Abstract: To study the EC ensemble forecast method, the authors took daily precipitation process in 2014 flood season in Inner Mongolia as an example, tested the optimization method and compared it with the ensemble average forecast, based on temperature of EC 51 member ensemble forecast system and IDL lattice data of precipitation. The results showed that: (1) in the short-term forecast, when the number of ensemble members increased to 3 or 4, it had the highest correlation coefficient, when the number of ensemble members increased more or less, the average correlation coefficient decreased; (2) the members with bigger coefficients were selected as the members for the optimization ensemble forecast, results showed that the forecast effect was improved, and it was better than the ensemble average method, therefore, it was necessary to study the selection of number of ensemble forecast member; (3) either the ensemble average method or the optimization method, the forecast effect was bad for heavy rain or above level. Therefore, it is necessary to develop a better ensemble forecasting method which could be more applicable to disastrous weather forecast.