• Bogner, P. B., , Barnes G. M. , , and Franklin J. L. , 2000: Conditional instability and shear for six hurricanes over the Atlantic Ocean. Wea. Forecasting, 15 , 192207.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bosart, B. N., , Lee W-C. , , and Wakimoto R. M. , 2002: Procedures to improve the accuracy of airborne Doppler radar data. J. Atmos. Oceanic Technol., 19 , 322339.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brandes, E. A., 1984: Relationships between radar-derived thermodynamic variables and tornadogenesis. Mon. Wea. Rev., 112 , 10331052.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bunkers, M. J., , Klimowski B. A. , , Zeitler J. W. , , Thompson R. L. , , and Weisman M. L. , 2000: Predicting supercell motion using a new hodograph technique. Wea. Forecasting, 15 , 6179.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Curtis, L., 2004: Midlevel dry intrusions as a factor in tornado outbreaks associated with land-falling tropical cyclones from the Atlantic and Gulf of Mexico. Wea. Forecasting, 19 , 411427.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davies, J. M., 1990: Midget supercell spawns tornadoes. Weatherwise, 43 (10) 260261.

  • Davies, J. M., 2006: Hurricane and tropical cyclone tornado environments from RUC proximity soundings. Preprints, 23rd Conf. on Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., 12.6A. [Available online at http://ams.confex.com/ams/pdfpapers/115483.pdf.].

    • Search Google Scholar
    • Export Citation
  • Davies-Jones, R., 1984: Streamwise vorticity: The origin of updraft rotation in supercell storms. J. Atmos. Sci., 41 , 29913006.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davies-Jones, R., , Trapp R. J. , , and Bluestein H. B. , 2001: Tornadoes and tornadic storms. Severe Convective Storms, Meteor. Monogr., No. 50, Amer. Meteor. Soc., 167–221.

    • Search Google Scholar
    • Export Citation
  • Doswell C. A. III, , , and Burgess D. W. , 1993: Tornadoes and tornadic storms: A review of conceptual models. The Tornado: Its Structure, Dynamics, Prediction and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 161–172.

    • Search Google Scholar
    • Export Citation
  • Droegemeier, K. K., , Lazarus S. M. , , and Davies-Jones R. , 1993: The influence of helicity on numerically simulated convective storms. Mon. Wea. Rev., 121 , 20052029.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Edwards, R., , and Pietrycha A. E. , 2006: Archetypes for surface baroclinic boundaries influencing tropical cyclone tornado occurrences. Preprints, 23rd Conf. on Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., P8.2.

    • Search Google Scholar
    • Export Citation
  • Franklin, J. L., , Pasch R. J. , , Avila L. A. , , Beven J. L. II, , Lawrence M. B. , , Stewart S. R. , , and Blake E. S. , 2006: Atlantic hurricane season of 2004. Mon. Wea. Rev., 134 , 9811025.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fujita, T. T., , Watanabe K. , , Tsuchiya K. , , and Shimada M. , 1972: Typhoon-associated tornadoes in Japan and new evidence of suction vortices in a tornado near Tokyo. J. Meteor. Soc. Japan, 50 , 431453.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gamache, J. F., 1997: Evaluation of a fully three-dimensional variational Doppler analysis technique. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 422–423.

    • Search Google Scholar
    • Export Citation
  • Gamache, J. F., , Marks F. D. Jr., , and Roux F. , 1995: Comparison of three airborne Doppler sampling techniques with airborne in situ wind observations in Hurricane Gustav (1990). J. Atmos. Oceanic Technol., 12 , 171181.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gentry, R. C., 1983: Genesis of tornadoes associated with hurricanes. Mon. Wea. Rev., 111 , 17931805.

  • Greene, D. R., , and Clark R. A. , 1972: Vertically integrated liquid water—A new analysis tool. Mon. Wea. Rev., 100 , 548552.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hagemeyer, B. C., 1998: Significant tornado events associated with tropical and hybrid cyclones in Florida. Preprints, 16th Conf. on Weather Analysis and Forecasting, Phoenix, AZ, Amer. Meteor. Soc., 4–6.

    • Search Google Scholar
    • Export Citation
  • Hill, E. L., , Malkin W. , , and Shulz W. A. Jr., 1966: Tornadoes associated with cyclones of tropical origin—Practical features. J. Appl. Meteor., 5 , 745763.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jorgensen, D. P., 1984: Mesoscale and convective-scale characteristics of mature hurricanes. Part I: General observations by research aircraft. J. Atmos. Sci., 41 , 12681285.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jorgensen, D. P., , Hildebrand P. H. , , and Frusch C. L. , 1983: Feasibility test of an airborne pulse-Doppler meteorological radar. J. Climate Appl. Meteor., 22 , 744757.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kennedy, P. C., , Westcott N. E. , , and Scott R. W. , 1993: Single-Doppler radar observations of a minisupercell tornadic thunderstorm. Mon. Wea. Rev., 121 , 18601870.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Klemp, J. B., 1987: Dynamics of tornadic thunderstorms. Annu. Rev. Fluid Mech., 19 , 369402.

  • Klemp, J. B., , Wilhelmson R. B. , , and Ray P. S. , 1981: Observed and numerically simulated structure of a mature supercell thunderstorm. J. Atmos. Sci., 38 , 15581580.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Markowski, P. M., , Rasmussen E. N. , , and Straka J. M. , 1998a: The occurrence of tornadoes in supercells interacting with boundaries during VORTEX-95. Wea. Forecasting, 13 , 852859.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Markowski, P. M., , Straka J. M. , , Rasmussen E. N. , , and Blanchard D. O. , 1998b: Variability of storm-relative helicity during VORTEX. Mon. Wea. Rev., 126 , 29592971.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Markowski, P. M., , Straka J. M. , , and Rasmussen E. N. , 2002: Direct surface thermodynamic observations within the rear-flank downdrafts of nontornadic and tornadic supercells. Mon. Wea. Rev., 130 , 16921721.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Markowski, P. M., , Hannon C. , , Frame J. , , Lancaster E. , , Pietrycha A. , , Edwards R. , , and Thompson R. L. , 2003: Characteristics of vertical wind profiles near supercells obtained from the Rapid Update Cycle. Wea. Forecasting, 18 , 12621272.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marks F. D. Jr., , , Houze R. A. Jr., , and Gamache J. F. , 1992: Dual-aircraft investigation of the inner core of Hurricane Norbert. Part I: Kinematic structure. J. Atmos. Sci., 49 , 919942.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , 1987: Observations of the hurricane “Danny” tornado outbreak of 16 August 1985. Mon. Wea. Rev., 115 , 12061223.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , 1991: Buoyancy and shear characteristics of hurricane-tornado environments. Mon. Wea. Rev., 119 , 19541978.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , 1993: Observations and simulations of hurricane-spawned tornadic storms. The Tornado: Its Structure, Dynamics, Prediction and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 119–142.

    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , , and Weisman M. L. , 1996: Simulation of shallow supercell storms in landfalling hurricane environments. Mon. Wea. Rev., 124 , 408429.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , , and Weisman M. L. , 2001: The sensitivity of simulated supercell structure and intensity to variations in the shapes of environmental buoyancy and shear profiles. Mon. Wea. Rev., 129 , 664687.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , , Buechler D. E. , , Goodman S. J. , , and Cammarata M. , 2004: Doppler radar and lightning network observations of a severe outbreak of tropical cyclone tornadoes. Mon. Wea. Rev., 132 , 17471763.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Molinari, J., , and Vollaro D. , 2008: Extreme helicity and intense convective towers in Hurricane Bonnie. Mon. Wea. Rev., 136 , 43554372.

  • NCDC, 2004: Storm Data. Vol. 46, No. 9, 262 pp. [Available from National Climatic Data Center, 151 Patton Ave., Asheville, NC 28801-5001.].

  • Novlan, D. J., , and Gray W. M. , 1974: Hurricane-spawned tornadoes. Mon. Wea. Rev., 102 , 476488.

  • Pearson, A. D., , and Sadowski A. F. , 1965: Hurricane-induced tornadoes and their distribution. Mon. Wea. Rev., 93 , 461464.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Powell, M. D., , and Houston S. H. , 1998: Surface wind fields of 1995 Hurricanes Erin, Opal, Luis, Marilyn, and Roxanne at landfall. Mon. Wea. Rev., 126 , 12591273.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rao, G. V., , Scheck J. W. , , Edwards R. , , and Schaefer J. T. , 2005: Structures of mesocirculations producing tornadoes associated with Tropical Cyclone Francis (1998). Pure Appl. Geophys., 162 , 16271641.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rasmussen, E. N., , and Blanchard D. O. , 1998: A baseline climatology of sounding-derived supercell and tornado forecast parameters. Wea. Forecasting, 13 , 11481164.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reasor, P. D., , Montgomery M. T. , , Marks F. D. Jr., , and Gamache J. F. , 2000: Low-wavenumber structure and evolution of the hurricane inner core observed by airborne dual-Doppler radar. Mon. Wea. Rev., 128 , 16531680.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reasor, P. D., , Eastin M. D. , , and Gamache J. F. , 2009: Rapidly intensifying Hurricane Guillermo (1997). Part I: Low-wavenumber structure and evolution. Mon. Wea. Rev., 137 , 603631.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schneider, D., , and Sharp S. , 2007: Radar signatures of tropical cyclone tornadoes in central North Carolina. Wea. Forecasting, 22 , 278286.

  • Smith, J. S., 1965: The hurricane-tornado. Mon. Wea. Rev., 93 , 453459.

  • Spratt, S. M., , Sharp D. W. , , Welsh P. , , Sandrik A. , , Alsheimer F. , , and Paxton C. , 1997: A WSR-88D assessment of tropical cyclone outer rainband tornadoes. Wea. Forecasting, 12 , 479501.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Suzuki, O., , Niino H. , , Ohno H. , , and Nirasawa H. , 2000: Tornado-producing minisupercells associated with Typhoon 9019. Mon. Wea. Rev., 128 , 18681882.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Testud, J., , Hildebrand P. H. , , and Lee W-C. , 1995: A procedure to correct airborne Doppler radar data for navigation errors using the echo returned from the earth's surface. J. Atmos. Oceanic Technol., 12 , 800820.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., , Edwards R. , , Hart J. A. , , Elmore K. L. , , and Markowski P. , 2003: Close proximity soundings with supercell environments obtained from the Rapid Update Cycle. Wea. Forecasting, 18 , 12431261.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., , Edwards R. , , and Mead C. M. , 2004: An update to the supercell composite and significant tornado parameters. Preprints, 22nd Conf. on Severe Local Storms, Hyannis, MA, Amer. Meteor. Soc., P8.1. [Available online at http://ams.confex.com/ams/pdfpapers/82100.pdf.].

    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., , and Klemp J. B. , 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110 , 504520.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., , and Klemp J. B. , 1984: The structure and classification of numerically simulated convective storms in directionally varying wind shears. Mon. Wea. Rev., 112 , 24792498.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Tornado reports (derived from NCDC 2004) from 0800 UTC 15 Sep through 1400 UTC 16 Sep 2004. Of the 31 tornado reports, 20 were rated F0 in intensity, 9 were rated F1, and 2 were rated F2. Rectangles enclose the reports issued from four radar-analyzed storm cells that crossed over from sea to land (Fig. 7). The reports are numbered chronologically. The letters correspond to the cells identified in Fig. 6 (for A and B) and Fig. 7.

