1. Introduction

Together with the National Aeronautics and Space Administration (NASA) at the Kennedy Space Center, FL (KSC), the National Weather Service (NWS) at Melbourne, FL (MLB) configured a local adaptation of the Advanced Regional Prediction System (ARPS) Data Analysis System (ADAS) to facilitate short-term forecasting efforts during hazardous weather situations. The ADAS was modified to assimilate nationally and locally available in-situ and remotely-sensed observational data into a series of high-resolution gridded analyses centered around east central Florida. It is initialized with the 40-km Rapid Update Cycle Model (RUC, now 20-km RUC) as the background and assimilates a variety of information such as visible and infra-red satellite, Doppler radar, area surface observations (from several networked sources), ascent/descent aircraft observations, and data from the NASA/Kennedy Space Center mesonet, (including tower winds, profilers, etc.) for a full volume depiction of the local atmosphere. The MLB ADAS was configured with nested domains with spatial resolutions of 10 km and 2 km respectfully, and with a temporal resolution of 15-min. Refer to Case et al. (2002) for a thorough description of the MLB ADAS configuration.

Operationally at MLB, ADAS diagnostics have been used to increase forecaster confidence regarding the threat of severe local storms within the 0 to 3 hour time frame by providing detailed analyses of the pre-storm environment. This has greatly enhanced the local outlook process. More so, it has proven to be a valuable contribution to the warning decision making process within the 0 to 60 minute time frame, depicting the evolution of the near-storm environment during difficult severe local storm situations.

On the morning of 14 September 2001, four tornadoes struck Brevard County in east central Florida resulting in several swaths of damage. These tornadoes were associated with Tropical Storm Gabrielle which tracked west to east across the Florida peninsula from near Sarasota to Cape Canaveral. Three of the tornadoes were F0 on the Fujita scale, and one reached F1 intensity as it moved through the town of Cocoa, FL, causing significant damage to multiple homes and business, while totally unroofing one house. Importantly, three of the tornadoes were associated with the same mini-supercell as it moved onshore and through Brevard's coastal zone, crossing barrier islands and an intracoastal waterway while advancing several miles inland (1125-1248 UTC). At one marina, significant damage occurred as a large section of the main boat house roof was peeled back and many boats were tossed about. Confirming eye-witness reports were received from storm spotters and weather observers at Patrick Air Force Base, FL. These tornadoes occurred within the leading outer convection of Gabrielle.

In the hours and moments leading up to the tornadoes, ADAS output was used to facilitate the assessment of the local tornado threat by revealing the evolution of various stability and shear parameters. In post analysis, 0-3 km helicity depictions were generated to examine potential utility for future tropical cyclone events. This study will examine the value of real-time high-resolution diagnostics during the Gabrielle tornado event as it occurred over east central Florida and their contribution to the local outlook and warning efforts.

2. Tropical Cyclone Gabrielle

Reviewing the synoptic history of Gabrielle prior to 14 September 2001 reveals that the cyclone was non-tropical in origin (Lawrence and Blake 2001). Its origin was associated with a mid-level cut-off low over Florida, where a subsequent surface low formed over the southeastern Gulf of Mexico. The resulting convection became sufficiently organized to classify the system as a depression on 11 September. Under weak steering currents, the circulation moved in a small counterclockwise loop over the southeastern Gulf of Mexico for two and a half days while it gradually strengthened. It then reached tropical storm intensity on the 13^th as a hybrid cyclone. At that time, a mid-level trough in the westerlies was moving into the eastern United States and Gabrielle began moving northeast with  gradual increasing speed and was a contributing factor in the 0900 UTC official forecast (Fig. 1). At that time the center position was about 140 km (85 nmi) west of Naples, FL and forecast to move northeast at 17 km hr^-1 (10 kt). The maximum sustained winds were near 110 km hr^-1 (60 kt) with no significant change expected in the intensity before landfall.

3. Facilitating the Local Tornado Outlook

Both historically and statistically, tropical cyclones which pass through the eastern Gulf of Mexico, especially hybrid storms such as Gabrielle, present themselves as tornado threats to the Florida peninsula (Hagemeyer 1997). In fact, depending on the exact track and intensity, the threat of tornadoes may be the primary hazard for certain locations, particularly at greater distances from the cyclone center in vicinity of the outer rainbands. When employing the TPC's 0900 UTC official forecast, the area of east central Florida between Orlando to Cape Canaveral (north) and Lake Okeechobee to Jupiter Inlet (south) was situated in the favored area for tornadoes within the right front quadrant and about 200 km from the cyclone center (Fig. 1). A dynamic and thermodynamic inspection of the intermediate
0600 UTC upper air soundings (not shown) from Tampa Bay (TBW) and Cape Canaveral (XMR) identified an environment marked by deep moisture and increasing parcel buoyancy with increasing distance from the storm center. The available positive buoyant energy above the level of free convection (CAPE) was only 10 J kg^-1 at TBW, yet was remarkably higher at XMR at 1119 10 J kg^-1. With the tropical storm force winds still over the eastern Gulf of Mexico, the 0-6 km average wind was only about 8 m s^-1 (~15 kt) for both locations. To obtain a better understanding of the local environment and its near-term ability to promote (tornadic) supercell development, ADAS analyses were considered between 0600-0900 UTC at 15-min intervals, showing the spatial distribution and temporal evolution of the 560 m wind field, CAPE, 10 m wind, and 0-3 km helicity. Selected images at 0900 UTC have been provided to show the condition of the local environment. Fig. 2 shows the encroachment of tropical storm force winds toward east central Florida from the west with the inference of increasing shear in the lower-levels. Fig. 3 shows the distribution of CAPE with a maximum around 2000 J kg^-1 over the Atlantic waters and a minimum over land, but with a sharp gradient within the east central Florida coastal zone. Finally, Fig. 4 shows the 10 m wind and the pseudo-warm front (gradually lifting north) and separating stratiform from convective precipitation. It also depicts an emerging favorable helicity field with higher values (near 450 m² s^-2) moving toward east central Florida. With a Tornado Watch already issued by the Storm Prediction Center, these analyses enabled MLB forecasters to refine the threat for tornadoes, with particular focus on coastal sections from Cape Canaveral south and in vicinity of the lifting front.

