In view of the chaos and randomness of precipitation and the high difficulty of accurate long-term precipitation prediction, the multiple time-varying sea-temperature index and factor prediction opinion index were proposed. Based on the atmospheric circulation and sea surface temperature (SST) field, a long-term combined precipitation forecasting model was constructed, which is verified by taking the Three Gorges Reservoir Basin as an example.The results show that the combined forecasting model has good applicability in the monthly precipitation forecast of the flood season from 1961 to 2020 in the Three Gorges Reservoir Basin, especially in June and September.Compared with multi-factor regression, random forest mathematical statistical model and CFSv2, ECMWF system 4 dynamic numerical models, this model is more robust and has significantly improved prediction accuracy.