一种改进的参数区间选取方法及在WEP-L模型中的应用  

An improved method of parameter range selection and its application to WEP-L model

作　　者：段浩[1] 赵红莉[1] 蒋云钟[1] 贾仰文[1] 杜军凯[1] 刘佳嘉[1] 徐飞[1] DUAN Hao;ZHAO Hongli;JIANG Yunzhong;JIA Yangwen;DU Junkai;LIU Jiajia;XU Fei(China Institute of Water Resources and Hydropower Research,Beijing 100038,China)

机构地区：[1]中国水利水电科学研究院水资源研究所,北京100038

出　　处：《南水北调与水利科技》2018年第4期50-57,共8页South-to-North Water Transfers and Water Science & Technology

基　　金："十三五"国家水资源立体监测体系与遥感技术应用(2017YFC0405804);中国水利水电科学研究院基本科研专项(WR0145B012017)~~

摘　　要：分布式水文模型由于参数众多,各参数之间不能完全独立,在参数自动率定中存在部分参数的优选过程相互抑制情况。以WEP-L(Water and Energy transfer Process in Large river basins)分布式水文模型为例,基于GLUE算法推求模型参数的后验分布,将参数全局敏感性分析的Sobol方法与概念性水文模型的参数区间优选方法相结合,给出了在较少参数采样次数条件下分布式水文模型不完全独立的多参数自动优选方案,并在白河流域进行了应用。结果表明:(1)用GLUE算法推求参数后验分布,结合Sobol分析及参数的区间优选方法,可对WEP-L实现参数自动率定,模拟精度能达到0.633;(2)将Sobol全局敏感度分析与参数的区间优选方法相结合,可进一步优化参数区间确定的原则,提高模拟的精度。研究结果可为分布式水文模型的参数率定提供借鉴。There are many parameters in a distributed hydrological model and the parameters are not completely independent from each other.During model calibration,there is mutual restrain of the selection of some parameters.The WEP-L,a distributed hydrological model,was taken as an example to validate the efficacy of the GLUE method for parameter calibration.The Sobol method and the parameter range selection method of conceptual hydrological models were combined to provide a scheme of parameter calibration for WEP-L.The scheme was applied to the Baihe basin.Results showed that the parameters of WEP-L model could be auto-calibrated with the help of GLUE method,Sobol analysis,and parameter range selection.The simulation precision was 0.633.The Sobol analysis can optimize the principle of parameter range selection and improve the simulation precision.The results can provide useful reference for the parameter calibaration of distributed hydrological models.

关 键 词：GLUE 参数区间 WEP-L 白河流域 Sobol分析 

分 类 号：TV11[水利工程—水文学及水资源]