BTOPMC模型参数敏感性的定性及定量分析

Qualitative and Quantitative Sensitivity Analysis of Parameters in the BTOPMC Model

敏感性分析(SA)是用于识别模型重要参数的最具有影响力的方式。以富士川流域作为研究区,对包含5个具有明确物理意义的参数的分布式水文物理模型BTOPMC进行敏感参数筛选。利用Morris One At A Time(MOAT)方法,在适当的采样设计和有效筛选敏感参数所需的足够样本大小条件下定性分析BTOPMC模型参数敏感性。根据输出结果发现,该模型中的敏感变量有9~12个,涉及到的参数有Srmax(m)、m(m)和n0。此外,采用基于导数的全局敏感度量(DGSM)这一定量方法验证定性分析的结果并量化所有参数的重要性,该方法所得结果与定性分析结果基本一致。

SA(Sensitivity analysis) is the most efficient way to identify important parameters in a model. In this paper, sensitive parameters of a physically based distributed hydrological model, BTOPMC, are identified by using SA methods. And the Fujikawa Basin is chosen as the research area. Firstly, MOAT(Morris One At A Time) method is used to qualitatively analyze the sensitivity of BTOPMC model parameters under appropriate sampling design and sufficient sample size required for effective screening of sensitive parameters. According to the outputs, it has been found that there are 9~12 sensitivity variables in BTOPMC, and the involved parameters are Srmax(m), m(m) and n0. In addition, a quantitative method, DGSM(Derivative-based Global Sensitivity Measures), is used to verify the results of qualitative analysis, and the importance of all parameters is quantified. The result of quantitative SA is basically consistent with that of qualitative SA.