Abstract: Using WRF V4.1.3 model and ERA5 reanalysis data, analyses and numerical simulation tests were conducted for a weather event of Meiyu front torrential rainfall occurred in central China on 7 July 2020 to explore the effects of the combinations of different vertical and horizontal resolutions on the rainfall simulation. Based on results from the above analyses, then investigations are carried out on whether applying cumulus parameterization at a horizontal resolution of 3 km would influence on the simulation of precipitation. The simulated results show that the features of circulation at middle and low levels, characteristics of rain clusters and rainfall distribution largely conform to the actual situations, although there are some differences among different combinations of tests. For the evolution of hourly precipitation at 3 selected relatively heavy precipitation areas, the simulation results of each simulation test are greatly deviated from the actual situation. Results from tests with the same horizontal resolution and cumulus parameterization are closer to each other. According to TS, ETS, BS and SAL scores, the simulation results of the 3 km tests were generally better than those of the 12 km tests, indicating that finer horizontal resolution of the model could improve the precipitation forecasts. Turning off cumulus parameterization at horizontal resolution of 3 km obviously reduces the forecasting skill, especially for events with the magnitude of heavy rain and above. However, for precipitation prediction with different threshold values, the influence of cumulus parameterization is different. The effect of changing vertical resolution on prediction skills is not as significant as that of changing horizontal resolution and the inclusion of cumulus parameterization. In the 3 km horizontal resolution model, the adoption of cumulus parameterization scheme can significantly improve the prediction skill of rainfall intensity.