To evaluate the effect of various schemes on rainstorm forecast in summer in Hubei and its surrounding areas,eight microphysical schemes and six cumulus parameterization schemes of WRFV3.6 are used to forecast 12 rainstorm cases.The results show that the prediction effects of different schemes are not the same on different scales.For the eight microphysical schemes (Lin,WSM6,Thompson,Morrison 2-mom,CAM5.1,WDM5,WDM6 and NSSL 2-mom) and KF cumulus parameterization scheme,combined with TS scores,anomaly correlation coefficient(ACC) and root mean square error(RMSE),it can be concluded that CAM5.1 is better than others,followed by Morrison 2-mom.Under the condition of CAM5.1,the sensitivity test results of six cumulus parameterization schemes (KF,BMJ,GD,SAS,G3D and Tiedtke) indicate that the six schemes both have advantages and disadvantages in the precipitation prediction of different magnitudes.GD,SAS and Tiedtke are better than the other through synthetical consideration.On these bases,the ensemble forecasting experiments are carried out by the ensemble mean (EMN).The results show that the method can reduce the forecast error and the uncertainty of individual member forecasts.