Sensitivity of a flash flood simulation to convective and boundary layer parameterization schemes

On 10 November 1992 at 1200 UTC, a quasi-stationary surface front set up across the central United States, oriented from north-northeast to south-southwest and west. This cold front was accompanied by a slowly moving upper-level trough and strong moisture convergence from the Gulf of Mexico across Oklahoma. By 1900 UTC, a thunderstorm developed in southwestern Oklahoma, and by 2100 UTC a line of convection organized producing hail. This pattern re-developed continuously over the same location for several hours. At 1235 UTC, 11 November 1992, the line began to propagate, still organized, toward the northeast. This train echo pattern persisted long enough to cause a flash flood event primarily over southwestern to northeastern Oklahoma. This relatively stationary line of convection dumped 9.32 inches of rain in 24 hours in Grandfield, Oklahoma and left many impassible roadways and water-logged structures throughout Oklahoma in its wake.

This case study will be simulated on a storm-resolving scale using the PSU-NCAR mesoscale model (MM5). Various parameterization schemes will be examined to test the sensitivity of the model during high-resolution runs. In particular, the Betts-Miller and Kain-Fritsch convective schemes and the Blackadar and Burk-Thompson planetary boundary layer (PBL) schemes will be used. Combinations of these schemes will be used to create four model runs to determine if there is a profound effect on the simulation of this event.

Since each scheme is a reasonable approximation of the process it represents, a detailed analysis will be needed to see which combination of the schemes provides a more realistic simulation of this train echo resulting in a flash flood. The better the parameterizations within the model, the better the simulations, and hopefully these results will provide a better understanding for the mechanisms that cause these events.