  • View in gallery

    Selected track segments and maximum surface winds for Hurricanes Ivan and Jeanne. The date and hour (dd/hh) are shown for selected 6-h locations (filled circles) for each storm. The maximum surface winds (m s−1) are shown for each 6-h location during the prelandfall periods examined for Ivan (red) and Jeanne (blue).

  • View in gallery

    Regional WSI NOWRAD composite of radar reflectivity (dBZ) (a) for Hurricane Ivan at 2100 UTC 15 Sep 2004 and (b) for Hurricane Jeanne at 0000 UTC 26 Sep 2004.

  • View in gallery

    Base-scan radar reflectivity (dBZ) from (a) KTLH at 1804 UTC 15 Sep 2004 and (b) KTBW at 1803 UTC 15 Sep 2004. The location of the dual-Doppler analysis in Fig. 5 is highlighted with a black square.

  • View in gallery

    (a) Radar reflectivity (color shading) and storm-relative dual-Doppler wind vectors at 2-km altitude in the target outer rainband of Hurricane Ivan at 1804 UTC 15 Sep 2004 located over 100 km offshore. (b) Radar reflectivity (grayscale shading), upward vertical velocity (red contours), and positive vertical vorticity (blue contours) at 2-km altitude. (c) Vertical velocity (red contours), storm-relative winds (black vectors), and horizontal vorticity (blue vectors) at 1-km altitude. Vectors are shown at every other grid point. (d) Vertical cross section along N–S of radar reflectivity (gray shading), vertical velocity (red), and vertical vorticity (blue). Radar reflectivity is shaded at the 0-, 10-, 20-, 30-, and 40-dBZ levels; vertical velocity is contoured at 2, 4, 6, and 8 m s−1; and vertical vorticity is contoured at 3, 5, 7, and 9 × 10−3 s−1. The three black boxes denote minisupercells A, B, and C (as referred to in the text; cf. Fig. 6). Reference vectors for the wind and horizontal vorticity are also shown.

  • View in gallery

    (a) Locations of minisupercells A, B, and C (cf. Fig. 5) between 1800 and 2100 UTC 15 Sep 2004 as determined from the KTLH WSR-88D. Radar reflectivity at 0.5° elevation for cells A, B, and C as viewed from KTLH during (b) the dual-Doppler analysis at 1804 UTC, (c) the first mesocyclone detection (in cell B) at 1945 UTC, and (d) the first tornado reported (from cell A) at 2044 UTC. Zoomed view of (e) radar reflectivity and (f) Doppler velocities at 0.5° elevation in cell A at 2044 UTC. The black box in (a) denotes the approximate dual-Doppler domain (Fig. 5). The black circulation markers (circles) denote radar-detected mesocyclones.

  • View in gallery

    (top) Temporal evolution of range from the KTLH radar and (bottom) 30-min-averaged midlevel rotational velocity (Vrot) for the four analyzed tornadic cells shown in Fig. 1. Cells A and B are those shown in Fig. 6; cells D and E occurred several hours later and are not shown in the other radar figures. The tornadoes produced by these cells are outlined in Fig. 1. For Vrot, the mean trend line is shown, with vertical bars indicating the range of instantaneous values. Time = 0 min represents the time of each cell’s landfall, with negative values representing min before cell landfall. The small triangles along the time axis point to the times of initial tornado reports after cell landfall for all four analyzed supercells. The large triangle represents the average initial tornado report time after cell landfall.

  • View in gallery

    Distribution by quadrant of the soundings used for the thermodynamic and wind analyses. The soundings are divided into two time windows (approximately 6 h each) for both Ivan and Jeanne. Within each quadrant, the number of land-based rawinsondes (from stations on the windward coast in the right-front quadrant only) and the number of sea-based research dropsondes are indicated, along with the group’s average distance from the storm center. Quadrants are defined with respect to the observed TC motion vectors; the right-front quadrant is highlighted since it is the basis for most of the comparisons in the text. The data for subsets of these soundings are in Figs. 9 –12 and Tables 1, 2, 4, and 5.

  • View in gallery

    Averaged hodographs (m s−1) for quadrants relative to storm motion of Ivan (sea only). Plotted numbers represent observing heights in km AGL. Related indices are given in Table 1.

  • View in gallery

    Averaged hodographs (m s−1) for distance (km) away from storm center in the RF quadrant of Ivan. Plotted numbers represent observing heights in km AGL. The range bins were defined such that there were a roughly even number of soundings in each bin, and so that the increments among the bins’ mean ranges were roughly the same. Related indices are given in Table 2. Note: the weakly negative (anticyclonic) near-surface values of Vtan at long range are mainly due to the subtraction of Ivan’s forward motion from RF quadrant soundings on the periphery of Ivan’s cyclonic circulation.

  • View in gallery

    Averaged hodographs (m s−1) for sea and land regimes in the RF quadrant of Ivan. Plotted numbers represent observing heights in km AGL. Related indices are given in Table 4. Note: the weakly negative near-surface values of Vtan over land are partly due to the subtraction of Ivan’s forward motion from several RF quadrant soundings at greater distance from Ivan’s center (cf. Fig. 10).

  • View in gallery

    Averaged hodographs (m s−1) for early (sea and land) and late (land only) time regimes (UTC) in the RF quadrant of Jeanne. Plotted numbers represent observing heights in km AGL. Related indices are given in Table 5.

  • View in gallery

    GOES water vapor imagery for (a) Hurricane Ivan at 1815 UTC 15 Sep 2004 and (c) Hurricane Jeanne at 0015 UTC 26 Sep 2004. Skew T diagrams for (b) the global positioning system (GPS) dropsonde deployed at 1807 UTC 15 Sep and (d) the Jacksonville, FL (KJAX), rawindsonde launched at 0000 UTC 26 Sep within the respective dry air intrusions. Black circles in (a) and (c) denote locations of representative soundings.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 157 157 12
PDF Downloads 111 111 10

Environmental Ingredients for Supercells and Tornadoes within Hurricane Ivan

View More View Less
  • 1 Department of Marine, Earth, and Atmospheric Sciences, North Carolina State University, Raleigh, North Carolina
  • | 2 Department of Geography and Earth Sciences, University of North Carolina at Charlotte, Charlotte, North Carolina
© Get Permissions
Full access

Abstract

Hurricane Ivan (2004) was a prolific producer of tornadoes as it made landfall on the U.S. Gulf Coast. Prior researchers have revealed that the tornadic cells within tropical cyclone (TC) rainbands are often supercellular in character. The present study investigates the utility of several common midlatitude, continental supercell and tornado diagnostic tools when applied to Hurricane Ivan’s tornado episode.

The environment within Hurricane Ivan was favorable for storm rotation. While well offshore, the bands of Hurricane Ivan possessed embedded cells with mesocyclones of moderate intensity. A dual-Doppler analysis reveals that the updrafts of these cells were highly helical in the lower troposphere, suggesting significant ingestion of streamwise environmental vorticity. These coherent cells were long lived and could be tracked for multiple hours. As the supercells over the Gulf of Mexico approached the coast during Ivan’s landfall, rapid increases in midlevel vorticity and vertically integrated liquid (VIL) occurred. Based on compiled severe weather reports, these increases in storm intensity appear often to have immediately preceded tornadogenesis.

The local environment for supercells in Ivan’s interior is evaluated through the use of 62 soundings from the operational land-based network and from research flights. There were substantial differences in the thermodynamic profiles and wind profiles at differing ranges from Ivan’s center, from quadrant to quadrant of Ivan’s circulation, and between land and sea. The most optimal environment for supercells and tornadoes occurred in the most interior section of Ivan’s right-front quadrant, with conditions being even more favorable over land than over the sea. For contrast, comparable values are presented for Hurricane Jeanne (2004), which was similar to Ivan in several respects, but was not a prolific tornado producer at landfall. Although both storms provided environments with comparable shallow—and deep—layer vertical wind shear, the Ivan environment had notably more CAPE, likely due to a prominent dry air intrusion. This increase in CAPE was reflected in substantial increases in common operational forecasting composite indices. The results suggest that the conventionally assessed ingredients for midlatitude continental supercells and tornadoes can be readily applied to discriminate among TC tornado episodes.

Corresponding author address: Dr. Matthew Parker, North Carolina State University, Campus Box 8208, Raleigh, NC 27695-8208. Email: mdparker@ncsu.edu

Abstract

Hurricane Ivan (2004) was a prolific producer of tornadoes as it made landfall on the U.S. Gulf Coast. Prior researchers have revealed that the tornadic cells within tropical cyclone (TC) rainbands are often supercellular in character. The present study investigates the utility of several common midlatitude, continental supercell and tornado diagnostic tools when applied to Hurricane Ivan’s tornado episode.

The environment within Hurricane Ivan was favorable for storm rotation. While well offshore, the bands of Hurricane Ivan possessed embedded cells with mesocyclones of moderate intensity. A dual-Doppler analysis reveals that the updrafts of these cells were highly helical in the lower troposphere, suggesting significant ingestion of streamwise environmental vorticity. These coherent cells were long lived and could be tracked for multiple hours. As the supercells over the Gulf of Mexico approached the coast during Ivan’s landfall, rapid increases in midlevel vorticity and vertically integrated liquid (VIL) occurred. Based on compiled severe weather reports, these increases in storm intensity appear often to have immediately preceded tornadogenesis.

The local environment for supercells in Ivan’s interior is evaluated through the use of 62 soundings from the operational land-based network and from research flights. There were substantial differences in the thermodynamic profiles and wind profiles at differing ranges from Ivan’s center, from quadrant to quadrant of Ivan’s circulation, and between land and sea. The most optimal environment for supercells and tornadoes occurred in the most interior section of Ivan’s right-front quadrant, with conditions being even more favorable over land than over the sea. For contrast, comparable values are presented for Hurricane Jeanne (2004), which was similar to Ivan in several respects, but was not a prolific tornado producer at landfall. Although both storms provided environments with comparable shallow—and deep—layer vertical wind shear, the Ivan environment had notably more CAPE, likely due to a prominent dry air intrusion. This increase in CAPE was reflected in substantial increases in common operational forecasting composite indices. The results suggest that the conventionally assessed ingredients for midlatitude continental supercells and tornadoes can be readily applied to discriminate among TC tornado episodes.

Corresponding author address: Dr. Matthew Parker, North Carolina State University, Campus Box 8208, Raleigh, NC 27695-8208. Email: mdparker@ncsu.edu

1. Introduction

For some time it has been known that most landfalling tropical cyclones (TCs) are associated with outbreaks of tornadoes, with episodes ranging from several tornadoes to several dozen [e.g., the reviews of Hill et al. (1966), Novlan and Gray (1974), Gentry (1983), and McCaul (1993)]. The task of forecasting and warning for TC tornadoes is difficult, as rotational features in radar imagery may be more subtle than their midlatitude, continental counterparts (e.g., Spratt et al. 1997) and lead times may be comparatively short. The landfall of Hurricane Ivan along the coast of Alabama was associated with 31 tornado reports across portions of Florida, Alabama, and Georgia from 0800 UTC 15 September 2004 to 1400 UTC 16 September 2004 (Fig. 1; NCDC 2004). A large number of the associated tornadoes occurred very near the coastline in Florida, an observation that motivated our present study. This research focuses upon the trends and features of supercells, embedded within Ivan’s rainbands, which moved onshore from sea (the Gulf of Mexico) to land (the Florida panhandle), and upon the locally favorable versus unfavorable thermodynamic and kinematic environments within Ivan’s interior during landfall.

