The impacts of soil moisture(SM) on heavy rainfall and the development of Mesoscale Convection Systems(MCSs) are investigated through 24-h numerical simulations of two heavy rainfall events that occurred respectively on28 March 2009(Case 1) and 6 May 2010(Case 2) over southern China. The numerical simulations were carried out with WRF and its coupled Noah LSM(Land Surface Model). First, comparative experiments were driven by two different SM data sources from NCEP-FNL and NASA-GLDAS. Secondary, with the run driven by NASA-GLDAS data as a control one, a series of sensitivity tests with different degree of(20%, 60%) increase or decrease in the initial SM were performed to examine the impact of SM on the simulations. Comparative experiment results show that the 24-h simulated cumulative rainfall distributions are not substantially affected by the application of the two different SM data,while the precipitation intensity is changed to some extent. Forecast skill scores show that simulation with NASA-GLDAS SM data can lead to some improvement, especially in the heavy rain(芏50 mm) forecast, where there is up to 5% increase in the TS score. Sensitivity test analysis found that a predominantly positive feedback of SM on precipitation existed in these two heavy rain events but not with completely the same features. Organization of the heavy rainfall-producing MCS seems to have an impact on the feedback process between SM and precipitation. For Case 1, the MCS was poorly organized and occurred locally in late afternoon, and the increase of SM only caused a slight enhancement of precipitation. Drier soil was found to result in an apparent decrease of rainfall intensity,indicating that precipitation is more sensitive to SM reduction. For Case 2, as the heavy rain was caused by a well-organized MCS with sustained precipitation, the rainfall is more sensitive to SM increase, which brings more rainfall. Additionally, distinctive feedback effects were identified from different stages and different organization of MCS, with strong feedback between SM and precipitation mainly appearing in the early stages of the poorly organized MCS and during the late period of the well-organized MCS. The impacts of soil moisture (SM) on heavy rainfall and the development of Mesoscale Convection Systems (MCSs) are investigated through 24-h numerical simulations of two heavy rainfall events that occurred respectively on March 2009 (Case 1) and 6 May 2010 (Case 2) over southern China. The numerical simulations were carried out with WRF and its coupled Noah LSM (Land Surface Model). First, comparative experiments were driven by two different SM data sources from NCEP-FNL and NASA-GLDAS. Secondary, with the run driven by NASA-GLDAS data as a control one, a series of sensitivity tests with different degree of (20%, 60%) increase or decrease in the initial SM were performed to examine the impact of SM on the simulations. Comparative experiment results show that the 24-h simulated cumulative rainfall distributions are not substantially affected by the application of the two different SM data, while the precipitation intensity is changed to some extent. Forecast skill scores show that simulation with NASA-GLDAS SM data can lead to some improvement, especially in the heavy rain (芏 50 mm) forecast, where there is up to 5% increase in the TS score. Sensitivity test analysis found that a predominantly positive feedback of SM on precipitation existed in these two heavy rain events but not with completely the same features. Organization of the heavy rainfall-producing MCS seems to have an impact on the feedback process between SM and precipitation. For Case 1, the MCS was poorly organized and occurred locally in late afternoon, and the increase of SM only caused a slight enhancement of precipitation. Drier soil was found to result in an apparent decrease of rainfall intensity, indicating that precipitation is more sensitive to SM reduction. For Case 2, as the heavy rain was caused by a well-organized MCS with sustained precipitation, the rainfall is more sensitive to SM increase, which brings more rainfall. ages and different organization of MCS, with strong feedback between SM and precipitation mainly appearing in the early stages of the poorly organized MCS and during the late period of the well-organized MCS.