The sensitivity of the PSU-NCAR model (MM5) to cumulus parameterization in simulating the mesoscale environment associated with 2 June 1995 West Texas tornado outbreak

Abstract

On 2 June 1995, many supercede thunderstorms were graphics. observed in West Texas between Lubbock and Amarillo under the synoptic and mesoscale environment which was increasingly more supportive of severe convection. Of the storms, those which crossed a particular outflow boundary, generated by earlier convection, produced violent tornadoes near Friona and Dimmitt, Texas. As documented in numerous other studies, the outflow boundary generated from the earlier convection seemed to play a important role in producing tornadic supercedes. This study presents observational features of the event, performs model simulations with three disparate cumulus parameterization schemes, and does a careful comparison between the simulations and observations. This study tries to understand the sequence of events which preceded the severe storm outbreak. Particularly, mesoscale features such as the dry line and outflow boundaries are carefully documented. Because of the significant impact of the convective outflow boundaries, this study tries to examine the sensitivity of the PSU/NCAR three dimensional nonhydrostatic mesoscale model (MM5) to the choice of cumulus parameterization scheme. The schemes used in this study include Kain-Fritsch, Fritsch-Chappell, and Grell schemes. The model simulation has a 67 x 67 grid domain centered at 35[] N and 102[] W with 27 km grid spacing, and starts at 12 UTC 2 June for 12-h forecast. The simulation results showed that, even though the general features agreed well among the three different simulations, the mesoscale features such as the outflow boundary, convective rain, temperature gradient, and pressure gradient associated with tee convective outflow were somewhat different among the different cumulus parameterizations. It seemed that the different behavior of the simulations was strongly dependent on the components constructing each cumulus parameterization. Despite the limitation of evidence suggested in this study, the Kain-Fritsch scheme appeared to be most suitable for the simulation of meso-[] scale features.