Observations (e.g., Fujita et al. 1972; McCaul 1987; Spratt et al. 1997; Suzuki et al. 2000; McCaul et al. 2004) have shown that tornadoes within the rainbands of landfalling hurricanes may often be produced by embedded supercells (rotating parent thunderstorms). Often (e.g., McCaul 1991, 1993) these are in fact miniature supercells, that is, supercells with comparatively small depth (e.g., Davies 1990; Kennedy et al. 1993). A subsequent modeling study by McCaul and Weisman (1996) demonstrated that a representative high-shear, low-buoyancy environment from a hurricane’s interior was favorable for the development of these shallow supercells.

Perhaps the role of supercells in TC tornadoes is not surprising considering that many, if not most, midlatitude continental tornadoes, especially significant tornadoes, are associated with supercellular thunderstorms (e.g., Davies-Jones et al. 2001). Recent efforts have emphasized an understanding of the interior TC environment in terms of the ingredients that are conventionally thought to favor midlatitude supercell development (e.g., Davies 2006; Edwards and Pietrycha 2006). Here, we review in turn some of these principal supercell and tornado ingredients.

The simulations of Weisman and Klemp (1982, 1984) made clear that large vertical wind shear in the lower and middle troposphere is important to the formation of supercells. Operationally, this is commonly assessed in terms of the 0–6-km AGL shear vector magnitude, using a minimal threshold of roughly 20 m s−1 (e.g., Rasmussen and Blanchard 1998; Thompson et al. 2003). The importance of storm-relative helicity (SRH) to updraft rotation was elucidated by Davies-Jones (1984). For the formation of supercells, this parameter is now conventionally assessed in the lower troposphere [e.g., 0–3 km AGL; Thompson et al. (2003)]. These parameters, along with the presence of potentially buoyant air parcels (i.e., CAPE), have been usefully combined into a supercell composite parameter (SCP; Thompson et al. 2003) that is quite effective at separating supercellular from nonsupercellular storm environments (using a minimal threshold of 1.0):
i1520-0434-24-1-223-e1
where MUCAPE is the CAPE value of the most unstable parcel in the lowest 300 hPa and BRN shear is one-half of the squared difference between the density-weighted mean winds in the 0–6-km and 0–500-m layers. In recent years the SCP has found wide use for midlatitude continental supercell forecasting, but it has not been widely applied in the TC-tornado literature.
In the past decade, several key features of tornadic environments have also emerged. Markowski et al. (2002) revealed that the environments of tornadic supercells typically possess higher boundary layer relative humidity than those of nontornadic supercells. In addition, Markowski et al. (2003) found that the principal differences between tornadic and nontornadic supercellular environmental wind profiles were in the lowest 1 km AGL. Tornadic storms were typically associated with much larger 0–1-km AGL shear vector magnitudes, and much larger 0–1-km SRH. These attributes have been combined with the preceding ingredients for supercells to yield another composite index—the significant tornado parameter (STP; Thompson et al. 2003), again using a minimal threshold of 1.0:
i1520-0434-24-1-223-e2
wherein 0–6-km shear refers to the shear vector magnitude between the surface and 6 km AGL, MLCAPE is the CAPE value of a parcel with the mean properties of the lowest 100 hPa of the sounding, and MLLCL is the height of the lifted condensation level (LCL) for that same mean parcel. Much like the SCP, the STP has also proven to be operationally useful;1 it clearly segregates the available climatological data for tornadic versus nontornadic supercells. Again, however, it has not yet been applied in the TC-tornado literature.

The kinematic and thermodynamic environments of TCs have received some attention, including the large climatological datasets studied by Novlan and Gray (1974), McCaul (1991), and Bogner et al. (2000). These studies have repeatedly shown that the interior environments of landfalling hurricanes are typified by high moisture content, limited CAPE, and very large lower-tropospheric wind shear. A recent study by Molinari and Vollaro (2008) reported some truly remarkable values of SRH within Hurricane Bonnie (one sounding with 0–3-km SRH in excess of 1300 m2 s−2). Novlan and Gray (1974) found that tornadic TC landfalls are characterized by significantly more vertical wind shear in the lowest 1–2 km than are nontornadic cases. McCaul (1987) further showed that this low-level wind shear is associated with significant storm-relative helicity, which contributes to low-level updraft rotation (e.g., Davies-Jones 1984). Another well-known finding of these climatologies [dating back at least to the work of Smith (1965) and Pearson and Sadowski (1965)] is that the preponderance of TC-spawned tornadoes occurs in the right-front (RF) quadrants of TCs [which should be defined with respect to the storm’s motion, as originally noted by Hill et al. (1966)]. McCaul (1991) found that the vertical shear and helicity of the wind profile were indeed greatest in the RF quadrant, and attributed this to the implied presence of background environmental shear along the TC’s direction of travel (associated with a steering level flow in the middle troposphere).

In addition to the preceding ingredients for supercells and tornadoes, several other environmental features may be important to TC tornadogenesis. Persistent inhomogeneities in the background environment, which can be associated with gradients in the above ingredients, include the coastline (dividing the oceanic- from the land-influenced boundary layer), preexisting synoptic baroclinic boundaries, and midlevel incursions of dry air into mature TCs. We conclude our review by briefly discussing these in turn.

First, it is clear that the sea-to-land transition is a fundamental part of a TC’s evolution during landfall. The same must also be true of the mesoscale environment for storms within a TC’s rainbands. Our study was motivated by the hypothesis that, at the time of landfall, there are fundamental differences in the local environment for supercells and tornadoes between the land and the sea. For example, Rao et al. (2005) found that a trackable “mesocirculation” existed for some time over the ocean and produced a tornado very soon after crossing onto land. As detailed in section 2, Ivan produced long-lived supercells well offshore. We cannot be certain of whether or not these supercells produced any tornadoes offshore (owing to the lack of observers and damage indicators), but our analysis shows that the cells intensified and produced tornadoes shortly after they came onshore. We are not aware of any studies that have specifically detailed the differences between continental versus maritime soundings during a tornadic TC landfall. We make such a comparison for Ivan in section 3.

Second, baroclinic boundaries may contribute to localized ascent, often possess enhanced lower-tropospheric vertical wind shear and SRH (e.g., Markowski et al. 1998a,b), and also commonly represent transitions in stability. The possible role of localized baroclinity in TC tornadoes has received increased attention of late (e.g., McCaul et al. 2004), and Edwards and Pietrycha (2006) identified recurring patterns in which TC tornadoes were concentrated along, or preferentially on one side of, preexisting boundaries. In the case of Hurricane Ivan, a number of tornadoes were indeed observed after landfall along a baroclinic boundary located inland in Georgia. However, the primary focus of the present study is the evolution of the convective storms and their environment near the coast at the time of landfall. Therefore, we set aside for now the problem of TC interactions with midlatitude baroclinic zones.

Finally, the entrainment of middle-tropospheric dry air into the TC’s circulation may be a key contributor to tornado outbreaks (Hill et al. 1966; Novlan and Gray 1974; McCaul 1987; Curtis 2004). The importance of dry air to TC tornadoes has been attributed alternatively to the existence of potential instability (e.g., Hill et al. 1966), the generation of low-level baroclinity (e.g., McCaul 1987; Curtis 2004), and to evaporative cooling that increases with height (e.g., McCaul 1987). Curtis (2004) found that 11 of the 13 tornado outbreaks he studied exhibited clear evidence of a dry intrusion. A vast majority of the tornadoes in the dry intrusion cases were diurnal, suggesting that the dry intrusion may favor enhanced insolation, increased CAPE in zones of clearing, and the generation of baroclinity along any cloudy–clear boundaries. Differences in CAPE have been broadly dismissed as unimportant to TC tornadoes, given that prevailing CAPE values are typically somewhat low in TC interiors. However, as discussed later, our study reveals large differences in CAPE between a prolific (Ivan) and nonprolific (Jeanne) tornado-producing landfall; Ivan appeared to have been more favorably influenced by a dry intrusion than Jeanne, which may account for these differences.

The primary aim of the present research was to understand the degree to which commonly used midlatitude paradigms for supercells and tornadoes can be applied in landfalling hurricanes. Section 2 of this paper outlines an investigation of radar data from the time period of Ivan’s approach and subsequent landfall along the Gulf Coast. The results of this survey are the motivation for the analysis of regional rawinsonde and dropsonde data, presented in section 3. Section 4 then presents some comparable data from Hurricane Jeanne, which was similar to Ivan in many respects, and yet was not a prolific tornado producer during landfall. The paper then concludes with a synthesis and some ideas for future research.

2. Hurricane Ivan

a. Synoptic overview

Hurricane Ivan developed from a strong African easterly wave, attaining hurricane status on 5 September 2004 as the system crossed the central Atlantic (Franklin et al. 2006). Ivan’s peak 1-min sustained winds of 75 m s−1 occurred with a central pressure of 910 hPa (category 5 on the Saffir–Simpson scale) on 12 September as it passed south of Grand Cayman. Soon afterward, the system began to move northward (Fig. 2), entering the Gulf of Mexico on 14 September as a strong category 4 system after clipping the western tip of Cuba. Ivan made landfall near Gulf Shores, Alabama, at approximately 0700 UTC on 16 September with maximum sustained winds of 54 m s−1 and a central pressure of 943 hPa [see Franklin et al. (2006) for a complete and detailed account of Ivan’s life span].

In the 26 h prior to landfall, multiple research and reconnaissance aircraft observed Ivan’s environment (both exterior and interior). During this period, Ivan was moving ≈5.7 m s−1 toward the north and was slowly weakening (maximum winds decreased from 62 to 54 m s−1) in response to increased vertical shear associated with an approaching midlatitude trough and the ingestion of dry air into its core. At the same time, a prominent outer rainband developed ≈450 km from the storm center, extending through Ivan’s entire northeastern quadrant (Figs. 3a and 4). The band was situated just outside a dry air intrusion that wrapped around the southern and eastern quadrants. Examination of animated satellite and land-based radar imagery revealed that the band was composed of numerous deep convective cells that initially formed well offshore (southeast of the storm center), moved along the band (through the eastern quadrant), and then moved onshore into the Florida panhandle (northeast of the center). Many of these cells exhibited supercellular features (e.g., hook echoes and mesocyclones) well offshore and maintained such structures as they transitioned from offshore to onshore regimes.

b. Radar data and techniques

Numerous supercells developed well offshore (>150 km from the coastline) within Hurricane Ivan’s prominent outer rainband and persisted for several hours as they moved along the rainband. The supercells then appeared to intensify as they moved onshore, with several producing tornadoes within 50 km of the coastline (Fig. 1). To document the structure and evolution of the supercells prior to tornadogenesis, we examined both airborne (i.e., offshore) and land-based Doppler radar data.

1) Airborne radar observations and analysis methods

A National Oceanic and Atmospheric Administration (NOAA) WP-3D research aircraft crossed Hurricane Ivan’s prominent outer rainband between 1750 and 1815 UTC on 15 September en route to the eyewall (the primary mission goals were unrelated to the outer rainband). The aircraft was equipped with a 5.5-cm wavelength lower-fuselage radar and a 3.2-cm-wavelength tail Doppler radar (Jorgensen 1984). During the rainband crossing at ≈2.5 km altitude, the tail radar employed the fore–aft scanning technique (FAST; Gamache et al. 1995), thus enabling one dual-Doppler analysis of the three-dimensional wind field (Fig. 5) for the domain shown in Fig. 4. The geometry of the FAST scanning strategy dictates that a time lag occurs between each radar measurement at a given location within the domain. In the present analysis, the maximum time lag was ≈3 min and occurred along the outer edges of the domain. The lags were much smaller near the center of the domain, and thus the structural details of individual convective cells were well represented.

