长江三峡区间因暴雨形成的洪水峰高量大， 对三峡水库防洪安全和运行调度的影响不容忽视。依据三峡地区的地形地貌特征，采用基于 GIS的分布式水文模型来模拟区间洪水过程，以尽量减小洪水预报中的不确定性。研究表明 ，降雨信息缺失是洪水预报不确定性的主要来源，尤其是在雨量站稀疏的山区，雨量站观测 的降雨信息难以充分表征降雨空间变化。为此，进一步尝试利用雷达测雨数据作为分布式水 文模型的输入，通过对2个小流域的示例研究，结果表明雷达测雨能捕获降雨的空间分布， 它与分布式水文模型相结合，是提高洪水预报精度的有效途径之一。

The regional floods occurring in the Three Gorges of the Yangtze River due to rainstorm have most important influence on the safety a nd operation of the Three Gorges Reservoir. For reducing the uncertainty of floo ds forecast as much as possible, a distributed hydrological model based on the c haracteristics of the catchment geomorphology has been developed to simulate reg ional floods processes. It is found that the inadequate information of rainfall from rain gauges is a dominant factor which brings about great uncertainty for f lood forecast, particularly in the mountainous area where the rainfall distribut ion greatly varies in space. The paper further tries to employ the radar rainfal l data obtained from rain gauges as the input of the model. The results of case studies in two small catchments of Three Gorges region show that the distributed hydrological model combining with radar rainfall data is a better approach to improve the accuracy of flood forecast.