摘要
针对水文模型参数不确定问题,选取干旱半干旱地区开都河流域为研究对象,提出了基于多项式混沌展开的水文模型参数敏感性分析方法。首先在开都河流域多年气象、水文观测数据的基础上建立SLURP水文模型;然后利用稀疏网格配置法获得配置点对模拟径流量进行多项式混沌展开;最后根据混沌系数计算Sobol指标评估模型参数及其交互效应对径流的影响。结果表明,在参数主效应中选取的参数中Sobol敏感度指标最大的是降雨系数,最小的是深层地下水截留常数;在交互效应中,浅层土壤蓄水容量和降雨系数之间交互效应最强,Sobol敏感度指标最大,浅层土壤截留常数与深层地下水截留常数之间交互效应最弱,敏感度指标最小。
Aiming at the parameter uncertainty of hydrological model,apolynomial chaos expansion method was proposed for parameter sensitivity analysis of SLUPR model in Kaidu River Basin,China.Firstly,the spatial and temporal variations associated with system components such as precipitation,topography,and vegetation were tackled using the SLURP hydrological model.Then,the collocation points were obtained through the sparse grid based stochastic collocation method and the response surface model was constructed using polynomial chaos expansion.Finally,the Sobol sensitivity indices were calculated to reflect the sensitivity of parameters and their interactions on modelling outputs.Results indicate that PF has the largest sensitivity indices and plays an important role in runoff simulation;the interaction between MF and PF should be paid more attention to enhance the model performance.
出处
《水电能源科学》
北大核心
2016年第10期14-18,共5页
Water Resources and Power
基金
青海省科技支撑计划(2015-SF-130)
中央高校基本科研业务费专项资金项目(2015XS100)
关键词
开都河
水文模型
稀疏网格
多项式混沌展开
敏感性分析
Kaidu River
hydrological model
sparse grid
polynomial chaos expansion
sensitivity analysis
