“75.8" heavy rain is a disaster, which caused a large loss to the soci ety. Many meteorologists have done a lot of research works of this event, and pa id much attention to the predictability.“75.8"heavy rain in Henan Province was particularly chosen for the study of the mesoscale ensemble forecast. Two experiments were implemented. The first one especially focused on the relati ons between cumulus convective parameterizations and the heavy rain. For this pu rpose, the KainFristch ETA scheme was chosen. The interaction between the trig g er function and the mass flux calculations in large scale environment was taken into consideration. Some empirical factors have been perturbed in a reasonable r ange to construct the ensemble members to describe the uncertainty of model. The results indicate that different factors in cumulus convective parameterization have different impacts on the precipitation. The location of the rainfall is les s influenced by different factors, but the intensity of the heavy rain is improv ed, the precipitation forecasted by ensemble mean increased by 30%. Meanwhile, d ifferent factors would have different impacts on the time when the convection ha ppens, the rainy duration as well as the development of convection. The second e xperiment adopted different resolution of model and multiparameterization to f o rm ensemble members. The results prove that the ensemble mean would improve the intensity forecast to some extent. Compared with the results coming from model w ith high resolution, the precipitation forecasted by ensemble increases by 20%-3 0%. Ensemble forecast would decrease the uncertainties from the model itself. Th e differences among various cumulus convective parameterizations, empirical fact ors and model resolutions indeed have a certain impacts on the simulated results of model.