4. Facilitating Local Tornado Warnings

Ideally, frequent inspection of the 10 km ADAS analyses should lead to better situational awareness and severe local storm anticipation. This is also true during the warning phase when using the 2 km ADAS analyses.

By 1048 UTC as seen by MLB radar (not shown), several shallow mesocyclones were apparent within and near a dominant outer rainband just offshore, with another developing mesocyclone located near the intersection of the coast-line and lifting front. These mesocyclones were similar in physical dimension to the typical "mini-supercells" associated with tropical cyclones (McCaul 1990; McCaul 1991; Spratt et al. 1997; Sharp et al. 1997) and present an inherent warning challenge.

By 1115 UTC, 10 minutes prior to the time of the first tornado, the near-storm 0-3 km helicity had greatly increased to 300-400 m² s^-2 (Fig. 5) while area CAPE remained modest but sufficient (not shown). These ADAS depictions clearly show how rapidly the local environment can evolve and the sharp gradients of CAPE and helicity that can exist near tropical cyclone outer rainbands. The developing mesocyclone along the coast continued to move through the area of higher tornado potential, as made evident by a near-storm environmental analysis which closely matches the storm scale. Then by 1208 UTC, it had acquired a small but classic low-level "hook" reflectivity signature and attained a sampled shear of 0.120 s^-1 across a tight cyclonic couplet as it passed within a few km of the MLB WSR-88D during the time of the second F0 tornado.

5. Summary and Conclusions

It has been shown that the real-time use of high-resolution diagnostics can significantly enhance both the local outlook and local warning efforts during difficult severe local storm situations. With high-density observational data being assimilated and analyzed every 15 minutes, especially at temporal and spatial resolutions that closely match the storm scale, there exists genuine opportunities for locally improved warning services and verification scores.

6. References

Case, J.L., J. Manobianco, T.D. Oram, T. Garner, P.F. Blottman, and S.M. Spratt, 2002: Local Data Integration over East-Central Florida Using the ARPS Data Analysis System, Wea. Forecasting, 17, 3-26.

Hagemeyer, B. C., 1997: Peninsular Florida tornado outbreaks. Wea. Forecasting, 12, 399-427.

Lawrence, M.B. and E. S. Blake, 2002: Tropical Cyclone Report Hurricane Gabrielle, National Hurricane Center, WWW version.

McCaul, E. W., Jr., 1990: Simulations of Convective Storms in Hurricane Environments, Preprints, 16th Conference on Severe Local Storms, Amer. Meteor. Soc., Alberta, Canada, pp. 334-339.

McCaul Jr., Eugene W., 1991: Buoyancy and Shear Characteristics of Hurricane-Tornado Environments. Monthly Weather Review: Vol. 119, No. 8, pp. 1954�1978.

Sharp, D. W., J. Medlin, S.M. Spratt, and S.J. Hodanish, 1997: A Spectrum of Outer Spiral Rain Band Mesocyclones Associated with Tropical Cyclones, Preprints, 22nd Conference on Hurricanes and Tropical Meteorology, Amer. Meteor. Soc., Ft. Collins, CO, pp 117-118.

Spratt, S. M., D. W. Sharp, P. Welsh, A. Sandrik, F. Alsheimer, and C. Paxton, 1997: A WSR-88D assessment of tropical cyclone outer rainband tornadoes. Wea. Forecasting, 12, 479-501.

Fig. 1.  The official forecast of Tropical Storm Gabrielle at 0900 UTC on 14 September 2001 issued by the Tropical Prediction Center.  The box highlights the favored area for tornadoes within east central Florida.  The dots show the tornadoes associated with the main supercell between 1125-1248 UTC.


Fig. 2.  The 10 km ADAS analysis of the 560-m winds and streamlines associated with Gabrielle at 0900 UTC on 14 September 2001.


Fig. 3.  The 10 km ADAS analysis of Convective Available Potential Energy at 0900 UTC on 14 September 2001.


Fig. 4.  The 10 km ADAS analysis of 10 m winds and 0-3 km helicity at 0900 UTC on 14 September 2001.


Fig. 5.  The 2 km ADAS analysis of 10 m winds and 0-3 km helicity at 1115 UTC on 14 September 2001.