Calibrated Probabilistic Precipitation Forecast Using the WRF and MM5 Ensemble over Iran

The output of an ensemble for country-wide daily precipitation probabilistic forecasts were calibrated with two models of WRF and MM5 with respectively 5 and 3 different configurations. The cumulative precipitation of 257 synoptic stations in the country has been used from 1st of November 2008 to 30th of April 2009. These data have been divided into two three-month periods which has been used for training and evaluating. The ensemble's rank histogram in training period, has been divided into two sets with the standard deviations of (0< s <0.45) and (s >0.45). Finally, daily precipitation forecast has been calibrated for thresholds p≤0.1, 0.1≤ p<10, and p>10 millimeters at each day of the evaluating period. This was done by means of the rank histogram produced by training period and probabilistic precipitation standard deviation in the same day. For different verification tools it has been shown that calibration with rank histogram leads to an improvement in probabilistic forecasts of daily precipitation (especially in heavy precipitation categories).