Between 0000 UTC and 0600 UTC October 22, 1996, bands of convection developed in an environment characterized by surface temperatures dropping from 10C to around 0C with strong cold air advection continuing at the surface. The storms rapidly intensified, eventually dropping dime to nickel size hail, copious amounts of sleet, and light snow over parts of west-central Texas before dissipating shortly after midnight. A review of this event revealed several interesting features, including evidence of simultaneous elevated upright convection and conditional symmetric instability (CSI), and a succession of ducted gravity waves, that combined to produce severe weather.
2. The synoptic situation
By 0000 UTC October 22, a surface cold front was located from southeast Oklahoma through north-central Texas, into the Big Bend region of southwest Texas Fig.1. Cold air advection dominated from the surface to the top of the frontal inversion. Above the front, strong warm air advection was ongoing in the mid levels of the atmosphere, ahead of a potent long-wave trough moving across New Mexico Fig.2. The trough was accompanied by a strong jet streak of 55 ms-1 extending from southern New Mexico across west-central Texas and into southern Kansas.
3. The mesoscale situation
The surface cold front had passed through Abilene by 1800 UTC October 21, 1996, and through San Angelo by 2100 UTC, dropping temperatures by 7C in one hour. By late afternoon, temperatures in the northwest sections of the NWSO San Angelo county warning area (CWA), where the convection would first develop, had fallen to around 5C, and light rain was falling. The first thunderstorm developed around 2300 UTC, dropping pea size hail west of Abilene. Over the next two hours several bands of precipitation developed, with one dominant band of thunderstorms producing dime size hail which covered the ground, along with frequent lightning and sleet, in Abilene.
After a brief respite between 0300 UTC and 0400 UTC, when an area of light to moderate rainfall remained but the convection had largely subsided, another band of thunderstorms began to form closer to San Angelo. This second band was even stronger than the first, with many reports of nickel-size hail before the convection ended around 0600 UTC. The cold air advection at the surface had continued through the evening, with temperatures dipping to near 0C by the end of the event.
With the approaching long-wave trough and the associated warm advection aloft, precipitation chances were correctly forecast to be quite high. However, the chance for convection, especially reaching severe criteria, appeared much smaller. A review of the San Angelo WSR-88D data as well as the Eta and NGM model data revealed several additional factors that contributed to the convective situation, helping to support the formation of large hail.
4. Model data
A post-storm analysis of the NGM and Eta models clearly showed the potential for CSI and elevated upright convection across parts of west-central Texas between 0000 UTC and 0600 UTC on October 22. CSI exists when parcels of air that are normally stable when subjected to either horizontal or vertical displacements are unstable when displaced both horizontally and vertically simultaneously in a nearly saturated environment. One common location for the formation of CSI is near strong jet streaks where strong vertical wind shear exists (Dankers 1994). The existence of a potent upper-level jet streak across west-central Texas has already been mentioned. The San Angelo WSR-88D Velocity Azimuth Display (VAD) Wind Profile also showed this shear increasing through the evening as the jet streak approached.
One way to diagnose the location of CSI is through the use of equivalent potential temperature (e) and momentum cross sections drawn perpendicular to the thermal wind or thickness contours. Areas where isopleths of e are more vertical than isopleths of momentum in a layer where relative humidities are above 70 percent are prime locations for CSI to develop. Cross sections were produced of the 0000 UTC October 22 Eta initialization and of the 6-hour forecast at 0600 UTC. As can be seen on the 0000 UTC cross section Fig.3a, CSI was possible across much of west-central Texas, where the nearly vertical e isopleths cross the isopleths of momentum. Also note the folded shape of the e isopleths between 500 and 700 mb, indicative of an elevated convectively unstable layer. In areas where CSI and convective instability coexist, the convective instability will dominate due its faster growth rate (Moore and Lambert 1993). Moore also speculates that, although convective instability will dominate, CSI may act to organize the convection that develops into banded features that look very similar to CSI alone.
Drastic changes from 0000 UTC are noted on the 0600 UTC cross section Fig.3b. The folded look to the e isopleths is now gone, indicating the loss of the elevated convective instability. This makes sense as the approaching mid level cold pool with the shortwave had decreased the mid level lapse rates in the area where convective instability had dominated. However, the CSI was still prevalent, although a little weaker. The showery activity continued for several hours, but the strong convection had all but ended by 0600 UTC, suggesting that the CSI was not the main factor in the thunderstorm formation and could not sustain the convection already present without the elevated instability.
With the passage of the long wave trough shortly after midnight, drier air began to filter in across all of west-central Texas, ending all precipitation across the area by 0900 UTC.
