Precursors to Southwest Florida Warm Season Tornado
Development
Jennifer M. Collins1,
Charles H. Paxton1,2 and
Alicia N. Williams1
1
University of South
2 National
Weather Service,
(Manuscript received 29 May 2009; in final form 16 January
2010)
ABSTRACT
Predicting
and warning for tornadoes developing near the complex coastline of urban Lee
and Charlotte counties in Southwest Florida is often a challenge. The closest National Weather Service Doppler
radar is 130-180 km to the northwest.
Four warm season tornado days were initially examined from coastal
southwest Florida. Those cases showed a
number of similarities in the synoptic and mesoscale patterns and processes
leading to tornado development. Easterly
flow and gulf coast sea breeze development interacted with local topography to
create cyclonic mesoscale circulations around 50 km in diameter. These circulations likely lead to more
predictable boundary collisions and enhanced convection with strong updrafts
capable of supporting brief tornadoes. An
additional nineteen case days were gathered by collecting dates of tornadoes in
Lee and Charlotte counties and parsed to include only those days with similar
vertical wind profiles. To gain more insight into the patterns of various
interactions, the cases were composited to show ambient flows and the degree of
instability. Thirty warm season tornado events that occurred on 23 different
days from 1980-2008 are examined from coastal southwest Florida to identify
synoptic and mesoscale environments associated with tornado development. The results of this study should help
forecasters identify the profiles conducive to southwest Florida tornado
development.
_______________
1.
Introduction
Predicting and warning for warm season (May-Sep)
tornadoes developing near the complex coastline of urban Lee and Charlotte
counties in southwest Florida is often a challenge. Sea breeze circulations develop along the
east and west coasts of the
When the subtropical high that typically extends
across the state of Florida during the summer months is north of Lee and
Charlotte counties, easterly flow develops and sea breeze mergers occur toward
the southwest portion of the peninsula.
The nearest National Weather Service (NWS) WSR-88D radar at Ruskin FL near
Tampa Bay (TBW – see Fig. 1) is 130-180 km from the
tornado locations in Lee and Charlotte counties. Although radar reflectivity and velocity
resolution improved to 0.25
km by 0.5 degree in 2008, at this distance, smaller
tornado related circulations may not be well resolved by the radar. In addition, the radar beam overshoots the
lower levels of convection. In the
absence of spotter reports as the tornado is developing, warnings may not be
issued. Many of the tornadoes reported
over southwest Florida develop as waterspouts (Golden 1971) and move onto the
populated land area causing localized damage (Fig. 2). Typically,
Thirty warm season tornado events that occurred on
23 different days from 1980-2008 are examined from coastal southwest Florida to
identify synoptic and mesoscale environments associated with tornado
development. These cases were
characterized by similar flow regimes with significant convective development
during the afternoon and evening hours. Interestingly, none of the May cases
fit this pattern. Figure 3 shows the touchdown locations and paths of
the longer track tornadoes, and some symbols represent more than one tornado
touchdown location. An understanding of
these cases is expected to lead to improved tornado forecasts and warnings by
operational meteorologists.
2.
Data collection and methodology
Objectively analyzed
maps were obtained from Plymouth State University Weather Center (2009) to
determine surface conditions. Local
Analysis and Prediction (LAPS) software ingests surface mesonet data to produce
the wind plots and streamlines analyses.
The LAPS analyses are less reliable at places and times where data are sparse
such as over the Gulf of Mexico or prior to the widespread use of automated surface
observing systems. Skew-T atmospheric sounding data were
collected from the University of Wyoming (2009) for the Tampa Bay radiosonde
site at Ruskin, FL (TBW – see Fig. 1) area to examine
sounding characteristics. A composite of
the sounding data was created by taking the mean of all observations within +/-
5 hPa for every 10 hPa level. No
sounding observations were used twice when more than one tornado occurred on a
case day. We analyzed common variables
that indicate moisture and instability, including virtual convective available
potential energy (CAPEv) that lifts
the mean parcel in the lowest 500 m above ground level (Doswell and Rasmussen
1994), the virtual lifted index (LIv),
precipitable water (PW, Benwell 1965), and vertical wind profiles. Mean vertical vector winds were calculated
using a vector average of the significant levels in the layers. The mean wind values are a vector mean of
both mandatory and significant levels.
