Abstract: The 130 climate system monitoring index issued monthly by the National Climate Center and the precipitation data from 1954 to 2016 recorded by Jingzhou climate station were analyzed to find the correlation coefficient between these climate index and the flood season precipitation anomaly percentage as well as the main flood season precipitation anomaly percentage. Using stepwise regression method， the index with the correlation coefficients larger than 0.3 were chosen as predictor group to establish the Jingzhou precipitation forecast models of flood season and main flood season. The statistical results showed that the correlation coefficients of precipitation forecast models of flood season and main flood season were 0.874 and 0.914 respectively， which were significantly higher than the coefficients of single factor. The flood season and main flood season precipitation anomaly percentages of Jingzhou in 2016 were forecasted to be more than the normal 17.3% and 223.2% with those models. Comparing to the actual value of more than 6.2% and 30.2%， the flood season model gave better forecasting result and the main flood season model gave same trend but too large value. The large forecasting error of main flood season model may be related to the shortage of effective data of last November India subtropical high pressure zone index and larger coefficient. More data was needed to tune the relevant coefficient to improve the forecasting accuracy.

Key words: flood season precipitation, climatic monitoring index, predictive factors, prediction model