Abstract: Based on the 24-hour accumulated precipitation forecasts produced by SouthWest Center-WRF ADAS Real-time Modeling System(SWC-WARMS) initialized at 0000 UTC during the period of May 1, 2019 to August 31, 2019, along with the corresponding precipitation observation data from meteorological stations in Sichuan Province, the Frequency Matching Method (FMM) was applied to calibrate the precipitation forecasts from June 1, 2019 to August 31, 2019. Results show that the 24-hour accumulated precipitation forecasts from SWC-WARMS perform with wet bias on the whole. The averaged absolute errors of precipitation forecasts are reduced after calibration. The bias scores of heavy rain and rainstorm are improved to a large extent. In addition, the TSs of little, medium and heavy rain are improved a lot while the TS of rainstorm is inferior to the raw forecasts. The false alarm rates of all magnitudes of precipitation are decreased, the miss rates of light and medium rain are decreased, while the miss rates of heavy rain and rainstorm are increased, which is more obvious in rainstorm. The FMM can improve (decrease) the TS of raw SWC-WARMS forecast when the SWC-WARMS predicts the rainstorm precipitation area well (poorly).