No vertical weighting scheme was used, therefore each value ranks
equally. The circulations which produced
tornados were examined by viewing WSR-88D radar storm relative motion (SRM)
using the NWS Weather Event Simulator (Magsig and Page 2002) and GR2Analyst
(Gibson 2008). GR2Analyst is a
commercial radar analysis software package that incorporates level II WSR-88D
radar data and is widely used by NWS forecasters. The radar data were collected from the
NCDC. The SRM product
which was used subtracts the average motion of all radar identified storm cells
from the preceding volume scan to better highlight storm scale
circulations. When the radar calculated
storm motion is not representative of some of the convective elements, then
storm scale circulations, particularly weaker ones, may not be as
apparent. For the cases shown, the
default storm motion did not affect identification of storm scale circulations
using the SRM. Visible
satellite data were obtained via email from the Cooperative Institute for
Meteorological Satellite Studies (CIMSS).
Composite plots of
geopotential heights at 1000, 500, and 300 hPa were produced from NCEP/NCAR
Reanalysis data (Kalnay et al. 1996) through the NOAA/ESRL Physical Sciences
Division interactive plotting and analysis pages (PSD 2009).
3. Analysis and Discussion
Four warm season tornado days from 2006-2008 were
initially examined from coastal southwest Florida. The synoptic pattern leading
to tornado development for these cases occurs with dominant low level (surface
to 700 hPa) easterly flow (Fig. 1). This easterly flow shown at 1600-1700 UTC 08
June 2008 (Fig. 4a-b) weakens along Florida's west
coast in response to diurnal warming over land areas and a sea breeze
circulation begins from 1800-1900 UTC (Fig. 4c-d). A northwesterly sea breeze develops to the
north while to the south the coastline shape creates a southerly sea
breeze. This regime creates a convergent
pattern that sometimes evolves into a mesoscale (~50 km) cyclonic circulation
near the coast as shown from 2000-2100 UTC (Fig. 4e-f).
In the cases with a sufficient density
of surface observations, the mesoscale circulation is evident up to 1.5 hours
prior to tornado development. As the
west coast and east coast sea breezes merge, the cyclonic circulation
fades. An additional nineteen case days
from 1980 to 2005 were
gathered by collecting dates of tornadoes in Lee and Charlotte counties and parsed
to include only those days with similar vertical wind profiles. Older cases with a lower density of surface
observations may not resolve the circulation.
The mesoscale circulation is not easily identifiable using satellite
data. This circulation likely enhances
convection at the sea breeze interface and appears to be a precursor to
tornadic development.
a. Synoptic conditions
Surface synoptic conditions varied for the cases but
the common factor was an inverted low pressure trough intersecting the
sub-tropical ridge over peninsular
1. 1800 UTC 16 Sep 2007 (Fig. 5a.):
Synoptic conditions indicate a moderate high pressure ridge which extends
southward along the Mississippi valley.
An inverted trough of low pressure is noted along the southwest coast of
Florida.
2.
1800 UTC 08 Jun 2008 (Fig.5b.): A
high pressure ridge across northeast Florida is intersected by a weak trough
along the west coast.
3. 1800 UTC 13 Jun 2008 (Fig.5c.):
A high pressure ridge extends along the eastern
seaboard across the Florida panhandle and over the Gulf, with an inverted
trough along the west coast of Florida.
East to southeast surface flow continued during this period with the
surface ridge axis north of the area over northern Florida.
b.
