Introduction

There have been several studies indicating that during the collapse phase of some supercell thunderstorms, a decrease in VIL values and an associated increase in rotational velocities, may be an indicator to the onset of a tornado. One of these studies included a VIL velocity comparison of the Palm Sunday outbreak of March 1994 (Murphy, R., K. Pence, J. Westland, and R. Kilduff, 1994). This study is a brief, but similar look at two tornadic supercells. It is a comparison of VIL, rotational velocity, and storm top associated with tornado occurrence. One storm, on April 8, 1998, produced an F5 killer tornado. The other, on May 9, 1998, only produced two short F0 tornado tracks with relatively minor damage.(map showing tracks, size = 38 kb ).

Methodology

This study used archived data from the NWS Birmingham, Alabama WSR-88D (KBMX) site. The two storms occurred near the same time of day and about one month apart. The data used were from 2330 GMT on 4/8/98 through 0130 GMT on 4/9/98, and 2330 GMT on 5/9/98 through 0145 GMT on 5/10/98. The radar was operating normally and in VCP 11 during both events.

Maximum values of VIL (kg/m²) were obtained from the VIL product, and storm-relative velocities (Vrot) from the SRM product for the April storm and a combination of SRM and base velocity for the May storm. Rotational velocities and shear were obtained at 0.5 and 1.5 degrees using the Doppler radar VR Shear function. Storm top was obtained for each storm from the attributes table.

The Environment

The airmass was warm, moist, and potentially unstable on both April 8 and May 9, 1998. The Storm Prediction Center had placed much of north and central Alabama under a high risk of severe weather on April 8, and under a moderate risk on May 9. On April 8 (0000 GMT April 9), the unmodified Birmingham, Alabama (BMX) sounding, (figure 1a and 1b) indicated an observed CAPE (convective available potential energy) of 2178 J/kg, and a cap of just 0.8 degrees C. The mean wind in the lowest 6 km (at BMX) was 41 kt out of the southwest. The storm-relative helicity (0-3km) was 220 (m/s)², which resulted in an energy/helicity index of 2.82. There also appeared to be a weak east-west thermal boundary across central sections of Alabama resulting from partial cloud cover earlier in the day.

On May 9 (0000 GMT May 10), the unmodified Birmingham, Alabama (BMX) sounding (figure 2a and 2b) indicated an observed CAPE (convective available potential energy) of 438 J/Kg. The mean wind in the lowest 6 km (at BMX) was 34 kt out of the southwest. The storm-relative helicity (0-3km) was 384 (M/S)², which resulted in an energy/helicity index of 0.94. There was a strong cap of 6.3 degrees C, and there was a warm front approaching the area from the south at the time of the sounding.

Although the shear was greater on May 9 than April 8, the April event had a higher mean wind in the lowest 6 km and it was much more unstable. The energy/helicity index of 2.82 during the April event indicated it had much greater potential for the development of severe thunderstorms and tornadoes. Both days had a sheared environment favorable for the development of supercells, and on both days the supercells moved along a thermal boundary. Hart and Korotky (1991) suggested that an energy helicity index of 1.0 is indicative of a significant tornado potential, while Davies (1993) suggested that 2.0 or 2.5 may be a better indicator of a significant tornado event.

Findings

The radar reflectivity displays of the two supercell thunderstorms appeared remarkably similar. They traveled almost the same path near the same time of day, and each displayed an impressive looking hook. If you viewed only the reflectivity display you would think both storms produced the same result, but this was not the case. They both produced tornadoes, but at opposite ends of the Fujita Tornado Intensity Scale. These two storms exhibited a significant decrease in VIL values with a corresponding increase in Vrot before the onset of a tornado, as indicated in other studies. However, this correlation was much more dramatic in the deadly April 8 storm.

Both storms began in Mississippi and moved rapidly eastward (about 45 to 50 kt) into Alabama and across Pickens, Tuscaloosa, and Jefferson counties. The April storm continued strong and produced additional tornadoes, while the May storm diminished over Jefferson county and eventually became non severe. The 0.5 degree center beam height (MSL) of the KBMX WSR 88D radar ranged from about 8500 feet over Pickens county decreasing to 2000 feet over Jefferson county, while at 1.5 degrees it ranged 15500 to 4000 feet. The distance from radar ranged from about 78 nmi to 15 nmi for both storms. To view a 0.5 degree animated radar reflectivity loop of both storms click here, size=1.9 mb. To just see the loops individually click on April 8, 1998 or May 9, 1998 .

To make it easier to relate changes in the various parameters to the onset of a tornado, all the graphics of VIL, Vrot, shear, and storm top, shown in the following sections, also include purple triangles at the bottom of each graphic. These triangles approximate the lifetime of each tornado.

A comparison of VIL, Vrot, and storm top also shows an interesting, but expected correlation. The storm top figure 4 shows a slight decrease during the time of the first tornado, but shows a dramatic decrease during the time the F5 tornado was occurring. It reached a minimum (37,000 ft) near the time of the worst damage. This collapse phase of the supercell correlates well with the decrease in VIL. To reduce clutter, the 1.5 degree Vrot was not included in this graph.

A comparison of shear is shown in figure 5. As expected, it shows the same trend as Vrot. The general increase in shear with time was partially due to the fact that the storm was moving closer to the radar. However the pattern of increasing and decreasing shear closely followed that of the velocity. The greatest shear was observed near the maximum damage on the ground, and was an amazing 0.074/s.

A comparison of VIL, Vrot, and storm top figure 7 shows a correlation between decreasing storm top, decreasing VIL and increasing Vrot, although weaker than the April storm, and the storm top exhibited an overall decreasing trend as the storm moved eastward. One of the strongest correlations with this storm was near 0100 GMT over eastern Tuscaloosa county, but again, this was an area that was not completely surveyed. After about 0140 GMT, the storm top VIL and Vrot all dropped off dramatically.

Conclusion

Both supercells showed a decrease in VIL near or at the time of a significant rise in the Vrot. As suggested in other studies, and also by the decrease in storm top at the same time, this is likely related to the collapse phase of the storms. During the collapse phase of the supercell, there is an increase in the storm's downdrafts and a corresponding decrease in the updraft. It is at this time that you will likely find the strongest straight-line winds and tornadoes.

In the April storm the correlation between VIL, Vrot, storm top, and tornado occurrence was most dramatic. Figure 3 shows that the highest Vrot, and the tornado, both occurred after the VIL values decreased. This was most evident at the time the F5 tornado was occurring, and the sharply decreasing storm top is consistent with the collapse phase of the supercells. These trends in VIL, Vrot, and storm top were not as evident in the May storm, and may not always be present. They may also be difficult to detect in real time, but perhaps this could one day become part of one of the WSR-88D algorithms.

References

Davies, J. M., 1993: Hourly Helicity, Instability, & EHI in Forecasting Supercell Tornadoes. Preprints, 17th Conference on Severe Local Storms, Saint Louis, Missouri, Amer. Meteor. Soc., 107-111.

Hart J. A., Korotky W. D., and G. Jackson, 1998 : The SHARP Workstation-v2/98. A Skew T/hodograph analysis and research program for the IBM and compatible PC.

Murphy, R., K. Pence, J. Westland, and R. Kilduff, 1994: A comparison study of VIL vs rotational velocity associated with tornadic thunderstorms. Postprints, The First WSR-88D Users' Conference, Norman, Oklahoma, OSF/JSPO, 259-265.