The raw reflectivity and Doppler velocity data from the tail radar were corrected for aircraft motion, and spurious echoes (e.g., sea clutter) were removed (Testud et al. 1995; Bosart et al. 2002). The edited fields were then interpolated to a band-centered 90 km × 90 km Cartesian domain extending from 0- to 18-km altitude with uniform horizontal and vertical grid spacings of 1.5 and 0.5 km, respectively. The values at each grid point were determined through a Gaussian weighting of all nearby observations using a cutoff radius of 3 km (1 km) and an e-folding distance of 0.75 km (0.25 km) in the horizontal (vertical) direction. The Doppler radar analysis methodology of Gamache (1997) was used to simultaneously solve the radar projection equations and anelastic mass continuity equation for the final three-dimensional wind field. The solution was subject to a zero vertical velocity constraint at the surface and just above echo top.

In application, Gamache (1997) found the analysis technique superior to traditional methods (e.g., Jorgensen et al. 1983; Marks et al. 1992), but time evolution of the wind field, beam-filling issues, and variable cell motions still make significant contributions to the error. Thus, the quality of the Doppler analysis was assessed via a gridpoint by gridpoint comparison of flight-level measurements and the Doppler-derived winds at 2.5-km altitude following Gamache et al. (1995). Summary error statistics (not shown) computed for each wind component (zonal, meridional, and vertical) were similar to previously reported values in hurricanes (Marks et al. 1992; Gamache et al. 1995; Reasor et al. 2000, 2009).

2) Land-based radar observations and analysis methods

We also incorporated reflectivity and radial velocity data from the region’s operational Weather Surveillance Radar-1988 Doppler radars (WSR-88Ds). The Tampa Bay, Florida (KTBW), radar made distant measurements of the offshore cells within Ivan’s rainbands, complementing the analysis from the airborne research radars. The Tallahassee, Florida (KTLH), radar provided an excellent depiction of the supercells that made landfall in the Florida peninsula. Individual storms were tracked during the parts of their lives that fell within the KTLH radar volume. We recorded the times at which individual storms made landfall, and also carefully compared the times and positions of the storms with reports of tornadoes from Storm Data (NCDC 2004).

Four distinct storms were identified as both making landfall and producing at least one tornado (Fig. 1). Reflectivity and radial velocity data from KTLH were analyzed to discern trends in cell structure and associated mesocyclone intensity. Values of rotational velocity (Vrot, one-half of the greatest observed difference between radial velocities from within the mesocyclone) were calculated for the midlevels [15 000 ft (≈4.5 km)]. Additional structures such as weak-echo regions and hook echoes were also noted and catalogued. Altogether, trends in these cell attributes were used to infer the evolution of the supercells during the period prior to and during landfall, and around the time of tornadogenesis.

c. Evolution of convective cells

1) Offshore structure

Within the main rainband in Ivan’s RF quadrant (Fig. 3a), coherent convective cells with weak rotational signatures were evident well offshore. These cells were observed by both the KTLH and KTBW radars (Fig. 4), as well as the NOAA WP-3D airborne radars. Strong reflectivity gradients adjacent to the precipitation-free dry air intrusion marked the western edge of the rainband, while the eastern side was dominated by stratiform precipitation (Figs. 3a and 4). Three distinct high-reflectivity cells were evident along the western edge, and the dual-Doppler analysis of the airborne radar data was able to resolve the mesoscale-convective structure of these cells (labeled A, B, and C in Figs. 5 and 6). At this time the cells were located >150 km offshore and moving at 24 m s−1 toward the north-northeast. Each cell contained modest cyclonic rotation in the storm-relative winds (i.e., a mesocyclone; Figs. 5a and 5c) and a hooklike reflectivity appendage (Figs. 5a and 5b). Each storm possessed a vertical vorticity maximum (ζ > 5 × 10−3 s−1) collocated with a vertical velocity maximum (w > 6 m s−1), that is, a strong rotating updraft.2

Figure 5d shows a north–south cross section through the center of cell B (see Fig. 5b). The mesocyclone (defined here by the 3 × 10−3 s−1 vertical vorticity contour) extends from the bottom of the analysis domain upward to nearly 4 km AGL with a mean diameter of approximately 4–5 km. The updraft (defined by the 2 m s−1 vertical velocity contour) extends roughly from 1 km upward to 4 km AGL, with a mean diameter of approximately 5–6 km. Cross sections through cells A and C (not shown) exhibited similar structure. Such structure is qualitatively consistent with previous observations of hurricane minisupercells that were observed by land-based radars (Spratt et al. 1997; McCaul et al. 2004; Schneider and Sharp 2007).

Prior studies have suggested the development of such minisupercells is linked to the increase in friction (and thus low-level vertical shear) as the cells transition from an offshore to an onshore environment (e.g., Gentry 1983). Clearly, the cells within Ivan’s rainband contradict this traditional conceptual model. In section 3, we document that the offshore environment was already quite conducive to supercell formation. Here, we briefly demonstrate that the tilting and stretching of vorticity likely contributed to supercell development and maintenance. Shown in Fig. 5c are the cell-relative wind and the horizontal vorticity vectors at 1.0 km AGL. The flow entering the base of each updraft is roughly aligned with the horizontal vorticity vector, suggesting that the tilting of streamwise vorticity into the vertical by the updraft contributes to mesocyclone formation. The vorticity maximum (≈7 × 10−3 s−1 at 1.5 km AGL) in cell B is located below the vertical velocity maximum (≈6 m s−1 at 2.5 km AGL), which suggests that vorticity stretching may also have enhanced the low-level vertical vorticity maximum and mesocyclone. Such formation processes are qualitatively consistent with prior observations and numerical simulations of supercells (e.g., Klemp et al. 1981; Davies-Jones 1984; Klemp 1987; Droegemeier et al. 1993; McCaul and Weisman 1996). Additional analyses of the WP-3D airborne Doppler radar observations are currently under way, but these are beyond the scope of the present study.

2) Structure during transition onshore

Using the Tallahassee (KTLH) WSR-88D, the three minisupercells were tracked from their offshore dual-Doppler positions at 1804 UTC to their landfall near Panama City Beach, Florida, at 2045 UTC. Figure 6a shows the location of each cell (defined by the maximum reflectivity) at ≈5 min intervals during this period. At 1804 UTC the cells were located ≈220 km south of Tallahassee. Despite clear supercellular structure in the dual-Doppler analysis below 4 km AGL, the cells exhibited little evidence of such structure when viewed from KTLH (Fig. 6b). However, it is important to note that the base-scan elevation was at ≈5 km AGL and Doppler velocity data were not available at this range.

From the KTLH perspective, cells C and B first exhibited prominent hook echoes near 1930 UTC when they were located ≈25 km offshore and ≈140 km from the radar. The hook echoes persisted for approximately 40–50 min as the cells paralleled the coast (≈15 km offshore and ≈130 km from the radar). No tornadoes or radar-detected mesocyclones were observed in cell C during this period. A weak mesocyclone was briefly detected within cell B at 1945 UTC (Fig. 6c) as it was located over a barrier island (<10 km offshore and ≈125 km from the radar). Cell B spawned one F0 tornado at 2040 UTC as the cell moved onshore (tornado 4 in Fig. 1).

Cell A first exhibited a hook echo at 1934 UTC when it was located ≈32 km offshore and ≈130 km from the radar. The hook echo persisted over the next 90 min as the cell paralleled the coast and then moved onshore near Panama City at 2044 UTC (Figs. 6d–f). A weak mesocyclone was continuously detected between 2022 and 2044 UTC as the cell moved from ≈10 km offshore to ≈20 km onshore (at ≈120 km from the radar). Cell A spawned two tornadoes between 2040 and 2050 UTC (tornadoes 3 and 6 in Fig. 1); the most intense produced F1 damage and a fatality.

Four landfalling tornadic storm cells were analyzed for possible trends in Vrot in the KTLH radar data. These cells were responsible for the storm reports outlined by the four rectangles in Fig. 1. After calculating running averages of midlevel Vrot for all four storm cells (Fig. 7), several interesting points emerged. The most rapid increase in mean midlevel Vrot occurred between 11 and 5 min before landfall of the cells, with the maximum Vrot roughly concurrent with the cell landfall. There was then a much slower decrease in mean midlevel Vrot after landfall.

Because the cells were moving closer to KTLH as their values of Vrot increased (top panel in Fig. 7), we tried to determine the degree to which differences in radar resolution with range impacted the measured Vrot. We performed a simple calculation assuming that the mesocyclone could be represented as a Rankine combined vortex, with a mesocyclone diameter of 4 km (cf. Fig. 5), and a 1° radar beamwidth. For cell D, which moved from roughly 145 to 90 km from KTLH (Fig. 7), we computed an increase in Vrot of approximately 14% owing to the improved resolution at closer range. For larger mesocyclones, and for cells that began closer to the radar, the effect would be smaller (because the initial resolution would be better). In short, the differences in resolution due to range from the radar3 explain only a small part of the doubling in Vrot seen in Fig. 7. A similar conclusion follows from the mesocyclone strength nomograms presented by Spratt et al. (1997, their Fig. 2).

For the most part, the increase in mesocyclone strength associated with the cells’ transition from sea to land appears to be real (not a measuring artifact). The initial tornado reports (Fig. 7) occurred not long thereafter (although it is possible that there were earlier, offshore tornadoes that were not reported). During the 35–40 min prior to cell landfall, there was also a corresponding increase in the cells’ intensity as measured by vertically integrated liquid (VIL; Greene and Clark 1972). This trend continued until a peak in VIL approximately 10 min after cell landfall (not shown). Although caveats apply to both the radar measurements and tornado reports, we conclude that the embedded cells became stronger as they made landfall, increased in rotational velocity, and soon produced their first documented tornadoes.

3. Storm mesoscale environments within Hurricane Ivan

Motivated by the large number of offshore supercells in Ivan’s main rainband, and by the cells’ apparent transition as they moved onshore, we investigated the mesoscale environment within Ivan’s interior.

a. Sounding data and techniques

Our analysis focused on 62 vertical soundings from within 1250 km of Ivan’s center between 0547 UTC 15 September 2004 and 0053 UTC 16 September 2004 (top row in Fig. 8). Of the 62 soundings, 11 were operational rawinsondes launched over land and 51 were dropsondes from a pair of research aircraft flights through Hurricane Ivan. The dropsondes were well distributed throughout all four quadrants of the storm [right front (RF), right rear (RR), left rear (LR), and left front (LF), relative to the observed storm motion; Fig. 8]. Based upon our analysis of the sea-based dropsondes, we then chose to focus on analysis of land-based soundings from stations on the windward coast in the storm’s RF quadrant as it approached landfall and, thus, incorporated rawinsonde observations from Jacksonville, Tampa Bay, Tallahassee, and Cape Canaveral, Florida.

To easily compare composited soundings across times, quadrants, and different hurricanes, we translated and vertically interpolated the data. All winds were converted to a TC-relative reference frame by subtracting the observed TC motion. The wind data were then converted into radial (hereafter Vrad) and tangential (hereafter Vtan) components. All wind and thermodynamic data were then linearly interpolated to a grid with a vertical spacing of 200 m, and were averaged for a variety of quadrants, times, ranges, etc. Interpolation was performed across data gaps of less than 1000 m; for vertical gaps of 1000 m or more, the data were treated as missing and were excluded from the averages. If data were missing at the surface, then values from the lowest level below 200 m AGL were used; if there were no data below 200 m, then all data below the lowest reported value were excluded from the averaging. In computing Vrad and Vtan (and assigning a sounding to one of the four storm quadrants), we used the position of the sonde at the surface. All but two of our soundings were more than 200 km from the storm center (with the closest having a range of 163 km), such that cyclonic advection of the soundings by the TC circulation was almost always less than 5° in azimuth, and had only a minimal impact on partitioning between Vrad and Vtan.

