中文摘要：

降雨径流模型在防洪预报调度中具有重要作用,然而由于模型参数具有不确定性,即使在观测到实际降雨的情况下也可能产生较大的预报误差,因此,根据预报结果作出的防洪调度决策存在一定的风险。为定量分析预报模型参数不确定性带来的防洪调度风险,本文提出结合普适似然不确定估计（Generalized Likelihood Uncertainty Estimation,GLUE）方法和模拟调度的防洪风险分析方法,以大量同效参数获得可能入库洪水过程集合,进而将模拟结果导入调度决策得到风险事件概率。在双牌流域采用新安江模型进行预报的实例研究说明了方法的有效性,对水库洪水预报调度过程中的风险决策具有应用价值。

英文摘要：

Rainfall-runoff models are important for flood forecasting and control, but large errors in their predictions could be produced even using the data of in-situ rainfall observation due to uncertainties in their parameters. Therefore, certain risks exist in the flood control decision making that is based on the forecasting results. For a quantitative analysis of such risks, this paper describes a flood risk analysis method that integrates a generalized likelihood uncertainty estimation（GLUE） method and simulations of reservoir operation. By using a large number of equivalent parameters, this method calculates a set of reservoir inflows for possible flood processes and estimates their effects on scheduling decision so as to obtain the probability of risk events. Its application in a case study of the Shuangpai basin together with the Xin＇anjiang model shows satisfactory effects in reducing forecast errors and advantages in the decision making of flood forecasting and control.