Sounding Data
Our area of interest is approximately 120 km from
TBW where the soundings were taken. The
1200 UTC TBW soundings (Table 1) for the twenty-three
case days indicated that the CAPEv
ranged from 31 to 2723 Jkg-1. The LIv
ranged from -7.52°C to 0.32°C. PW ranged from 26 to 52 mm with most of the
cases having mid level (700 to 300 hPa) dry layers. These broad ranges and sometimes low values
indicate that soundings taken hours earlier and up to 120 km away are not
necessarily representative of conditions during tornadogenesis, however they do
establish a general atmospheric pattern on tornado days. The mean vertical winds were more consistent
considering both direction and speed.
Low level vertical wind profiles were generally southeast to northeast
while mid level flow ranged from north to east.
The winds were light averaging 4 ms-1 in the low levels and 5.5
ms-1 in the mid levels. Variability
exists in the soundings but all 23 case days had easterly wind flow in the
lower levels, a northerly wind component in the mid to upper levels, some
instability (CAPEv>0), and the mid
levels had dry layers with 20 to 60 degree Celsius dewpoint depressions. Figure 6a is a TBW sounding that is representative
of the case profiles. The composite
sounding plot of all case days between 1980 and 2008 (Fig.
6b) may be used as a guide for providing tornado outlook information when
that general pattern is met. Low-level
wind flow was east (93 degrees) at 4 ms-1 while mid-level flow had a
variable north to east component averaging 52 degrees at 5.5 ms-1. Mid-level dry layers are noted in the
composite. Other mean parameter values
in the composite sounding are CAPEv
1433 Jkg-1, LIv -4.0°C,
and PW 38 mm.
c.
Radar, Streamlines, and Tornado Damage Descriptions
Three sample cases of WSR-88D Doppler radar storm
relative motion (SRM) and mesoscale circulations depicted in wind plots and
streamline analyses from LAPS are shown in Figs. 7a-c. These cases show similar precursor patterns
up to an hour or more in advance of tornadogenesis, as indicated by the LAPS
overlays which precede the reflectivity times.
Easterly gradient flow and gulf coast sea breeze development interact
with local geography to create a brief, up to an hour long, cyclonic coastal
mesocirculation. The following
descriptions are based on Storm Data (NCDC 2009).
The 2100 UTC 16 September 2007 streamlines (Fig. 7a) show a
broad area of cyclonic circulation centered within the red circle During the next hour thunderstorms developed rapidly
northward along the sea breeze producing a waterspout that moved onto Fort
Myers Beach at 2225 UTC and traveled north until 2227 UTC causing damage to a
hotel and tiki bar. This funnel
retracted and reappeared six minutes later over
The 2100 UTC 08 June 2008 streamlines (Fig. 7b) show the precursor mesocyclonic circulation
centered over northeastern Lee County. Under easterly surface flow, convection began
along the convergent Lake Okeechobee breeze boundaries, creating an outflow
boundary that propagated west. The outflow boundary then intersected the west
coast sea breeze briefly amplifying convection that moved north and produced
the tornado. The 2102 UTC SRM (Fig. 7b) shows a rotational velocity couplet associated
with the tornado that was produced by two intersecting boundaries over northern
Lee County near Bayshore Manor. Tornado
wind speeds were estimated as EF0, but with a small pocket of EF1 damage,
mostly occurring in a mobile home community.
One injury occurred. Minor damage
was sustained to carport roofs and sheds.
The tornado path length was 8 km, 30 m wide, and lasted from 2055
UTC-2115 UTC. In this case the
mesocyclonic circulation developed within the hour preceding the tornado but
observation data didn’t show the complete picture until just before the tornado
developed.
The 2000 UTC 13 June 2008 streamlines (Fig.
7c) again show a
precursor cyclonic circulation in the vicinity of the tornado. This tornado occurred around 2045 UTC on the leading edge of a
thunderstorm gust front. As the outflow boundary intersected the gulf coast sea
breeze convective development toward the north occurred. The brief touchdown which lasted from
2045-2048 UTC affected 12 homes with EF0 damage to shingles and a pool
cage. One home suffered 50% roof damage
and another home sustained minor damage, including a fallen fence, and some
roof damage which led to minor flooding.
