In this study we investigate the skill of a mesoscale model in predicting the amount and spatial distribution of precipitation in wintertime orographic precipitation events in the Sierra Nevada, and the sensitivity of this forecasting skill on the choice of a microphysical parameterization. During winter, the Sierra Nevada, a quasi-two dimensional mountain range with a half-width of 100 km, an average crest height of 2 km, and a gentle (2%) upwind slope, is located in the path of Pacific storms, with the precipitation produced mainly by large-scale forced orographic lifting.

The mesoscale model used in the study is MM5. High-resolution model simulations have been carried out on four nested domains with the highest horizontal grid increment of 1.5 km. The simulations were varied only in the choice of microphysical parameterization. Four microphysical parameterizations employed in this study are: (i) Dudhia ice scheme, and (ii) Schultz, (iii) Reisner, and (iv) GSFC mixed-phase schemes. The verification data set consists of a selected number of relatively high-impact precipitation events that were observationally documented during the Sierra Co-operative Pilot Project (SCPP) in the 1980s.

Model-predicted 24-h precipitation accumulations were evaluated against the precipitation observations at SCPP stations, the majority of which are located on the windward slopes. We find, in general, that regardless of the choice of microphysical scheme, the model over-predicts the precipitation amounts on the windward slopes and under-predicts them on the lee slopes. From the statistical analysis based on the contingency table approach with light (0.01-0.5 in), moderate (0.5-1.5 in) and strong (1.5-2.5 in) precipitation thresholds, the best and the worst skill scores for all the schemes are achieved for precipitation amounts in the moderate and strong classes, respectively. In the presentation, we will discuss differences in the statistical skill of precipitation forecasts obtained with different microphysical parameterizations.