To enhance the accuracy of heavy rainfall simulation and prediction in Wanzhou section of the Three Gorges Reservoir area, we employed the mesoscale weather research and forecasting (WRF) model along with FNL reanalysis data for the prediction. 15 physical parameterization scheme combinations were designed by integrating five cloud microphysics schemes, three cumulus convection schemes, the RRTM/Dudhia longwave and shortwave radiation schemes, the YSU planetary boundary layer scheme, and the Noah land surface scheme. We systematically analyzed the performance of different parameterization scheme combinations in the simulation of spatio-temporal characteristics of precipitation and identified the optimal scheme, aiming to develop a WRF model suitable for the study area with high simulation accuracy. The results show that the WRF model can reasonably reproduce the spatial and temporal variations in precipitation; however, it exhibits a tendency of predicting an earlier onset of heavy rainfall and generally underestimates precipitation amounts. The mean hourly precipitation bias ranges from -0.62 to -0.12 mm, while the relative error in the spatial distribution of 48-hour accumulated precipitation lies between -66.90% and 16.75%. Among all the scheme combinations, the WSM6-GD-Noah-RRTM / Dudhia scheme (A5) demonstrated the best overall performance, with a simulation relative error of only 2.79% for the 48-hour accumulated precipitation. Moreover, it achieved the highest average threat score (TS) of 0.26 for heavy and torrential rainfall and showed a significantly improved simulation of rainfall intensity at the storm center compared to other combinations. The precipitation characteristics simulated by scheme A5 closely aligned with the observed data, indicating the suitability of this scheme for simulating precipitation in this region. These findings can contribute valuable insights into meteorological modeling and disaster prevention and mitigation in the Three Gorges Reservoir area.