The tornado path length was estimated about 500m and the width at 25 m. The 2047 UTC SRM (Fig. 7c)
shows a rotational velocity couplet associated with the tornado over the
Cape Coral area in Lee
County.
The radar loop
of reflectivity and SRM for 16 September 2007 (Fig. 7d)
from GR2Analyst shows convective development northward as the east and west
coast sea breezes merge. This is seen in
the reflectivity loop with propagation of stronger signatures northward. The SRM loop shows the brief thunderstorm
circulation associated with the tornado moving from south to north as the
convection builds northward.
d. GOES
Satellite data.
The
sequence of GOES images from 16 September 2007 (Figs. 8a-c)
shows strong convection propagating northward along the coast and over the Cape
Coral area. This pattern of south to
north convective development and propagation was common in many of the cases
even though the winds from the surface to 500 hPa were typically from the east
to northeast. Satellite loops (Figs. 8d-f) of development for the most recent three cases
illustrate the development over coastal southwest Florida. The three loops of 1 km resolution visible
imagery show similar scenarios with the east coast sea breeze and associated
convection moving across south Florida and merging with the west coast sea
breeze creating convective development thereafter. Localized circulations
associated with the tornadoes that were seen in radar velocity data appear to
be linked with converging outflow boundaries seen in each of the satellite
loops. The east coast sea breeze interacts with Lake Okeechobee and convective
outflow boundaries and becomes more irregularly shaped as it moves
westward. Also evident in the satellite
loops is the stabilizing effect of
e. Composite synoptic patterns.
An examination of the twenty-three
warm-season tornado days between 1980 and 2008 revealed similar surface
patterns as detailed in the composite plots of 1200 UTC geopotential
heights at 1000, 500, and 300 hPa (Fig. 9a-c). Figure
9a shows heights at 1000 hPa associated with a ridge over the mid Atlantic
states, with a long fetch of easterly flow over Florida. This differs from the inverted trough seen in
the 1800 UTC case examples (Fig. 5a-c.) that were 6 hours later. Higher in the atmosphere at 500 hPa (Fig. 9b), the
center of the high is over the lower Mississippi valley ridging eastward over
Florida with weak flow over the Florida peninsula. At 300 hPa (Fig. 9c), the ridge is farther southwest
with a northerly wind component over Florida.
4.
Conclusions
A sampling of recent warm season tornado cases
examined from coastal southwest Florida show
a number of similarities in the synoptic and mesoscale pattern and processes
leading to tornado development. Easterly gradient flow and gulf coast sea
breeze development interact with local geography to create a cyclonic coastal
mesocirculation. This circulation likely
leads to more predictable boundary collisions and enhanced convection with
strong updrafts capable of supporting brief non-supercell tornadoes. Easterly low level flow, east-northeast mid
level flow, and instability to support strong convection are the common
attributes of these cases. Ambient flow
direction and magnitude, and degree of instability are important factors in the
timing of various boundary interactions and resulting circulations, leading to
tornado development along the complex coastline. This populated urban area is particularly challenging
to warn for convection growth and tornado development. Based on preliminary results from this study,
forecasters responsible for issuing warnings for southwest Florida have a
pattern to recognize that develops prior to warm season tornadogenesis.
The next step in this research is to find null cases
over Lee and Charlotte counties where the pattern did not produce tornadoes. Then the Weather Research and Forecasting (WRF)
model will be utilized to examine moisture transport, stability profiles, and
wind profiles as well as explicit predictions of thunderstorm development and
evolution. The WRF model may also provide more insight into the influence of
Lake Okeechobee.
Acknowledgements.
