Abstract: Cloud physics is one of the most important non-adiabatic heating processes in mesoscale numerical model.Cloud and the occurrence of rainfall take effect on large-scale circulation as feedback through the sensible heat,latent heat and momentum transfer effects,and play a key role in determining the vertical structure of atmospheric temperature and humidity.Also it is the things that people are most concerned about precipitation forecast.Therefore,accurately description of cloud physics will improve the capability of precipitation forecasting in mesoscale numerical weather prediction models greatly.In this paper,we choose 3 microphysics parameterization schemes that was new in WRF model after version 3.0 and use them to simulate a heavy rainfall process in Sichuan on 14-18 July,2010 respectively.The 3 schemes include WDM6 scheme,New Thompson scheme and SBU-YLIN scheme.First we analyze the difference between the real stat and the 3 experiments,and the differences between the 3 experiments in the entire precipitation process.In the early precipitation stage,to the center of the precipitation in the eastern Sichuan basin,WDM6 scheme's result is on southern side a little compared with the real,the results of New Thompson's scheme and SBU-YLin seem good.New Thompson's result is a little weak compare to real in the precipitation of Guangyuan area.Overall,New Thompson scheme and SBU-YLin scheme are good and quite near,WDM6 scheme slightly worse in the early stage of the case.In the period of wide range of precipitation occurred,the effect of the 3 schemes are very close.And the simulation of the3 schemes all strengthen the rainband from Leshan to Yibin,the result from 18:00 July 16th to 06:00 July 17th is the best.WDM6 scheme is the best in simulate the magnitude of the rainfall.For the precipitation area simulation,SBU-YLin scheme is the best.New Thompson scheme's capability is between the other 2 schemes.At last,we analyzed why there is difference between 3 schemes,which maybe the treatment of the water particles differently.