Abstract:According to the annual warning water levels from the Jinchengjiang Guaging Station on the Longjiang River, with 500 hPa circulation index, circulation characteristic amount, local climate and etc., analysis can be made with variance cycle, multiple linear regression, stepwise regression procedure. Then, the predicted values can be taken while error values be received to make analysis and test through the integrated model of neural network. The results show that the model has achieved a remarkable effect in weather forecasting, which can be used for drought and flood predicting.