The authors gratefully thank
the University Corporation for Atmospheric Research (UCAR) and the National Weather
Service (NWS) Cooperative Program for Operational Meteorology, Education and
Training (COMET) for providing funding for this project. The authors would like to acknowledge the use
of the Linux cluster provided by Research Computing, University of South
Florida (USF). Many thanks for Dan
Noah’s initiation of this project who worked alongside Dr. Collins to prepare
the grant application. The authors
appreciate the assistance of Tom Whittaker from the Cooperative Institute for
Meteorological Satellite Studies (CIMSS) who assisted us in obtaining the
high-resolution satellite imagery.
Special thanks to David Roache from The University of South Florida who
assisted us with displaying the Skew-T data.
The authors would like to
thank Steven Weiss (Storm Prediction Center) and Andrew Devanas (NWS Key West)
for their suggestions and comments.
REFERENCES
Benwell,
G.G.R., 1965: Estimation and variability of precipitable water. U.S. Dept.
Commerce, Wea. Bur. Tech. Pap. No. 10, 48 pp.
Brady,
R.H., and E. J. Szoke, 1989: A case study of nonmesocyclone tornado development
in northeast
Collins,
W.G., C.H. Paxton, and J.H. Golden, 2000: The 12 July 1995
Doswell,
C.A. III and E. N. Rasmussen, 1994: The effect of neglecting the virtual
temperature correction on
Gibson,
M. S. 2008: GR2Analyst. Available at http://www.grlevelx.com.
Golden,
J.H., 1971: Waterspouts and tornadoes over south
Kalnay, E., and collaborators,
1996: The NCEP/NCAR 40-year reanalysis project. Bull. Am. Meteor. Soc., 77,
437-471.
Lee,
B.D., and R.B. Wilhelmson, 1997: The numerical simulation of nonsupercell
tornadogenesis. Part II: Evolution of a family of tornadoes along a weak
outflow boundary. J.
Atmos. Sci., 45,
2387-2415.
Magsig,
M.A., and E.M. Page, 2002: Using the warning event simulator. Preprints, Interactive Symposium on AWIPS, Orlando,
FL. Amer. Meteor. Soc. J212-J213.
McDonald,
J.R., K.C. Mehta, and S. Mani, 2006: A recommendation for an enhanced Fujita
scale (EF-Scale), revision 2. Wind Science and Engineering,
NCDC, 2009: http://www4.ncdc.noaa.gov/cgi-win/wwcgi.dll?wwevent~storms.
Plymouth State Weather Center, 2009: http://vortex.plymouth.edu
Pielke, R.A.,
1974: A comparison of three-dimensional and two-dimensional
numerical predictions of sea breeze, J.
Atmos. Sci., 31,
1577-1585.
PSD,
2009: http://www.esrl.noaa.gov/psd.
University
of Wyoming, Atmospheric Soundings, 2009: http://weather.uwyo.edu/upperair/sounding.html.
Wakimoto,
R., and J.W. Wilson, 1989: Non-supercell tornadoes. Mon. Wea. Rev., 117, 1113-1140.
Wolf,
R.A., 2004: Tornadogenesis and the role of a lake breeze boundary on 14 June
2003 in northwest
TABLES AND FIGURES
Table
1.
1200
UTC TBW sounding data for
23 warm season tornado days in Lee and Charlotte counties during 1980-2008 including
low level (LL) mean wind (surface to 700 hPa), mid level (ML) mean wind (700 to
300 hPa), CAPEv, virtual lifted index
(LIv) and precipitable water (PW).
