1. Introduction

During the pre-dawn hours of Wednesday 6 March 1996, two separate HP supercells produced tornadoes in central Alabama that resulted in six fatalities and extensive property damage to the cities of Selma (Sel) and Montgomery (Mgm). Post-storm aerial and ground surveys revealed that the Dallas County F3 tornado, which dissipated soon after reaching Selma, touched down at approximately 0945 UTC (Fig. 1). The Montgomery damage was caused by a series of downbursts and weak tornadoes, which started at approximately 1110 UTC. This damage included an F2 tornado.

While severe weather was expected across Alabama on 6-7 March, model data indicated it would not begin until late in the day on 6 March. The Storm Prediction Center's Severe Weather Outlook, which was issued on the afternoon of 5 March, and was valid through 1200 UTC 6 March (Fig. 2), paralleled this time frame and placed a slight risk area only as far east as a Tupelo to Jackson, Mississippi line. The slight risk area was shifted eastward to a Huntsville to Tuscaloosa, Alabama line at the next update (06/0130 UTC). The purpose of this paper is to describe what environmental conditions led to the initiation of HP supercells and whether model gridded data gave any indications that such conditions would be present.

2. Synoptic and Mesoscale Features

Model runs for several days prior to the event indicated there would be favorable environmental conditions for strong to severe thunderstorms across Alabama from 0000 UTC 7 March-0000 UTC 8 March. A zonal flow over most of the United States on 5 March would become highly amplified over the next several days as a 500 mb trough moved across the Southern Plains States and into the Lower Mississippi Valley. Ahead of the main upper trough, a short-wave trough moved into north Alabama around 0600 UTC 6 March (Fig. 3). An area of showers and thunderstorms increased rapidly over north Alabama shortly after 0000 UTC 6 March, out ahead of the approaching short-wave trough.

By 0930 UTC 6 March, just prior to the first tornado touchdown, the Storm Total Precipitation Product from the Maxwell AFB (Montgomery) WSR-88D (KMXX) showed 1 to 2 inch rainfall totals across central Alabama (Fig. 4), just north of Selma and Montgomery. A boundary formed along the south edge of the rain mass separating relatively stable air to the north and unmodified air to the south. In the unmodified airmass, southerly winds in excess of 30 knots within the lowest 2 km above the surface were advecting very moist air (surface Td>68F) toward the boundary.

The 1200 UTC 6 March Birmingham (BMX) sounding was more representative of the modified airmass and was only slightly unstable (LI = -1, CAPE = 267 J kg^-1, Energy/Helicity Index = 0.53). By using surface observations from an Automated Surface Observing System (ASOS) in Montgomery and environmental wind data from KMXX (located 30 miles northeast of Montgomery), the BMX sounding was modified in order to get a better representation of the environment near the location of the tornadoes. The modified sounding was much more conducive for supercell formation (LI = -2, CAPE = 821 J kg^-1, Energy/ Helicity Index = 2.05). The Storm Relative Helicity (0-3 km) also increased dramatically from a respectable 342 m² s^-2 to an incredible 539 m² s^-2.

3. Gridded Data

Atmospheric stability over Alabama was examined by utilizing gridded ETA and NGM models. 24-hour gridded data fields from the 1200 UTC 5 March models were compared with the analysis of the 00-hour 1200 UTC 6 March models. In order for HP supercells to develop, three atmospheric conditions are typically present: instability, a deep layer of moisture, and strong vertical shear. A trigger must also be available (i.e. low-level convergence) to initiate convection.

Instability was examined by using the 4-layer lifted index. This index provides a better representation of instability during the cool season when low-level inversions are more common, and will not have the diurnal swings as does the surface-based lifted index. Values at or less than zero indicate a parcel of air, when lifted above the level of free convection, will have sufficient buoyancy to produce updrafts for convective development. Forecast lifted indices were less than zero across central Alabama (Fig. 5a), which compared well with the 00-hour indices (Fig. 5b).

Deep-layered moisture can be determined by looking at several different fields: precipitable water, 1000-500 mb mean relative humidity, theta-e at several pressure layers. The K-index, widely used as a stability index, is also a good measure of mid-level moisture between 850 and 700 mb.

Analysis of K-indices during the cool season can indicate areas of mid-level instability and moisture associated with fast moving short-wave troughs. The models predicted an area of maximum K-indices (Fig. 6a) moving across Alabama that morning. However, both models placed the axis of maximum indices across northern Alabama, around 150 miles north of the 00-hour maximum axis (Fig. 6b). This northward displacement was caused by the short-wave trough moving faster than predicted by the models.

Vertical shear, especially in the lowest 3 km, is important to the formation of mesocylones and tornadic storms. One tool to measure this turning of the winds is helicity. With access to gridded data fields, forecasters have the ability to easily compute helicity. More importantly, Storm-Relative Helicity (SRH) can be computed by subtracting storm motion. This type of helicity gives a better representation of the rotation within the storm. Storm motion was calculated from the gridded data by using the following assumption: 30 degrees to the right/75% of the 850-300 mb mean wind. Both models forecasted an area of high helicity across central Alabama at 1200 UTC (Fig. 7a), even though again, the axis of maximum values was displaced about 150 miles north of the 00-hour maximum axis (Fig. 7b).

It is vital to be aware of mesoscale features in an environment which will support severe thunderstorms. Even though atmospheric conditions were ripe for the development of thunderstorms across central Alabama early that morning, I am confident that supercell thunderstorms would not have developed if there had not been a mesoscale focusing mechanism in place to initiate a concentrated area of low-level convergence. The boundary which formed along the south edge of the rain mass created an enhanced area of low-level convergence and baroclinicity needed to initiate horizontal helicity. This created an environment which produced vertical shear necessary for rotating updrafts.

I was working the day shift public forecast desk on Tuesday, 5 March, and was focused on the severe weather event which was expected to occur late in the day on Wednesday. As I prepared my forecast I did not see any indication of an early morning event in the models. I was shocked at the timing and magnitude of the storms which occurred and wanted to determine the conditions which produced them.

After reviewing the gridded data to prepare this paper, I realized that the synoptic-scale models had indeed suggested some potential for severe thunderstorms. However, the location of the severe weather was shifted farther south than indicated by the models due to the rain-induced mesoscale boundary. The synoptic-scale sets the stage for severe weather, while the timing and location is determined by mesoscale features.

Also see,

The Central Alabama Tornadoes of 6 March 1996 for radar images.

Back to home page

Weather Hazards | Graphical Forecast | Present Weather
Past Weather | Tropical Weather | Contact Us