For each averaged set of soundings, the following mean values were computed: range from storm center, surface-based CAPE, surface-based convective inhibition (CIN), LCL height, 0–3-km CAPE, 0–1- and 0–6-km shear vector magnitudes (hereafter “0–1-km shear” and “0–6-km shear”, respectively), 0–1- and 0–3-km SRH, supercell composite parameter [SCP; Eq. (1)], and significant tornado parameter [STP; Eq. (2)]. In computing the storm-relative helicities, supercell storm motions were approximated using the technique of Bunkers et al. (2000). We also visually inspected averaged hodographs for each population in order to identify any commonalities and differences that might be masked by the basic wind shear indices above.

When we grouped the available soundings by time (into windows of 0547–1235 and 1750–2455 UTC; Fig. 8), we found that there were no marked differences in the vertical wind shear nor thermodynamic indices. Therefore, in order to increase the sample size for the following analysis, we treat these two time windows together as one dataset.

b. Spatial variability

Upon grouping soundings by quadrants relative to Ivan’s motion (Fig. 9, Table 1), it became evident that the hodographs’ lengths were much greater on the storm’s right side, especially in the RF quadrant. McCaul (1993) attributed this recurring observation to the superposition of the hurricane’s circulation (which can be conceptualized as axisymmetric) upon the sheared base-state environment (the environmental deep-layer shear vector and the environment’s middle-tropospheric steering current are likely to be nearly parallel). The RF quadrant’s long hodographs corresponded to maximal values of both 0–1- and 0–6-km vertical wind shear, as well as much greater SRH values (Table 1). As noted by McCaul (1991), the observed values of low-level shear in the RF quadrant are large within the context of values typically observed for midlatitude continental supercells. CAPE and CIN values did not vary dramatically among the quadrants (Table 1), although the lower LCL heights on the storm’s right flank represent the northward advection of moist maritime air from the deep tropics. The moderately high CAPE and vertical wind shear contributed to a SCP of >1.0 in the RF quadrant (Table 1). The addition of significant 0–1-km shear and low LCL heights further led to an STP of >1.0 in the RF quadrant (Table 1). The parameters for supercells and tornadoes were so clearly maximized in the RF quadrant that we focus the remainder of our analyses on only that quadrant. The propensity for tornadoes to occur predominantly in the RF quadrant has long been known (e.g., Pearson and Sadowski 1965), and all of Ivan’s tornadoes were also observed within the RF quadrant.

Within the RF quadrant, averaged hodographs grouped by range bins of distance away from the center of Ivan (Fig. 10, Table 2) show that the most favorable environments for supercells and tornadoes by far were located nearest to Ivan’s center. CAPE decreased gradually inward, although significant lower-tropospheric CAPE remained (Table 2). Other thermodynamic differences (e.g., CIN and LCL height) were minimal. Meanwhile from Ivan’s outer reaches to its interior (i.e., the inner versus the outer range bin), hodograph lengths increased dramatically, especially in the lowest levels. Studies by McCaul (1991) and Bogner et al. (2000) have similarly shown a tendency for CAPE to increase with range and for the vertical wind shear to decrease with range. In terms of Ivan’s observed SCP and STP composite parameters, the substantial interior increases in shear and SRH more than offset the decreases in CAPE. The enhanced mean 0–1- and 0–3-km SRH values imply that significant streamwise vorticity was available for tilting into the vertical.

The primary rainband responsible for the supercells and tornadoes during Ivan’s landfall was located between roughly 250- and 450-km range. Thus, the first range bin is representative of the local environment for the convective cells analyzed above. In comparison to other studies (Table 3), we find the mean CAPE value in Ivan’s RF quadrant to be somewhat high, although comparable to the data for offshore hurricanes collected by Bogner et al. (2000). Ivan’s mean 0–3-km SRH values are smaller (Table 3) than those from near intense cells in Hurricane Bonnie as reported by Molinari and Vollaro (2008), but considerably larger than those reported by McCaul (1991) and Bogner et al. (2000), although those latter studies admittedly used estimated supercell storm motion vectors that would have dramatically underestimated the amount of SRH. All of the studies have consistent values for 0–1-km shear vector magnitudes (Table 3), while the 0–6-km shear vector magnitudes vary greatly among the studies (Table 3), likely as a consequence of varying averaging techniques and sample sizes. Based on the extant literature, Ivan’s CAPE and shear values appear to be at or above the norm. Although most prior TC–tornado studies have not reported the SCP nor STP due to their relative newness, we note that, for Hurricane Bonnie, Molinari and Vollaro (2008) reported a mean SCP of 2.4 from soundings in the vicinity of intense convective cells, which was actually exceeded in the interior of Ivan’s RF quadrant.

The principal difference with range among Ivan’s averaged hodographs was in their overall size; across all ranges in the RF quadrant there were commonalities in hodograph shape. There was significant vertical shear in Vtan (and minimal shear in Vrad) over the lowest 1 km AGL. This decrease in Vtan toward the surface is largely attributable to surface friction. In the 1–5-km layer, the wind shear was weaker and was predominantly due to Vrad, with radial inflow noted below approximately 2 km AGL and radial outflow above. This shear in Vrad is associated with the hurricane’s basic meridional circulation. The combination of friction’s impact on Vtan and the parent storm’s radial circulation help to explain the presence of moderate deep-layer shear and large hodograph curvature in the lower troposphere, which together favor supercells and tornadoes.

c. Contrasts between land and sea

Motivated by the apparent rapid increase in storm rotation around the times that individual cells made landfall on 15 September, we sought to understand whether the observed convective environment was appreciably different between land and sea within the RF quadrant. The land-based soundings had comparable CAPE, although somewhat less lower-tropospheric CAPE and somewhat more CIN (Table 4). The lower (by approximately 35%) 0–3-km CAPE value over land is noteworthy in that McCaul and Weisman (1996, 2001) suggested that updraft strength and vorticity were both enhanced when buoyancy is concentrated in the lower levels of environments with modest total CAPE. The hodographs nearly overlay one another above roughly 4 km AGL (Fig. 11), and the associated values for the deep-layer shear were also comparable (Table 4). The principal difference between the two regimes was in the dramatically increased hodograph length over land below 3 km AGL, and especially in the 0–1-km layer (Fig. 11). The bulk 0–1-km shear vector magnitude was, on average, roughly 50% greater over land (Table 4); associated with this, the 0–1-km SRH was also roughly 50% greater. Climatological studies suggest that increased lower-tropospheric shear is often associated with tornadoes (Markowski et al. 2003). As is evident in Fig. 11, the land-based soundings exhibit slower near-surface flow that is dominated by the radial wind component. This slowing and backing of the wind can be attributed to the impacts of enhanced surface friction over the land versus the water.

It is reasonable to wonder about the horizontal distance over which the environmental winds might vary from a sealike profile to a landlike profile. Some researchers (e.g., Powell and Houston 1998) have suggested that there may be changes in surface wind speed as large as 8–10 m s−1 across horizontal distances of roughly 10 km. Although we have no such measurements for Hurricane Ivan, based on the findings of Powell and Houston it seems plausible that a rapidly moving embedded supercell could indeed begin to experience a dramatically different, frictionally modified environmental wind profile within the span of 10–15 min.

Notably, the supercell and tornado composite parameters (SCP and STP) were still favorable (>1.0) in the sea-based RF quadrant soundings. This is consistent with the observations of supercells well offshore in Ivan’s rainbands. It appears that the large-scale pattern responsible for steering Ivan to the north [a relatively strong anticyclone to the east-southeast of Florida and a weak trough to the west of Ivan; Franklin et al. (2006)] set up an east-to-west pressure gradient that led to easterly low-level background flow over the Gulf of Mexico, leading to enhanced wind profiles throughout Ivan’s RF quadrant. In summary, the transition of Ivan’s RF quadrant from sea to land augmented an environment that was already quite favorable for supercells and tornadoes.

4. Comparison to Hurricane Jeanne

It was instructive to compare Ivan to another TC that was a far less prolific tornado producer at landfall. Hurricane Jeanne made landfall on the Atlantic coast of Florida at a similar time of day (Ivan, 0700 UTC; Jeanne, 0400 UTC), was similar in intensity at the time of landfall (Ivan, 943 hPa central pressure and 54 m s−1 peak wind; Jeanne, 951 hPa central pressure and 54 m s−1 peak wind), and was moving at a very similar forward speed (Ivan: 5.7 m s−1 and nearly perpendicular to the coastline; Jeanne: 5.9 m s−1 and nearly perpendicular to the coastline). In short, the two storms were similar in terms of several common attributes that have been hypothesized to impact the production of supercells and tornadoes around the time of landfall (e.g., as investigated by McCaul 1991).

Despite these similarities, Jeanne produced only eight tornadoes in Florida on the day of its landfall4 (0000 UTC 26 September–0000 UTC 27 September; NCDC 2004). In reviewing the operational Doppler radar data for the positions and times of the tornado reports, it appears that only one of these eight tornadoes was associated with an obvious supercell. We consider six of the eight reports to be dubious based upon radar data, owing to the lack of a convective storm at the location and/or a complete lack of a rotational signature in the velocity data. Indeed, several of the official storm data refer to “tornado-like events” (based on damage surveys) near Jeanne’s eyewall (NCDC 2004). It is possible that errors in the reported times and locations of the reports contributed to the discrepancies with the radar data [tornado reports are known to be somewhat uncertain in the vicinity of hurricane landfalls, e.g., McCaul (1987); McCaul et al. (2004)]. It is also possible that the small scale of the tornadoes (all of the reported path widths were 15 m or less) caused radar sampling (“beam filling”) problems. At the very least, if we assume that most of the reports are legitimate, we conclude that there were not long-lived mesoscyclones associated with the majority of Jeanne’s landfall tornadoes.

a. Synoptic overview

Hurricane Jeanne initially developed from an African easterly wave, attaining hurricane classification for the second time on 20 September 2004 as it began a slow anticyclonic loop roughly 900 km northeast of the Bahamas (Franklin et al. 2006). By 23 September a strong anticyclone had developed to the north and Jeanne began moving westward. Figure 2 shows the best-track positions and selected intensities for Jeanne after 1200 UTC on 24 September as it crossed the northern Bahamas, central Florida, and the southeast United States. Jeanne made landfall near Stuart, Florida, at 0400 UTC on 26 September with maximum sustained wind speeds of 54 m s−1 and a central pressure of 951 hPa. Franklin et al. (2006) provided a complete history of Jeanne.

b. Storm mesoscale environments

In the 15 h prior to landfall, a research reconnaissance aircraft observed Jeanne’s synoptic environment and mesoscale inner core. During this period, Jeanne was moving at ≈5.9 m s−1 toward the west and slowly intensifying in a low-shear environment. As in Ivan, a dry air intrusion was observed in association with Jeanne, roughly 400–500 km to the north and west of the storm center (see Fig. 3b). To understand the discrepancy in supercell and tornado occurrences between Hurricanes Ivan and Jeanne, the sounding procedures described in section 3a were also performed for Jeanne (bottom row in Fig. 8). A total of 37 soundings from Jeanne were utilized, including 27 sea-based dropsondes and 10 land-based rawinsondes (again from stations on the windward coast in the storm’s right-front quadrant as it approached landfall: Cape Canaveral and Jacksonville, Florida, and Charleston, South Carolina). For both storms, the data were separated into two time periods of roughly 6 h each. For Ivan, the observing windows ran from roughly landfall – 25 h to landfall – 19 h and from landfall – 13 h to landfall – 6 h; for Jeanne, the observing windows ran from roughly landfall – 22 h to landfall – 16 h and from landfall – 10 h to landfall – 4 h. In short, observations from the two storms should be readily comparable. In the case of Jeanne, only one research flight occurred, and so there are only sea-based dropsondes from the earlier of the two time windows.

