The circulation classification of 850 hPa hourly wind field in the rainy season in Hunan Province from April to September in 2021 is carried out by using the T-mode oblique rotating principal component analysis method. On this basis, the evaluation is carried out for the hourly precipitation forecast of Rapid Updating Cycle Assimilation and Forecasting System in South China (CMA-GD-R3) in the same period. The results show that the main circulation types affecting the rainy season in Hunan in 2021 are the southwest vortex shear type, the shear type, the southerly type at the edge of the subtropical high and the easterly type at the periphery of the typhoon. The model shows good forecasting ability for the clear-rainy accuracy rate and graded precipitation. The daily variations of clear-rainy accuracy rate and TS of hourly precipitation forecast of CMA-GD-R3 are obvious. The clear-rainy accuracy rate is higher in the night time than that during the day, the peak value of TS score for graded precipitatin appears in the morning, and the effects of rainfall nowcasting of all circulation types are better. The southwest vortex shear type, which has the highest frequency of short-term heavy precipitation, has the lowest accuracy. The southerly type at the edge of the subtropical high has relatively poor performance in heavier precipitation. The SAL test shows that the rainfall location forecast of the model in the cases of the southwest vortex shear type and the shear type is closer to the observation. The intensity forecast is weak first but then strong. The precipitation location forecast in the cases of the southerly type at the edge of the subtropical high is scattered, and the location forecast is unstable. The overall intensity is obviously weaker than the observation. The location forecast in the cases of the easterly type at the periphery of the typhoon is close to the observation in the short-time forecast, and the intensity forecast is significantly weaker. The SAL method can more objectively reflect the structure, intensity, and location deviation in the hourly precipitation forecast.