THE MARCH 1994 TORNADO OUTBREAK IN THE SOUTHEAST U.S.: THE FORECAST PROCESS FROM AN SPC PERSPECTIVE

1. INTRODUCTION

On Palm Sunday, March 27, 1994 an outbreak of killer tornadoes raced east-northeastward across the southeast U.S. mainly from north-central Alabama and northern Georgia to the Carolinas(see Fig. 1). A total of 42 deaths and over 320 injuries were attributed directly to the storms and damage to property exceeded $100 million. Twenty people died in the tornado that struck the Goshen Methodist Church near Piedmont, Alabama. Eighteen fatalities occurred in Georgia and 2 in North Carolina. The supercell that produced the tornado near Piedmont tracked at least 200 miles from east-central Alabama into South Carolina. Not only did the supercells develop very early in the day (the first fatality occurred in Alabama before 1700 UTC), but the dozen or so tornado producing supercells traversed a narrow area typically less than two counties wide. They spawned tornadoes in parallel tracks, some of which were nearly overlapping.

The Storm Prediction Center (SPC) has responsibility for forecasting such events. Given the significance of the outbreak along with the atypical nature of the synoptic pattern, an examination of the SPC response is appropriate.

2. SYNOPTIC OVERVIEW

The outbreak of significant tornadoes across the southeast U.S. on Palm Sunday, March 27, 1994 was not as synoptically evident as many of the outbreak cases in the past. Some of the parameters that SPC looks for when forecasting tornado outbreaks were present, while others were missing. Features absent included a well defined surface low center and a prominent short wave trough. What did stand out as key to supercell formation were the warm sector air mass characteristics, particularly the wind fields.

An examination of the 500 mb chart at 1200 UTC 27 March (see Fig. 2) revealed no significant disturbances embedded in the mean flow. Rather strong west-southwest winds dominated the southeastern U.S. with the main trough position from the north-central U.S. southwestward into the southern Rockies. The tornado outbreak occurred on the anticyclonic periphery of the polar jet which was well to the north.

At the surface at 1500 UTC 27 March (Fig. 3), there was a well defined frontal zone which stretched from southern North Carolina west-southwestward, parallel to the upper flow, to a wave over northwest Alabama. Much of the severe weather focused along the boundary, however it is important to note that initiation of the first long track supercell occurred well south of the surface boundary in an area of very weak capping. The warm sector air mass was potent both thermodynamically and kinematically, and the characteristics of the air mass will be discussed further.

3. SPC PERFORMANCE

SPC began focusing on the affected area in the initial Day Two convective outlook issued at 0800 UTC March 26 (Refer to Table 1 for a chronology). The outlook highlighted the forecasted strong wind fields and indicated that instability would be adequate for a risk of severe thunderstorms over the area eventually affected. However, all numerical models under forecasted the amount of instability which actually became available for storm development. Another problem with the model solutions in this particular case was that they consistently over forecast the strength of a surface low over the Ohio Valley and Great Lakes. As a result, the models shifted the strongest low level wind fields northward away from the gulf coast during Sunday. As it turned out, a well developed surface low did not develop with this system. However, a frontal wave did develop over Mississippi Sunday afternoon and played an important role in the generation of a second round of severe thunderstorms.

The 2nd Day Two severe thunderstorm outlook issued early Saturday afternoon (1900 UTC) addressed the potential for supercell storms along with the possibility of tornadoes.

Early on Sunday morning, the Day One outlook (0700 UTC) again emphasized the strong wind fields favorable for severe thunderstorm development. The outlook indicated that a Moderate risk of severe thunderstorms existed from eastern Texas through the gulf states across Alabama. Though the emphasis was on damaging winds due to the expected rapid storm motion, the potential for supercells and tornadoes was also indicated. The lack of a well defined surface low and pronounced short wave trough moving across the warm sector during the day as well as a forecast of only moderate instability precluded a greater emphasis on the tornado potential. After analyzing the upper air data and examining the relevant soundings taken at 1200 UTC 27 March (Fig. 4), it became apparent that a very potent air mass had set up across the gulf states overnight. The hodographs and storm relative helicity calculations were classic for the development of supercell storms. Helicities ranged from around 300 m² /s² at Jackson, MS with moderate instability, to over 800 m² / s² with marginal instability at Athens, GA.

The interactive capabilities of the VDUC (VAS Data Utilization Center) work station used at SPC were invaluable in fully utilizing the data sources that were available on this day. By modifying the observed morning soundings and model forecast soundings with expected afternoon temperatures and dewpoints, a full appreciation of the potential of this air mass quickly became apparent. Overnight the entire warm sector, extending from Mississippi eastward through all of Georgia, had developed an explosive potential waiting for a trigger. The cap on this air mass was relatively weak and even though there were no discernible short wave troughs and developing synoptic surface lows in the area, SPC's concern was greatly increased based on the observed potent air mass structure. Since some of the parameters typically associated with tornado outbreaks (Miller 1972) were missing (coupled with a lack of mid level dry air and a rather weak cap), it was decided not to upgrade to a High Risk until after examining the morning model runs which would incorporate the more potent initial conditions. However, we felt the risk for significant tornadoes was great enough to considerably increase the emphasis on the tornado threat. As a result, the 1500 UTC outlook discussion strongly emphasized the threat for significant tornadoes. The Moderate risk was also extended eastward along the frontal boundary to include the western Carolinas because the Athens GA sounding indicated a very favorable wind profile for supercell storms and considerable destabilization was anticipated across the region during the day.

