Severe Thunderstorm Development along the Dryline: A Long-Distance Collaborative Project

Figure 1: MM5 model output vertical velocity (reds are downward and blues are upward) at 2 km MSL and 2:45 p.m. CST over a portion of the model domain in the eastern Texas Panhandle. The scale at left includes a range from -2 to +2 m s-1. Ground-relative horizontal wind vectors are also plotted (scale at lower left).

It is well known that the dryline is a favored location for the initiation of severe thunderstorms that affect the southern and central Great Plains during spring months. It is not so clear, however, what factors influence the along-line locations where storms form. On days when intensive data collection occurs observational data-sets provide certain clues concerning the reasons for location and timing of storm initiation along the dryline. These data, however, still lack the spatial and temporal resolution necessary to diagnose the details of the initiation process. Harald Richter and Carl Hane have used the MM5 model along with observations from a field program to try to better understand what factors control the location and timing of the dryline storm initiation process. One of the strengths of a numerical model (MM5, in this case) is its ability to provide high resolution information in both time and space. This strength was capitalized upon for the 15 May 1991 case when the dryline produced storms near the western Oklahoma border.

The model was run using a "nested" configuration of three grids with differing point spacings. Using this methodology, grids with progressively finer spacing cover smaller areas and are centered on the area of interest. In this case, the smallest grid (with 2 km grid point spacing) covered most of the Texas Panhandle and a portion of western Oklahoma. The model was initialized at 6 a.m. from larger scale observations and run forward in time for 12 hours.

The model run resulted in the initiation of convection along the dryline about 90 minutes earlier in the afternoon and about 30 km farther west than was observed in the eastern Texas Panhandle. The deep convection that developed in the model simulation was scattered in nature with similar spacing and line orientation to the observed convection. With this reasonably successful simulation in hand, a logical next step was to diagnose the model results to understand the convective initiation process in more detail. An analysis of the boundary layer vertical velocity field at a time just prior to and during the formation of the first deep convection (Figure 1) shows the vertical motion along the dryline within an extended sinuous band oriented from south-southeast to northnorthwest. West of the dryline are alternating bands of upward and downward motion generally oriented along the boundary layer wind direction. These bands resulted from circulations within horizontal convective rolls.

Where these bands intersect the dryline, both the dryline and the horizontal convective roll circulations appear to be changed (Figure 1). Especially evident are small-scale eastward bulges in the dryline that appear to result from enhanced downward transport of westerly momentum by the downward branches of roll circulations at the dryline. It is just ahead of each of these bulges that deep convective clouds first formed in the model, owing at least in part to enhanced low-level convergence and upward motion in the moist air. Local reorientations of the dryline near these bulges may also allow the air that reaches the roots of potential convective clouds to reside in moisture-rich air for a longer period. Current and future research on this project is focusing on the diagnosis of the mechanism that actually produces the development of sustained deep convection just east of the dryline.

This project began when Harald joined NSSL for one year as a National Research Council post-doctoral fellow and brought with him a working knowledge of the MM5 model. It continued by long distance after Harald took a position at the Bureau of Meteorology Research Center in Australia. It has also followed on the heels of a series of observational case studies over the last decade or so by Carl and other collaborators that were made possible by data collected in the COPS-91 field program over Oklahoma, Kansas, and Texas. Other researchers at NSSL, including Conrad Ziegler and collaborators, have also made good use of these data. Persistence in analysis of quality data from a single field program has paid significant dividends.