Rainfall is an important factor affecting agricultural production, how to forecast the rainfall become the guiding agriculture, water conservancy and other important indicators of science and technology. From the point of information utilization, the single forecasting model only use the part of the rainfall data, and the combination model will be complementary to the advantages of the single model, and get a better forecasting effect. The rapid development of neural network theory and Cascade neural network prediction model is widely used in all aspects and achieved good results. According to the characteristics of rainfall curve, Through the analysis of BP neural network and RBF neural network, we can find that BP neural network can be a good fit for the rainfall has a great impact on climate information and other factors, the output of the same type of rainfall impact information, and The characteristics of RBF network can be used to extract the features of the same kind of information, and the combination of the two can greatly improve the accuracy of rainfall prediction. In view of this, the BP-RBF cascade neural network is introduced into the study of rainfall prediction. The calculation results show that the proposed method is higher than the single neural network prediction accuracy, which proves that the method is reasonable and effective.