It was also indicated in the 1500 UTC Day One outlook that a Public Severe Weather Outlook (PWO) would be issued at 1600 UTC. In the PWO, the emphasis was on the expected widespread area of severe thunderstorms and tornadoes across the central Gulf Coast into the western Carolinas. The threat of particularly intense tornadoes was also mentioned to reflect the increased concern.

During the formulation of the 1500 UTC outlook it was noted on radar, satellite, and by the lightning detectors that thunderstorms were already developing in the warm sector across eastern Mississippi into western Alabama. Not only were the storms forming along the well defined east/west boundary across the northern portions of the two states, but also well to the south. A Mesoscale Discussion (MCD) was issued at 1505 UTC to inform the field offices of our concern regarding these newly developed thunderstorms, and that a tornado watch would be issued shortly.

Tornado Watch #41 was issued at 1518 UTC across eastern Mississippi and the northern half of Alabama. The storms rapidly increased to severe levels across northern Alabama and tornado warnings were issued prior to 1700 UTC by Birmingham and Hunstville, AL. Recognizing the rapid intensification of the storms with indications from the Maxwell Air Force Base, AL WSR-88D (KMXF) that the storms were rotating and likely producing tornadoes, an MCD was issued at 1651 UTC for Georgia indicating that a new tornado watch would soon be needed farther east. A valuable input at this time to the assessment of supercell potential was the very favorable wind structure observed by both the profilers and the WSR-88D VAD Wind Profile (VWP) product (Fig. 5).

At 1700 UTC Tornado Watch #42 was issued for all of northern Georgia and included enhanced wording for the potential for very damaging tornadoes reflecting our concern about the tornadic supercells moving out of Alabama.

While the initial area of tornadic storms was evolving over Alabama into northern Georgia, our attention shifted westward into Mississippi. At SPC we have available in the VDUC workstation the capability of analyzing the VWP data from the WSR-88D radars. We were carefully monitoring the wind profiles off the WSR-88D at Jackson, MS (KJAN) and noted that the helicities were steadily increasing, approaching 500 m2s-2. Meanwhile pressures were dropping rapidly in advance of the frontal wave moving out of Louisiana into central Mississippi. In addition, satellite water vapor imagery and profilers indicated the approach of a mid level(700 MB) short wave trough moving out of eastern Texas. Based on these factors, Tornado Watch #43 was issued for eastern Louisiana and portions of Mississippi. The enhanced wording was not incorporated in this watch because the strongest low level winds were farther east feeding the storms moving out of Alabama into Georgia.

At 1849 UTC an MCD was issued to alert the Carolinas of the serious storm conditions moving toward the region and that a watch would soon be issued. Also in the discussion it was noted that the upcoming Day One outlook would be upgraded to a High Risk.

At 1930 UTC the outlook was issued indicating that there was a High Risk of severe thunderstorms and that the severe thunderstorm outbreak underway would continue. The outlook emphasized the special 1800 UTC soundings taken by both JAN and CKL (Fig. 6) which indicated that winds at all levels were stronger than forecasted by the models, contributing to very high helicities. In addition, the air mass at each location was very unstable with CAPE values around 2500 JKg-1.

At 1926 UTC, Tornado Watch #44 was issued across portion of the western and central Carolinas as well as the remainder of northeast Georgia emphasizing the mesolow and supercells tracking out of Georgia.

Throughout the event, the KJAN and KMXF WSR-88D VWP's and the profilers at Okolona, MS and Winnfield, LA clearly depicted the stronger than forecasted mid level wind fields. These strong winds (70 kt at 700 mb) greatly improved environmental conditions favorable for supercell storms. The near real-time display of both lightning and satellite imagery in VDUC was crucial in observing the very early development of thunderstorms that led to the issuance of Tornado Watch #41. The two available 88D's KJAN and KMXF did an outstanding job detecting the many long-lived supercells during the day. They were also instrumental in confirming that a significant severe weather episode was unfolding. This message was conveyed in the products that were issued.

4. CONCLUSIONS

While classic (text book) tornado outbreaks do occur, they are rather infrequent. In many outbreaks one or more of the "apparently required" parameters is absent. In this respect, the March 27, 1994 outbreak was typical of many tornado episodes. The outbreak occurred without the presence of a deep surface low and strong upper trough. Nevertheless, the thermodynamics and kinematics of the atmosphere evolved into a favorable configuration for the formation of supercells and strong tornadoes. SPC was able to anticipate these important parameters and effectively forecast a significant event with non-classic characteristics.

5. REFERENCES

Miller, R.C., 1972: Notes on Analysis and Severe-Storm Forecasting Procedures of the Air Force Global Weather Central. Tech. Report 200(Rev.), Air Weather Service, Scott AFB, IL.

Natural Disaster Survey Report, Southeastern United States Palm Sunday Tornado Outbreak of March 27, 1994, NOAA Silver Springs MD.