The mean range of the Jeanne RF quadrant soundings was roughly 600 km (Table 5), meaning that they could not be directly compared to the 0–500-km range bin from Ivan’s RF quadrant (within which Ivan’s primary rainband of supercells was located). However, it is reasonable to compare the Jeanne soundings to the 500–750-km range bin from the RF quadrant of Ivan (mean range of 626 km; Table 2) as well as both the land and sea groupings for Ivan’s RF quadrant (mean ranges of 619 and 635 km, respectively; Table 4). These three samples are hereafter referred to as the Ivan subsets.

Over the lowest 1 km, the hodographs from Jeanne were of similar length to the Ivan subsets (cf. Figs. 10 –12), with 0–1-km shear vector magnitudes generally clustered around 8–10 m s−1 (cf. Tables 2, 4, and 5). The land-based soundings from Ivan had slightly more 0–1-km shear than the 1800–0000 UTC time window from Jeanne (all of which were land based). The 0–1- and 0–3-km SRH values were also similar between the two storms, although again Ivan’s land-based soundings had the highest. All of the Ivan subsets had more 0–6-km shear than either of the Jeanne subsets, but these differences were small. In short, although Ivan’s land-based soundings appeared to represent a somewhat more favorable low-level wind profile for rotating storms and tornadoes, the differences between the two storms generally were not dramatic.

However, the SCP and STP indices were nevertheless much lower for Jeanne than for Ivan; values much smaller than 1.0 are climatologically very unfavorable for supercells and tornadoes (Thompson et al. 2003). The low SCP and STP were largely attributable to the fact that the observed values of CAPE for Jeanne were a factor of 5–10 smaller than those observed in the Ivan subsets (cf. Tables 2, 4, and 5). The differences in shear contributed a small additional factor, as did the slightly higher mean LCL heights observed during Jeanne [which lower the STP; Eq. (2)].

McCaul and Weisman (1996) have pointed out that the environments of TC supercells need not have excessive CAPE, because the upward dynamic pressure gradient acceleration can compensate for it in the strongly sheared lower–middle-tropospheric environment of TCs. However, it is clear that at least some CAPE is necessary for the production of convective storms. Even McCaul and Weisman (1996)’s mean TC supercell sounding had in excess of 600 J kg−1 of CAPE. Subsequent work by McCaul and Weisman (2001) showed that, in environments with small CAPE and moderate vertical wind shear, storms tended to be weak unless the updraft buoyancy was concentrated in the low levels. This was especially so when the preponderance of the vertical shear and helicity was concentrated in the lowest levels (as they are in TC environments). In such cases, updraft air achieves large vertical accelerations close to the ground, thus providing for tilting and stretching of the ample horizontal vorticity into the vertical. It is therefore noteworthy that the 0–3-km CAPE values from Jeanne were a factor of 2–3 smaller than those from the Ivan subsets (cf. Tables 4 and 5). This may help to explain why so few coherent, long-lived storms were observed during Jeanne’s landfall.

Some authors (e.g., Gentry 1983; McCaul 1991; Hagemeyer 1998) have shown that Atlantic coast landfalling hurricanes in the United States typically produce fewer tornadoes than Gulf Coast landfalling hurricanes. McCaul (1991) reported that there were no statistically significant differences between the two populations, although we note that CAPE for the Gulf storms was roughly 50% higher in McCaul’s dataset. Given generally similar hurricane intensities, McCaul speculated that the principal difference between landfalls on the two coasts was attributable to the fact that, in many cases, Atlantic landfalling storms track along the eastern coast of the United States and thus their RF quadrants are mostly offshore. Although this was partly the case with Jeanne (Fig. 2), Jeanne’s landfall was at a nearly 90° angle to the Florida coast, with the RF quadrant having full exposure to land during and immediately following landfall.

Because the greatest differences between Ivan and Jeanne appear to have been related to the thermodynamic instability, we sought to understand the source of these differences. One possibility is that the somewhat warmer waters of the Gulf provided for boundary layer parcels with greater equivalent potential temperature (θe). The averaged RF soundings from Ivan did indeed have higher surface θe values than those from Jeanne (355 versus 344 K, respectively; not shown). However, in looking at the land-based surface observations (with their higher spatial and temporal resolutions) for each case, we found that along the coastline in the interior of both hurricanes the θe values were similar, peaking at 367 K. Therefore, it would appear that over at least part of the RF quadrant in both storms, air with similarly desirable thermodynamic properties was present. Therefore, it appears that at least some of the differences in CAPE (and in 0–3-km CAPE) must be attributable to differences in the lower–middle-tropospheric lapse rates.

c. Possible roles of dry intrusions

As seen in Geostationary Operational Environmental Satellite (GOES) water vapor imagery (Fig. 13), both Ivan and Jeanne contained prominent dry air intrusions in their RF quadrants. Curtis (2004) noted that TC rainbands adjacent to such intrusions are frequently associated with enhanced deep convection and increased tornadic activity. McCaul (1987) noted that, when dry air intrudes in the midlevels, the intrusion can result in a local increase in atmospheric instability (i.e., an increase in CAPE) due to midlevel evaporative cooling. Closer examination of representative soundings from each intrusion (Figs. 13b and 13d) reveals a more complex situation; there are potentially different configurations of such dry air intrusions. For example, in Ivan the dry air was observed below 4–5 km and the sounding remained fairly unstable. In contrast, the dry air intrusion in Jeanne was confined to the upper and middle levels, and the sounding was more stable (with lower CAPE) due to a prominent capping inversion at the base of the dry air. Thus, convection was limited.

One plausible explanation for these disparate dry air intrusions is the ability of evaporational cooling (via entrainment mixing with adjacent rainbands) to effectively remove any capping inversion as the intruding air sinks and adiabatically warms. In Ivan, the dry air intrusion was flanked by mature rainbands on both sides. In contrast, Jeanne’s intrusion was located along the outer edges of the hurricane circulation where there was likely less evaporational cooling to offset the subsidence warming. Thus, in addition to verifying the presence of a dry intrusion, forecasters may need to consider its location in the vertical, and how its horizontal position within the TC might impact the local stability. The latter may be related to how far into the storm circulation the dry air intrudes, and how long the intrusion exists (longer time periods would allow for more evaporative cooling aloft).

5. Conclusions

a. Summary

Hurricane Ivan (2004) produced more than 30 tornadoes as it made landfall, many of them in very close proximity to the coastline of the Florida peninsula. Previous studies have revealed that tornadoes within hurricane rainbands are often attributable to embedded supercells, and this was indeed the case during Ivan’s landfall. The principal rainband in Ivan’s RF quadrant, between 250 and 450 km from the storm’s eye, exhibited supercells over an extended period of time, including well offshore.

Analysis of conventional operational rawinsondes as well as airborne research dropsondes reveals that Ivan’s interior was characterized by moderate-to-high amounts of CAPE and high lower-tropospheric vertical wind shear, particularly in the storm’s RF quadrant and within 500 km of the storm center (the range where the principal rainband was located). The enhanced shear in the interior of hurricanes’ RF quadrants has been observed in numerous prior studies, and has been attributed to the superposition of the hurricane’s circulation and a sheared base-state wind profile (e.g., McCaul 1993). In the context of the literature, the values observed in Ivan appear to be comparable or slightly higher than what is common to strong TCs (Table 3).

Although it is challenging to conclusively demonstrate cause and effect with the available observations, the large instability (CAPE) observed in Ivan was collocated with a middle–upper-tropospheric dry intrusion, which we hypothesize to have enhanced the environmental lapse rates through evaporative chilling aloft (e.g., Curtis 2004) and, possibly, through the release of potential instability. The combination of shear and instability was associated with sufficient (i.e., >1.0) values of common midlatitude forecasting indices for supercells (i.e., the SCP) and tornadoes (i.e., the STP). In the interior of Ivan’s RF quadrant, values of SCP and STP (Table 2) were similar to those observed in the tornado-prone Great Plains.

In association with these favorable ingredients in Ivan’s RF quadrant, numerous supercells were observed over an extended period of time, including well offshore. Dual-Doppler analysis suggests that the significant streamwise vorticity present in the lower troposphere in the RF quadrant was tilted and ingested by the offshore cells. However, comparison of the land-based versus sea-based soundings revealed that the environment for supercells and tornadoes was even more favorable on shore than over the Gulf of Mexico (Table 4). Perhaps not surprisingly, radar observations showed rapid increases in storm intensity and updraft rotation as Ivan’s embedded supercells approached the coastline and made landfall. The possibility of small-scale gradients in the environments of storms along the coast is of interest, and warrants further attention.

A comparison of Ivan’s landfall to that of Jeanne is appropriate, considering that the two storms had almost the same intensity, forward speed, and track angle with respect to the coastline (each of which has been previously hypothesized to be important to TC tornado production). Jeanne was not a prolific tornado producer during landfall and, in our radar investigation, most of the eight reported tornadoes were not clearly associated with coherent supercells. Jeanne’s landfall was characterized by somewhat less lower-tropospheric shear and SRH than Ivan (when we compared similar mean ranges). However, the predominant difference between Jeanne and Ivan appeared to be the much smaller CAPE values that were present during Jeanne’s landfall, especially in the lower troposphere. We ascribe this difference partly to the comparatively warmer waters of the Gulf of Mexico (Ivan) versus the western Atlantic Ocean (Jeanne), but also to the more stable lapse rates observed in Jeanne’s interior. Because Jeanne’s dry intrusion was located along the hurricane’s periphery, we infer that the roles of evaporational cooling aloft and lifting of potentially unstable air were minimized in Jeanne’s interior. Jeanne’s dramatically smaller CAPE values were associated with much smaller values of SCP and STP (both were much <1.0, or well below the climatological threshold value).

A worthwhile operational result from our investigation is that the SCP and STP composite indices appeared to work quite well in indicating the potential (or lack thereof) for an outbreak of strong supercells and tornadoes during the Ivan and Jeanne landfalls. Forecasters should seriously consider using them to assess the comparative threat of tornadoes for different landfalling storms. The ingredients that are combined to form the SCP and STP are physically motivated, and correspond to hypothesized processes that are at work during the formation of supercells and supercellular tornadoes (e.g., Thompson et al. 2003, and references therein). To the extent that SCP and STP ingredients are proxies for such fundamental physical processes, one can infer that the supercell and tornado processes in landfalling TCs are likely the same as those that have been more widely studied over the midlatitude continents.

b. Future work

As with any study of limited scope, it would be worthwhile to broaden these results by including more storms, particularly TCs with more measurements at short range from the eye. It would also be useful to perform similar comparisons between groups of tornadic and nontornadic TCs that made landfall exclusively on the Gulf Coast. Only by comparison among numerous TC landfalls can the general applicability of the SCP and STP be upheld. We do note the existence of some other recent studies that have successfully applied analysis data from the Rapid Update Cycle (RUC) model to do just that [Davies (2006); as well as some unpublished studies by J. Case and R. Edwards (2008, personal communication)]. To the extent that the analogy between midlatitude continental supercells and TC rainband supercells holds, the addition of high-resolution ocean-based surface observations would also be of use in the assessment of the possible roles of thunderstorm outflows in organizing rainband convection and possibly in baroclinically generating horizontal vorticity.

