Abstract:Because the main earlier circulation indices with different monthly variation features have different influences on midsummer (July-August) precipitation in China, different methods should be used to predict the midsummer prediction anomalies with respect to the monthly continuity and transition features of the impact factors. The 78 monthly circulation indices from National Climate Centre and the 30 indices from CPC/NOAA are employed to construct two factor schemes to describe the continuity and transition of circulation anomaly characteristics respectively. With the above different schemes, the “Running Correlation， Stepwise Regression， Ensemble Analysis” method is used to set up two models to predict the midsummer precipitation of 160 stations over China. Therefore, different variation features of earlier circulation indices are used in our statistic prediction models and different effects of the above two schemes are well shown. The result indicates that both of the factor treating schemes have certain predictive abilities but the transition one shows better effect. Meanwhile, for single station, the ensemble of the indices with its cross modeling series closely correlated to the observation time series holds more potential prediction ability.