径流预测是水资源合理利用的前提。组合预测方法是一种综合预测技术，能利用各个体预测模型中有用的独立信息。贝叶斯分析可以充分吸取径流本身和预测模型等多种成分中的信息，考虑多种不确定性，既可进行预测，又可以对预测模型进行分析评价。两者的结合能进一步提高预测效果，从而获得更充分的信息。 Runoff prediction is the prerequisite for rational utilization of water resources. The combined forecasting method is a comprehensive forecasting technique that can make use of the individual independent information useful in each individual forecasting model. Bayesian analysis can fully absorb the information of many components such as runoff itself and forecast model, and consider a variety of uncertainties, both for prediction and for analysis and evaluation of prediction models. The combination of the two can further improve the predictive performance, so as to obtain more sufficient information.