Abstract:By using the prediction products of the Beijing Climate Centre Second-Generation Climate Prediction Model System (BCC-CPSv2) and a hybrid downscaling method as well as the Indian Ocean SST signals, this paper establishes a summer precipitation prediction model for eastern part of Northwest China. Relative to BCC-CPSv2 model, the prediction skill of this model is significantly improved for the summer precipitation in eastern part of Northwest China from 1991 to 2017. The spatial correlation coefficient increases from 0.42 to 0.75, and the root mean square error decreases obviously, down most by 80%. The model has better prediction ability for the spatial distribution pattern of precipitation anomaly percentage, such as for the distributions of the summer precipitation anomaly percentages in 1987 and 2010. By grasping the spatial distribution characteristics of meteorological variables, this prediction method can correct the prediction errors of dynamic model products and provide scientific basis and technical support for summer precipitation prediction in eastern part of Northwest China, so it is expected to have a good application prospect.