![[image1]](Fig1.jpg) Figure 1. Diagram of sea breeze regimes leading to tornado development over southwest Florida. Charlotte (a) and Lee (b) counties are shaded in green, and location of the NWS WSR-88D radar and radiosonde at Ruskin, FL is also shown. |
![[image2]](Fig2.jpg) Figure 2. Cape Coral tornado damage, 16 September 2007. Courtesy of Lee County Emergency Management. |
![[image3]](Fig3.jpg) Figure 3. Tornado touchdown locations and paths (indicated by arrows) of the longer track tornadoes based on NCDC data 1980-2008. |
![[image4a]](Fig4a.jpg) Figure 4a. LAPS and surface observations at 1600 UTC 08 June 2008. |
![[image4b1]](Fig4b.jpg) Figure 4b. LAPS and surface observations at 1700 UTC 08 June 2008. |
![[image4c]](Fig4c.jpg) Figure 4c. LAPS and surface observations at 1800 UTC 08 June 2008. |
![[image4d]](Fig4d.jpg) Figure 4d. LAPS and surface observations at 1900 UTC 08 June 2008. |
![[image4e]](Fig4e.jpg) Figure 4e. LAPS and surface observations at 2000 UTC 08 June 2008. |
![[image4f]](Fig4f.jpg) Figure 4f. LAPS and surface observations at 2100 UTC 08 June 2008. |
![[image5a]](Fig5a.jpg) Figure 5a. Surface sea level pressure (SLP) patterns 1800 UTC 16 September 2007. |
![[image5b]](Fig5b.jpg) Figure 5b. Surface SLP patterns 1800 UTC 08 June 2008. |
![[image5c]](Fig5c.jpg) Figure 5c. Surface SLP patterns 1800 UTC 13 June 2008. |
![[image6a]](Fig6a.jpg) Figure 6a. Ruskin, FL Skew-T 1200 UTC 13 Jun 2008 (University of Wyoming). |
![[image6b]](Fig6b.jpg) Figure 6b. Ruskin, FL composite 1200 UTC Skew-T. |
![[image7a]](Fig7a.jpg) Figure 7a. 16 Sep 2007 2213 UTC SRM and 2100 UTC surface streamlines of wind direction. Inbound (outbound) velocities are represented by shades of green (red) from greater than zero- 24.9 ms-1. The arrow indicates the tornado location and the red circle indicates the mesocyclonic circulation. |
![[image7b]](Fig7b.jpg) Figure 7b. 08 June 2008 2102 UTC SRM and 2100 UTC surface streamlines of wind direction. Inbound (outbound) velocities are represented by shades of green (red) from greater than zero- 24.9 ms |
![[image7c]](Fig7c.jpg) Figure 7c. 13 June 2008 2047 UTC SRM and 2000 UTC surface streamlines of wind direction. Inbound (outbound) velocities are represented by shades of green (red) from greater than zero-24.9 ms-1. The arrow indicates the tornado location and the red circle indicates the mesocyclonic circulation. |
![[image7d]](Fig7d.jpg) Figure 7d. 16 September 2007 2203 - 2253 UTC loop of two panel base reflectivity and storm relative motion. Arrow indicates the tornado location. Diamonds represent relative hail size estimations. |
![[image8a]](Fig8a.jpg) Figure 8a. GOES 12 visible satellite imagery 2131 UTC 16 September 2007. The arrow indicates the tornado location. |
![[image8b]](Fig8b.jpg) Figure 8b. GOES 12 visible satellite imagery 2201 UTC 16 September 2007. The arrow indicates the tornado location. |
![[image8c]](Fig8c.jpg) Figure 8c. GOES 12 visible satellite imagery 2231 UTC 16 September 2007. The arrow indicates the tornado location. |
![[image8d]](Fig8d.jpg) Figure 8d. GOES 12 visible satellite imagery 16 September 2007 loop. The arrow indicates the tornado location. |
![[image8e]](Fig8e.jpg) Figure 8e. GOES 12 visible satellite imagery 08 June 2008 loop. The arrow indicates the tornado location. |
![[image8f]](Fig8f.jpg) Figure 8f. GOES 12 visible satellite imagery 13 June 2008 UTC loop. The arrow indicates the tornado location. |
![[image9a]](Fig9a.jpg) Figure 9a. Composite 1000 hPa geopotential height. |
![[image9b]](Fig9b.jpg) Figure 9b. Composite 500 hPa geopotential height. |
![[image9c]](Fig9c.jpg) Figure 9c. Composite 300 hPa geopotential height. |