) StormMotionVector

25. Wicker (1996) Vertical Velocity and Horizontal Winds At 50 m, 110 min EnvironmentalWinds Storm Inflow

26. Wicker (1996) Pert. Temperature and Horizontal Vorticity At 50 m, 110 min FFD H adds to env. H FFD H adds to env. H H Parallel to H Parallel to inflow at FFD inflow at FFD Large streamwise Large streamwise vorticity vorticity X Updraft

27. Previous work has shown that supercells interacting with boundaries can produce long- lived tornadoes e.g., Maddox et al. 1980, Markowski et al. 1998, Rasmussen et al. 2000

28. Maddox et al. 1980 Along outflow boundary (C), streamwise vorticity is maximized

29. Climatology

30. Tornado Climatology Southeast US Domain

31. Storm Data Tornado Database 1950-2001Tornado Database 1950-2001 Corrected for duplicate reportsCorrected for duplicate reports Use only F2 or greater tornadoes for climatologyUse only F2 or greater tornadoes for climatology All tornadoes considered for compositesAll tornadoes considered for composites Cool season tornadoes (November – March)Cool season tornadoes (November – March)

33. By month By month By Hour (UTC) By Hour (UTC)

34. 1000 m Wind Climatology Pilot Balloon Stations

35. Pilot Ballon (Pibal) Dataset 1948-19571948-1957 4x/day (0300, 0900, 1500, and 2100 UTC)4x/day (0300, 0900, 1500, and 2100 UTC) Wind data for lowest ~ 3 kmWind data for lowest ~ 3 km No thermodynamic dataNo thermodynamic data No winds above cloud baseNo winds above cloud base

36. 1000 m Wind Climatology

38. Surface Wind Climatology Surface Stations

39. Surface Observations Locally archived at University at AlbanyLocally archived at University at Albany HourlyHourly Cool seasons 1995-96 to 1999-2000Cool seasons 1995-96 to 1999-2000 Subset of stations for which dataset is completeSubset of stations for which dataset is complete

40. Sample Windroses Coastal Station Coastal Station Inland Station Inland Station

41. Coastal Stations (n=16)

42. Inland Stations (n=17)

43. Hypothesis More cooling / higher pressures over land at nightMore cooling / higher pressures over land at night Less cooling / lower pressures over ocean at nightLess cooling / lower pressures over ocean at night Easterly geostrophic component at surfaceEasterly geostrophic component at surface

44. MORE COOLING AT NIGHT H LAND OCEAN LESS COOLING AT NIGHT L VgVg

45. Eta NARR Dataset 32 km reanalysis32 km reanalysis November – March 1999-2000November – March 1999-2000 0900-2100 UTC sea level pressure difference0900-2100 UTC sea level pressure difference

46. Nov-Mar 1999-2000 09–21 UTC SLP Diff. (hPa) – Eta NARR

47. Summary: Climatology Cool season tornado peak in southeast USCool season tornado peak in southeast US Nocturnal maximum is strongest near Gulf of Mexico coastNocturnal maximum is strongest near Gulf of Mexico coast 1000 m nocturnal LLJ is strongest near Gulf coast1000 m nocturnal LLJ is strongest near Gulf coast Surface winds along coastline back at night due to more (less) cooling over land (water)Surface winds along coastline back at night due to more (less) cooling over land (water)

48. Composites

49. Tornado Episode Composites NCEP/NCAR 2.5 x 2.5 Reanalysis datasetNCEP/NCAR 2.5 x 2.5 Reanalysis dataset All tornado episodes 1950-2001All tornado episodes 1950-2001 Episode-relative compositeEpisode-relative composite Grouped by start time: 0000-0600, 0600- 1200, 1200-1800, 1800-0000 UTCGrouped by start time: 0000-0600, 0600- 1200, 1200-1800, 1800-0000 UTC Will show only TWO:Will show only TWO: –DAY: 1800-0000 UTC (n=327) –NIGHT: 0600-1200 UTC (n=130)

