The C-band radar data is inverted with the LAPS model cloud analysis system, and the cloud microphysical field obtained from the inversion is introduced into the GRAPES mesoscale numerical model using Nudging technology, combined with a simulation experiment of a heavy precipitation weather process to study the influence of C-band radar data assimilation on the short-term precipitation forecast of GRAPES mesoscale numerical model. The results show that: ① Radar data assimilation can improve the precipitation forecast of the mesoscale model. The precipitation forecast correlation coefficient and TS scores of different grades of precipitation in the first 6 hours of the model have been improved. At the same time, the precipitation peak is advanced by 1 hour, which helps to alleviate the model spin -up problem. ②The improvement effect of model precipitation forecast is mainly contributed by heavy precipitation, and the maximum improvement effect is concentrated in 4~6 hours. ③ After assimilating the radar data, the number of forecasting stations for heavy precipitation above 25 mm is closer to the actual situation, the deviation of the falling area is smaller, and the precipitation falling area and intensity are adjusted to the actual direction. ④ After absorbing radar data for weak precipitation below 10 mm, the number of stations comparison control forecast increased more than the actual situation, the deviation range of the landing area increased, and there was a problem of excessive forecasting. The weak precipitation forecast is mainly concentrated in 4~6 hours, and the first 3 hours have a positive effect on the improvement of precipitation forecast.