Abstract:This study verified the whole 2016 year real time forecast and parts of 2013-2015 reforecast of heavy rain events and compared them with EC model and T639 model forecasts by using synoptic verification method. After summarizing up all verification results into several systematic biases, some conclusions were excavated to help improve the GRAPES GFS developments and operational applications. 38 heavy rainfall events were verified. Starting from the forecast quality of precipitation, synoptic weather systems and atmospheric physical factors were checked to find the direct causes of the precipitation biases and differences between other operational models. The results showed that some advances have been made in short range precipitation forecast, but still north bias exists in some convective rainband forecasts. Precipitation forecasts are weaker than observation in some convective cases which are under weak high level synoptic system background. Wet bias northern to the rainband and strong bias of subtropical vortex were also found in some cases while the precipitation was not over estimated.