Sensitivity of forecast rainfall in a Texas convective system to soil moisture and convective parameterization

The impact of soil moisture on the forecast of a small-scale convective system, and sensitivity of results to the convective parameterization used, are investigated through Eta Model simulations (run in an operational-like setting) of a convective system occurring on 27 May 1997 in Texas. The event was influenced by a southwestward-propagating gravity wave from early morning convection in Arkansas that intersected a slow-moving cold front, releasing extreme conditional static instability. Isolated heavy rainfall, over 100 mm, occurred in some regions. A control simulation with 22-km horizontal resolution reasonably simulated the event, even though mesoscale influences such as the gravity wave important to this event are often poorly captured by numerical models. A series of sensitivity tests were performed to examine the impact of soil moisture on the simulations. Two different convective parameterizations were used for the tests. Although domain average precipitation is found to generally vary in a straightforward way with soil moisture, peak precipitation in the regions of intense convection shows more complex behavior. Sensitivity of precipitation amounts to soil moisture differs significantly among runs having different convective parameterizations. For instance, with the Kain-Fritsch convective scheme, relatively dry soil is found to result in stronger convective outflows that converge with stronger ambient flow to greatly enhance the precipitation in the region where heaviest rainfall occurs. With the Betts-Miller-Janjic scheme, drier soil generally results in less precipitation than in the control run, although some enhancement in peak amount does occur within a narrow range of drying. The differences between the peak quantitative precipitation forecasts in the runs is primarily due to the inclusion of a convective downdraft in the Kain-Fritsch parameterization, and its impact on secondary convective development. Additional sensitivity tests find limited impact from prescribed vegetation coverage. A final sensitivity test shows that precipitation amounts are even more strongly affected by the vertical resolution of the data used to initialize the shallow but moist boundary layer than by variations in the soil moisture or vegetation fraction