50. DAYNIGHT 200 hPa height (m), isotachs (m s -1 )

51. DAYNIGHT 1000 hPa height (m), 1000-500 hPa thck. (dam), 700 hPa RH (%)

52. DAYNIGHT 700 hPa height (m), vertical motion (x 10 -3 hPa s -1 )

53. DAYNIGHT 850 hPa winds, 850 hPa e (K), 850-500 hPa lapse rate ( C)

54. DAYNIGHT CAPE 590 J kg -1 232 J kg -1 LCL 965 hPa 977 hPa 0-6 km shear 19 m s -1 20 m s -1 0-2 km shear 8 m s -1 8 m s -1

55. Surface Composite - Methodology Bin surface obs in 1° x 1° boxes relative to first tornado reportBin surface obs in 1° x 1° boxes relative to first tornado report Calculate temp/dew point anomalies relative to monthly climo for that stationCalculate temp/dew point anomalies relative to monthly climo for that station Composite u, v, PMSL, temp/dew point anomalies for each grid boxComposite u, v, PMSL, temp/dew point anomalies for each grid box

56. T PMSL Surface Composite – All Events X T T d v p u n 8°C T d 10°C T d

57. DAY NIGHT X X T T d v p u n T PMSL 8°C T d 10°C T d

58. Summary: Composites Strong signal in spite of large sample size:Strong signal in spite of large sample size: –ULJ entrance region at 200 hPa –Vigorous upstream trough at 500 hPa –Southwesterly LLJ at 850 hPa –Low-level e ridge Surface composites show 1 st tornado occurs:Surface composites show 1 st tornado occurs: –At strongest T –On northern edge of moisture surge/southerly flow

59. Case Study: 22-23 February 1998 Central Florida Tornado Outbreak

60. Case Overview Nocturnal outbreak over central Florida (~0000- 0600 UTC)Nocturnal outbreak over central Florida (~0000- 0600 UTC) 42 fatalities/260 injuries42 fatalities/260 injuries Ample shear/instability south of surface front in central FLAmple shear/instability south of surface front in central FL System was disorganized over Gulf of Mexico but rapidly intensified as it moved onshoreSystem was disorganized over Gulf of Mexico but rapidly intensified as it moved onshore Rapidly evolving threat + nocturnal nature made warning public difficultRapidly evolving threat + nocturnal nature made warning public difficult

62. Storm Reports + Wind > 26 m s -1 Hail > 2 cm Tornado TBW MCO GNV

63. 1000 hPa hgt, 1000-500 hPa THCK 500 hPa HGHT, AVOR 200 hPa HGHT, Isotachs 850 hPa HGHT, e, Isotachs 23 February 1998 0000 UTC

64. 22/1815 UTC

65. 23/0015 UTC

66. CAPE = 2891 J/kg LI = -9 C LCL = 962 hPa TBW Sounding and Hodograph 23/0000 UTC 925 850 700 500 T M G

67. SST and SST 22 February 1998

69. m s -1 dBZ 23/0156 UTC

70. m s -1 dBZ 23/0336 UTC

71. m s -1 dBZ 23/0515 UTC

72. Question Why did the disorganized convective line intensify so rapidly after making landfall on the Florida peninsula?Why did the disorganized convective line intensify so rapidly after making landfall on the Florida peninsula?

73. 23/0000 UTC Surface Analysis Lightning for ½ hour Ending 23/0000 UTC

74. GNV/MCO Meteograms e (K) T M G

75. Surface Frontogenesis 22/2100 UTC 23/0000 UTC

76. Surface Frontogenesis 23/0200 UTC 23/0400 UTC

77. Surface Vorticity 23/0000 UTC 23/0400 UTC

78. Maddox et al. 1980

79. Summary: Case Study Costliest/deadliest tornado outbreak in FL historyCostliest/deadliest tornado outbreak in FL history Convective intensity related to SST in Gulf of MexicoConvective intensity related to SST in Gulf of Mexico Rapid supercell development onshore related to diabatically induced frontRapid supercell development onshore related to diabatically induced front –Inferred enhanced ascent + backed winds Ample instability/strong shear south of front (TBW sounding)Ample instability/strong shear south of front (TBW sounding)

