1) The 3-year quasi-symmetric sliding window sampling method is used to collecttraining samples to calculate the OTS correction threshold, and the effect isbetter than that of the sampling method of collecting the samples from the sameseason in the previous 3 years. 2) The magnitude of the OTS correction thresholdF1 (i.e. elimination threshold) is closely related to the magnitude of 2 mtemperature. As the temperature rises, the elimination threshold first increasesand then decreases. Compared with the modeling scheme without grouping, thegrouping modeling based on the maximum 2 m temperature forecast can obtain theelimination threshold under different temperature conditions, which cansimultaneously reduce the false alarm ratio and missing ratio of light rain, andimprove the equitable threat score of light rain by 5.0%~8.2%. 3) The effect ofthe first scheme, which is to use the inverse distance weighted interpolationmethod to interpolate the model precipitation forecast, and then use the OTScorrection method to correct the precipitation forecast, is better than that ofthe second scheme, which is to use the nearest neighbor interpolation method tointerpolate the model precipitation forecast, and then use the OTS correctionmethod to correct the precipitation forecast.