Abstract: Base on the precipitation forecast of ECWMF model from 2016 to 2019 and the observation data, five schemes are designed to carry out revision experiment on numerical model clear-rainy forecast, including Optimal Percentile (OP), the Optimal Threat Score (OTS), the Probability Matching (PM), the sub-district OTS and the sub-district PM-OTS fusion scheme. The results show that: (1) The clear-rainy revised thresholds of OP and PM are static, and that of OTS is dynamic. The thresholds of the 5 schemes are all applicable to Area A(basin, northern Aba Prefecture and northern Garze Prefecture), among which the thresholds of PM and sub-district PM-OTS are more applicable to Area B(southern Ganzi Prefecture and panxi area) with obvious wet deviation of the numerical model. (2) Each scheme has obvious revised ability for ECWMF model clear-rainy forecast. The revised effect of 24 h is the best, and that of Area B is superior to Area B, autumn and winter is better than spring and summer. (3) Revised scheme scores after subdivision are better than before, and the sub-district PM-OTS is the best. Individual cases and batch tests show that the revised effects of the schemes in Area A are comparable, the rainfall distribution in Area B are more close to the observed precipitation after revised of PM, sub-district OTS and PM-OTS fusion schemes, among which the sub-district PM-OTS is the best.