From March to May every year, frequent cold fronts are usually seen in China. Be cause of active cold fronts during these 3 months, weather in china is character ized by sharp temperatures fluctuations and frequent dusty weather conditions. I n order to examine the forecasting effect of numerical model and accumulate for ecasting experiences, the 96hour forecasting product of T213 model were examin ed and analyzed and also compared its forecasting result with ECMWF model and Ja pan model. The result is: ECMWF model is more effective in midirange forecast. As for adjusting macroscale circulation and forecasting temperatures at 850hP a the ECMWF model has stronger predictive ability than the other 2 models. In ad dition, a dusty weather process occurred on 26-28 May 2008 was chosen, as a case study. Through analyzing this case, we found that Japan models were more effect ive than the other 2 models in midrange forecast of the strong surface winds, which caused this dusty weather process.