Abstract: Using hourly and daily precipitation observation data from surface observation stations in Sichuan Province during the 2022 flood season, multiple operational models, chiefly CMA-MESO_3KM, are evaluated for their forecast biases in various Sichuan regions. The results indicate: (1) The CMA-MESO_3KM model provided more accurate precipitation location forecasts compared to other models, but it overestimated precipitation amounts and coverage areas. (2) Comparing forecasts from different initialization times for the same period, forecast accuracy increased with shorter lead times, reflecting the advantage of rapid updates in the CMA-MESO_3KM model. (3) The CMA-MESO_3KM model was the closest to the actual situation in the diurnal variation of precipitation frequency, but the precipitation frequency forecast in the slope area and the latter half of the night in the Sichuan Basin was higher than the actual situation.