5. Gravity waves
Radar images from the San Angelo WSR-88D Fig.4and Fig.5indicate the existence of ducted gravity waves. Figure 4 shows a series of northwest to southeast oriented bands of alternating
inbound and outbound velocities located northwest of the radar. This is a pattern seen before with the passage of a series of gravity waves (Damiana and Marwitz 1995). The passage of gravity waves creates an atmosphere with localized areas of upward vertical motion, which may act to enhance the formation of convection (Bluestein 1993). The waves had wavelengths of approximately 28 km, with the wave train visible on the radar for nearly 8 hours.
In order for wave energy to be passed downstream from the generation source without loss of energy, a duct must be present. This duct may develop when a stable layer of sufficient depth is topped by a conditionally unstable layer, which acts to reflect the wave energy back into the stable layer (Lindzen and Tung 1976). Notice from the 0000 UTC sounding from Midland Fig.6 that with the passage of the cold front, a stable lower layer was formed from the surface up to 650 mb. Above this stable layer, a conditionally unstable layer existed from 650 mb to 575 mb. This unstable layer would have acted to reflect the gravity wave energy down into the stable layer. In fact, the waves visible on the San Angelo WSR-88D were best seen at a height of 3 to 4 km, about the height where this boundary would have been present. It is above this boundary level where any upward motion imparted by the gravity waves to the unstable layer would have resulted in continued upward motion and possible precipitation formation. Indeed, the San Angelo WSR-88D showed a series of 20 to 30 dBZ echoes, associated with the gravity waves, which persisted for nearly the entire evening. With the loss of the mid level instability, this condition necessary for the duct ceased and, like the precipitation, the gravity waves came to an end.
Without significantly more study, no definitive answer can be made to the question of wave genesis. However, one possible source might be simple topography. In circumstances where a stable lower layer is topped by an unstable layer downstream from a topographic obstruction, lee waves formed by flow over the obstruction may become "trapped" in the lower stable layer and propagate downstream (Holton 1992). Located along the path upstream from west-central Texas are a series of mountain ranges, the most prominent of which are the Davis Mountains, approximately 200 km from San Angelo. Strong southwesterly winds over the Davis Mountains often produce altocumulus standing lenticular clouds, characteristic lee waves, over southwest Texas.
6. The storms
In Fig. 5, several of the southwest to northeast-oriented convective bands associated with CSI are shown. The lighter reflectivities to the northwest of the radar are the echos associated with the gravity waves. The storms were quite low topped, with echo tops rarely exceeding 9 km (30,000 ft). While the general banded CSI activity drifted slowly southeast, the individual cells moved northeast at 15-20 ms-1. The use of the conventional Vertically Integrated Liquid (VIL) product would prove to be virtually useless in this situation, as is often the case with low topped fast moving storms. However, the Layer Reflectivity Maximum (LRM) would be invaluable. In virtually every case where the mid level LRM product showed a 50 dBZ core, severe hail was soon reported. The locations of severe hail reports are shown in Fig.7.
It has been known for years that CSI can cause bands of heavier precipitation to develop in an air mass usually thought of as too stable for convective activity. In this case, while the prospects of precipitation certainly looked impressive, a combination of CSI, elevated upright convection, and ducted gravity waves may have combined to produce a convective situation even more pronounced than the models suggested. With the low topped nature of the storms and the cold surface temperatures, the earliest convection across the northwest sections could have been dismissed as short-lived and non severe. After all, more "classic" type thunderstorms had developed in the warm sector ahead of the cold front across the southeast sections of the area, but had failed to generate any severe weather. However, it was the elevated convection farther northwest that produced large hail over the two largest cities in the warning area. Only by recognizing that the meteorological situation across the northwest half of the area was far more intense than first thought could accurate and timely warnings be issued.
Acknowledgments. I would like to thank Gregory E. Jackson (SOO, NWSO San Angelo) and Amy B. McCullough (forecaster, NWSO San Angelo) for their advice in preparing this paper and for helping work up the graphics. I would also like to thank Brian L. Francis (SOO, NWSO Midland) for providing the Midland sounding data, and Dottie Anderson (NOAA Central Library) for assisting me with a literature search.
Dankers, T., 1994: Observing CSI Bands Using the WSR-88D. Postprints of the First WSR-88D Users Conference, 159-168.
Moore, J.T. and T.E. Lambert, 1993: The Use of Equivalent Potential Vorticity to Diagnose Regions of Conditional Symmetric Instability. Weather and Forecasting, 8, 301-308.
Holton, J.R., 1992: An Introduction to Dynamic Meteorology. Academic Press, 507 pp.
Lindzen, R.S. and K.K.Tung, 1976: Banded Convection and Ducted Gravity Waves. Monthly Weather Review, 104, 1602-1617.
Bluestein, H.B., 1993: Synoptic-Dynamic Meteorology in Midlatitudes Volume 2: Observations and theories of Weather Systems. Oxford Press, 594 pp.
Damiana, T.A. and J. Marwitz, 1995: Development of a Gravity Wave Event During an Oklahoma Blizzard. Preprints of the 27th Conference on Radar Meteorology, 314-316.