Abstract: Using the four numerical model precipitation data of ECMWF fine grid, GRAPES _ GFS, GRAPES _ MESO, SWC _ WARM, as well as intelligent grid precipitation forecast products of Sichuan Meteorological Observatory, six regional rainstorm weather processes in Suining in July 2020 were analyzed. The results show that the four numerical prediction models have large deviations in the precipitation forecast of Suining, and the forecast values are generally less than 1 ~ 2 orders of magnitude. SWC _ WARM has good indicative significance for the forecast of falling area of warm area rainstorm. In the frontal precipitation process, ECMWF fine grid model has a greater reference value. The deviation of GRAPES _ GFS and GRAPES _ MESO is the largest, and the indication of rainstorm is poor. The intelligent grid precipitation forecast products of Sichuan Meteorological Observatory have a certain guiding role for rainstorm forecast, but the local forecasters need to further correct the falling area and magnitude.