Study on Interannual Increment Forecasting Approachfor Rainy Season Precipitation in Shaanxi
An interannual precipitation increment forecasting model with high predictive capability for forecasting the summer rainfall in Shaanxi Province is presented by analyzing the physical mechanisms of precipitation during flood season based on the interannual increment forecasting approach. The research results show that the interannual SST increment in middle and eastern equatorial Pacific, the 500 hPa height interannual increment and rainfall during flood season in Shaanxi have very good correlations. When the distribution pattern is north south “- + -” in interannual increment of SST in middle and eastern equatorial Pacific from the previous autumn to winter, there would be more rains in the summer. Conversely it would rain less. When the distribution pattern of the interannual increment at 500 hPa height is positive, showing zonal, near the equator from the previous autumn to winter, there would be more rains in the summer. Otherwise, it would be dry. In this research, we find a good correlation between the interannual increment of 74 circulation features from National Climate Centre, soil temperature at 0 cm depth and precipitation during flood season in Shaanxi. Based on the analysis of the physical meanings of predictors, predictors are introduced via the optimal subset regression, and 40 prediction models for rainy season precipitation and each month (June, July and August) of 10 climatic regions of Shaanxi are established by using the interannual increment approach. Cross validation testing of rainy season precipitation shows that the anomaly consistency rate reaches 78.4%. It is shown by the hindcasting results during 2010-2013 that the PS score of flood season precipitation is 75.8 and 66, separately. Thus, this work demonstrates that the interannual increment approach has good predictive skill. It can improve the level of predicting rainfalls in Shaanxi flood season and can be used as an effective way in forecasting operations.