During the rainfall, the intensity of precipitation varies. Changes in the amount of precipitation during an event of rainfall are effective in the resulting of flood and its intensity. Knowledge of how rainfall changes over time during rainfall is determined by temporal distribution pattern of rainfall. For this purpose, availability of short-term time scales rainfalls data are important that obtained by rain gauge stations. However, the low density of the rain gauge network and the lack of sufficient data from the time pattern of rainfall have always been a problem in determining storm patterns for executive plans. Therefore, the simulation of WRF numerical weather models can be used. The WRF model is one of the most responsive models for predicting precipitation, temperature and atmospheric elements that used in this study. In this paper, three great storm events on 15 December 2003, 24 - 26 December 2006 and 6-7 March 2007 have been selected in the Parsian dam basin and surrounding areas in south west of Iran. The result of WRF numerical weather prediction model for these great storms compared with data loggers. It showed that the WRF model was able to performance the heavy rainfall and simulates the rainfall pattern in these dates.