Abstract: The convective parameterization and microphysical scheme are two main approaches in a mesoscale atmospheric model to describing cloud and precipitation processes. When the grid scale of the model is small enough to distinguish the convective cloud scale, the microphysical scheme becomes more important. In this study, a case of heavy rainfall of Meiyu front happened on July 7-9, 2007 in China is simulated using the GRAPES mesoscale model. The effects of the different microphysical schemes (the Ncepcloud3 simple ice scheme, the Ncepcloud5 mixed phase scheme, the CAMS double moment mixed phase scheme) on heavy rainfall simulating are analyzed combining with the observations of precipitation, the radar echoes and hydrometeors patterns. The results show that for this heavy rainfall case, the simulations using the CAMS scheme are better for the classes of light rain and moderate rain, and the Ncepcloud3 scheme and the Ncepcloud5 scheme are of higher TS for the torrential rain and the class above it. The maximum of the heavy rainfall center simulated using the CAMS scheme is close to the observation. The simulated results of mesoscale convective system developing processes are different using these schemes from each other. The area of weak convective echoes simulated by the Ncepcloud5 scheme is small. The CAMS scheme simulated a bigger area of weak convective echo which is basically consistent with observations. The vertical distributions of the mass content and number concentration of ice particles calculated by the CAMS dual parameters scheme are similar to the CloudSat observations. The ice particles calculated by the Ncepcloud3 scheme and Ncepcloud5 scheme are larger than the observation with no number concentration of theirs covered. The reasons of the simulative difference caused by using different schemes are preliminarily analyzed. During the forming stage of convective system, the simulative difference is mainly produced by dynamical process. While during both the developing and mature stage, the microphysical processes affect simulation results significantly.