In order to improve the accuracy of numerical precipitation forecasting,forecast data from the European Centre for Medium-range Weather Forecasts (ECMWF)and Jiangxi meteorological observation stations were used in this study.A gamma function was used to simulate the precipitation cumulative probability of prediction and observation.The method was tested for a precipitation concentration period in Jiangxi province in 2017.The results show that the ECMWF forecast-observed precipitation probability matching dynamic correction method provides the latest real-time forecasts and observations for probability matching,and updates the precipitation correction values of all grades automatically according to the prediction adjustment and error of the recent model.It is found that the ECMWF's daily 12 h interval precipitation forecast is generally larger for precipitation 10 mm and below,and smaller for precipitation 25 mm and above.The precipitation forecast in the Jiujiang area along the Yangtze River and Jingdezhen is close to actual conditions.This precipitation forecasting method can improve the threat score of light and heavy rain,reduces the PO of heavy rain,and increases its POD.However,the correction effect of heavy rain and some moderate rain is not good;hence,the advantages and disadvantages should be considered in practice.