We intend to perform future research using numerical simulations to specifically isolate the possible roles of several features that appeared to be important during Ivan’s landfall. First, the role of the sea–land transition upon the local wind field merits attention. And, second, the impact of dry intrusions upon lapse rates within storms’ interiors deserves further study. Sensitivity tests, targeting both the broader mesoscale environment and the local convective processes, will be of great use in this regard. The overriding aim of such research is to improve our understanding of the unique processes governing TC supercells and tornadoes, and thereby enhance the situational awareness and skill of human forecasters.

Acknowledgments

This work was partially supported by a grant from the Faculty Research and Professional Development Fund at North Carolina State University, and by the National Oceanic and Atmospheric Administration's Collaborative Science, Technology & Applied Research (CSTAR) program under Award NA07NWS4680002. The GOES images are courtesy of the Naval Research Lab Tropical Cyclone Web site. The Storm Data (NCDC 2004) results were acquired and plotted using Severe Plot v. 2.0, which was created and is maintained by John Hart of the Storm Prediction Center. The authors acknowledge thoughtful reviews from Paul Markowski and two anonymous reviewers, as well as helpful consultation with Mark Powell.

REFERENCES

  • Bogner, P. B., , Barnes G. M. , , and Franklin J. L. , 2000: Conditional instability and shear for six hurricanes over the Atlantic Ocean. Wea. Forecasting, 15 , 192207.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bosart, B. N., , Lee W-C. , , and Wakimoto R. M. , 2002: Procedures to improve the accuracy of airborne Doppler radar data. J. Atmos. Oceanic Technol., 19 , 322339.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brandes, E. A., 1984: Relationships between radar-derived thermodynamic variables and tornadogenesis. Mon. Wea. Rev., 112 , 10331052.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bunkers, M. J., , Klimowski B. A. , , Zeitler J. W. , , Thompson R. L. , , and Weisman M. L. , 2000: Predicting supercell motion using a new hodograph technique. Wea. Forecasting, 15 , 6179.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Curtis, L., 2004: Midlevel dry intrusions as a factor in tornado outbreaks associated with land-falling tropical cyclones from the Atlantic and Gulf of Mexico. Wea. Forecasting, 19 , 411427.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davies, J. M., 1990: Midget supercell spawns tornadoes. Weatherwise, 43 (10) 260261.

  • Davies, J. M., 2006: Hurricane and tropical cyclone tornado environments from RUC proximity soundings. Preprints, 23rd Conf. on Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., 12.6A. [Available online at http://ams.confex.com/ams/pdfpapers/115483.pdf.].

    • Search Google Scholar
    • Export Citation
  • Davies-Jones, R., 1984: Streamwise vorticity: The origin of updraft rotation in supercell storms. J. Atmos. Sci., 41 , 29913006.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davies-Jones, R., , Trapp R. J. , , and Bluestein H. B. , 2001: Tornadoes and tornadic storms. Severe Convective Storms, Meteor. Monogr., No. 50, Amer. Meteor. Soc., 167–221.

    • Search Google Scholar
    • Export Citation
  • Doswell C. A. III, , , and Burgess D. W. , 1993: Tornadoes and tornadic storms: A review of conceptual models. The Tornado: Its Structure, Dynamics, Prediction and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 161–172.

    • Search Google Scholar
    • Export Citation
  • Droegemeier, K. K., , Lazarus S. M. , , and Davies-Jones R. , 1993: The influence of helicity on numerically simulated convective storms. Mon. Wea. Rev., 121 , 20052029.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Edwards, R., , and Pietrycha A. E. , 2006: Archetypes for surface baroclinic boundaries influencing tropical cyclone tornado occurrences. Preprints, 23rd Conf. on Severe Local Storms, St. Louis, MO, Amer. Meteor. Soc., P8.2.

    • Search Google Scholar
    • Export Citation
  • Franklin, J. L., , Pasch R. J. , , Avila L. A. , , Beven J. L. II, , Lawrence M. B. , , Stewart S. R. , , and Blake E. S. , 2006: Atlantic hurricane season of 2004. Mon. Wea. Rev., 134 , 9811025.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fujita, T. T., , Watanabe K. , , Tsuchiya K. , , and Shimada M. , 1972: Typhoon-associated tornadoes in Japan and new evidence of suction vortices in a tornado near Tokyo. J. Meteor. Soc. Japan, 50 , 431453.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gamache, J. F., 1997: Evaluation of a fully three-dimensional variational Doppler analysis technique. Preprints, 28th Conf. on Radar Meteorology, Austin, TX, Amer. Meteor. Soc., 422–423.

    • Search Google Scholar
    • Export Citation
  • Gamache, J. F., , Marks F. D. Jr., , and Roux F. , 1995: Comparison of three airborne Doppler sampling techniques with airborne in situ wind observations in Hurricane Gustav (1990). J. Atmos. Oceanic Technol., 12 , 171181.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gentry, R. C., 1983: Genesis of tornadoes associated with hurricanes. Mon. Wea. Rev., 111 , 17931805.

  • Greene, D. R., , and Clark R. A. , 1972: Vertically integrated liquid water—A new analysis tool. Mon. Wea. Rev., 100 , 548552.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hagemeyer, B. C., 1998: Significant tornado events associated with tropical and hybrid cyclones in Florida. Preprints, 16th Conf. on Weather Analysis and Forecasting, Phoenix, AZ, Amer. Meteor. Soc., 4–6.

    • Search Google Scholar
    • Export Citation
  • Hill, E. L., , Malkin W. , , and Shulz W. A. Jr., 1966: Tornadoes associated with cyclones of tropical origin—Practical features. J. Appl. Meteor., 5 , 745763.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jorgensen, D. P., 1984: Mesoscale and convective-scale characteristics of mature hurricanes. Part I: General observations by research aircraft. J. Atmos. Sci., 41 , 12681285.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Jorgensen, D. P., , Hildebrand P. H. , , and Frusch C. L. , 1983: Feasibility test of an airborne pulse-Doppler meteorological radar. J. Climate Appl. Meteor., 22 , 744757.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kennedy, P. C., , Westcott N. E. , , and Scott R. W. , 1993: Single-Doppler radar observations of a minisupercell tornadic thunderstorm. Mon. Wea. Rev., 121 , 18601870.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Klemp, J. B., 1987: Dynamics of tornadic thunderstorms. Annu. Rev. Fluid Mech., 19 , 369402.

  • Klemp, J. B., , Wilhelmson R. B. , , and Ray P. S. , 1981: Observed and numerically simulated structure of a mature supercell thunderstorm. J. Atmos. Sci., 38 , 15581580.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Markowski, P. M., , Rasmussen E. N. , , and Straka J. M. , 1998a: The occurrence of tornadoes in supercells interacting with boundaries during VORTEX-95. Wea. Forecasting, 13 , 852859.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Markowski, P. M., , Straka J. M. , , Rasmussen E. N. , , and Blanchard D. O. , 1998b: Variability of storm-relative helicity during VORTEX. Mon. Wea. Rev., 126 , 29592971.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Markowski, P. M., , Straka J. M. , , and Rasmussen E. N. , 2002: Direct surface thermodynamic observations within the rear-flank downdrafts of nontornadic and tornadic supercells. Mon. Wea. Rev., 130 , 16921721.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Markowski, P. M., , Hannon C. , , Frame J. , , Lancaster E. , , Pietrycha A. , , Edwards R. , , and Thompson R. L. , 2003: Characteristics of vertical wind profiles near supercells obtained from the Rapid Update Cycle. Wea. Forecasting, 18 , 12621272.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Marks F. D. Jr., , , Houze R. A. Jr., , and Gamache J. F. , 1992: Dual-aircraft investigation of the inner core of Hurricane Norbert. Part I: Kinematic structure. J. Atmos. Sci., 49 , 919942.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , 1987: Observations of the hurricane “Danny” tornado outbreak of 16 August 1985. Mon. Wea. Rev., 115 , 12061223.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , 1991: Buoyancy and shear characteristics of hurricane-tornado environments. Mon. Wea. Rev., 119 , 19541978.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , 1993: Observations and simulations of hurricane-spawned tornadic storms. The Tornado: Its Structure, Dynamics, Prediction and Hazards, Geophys. Monogr., Vol. 79, Amer. Geophys. Union, 119–142.

    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , , and Weisman M. L. , 1996: Simulation of shallow supercell storms in landfalling hurricane environments. Mon. Wea. Rev., 124 , 408429.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , , and Weisman M. L. , 2001: The sensitivity of simulated supercell structure and intensity to variations in the shapes of environmental buoyancy and shear profiles. Mon. Wea. Rev., 129 , 664687.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • McCaul E. W. Jr., , , Buechler D. E. , , Goodman S. J. , , and Cammarata M. , 2004: Doppler radar and lightning network observations of a severe outbreak of tropical cyclone tornadoes. Mon. Wea. Rev., 132 , 17471763.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Molinari, J., , and Vollaro D. , 2008: Extreme helicity and intense convective towers in Hurricane Bonnie. Mon. Wea. Rev., 136 , 43554372.

  • NCDC, 2004: Storm Data. Vol. 46, No. 9, 262 pp. [Available from National Climatic Data Center, 151 Patton Ave., Asheville, NC 28801-5001.].

  • Novlan, D. J., , and Gray W. M. , 1974: Hurricane-spawned tornadoes. Mon. Wea. Rev., 102 , 476488.

  • Pearson, A. D., , and Sadowski A. F. , 1965: Hurricane-induced tornadoes and their distribution. Mon. Wea. Rev., 93 , 461464.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Powell, M. D., , and Houston S. H. , 1998: Surface wind fields of 1995 Hurricanes Erin, Opal, Luis, Marilyn, and Roxanne at landfall. Mon. Wea. Rev., 126 , 12591273.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rao, G. V., , Scheck J. W. , , Edwards R. , , and Schaefer J. T. , 2005: Structures of mesocirculations producing tornadoes associated with Tropical Cyclone Francis (1998). Pure Appl. Geophys., 162 , 16271641.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Rasmussen, E. N., , and Blanchard D. O. , 1998: A baseline climatology of sounding-derived supercell and tornado forecast parameters. Wea. Forecasting, 13 , 11481164.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reasor, P. D., , Montgomery M. T. , , Marks F. D. Jr., , and Gamache J. F. , 2000: Low-wavenumber structure and evolution of the hurricane inner core observed by airborne dual-Doppler radar. Mon. Wea. Rev., 128 , 16531680.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Reasor, P. D., , Eastin M. D. , , and Gamache J. F. , 2009: Rapidly intensifying Hurricane Guillermo (1997). Part I: Low-wavenumber structure and evolution. Mon. Wea. Rev., 137 , 603631.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schneider, D., , and Sharp S. , 2007: Radar signatures of tropical cyclone tornadoes in central North Carolina. Wea. Forecasting, 22 , 278286.

  • Smith, J. S., 1965: The hurricane-tornado. Mon. Wea. Rev., 93 , 453459.

