基于LH-OAT的马斯京根模型参数全局敏感性分析

Analysis on Parameter Sensitivity of Muskingum Model Based on the LH-OAT

为定性了解参数C_0、C_1、C_2对马斯京根模型模拟洪水流量的影响程度,引入Latin-Hypercube抽样算法,并结合OAT方法,对马斯京根模型参数进行全局敏感性分析。改变了传统通过率定K、x值来计算C_0、C_1、C_2值的方式,直接对参数C_0、C_1、C_2进行全局敏感性分析,模型参数的敏感性分析分别以径流量和洪峰流量为评价目标。研究结果表明,对径流量:河道处于涨水阶段时,参数C_0较敏感,C_1、C_2一般敏感;河道处于落水阶段时,参数C_2较敏感,C_0、C_1一般敏感。对洪峰流量:参数C_0、C_1的敏感性相对前一时段的敏感性减弱,C_2的敏感性相对增加。改变某一组参数的扰动幅度(10%、20%、30%),参数C_0、C_1、C_2的敏感度随着扰动幅度的增大而增大。在敏感性分析的基础上,按敏感性顺序进行参数的率定,可提高69%的率定效率,减少模型参数率定的盲目性。

To quantitatively determine the influence of parameters on the Muskingum model for simulating the flood flow, the Latin- Hypercube sampling algorithm and the OAT method is used to analyze parameter＇ s sensitivity of the Muskingun model. The traditional method of calculating the C0 ,C1 ,C2 by using the rate fixed k and x value is changed, and the global sensitivity analysis of the parameter C0 , C1 ,C2 is carried out. The runoff and peak flow is the evaluation target of sensitivity analysis. The research results show that when the water line rises, Co is more sensitive than C1 ,C2.When the water line decreases, C2 is more sensitive than C0 ,C1 .For the peak flow, the sensitivity of C0 and C1 is weaker than that of the previous period, and the sensitivity of C2 is increased. The perturbation amplitude （ 10%, 20%, 30%） of a set of parameters is changed, and the sensitivity of the parameter C0 , C1 , C2 increases with the increase of the amplitude of the disturbance. On the basis of sensitivity analysis, the parameters calibration is based on the order of sensitivity, which can improve the efficiency of 69%, and the bhndness of the model parameter calibration is less.