80. Summary: Case Study Surface vorticity at front increased rapidly as convective line moved onshoreSurface vorticity at front increased rapidly as convective line moved onshore Hypothesis: Storms ingested inflow from highly helical (i.e. large streamwise vorticity) air as they came onshore at front rapid dev. of low-level rotationHypothesis: Storms ingested inflow from highly helical (i.e. large streamwise vorticity) air as they came onshore at front rapid dev. of low-level rotation Numerical models (e.g., Eta) failed to capture shallow but strong surface frontNumerical models (e.g., Eta) failed to capture shallow but strong surface front

81. Conclusions

82. Conclusions: Nocturnal cool season tornado phenomenon along Gulf coast existsNocturnal cool season tornado phenomenon along Gulf coast exists Climatologically favorable 0-1 km shear profile exists along coastline at night:Climatologically favorable 0-1 km shear profile exists along coastline at night: –Nocturnal LLJ maximized at coast at ~ 1000 m –Surface winds back along coastline at night This shear profile is enhanced when mid-latitude system moves across southeast US (e.g. isallobaric wind w/surface low to west enhanced backing in warm air + strong LLJThis shear profile is enhanced when mid-latitude system moves across southeast US (e.g. isallobaric wind w/surface low to west enhanced backing in warm air + strong LLJ

83. Conclusions contd: Composites show common features:Composites show common features: –ULJ entrance region at ~ 200 hPa –Vigorous upstream trough at 500 hPa –LLJ and e ridge at 850 hPa –Surface cyclone to west –Presence of highly anomalous (> 8 C) surface moisture –Surface front or boundary nearby –Stronger features during nocturnal episodes

84. Schematic Cool Season Southeast US Tornado Setup

85. Conclusions contd: Case study showed features similar to composite:Case study showed features similar to composite: –Strong upper-level system to west –Strengthening surface front (diabatically generated even after sunset) –Anomalous moisture & LLJ at 850 hPa

86. Cold Season (n=75)Warm Season (n=69) Johns et al. (1993)

87. Cold Season Warm Season (Southeast)(Midwest) Lower CAPELower CAPE Less steep lapse ratesLess steep lapse rates Higher shearHigher shear Highly anomalous low-level moistureHighly anomalous low-level moisture Higher CAPEHigher CAPE More steep lapse rateMore steep lapse rate Lower shearLower shear

88. Future Work More detailed investigation of null eventsMore detailed investigation of null events –Composite based upon high risk outlook location –What are good discriminators between null event/true threat? CAPE distribution (elevated vs. sfc. based)CAPE distribution (elevated vs. sfc. based) 0-1 km shear profile (degree of low-level backing)0-1 km shear profile (degree of low-level backing) Presence/orientation of sfc. boundariesPresence/orientation of sfc. boundaries What about other types of severe weather (e.g., high wind, hail, flash flooding)?What about other types of severe weather (e.g., high wind, hail, flash flooding)? How do ingredients necessary for southeast US nocturnal tornado episode compare to warm season Great Plains tornado episode?How do ingredients necessary for southeast US nocturnal tornado episode compare to warm season Great Plains tornado episode? –More moisture at low-levels –Lower CAPE, higher 0-6 km shear

89. Acknowledgements LanceLance Committee: Dan Keyser, John Molinari, Chris Thorncroft, Russ Schneider, Morris WeismanCommittee: Dan Keyser, John Molinari, Chris Thorncroft, Russ Schneider, Morris Weisman COMETCOMET Grant Participants: Russ Schneider, Steve Weiss, Bob Johns, Geoff Manikin, Pat WelshGrant Participants: Russ Schneider, Steve Weiss, Bob Johns, Geoff Manikin, Pat Welsh

90. Acknowledgements (contd) Celeste, Diana, Lynn, SharonCeleste, Diana, Lynn, Sharon Grad Students, past and presentGrad Students, past and present Sheryl (Honikman) Thorp, Eyad Atallah, Kristen Corbosiero, Mike Notaro, Teresa Bals-Elsholz, Dan Meade, Tom Galarneau, Anantha AiyyerSheryl (Honikman) Thorp, Eyad Atallah, Kristen Corbosiero, Mike Notaro, Teresa Bals-Elsholz, Dan Meade, Tom Galarneau, Anantha Aiyyer My Family: my parents and brotherMy Family: my parents and brother Tom and CatherineTom and Catherine