模型参数的敏感性分析是洪水预报的基础性工作。以中国洪水预报系统为平台，采用局部分析法，对MSKLOSS河道洪水演算模型参数进行敏感性分析，以提高洪水预报的精度。为提高MSKLOSS河道洪水演算模型的应用效率，考虑模型适用性、模拟分析方便性，选取海河南系河道下垫面接近现状条件的滹沱河黄壁庄-北中山、南运河岳城-蔡小庄“96?8”和大清河东茨村-新盖房“12?7”三场洪水资料对模型参数进行分析。分析表明：河道分段数、初始下渗率、下渗曲线指数和湿周等参数的敏感性较高，稳定下渗率的敏感性在洪水量级较小时相对较高，在参数优选和实时作业预报时需注意其初始范围的设定。

Sensitivity analysis of model parameters is the basic work of flood forecasting. Based on the platform of Chinese flood forecasting system, we conducted sensitivity analysis of the parameters of MSKLOSS river flood routing model using the local analysis method, so as to improve the accuracy of flood forecasting. In order to improve the application efficiency of MSKLOSS river flood routing model, considering the applicability of the model and the convenience of simulation and analysis, we selected three flood locations at the south of Haihe river, which were Huangbizhuang to Beizhongshan of Hutuo River, Yuecheng to Caixiaozhuang of South Canal in August 1996, and Dongcicun to Xingaifang of Daqing River in July 2012, where the underlying surface of the river is close to the present condition. Results showed that the number of river segments, initial infiltration rate, infiltration curve index, and wetted perimeter are highly sensitive. Steady infiltration rate is more sensitive when the flood magnitude is small. Attention should be paid to the initial range setting during parameter optimization and real-time operational forecast.