Evaluating the Performance of a WRF Physics Ensemble with 9 Schemes for Precipitation over Southwest of Iran in Winter

Mesoscale weatherforecast model(WRF) has a wide variety of physical parameterizations schemes and choosing a particular set of schemes can have a significant impact on the accuracy of model predictions. So according to weather systems we should select an appropriate options of these schemes. In this study¡ accuracy of 24 hours predictions in the south west of Iran post processed from December 2011 until February 2012 with 9 different WRF configurations has been considered. Post processing is done by 2×2 contingency table. Despite great difficulties in precipitation prediction¡ post processing of the bias quantity shows that WRF model overestimates forecast for all precipitation thresholds for all configurations except for 10-15 mm rainfalls thresholds for the 3rd configuration and over 15 mm rainfalls thresholds for the 2nd and3rd configurations that have underestimated the forecast. Proportion correct quantity (PC) for under and over 15 mm rainfalls shows respectively the 3rd and 2nd configurations are good by 91 and 97 percent accuracy. Comparing the bias quantity(B) for 9 configurations in 4 rainfalls thresholds indicates that the model has overestimated the forecast for the mild and moderate rains¡ severe rains except the 3rd configuration and heavy rains except the 2nd and 3rd configurations. This quantity shows that the 3rd configuration for mild to severe rainfalls and the 4th configuration for heavy precipitation are more appropriate. For the hit rate quantity (H)¡ the 2nd configuration for mild to mean rainfalls¡ the 4th configuration for severe rains and the 8th configuration for heavy precipitations¡ are more appropriate. Pierce skill scores (PSS) shows that for rainfalls thresholds below 10 mm the 2nd configuration is good. This score appropriates the 4th configuration for thresholds rainfalls between 10-15 mm and the 8th configuration for thresholds rainfalls over than 15 mm. Generally for light precipitations (0.1-5 mm) and moderate rainfalls (5-10 mm)¡ the 2nd and 3rd configurations have higher accuracy among the others. For severe precipitations (10-15 mm) the 3rd and 4th configurations have higher accuracy compared to the others. However¡ in the case of heavy rainfalls (more than 15 mm) the 2nd and 3rd configurations have similar results and a specify configuration can not be chosen between them.