年径流形成过程受到很多因素的影响,对其未来的精确描述十分困难,常用的成因预测法和统计预测法都有其一定局限性.以年径流作为自回归因子,降水、气温和用水做为回归因子,建立了年径流预测的混合回归模型,并以黄河三门峡站为例,对黄河三门峡站天然年径流未来变化进行了6种方案下的预测.研究表明,混合回归模型既能较好地模拟气温、降水及耗水变化对年径流量的影响,也能体现年径流自身的演变规律. The formation process of annual runoff is affected by many factors, and it is very difficult to accurately describe its future. The commonly used genetic predication and statistical forecasting have their own limitations. Taking annual runoff as an autoregressive factor, precipitation, temperature and water as Regression model was established to predict the annual runoff in the Yellow River Sanmenxia Station, and the prediction model of annual runoff was established under the six scenarios.The results show that the mixed regression model can not only better Simulated temperature, precipitation and water consumption on the annual runoff, but also reflect the evolution of annual runoff itself.