  • Spratt, S. M., , Sharp D. W. , , Welsh P. , , Sandrik A. , , Alsheimer F. , , and Paxton C. , 1997: A WSR-88D assessment of tropical cyclone outer rainband tornadoes. Wea. Forecasting, 12 , 479501.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Suzuki, O., , Niino H. , , Ohno H. , , and Nirasawa H. , 2000: Tornado-producing minisupercells associated with Typhoon 9019. Mon. Wea. Rev., 128 , 18681882.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Testud, J., , Hildebrand P. H. , , and Lee W-C. , 1995: A procedure to correct airborne Doppler radar data for navigation errors using the echo returned from the earth's surface. J. Atmos. Oceanic Technol., 12 , 800820.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., , Edwards R. , , Hart J. A. , , Elmore K. L. , , and Markowski P. , 2003: Close proximity soundings with supercell environments obtained from the Rapid Update Cycle. Wea. Forecasting, 18 , 12431261.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Thompson, R. L., , Edwards R. , , and Mead C. M. , 2004: An update to the supercell composite and significant tornado parameters. Preprints, 22nd Conf. on Severe Local Storms, Hyannis, MA, Amer. Meteor. Soc., P8.1. [Available online at http://ams.confex.com/ams/pdfpapers/82100.pdf.].

    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., , and Klemp J. B. , 1982: The dependence of numerically simulated convective storms on vertical wind shear and buoyancy. Mon. Wea. Rev., 110 , 504520.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Weisman, M. L., , and Klemp J. B. , 1984: The structure and classification of numerically simulated convective storms in directionally varying wind shears. Mon. Wea. Rev., 112 , 24792498.

    • Crossref
    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Tornado reports (derived from NCDC 2004) from 0800 UTC 15 Sep through 1400 UTC 16 Sep 2004. Of the 31 tornado reports, 20 were rated F0 in intensity, 9 were rated F1, and 2 were rated F2. Rectangles enclose the reports issued from four radar-analyzed storm cells that crossed over from sea to land (Fig. 7). The reports are numbered chronologically. The letters correspond to the cells identified in Fig. 6 (for A and B) and Fig. 7.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 2.
Fig. 2.

Selected track segments and maximum surface winds for Hurricanes Ivan and Jeanne. The date and hour (dd/hh) are shown for selected 6-h locations (filled circles) for each storm. The maximum surface winds (m s−1) are shown for each 6-h location during the prelandfall periods examined for Ivan (red) and Jeanne (blue).

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 3.
Fig. 3.

Regional WSI NOWRAD composite of radar reflectivity (dBZ) (a) for Hurricane Ivan at 2100 UTC 15 Sep 2004 and (b) for Hurricane Jeanne at 0000 UTC 26 Sep 2004.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 4.
Fig. 4.

Base-scan radar reflectivity (dBZ) from (a) KTLH at 1804 UTC 15 Sep 2004 and (b) KTBW at 1803 UTC 15 Sep 2004. The location of the dual-Doppler analysis in Fig. 5 is highlighted with a black square.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 5.
Fig. 5.

(a) Radar reflectivity (color shading) and storm-relative dual-Doppler wind vectors at 2-km altitude in the target outer rainband of Hurricane Ivan at 1804 UTC 15 Sep 2004 located over 100 km offshore. (b) Radar reflectivity (grayscale shading), upward vertical velocity (red contours), and positive vertical vorticity (blue contours) at 2-km altitude. (c) Vertical velocity (red contours), storm-relative winds (black vectors), and horizontal vorticity (blue vectors) at 1-km altitude. Vectors are shown at every other grid point. (d) Vertical cross section along N–S of radar reflectivity (gray shading), vertical velocity (red), and vertical vorticity (blue). Radar reflectivity is shaded at the 0-, 10-, 20-, 30-, and 40-dBZ levels; vertical velocity is contoured at 2, 4, 6, and 8 m s−1; and vertical vorticity is contoured at 3, 5, 7, and 9 × 10−3 s−1. The three black boxes denote minisupercells A, B, and C (as referred to in the text; cf. Fig. 6). Reference vectors for the wind and horizontal vorticity are also shown.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 6.
Fig. 6.

(a) Locations of minisupercells A, B, and C (cf. Fig. 5) between 1800 and 2100 UTC 15 Sep 2004 as determined from the KTLH WSR-88D. Radar reflectivity at 0.5° elevation for cells A, B, and C as viewed from KTLH during (b) the dual-Doppler analysis at 1804 UTC, (c) the first mesocyclone detection (in cell B) at 1945 UTC, and (d) the first tornado reported (from cell A) at 2044 UTC. Zoomed view of (e) radar reflectivity and (f) Doppler velocities at 0.5° elevation in cell A at 2044 UTC. The black box in (a) denotes the approximate dual-Doppler domain (Fig. 5). The black circulation markers (circles) denote radar-detected mesocyclones.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 7.
Fig. 7.

(top) Temporal evolution of range from the KTLH radar and (bottom) 30-min-averaged midlevel rotational velocity (Vrot) for the four analyzed tornadic cells shown in Fig. 1. Cells A and B are those shown in Fig. 6; cells D and E occurred several hours later and are not shown in the other radar figures. The tornadoes produced by these cells are outlined in Fig. 1. For Vrot, the mean trend line is shown, with vertical bars indicating the range of instantaneous values. Time = 0 min represents the time of each cell’s landfall, with negative values representing min before cell landfall. The small triangles along the time axis point to the times of initial tornado reports after cell landfall for all four analyzed supercells. The large triangle represents the average initial tornado report time after cell landfall.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 8.
Fig. 8.

Distribution by quadrant of the soundings used for the thermodynamic and wind analyses. The soundings are divided into two time windows (approximately 6 h each) for both Ivan and Jeanne. Within each quadrant, the number of land-based rawinsondes (from stations on the windward coast in the right-front quadrant only) and the number of sea-based research dropsondes are indicated, along with the group’s average distance from the storm center. Quadrants are defined with respect to the observed TC motion vectors; the right-front quadrant is highlighted since it is the basis for most of the comparisons in the text. The data for subsets of these soundings are in Figs. 9 –12 and Tables 1, 2, 4, and 5.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 9.
Fig. 9.

Averaged hodographs (m s−1) for quadrants relative to storm motion of Ivan (sea only). Plotted numbers represent observing heights in km AGL. Related indices are given in Table 1.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 10.
Fig. 10.

Averaged hodographs (m s−1) for distance (km) away from storm center in the RF quadrant of Ivan. Plotted numbers represent observing heights in km AGL. The range bins were defined such that there were a roughly even number of soundings in each bin, and so that the increments among the bins’ mean ranges were roughly the same. Related indices are given in Table 2. Note: the weakly negative (anticyclonic) near-surface values of Vtan at long range are mainly due to the subtraction of Ivan’s forward motion from RF quadrant soundings on the periphery of Ivan’s cyclonic circulation.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 11.
Fig. 11.

Averaged hodographs (m s−1) for sea and land regimes in the RF quadrant of Ivan. Plotted numbers represent observing heights in km AGL. Related indices are given in Table 4. Note: the weakly negative near-surface values of Vtan over land are partly due to the subtraction of Ivan’s forward motion from several RF quadrant soundings at greater distance from Ivan’s center (cf. Fig. 10).

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 12.
Fig. 12.

Averaged hodographs (m s−1) for early (sea and land) and late (land only) time regimes (UTC) in the RF quadrant of Jeanne. Plotted numbers represent observing heights in km AGL. Related indices are given in Table 5.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Fig. 13.
Fig. 13.

GOES water vapor imagery for (a) Hurricane Ivan at 1815 UTC 15 Sep 2004 and (c) Hurricane Jeanne at 0015 UTC 26 Sep 2004. Skew T diagrams for (b) the global positioning system (GPS) dropsonde deployed at 1807 UTC 15 Sep and (d) the Jacksonville, FL (KJAX), rawindsonde launched at 0000 UTC 26 Sep within the respective dry air intrusions. Black circles in (a) and (c) denote locations of representative soundings.

Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2222146.1

Table 1.

Inventory of averaged forecasting and nowcasting parameters, computed with sea-based soundings (only) from Hurricane Ivan’s four quadrants. The quadrants were defined with respect to Ivan’s overall motion. The distribution of the soundings is summarized by Fig. 8, and the text explains the selection of soundings and the computation of the composite indices. The corresponding hodographs are shown in Fig. 9. The original vs updated formulations of SCP–STP are discussed in footnote 1.

Table 1.
Table 2.

Inventory of averaged forecasting and nowcasting parameters, computed with soundings from Hurricane Ivan’s RF quadrant, and stratified by range from Ivan’s center. The range bins were defined such that there were a roughly even number of soundings in each bin, and so that the increments among the bins’ mean ranges were roughly the same. The corresponding hodographs are shown in Fig. 10. The original vs updated formulations of SCP–STP are discussed in footnote 1.

Table 2.
Table 3.

Values for the 0–500-km range bin in Ivan’s RF quadrant (cf. Table 2), in comparison to values reported by other studies of TCs. The values from McCaul (1991) are soundings from near TC tornadoes at ranges of less than 300 km and within a TC’s RF quadrant. The values from Bogner et al. (2000) are from their subset of category 4 hurricane soundings. The values from Molinari and Vollaro (2008) are from four soundings very near strong deep convective cells within Hurricane Bonnie (the soundings they identified as D2, D5, D9, and D10). For McCaul (1991) and Molinari and Vollaro (2008), the 0–1-km shear vector magnitudes were estimated from one of their figures. The same is true of the 0–6-km shear vector magnitudes from McCaul (1991) and Bogner et al. (2000).

Table 3.
Table 4.

Inventory of averaged forecasting and nowcasting parameters, computed with soundings from Hurricane Ivan’s RF quadrant, and stratified by location (land-based upsondes vs sea-based dropsondes). The corresponding hodographs are shown in Fig. 11. The original vs updated formulations of SCP–STP are discussed in footnote 1.

Table 4.
Table 5.

Inventory of averaged forecasting and nowcasting parameters, computed with soundings from Hurricane Jeanne’s RF quadrant, and stratified by time. As shown in Fig. 8 and discussed in the text, the earlier interval included both land-based upsondes and sea-based dropsondes. The later interval included only land-based upsondes. The corresponding hodographs are shown in Fig. 12. The original vs updated formulations of SCP–STP are discussed in footnote 1.

Table 5.

1

Thompson et al. (2004) presented even more skillful “updated” formulations of SCP and STP, which incorporate information about the depth of a storm’s “effective” inflow layer. For this study, we computed both the original and updated forms of SCP–STP. Because missing and quality-flagged data occurred in some of our dropsondes (section 3), there were undesirable uncertainties in the determination of this effective storm inflow layer. Even so, we report the updated (Thompson et al. 2004) values in our tables because they represent the current operational formulations used by the Storm Prediction Center. However, the values discussed in the text are exclusively those using the Thompson et al. (2003) formulations [shown in our Eqs. (1) and (2)], in which we have more confidence.

2

Doswell and Burgess (1993) defined supercells as those storms with “deep, persistent mesocyclones.” Since the work of Brandes (1984), ζ ≥ 10−2 s−1 has often been used as a threshold for the definition of a mesocyclone. Although the vertical vorticity maxima in Fig. 5 are a bit lower than this arbitrary threshold, it must be remembered that (a) these miniature supercells are smaller than typical Great Plains supercells (possibly leading to undersampling of the peak radial velocities in the mesocyclone) and (b) there is smoothing implicit in the dual-Doppler analysis, which reduces the magnitudes of spatial derivatives such as ζ.

3

We did not attempt to quantify the effect of increasing radar beam height with range, but because we always assessed Vrot at roughly 4.5 km AGL, the impact on our results should be minimal.

4

Notably, more than 30 tornadoes occurred in the Carolinas on 27–28 September in association with Jeanne (but long after landfall). We caution readers that our results do not address the potential for tornadoes on subsequent days from TCs that are continental.

Save