Abstract: Two heavy rainfall events occurred in the provinces of Jiangxi and Fujian, are investigated by comparing their environmental conditions and physical variable fields using the NCEP reanalysis data. Then the possible effects of land surface process on the heavy rainfall as well as its related physical mechanisms are explored by using the meso-scale model WRF V3.3. The results show that the Jiangxi heavy rainstorm event taking place on 12 May 2012 is dominated by large-scale precipitation in nature, which is featured by its large spatial scope and long lasting time. Another heavy rainstorm event occurring in Fujian Province afternoon on 23 August, 2011, is mainly under the control of the Subtropical High. The local sensible heat and latent heat fluxes evidently strengthen the instability of the lower atmosphere, which triggers the thermal convective precipitation. From the diagnostic results, it is concluded that the land-surface process has more important impacts on the heavy rainstorm event in Fujian compared to that on the heavy rainstorm in Jiangxi. After closing land-surface fluxes in the numerical experiments, it is found that land-surface process affect the heavy rain significantly, especially on the occurrence of the convective heavy rainfall in Fujian. Furthermore, several sensitivity experiments are performed by using the four land-surface parameterization schemes (LSPs) and the results suggest that the two heavy rainfall cases simulated are sensitive to the LSPs. The convective precipitation is more sensitive to the LSPs in the Jiangxi case, while the large-scale precipitation is more sensitive in Fujian case. On the whole, the heavy rainfall case in Fujian is more sensitive to the LSPs than that in Jiangxi. The mechanism responsible for the sensitivity is precipitation-type dependent. For large-scale rainfall, the sensitivity originates from the different responses of the convective system in the middle and upper troposphere to various LSPs. However, the sensitivity is mainly related to the different responses of the land surface-atmosphere flux to LSPs in the convective heavy rainfall case. Disturbance tests of the land-surface parameters are also carried out, and it is found that soil porosity and surface albedo play more important roles in the sensitivity of the convective precipitation to land-surface process compared with the surface roughness and the minimal stomatal impedance. The sensitivity of the model results to the land surface parameters disturbances strongly depends on the disturbance of the land surface-atmosphere fluxes in the nature. The responses of land surface-atmosphere fluxes to the land surface parameters are usually faster, larger in the spatial scale and more regular than that of the wind field. This study can provide a reference for our better understanding on the impacts of land surface process on the weather processes (such as torrential rainfall) and improvement of the performance of the numerical model in simulating the heavy rainfall events.