一种改进的参数区间选取方法及 在 WEP2L 模型中的应用

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 WEP2L( Water and Ener gy transfer Pr ocess in Lar ge 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 WEP2L 实现参 数自动率定, 模拟精度能达到 01 633; ( 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。

An improved method of parameter range selection and its application to WEP2L model

There ar e many par amet ers in a distributed hydrolog ical mo del and the parameters ar e no t completely independent from each other. During model calibr atio n, there is mut ual restrain o f t he selectio n o f some par ameters. T he WEP2L, a distribu2 ted hydro log ical model, w as taken as an example to v alidate the efficacy of the GLUE method fo r parameter calibratio n. T he So bo l method and the parameter range selectio n method of co ncept ual hydro log ical models w ere combined to pro vide a scheme of parameter calibratio n fo r WEP2L. The scheme w as applied to the Baihe basin. Results showed t hat the par ameters of WEP2L model could be auto2calibrated w ith the help of GLUE method, Sobol analy sis, and par ameter range selection. The simulatio n pr ecision w as 01 633. The Sobol ana lysis can o ptimize the principle of par ameter rang e selectio n and im pr ove the simulation pre2 cisio n. The results can pr ov ide useful refer ence for the par ameter calibar ation of distributed hydro log ical models.