A probability forecast model of heavy rainfall was constructed by developing probability of precipitation amount (PoPA) and based on probability of moderate precipitation.Step function based on estimating POP in suitable intervals more accurately describs nonlinear probability distribution and closely relates to probability of heavy rainfall.By using predictors transformed into step function and estimating correlation ratio between predictors and heavy rainfall event,the regression estimation of event probabilities (REEP) of PoPA was completed.By using multi time data and NWP products,multi time predictive equations were constructed,and an ensemble prediction that synthesized multi time predictions improved forecast effect.Forecasting tests indicate that the probability forecast model of heavy rainfall is efficacious.