摘要
目前,水文模型不确定性的量化问题在水文研究中受到很大关注,在一些文章中提到了许多不确定性量化的方法,其中,GLUE方法和标准Bayesian方法是两种最常用的方法。主要讨论这两种方法在研究TOPMODEL模型时计算有效性和不同之处.通过用GLUE和标准Bayesian方法估计TOPMODEL模型参数的不确定性和模拟的不确定性,对这两种方法的结果进行评价,并讨论产生不同的原因,研究的主要结果为:(1)由Bayesian方法得到的参数后验分布比GLUE方法得到的离散型小。(2)给定GLUE中阈值(=0.8)的情况下,由Bayesian方法得到模拟流量的不确定性置信区间与GLUE方法得到的很接近。
Quantification of uncertainties in the hydrological models has attracted much attention in the hydrologic research field in recent years.Many methods have been reported for the uncertainty quantification,of which GLUE and standard Bayesian method are the two most commonly used methods.This paper focuses on the computational efficiencies and differences of the two methods in the study of TOPMODEL model.The uncertainties of model parameters and simulation were evaluated using the GLUE and standard Bayesian method,which showed that(1)the parameter posterior distributions generated by the Bayesian method are slightly less scattered than those generated by the GLUE method; and(2) the confidence intervals calculated by the Bayesian method and the GLUE method are very similar when the threshold value(=0.8)is given in GLUE.
出处
《南水北调与水利科技》
CAS
CSCD
北大核心
2014年第6期44-48,共5页
South-to-North Water Transfers and Water Science & Technology
基金
国家重点基础研究发展计划(973计划)项目"冰冻圈变化及其影响研究"第六课题"寒区流域水文过程综合模拟与预估研究"(2013CBA01806)
国家自然科学基金面上项目(41371049)
重点实验室开放基金项目(IWHR-SKL-201213)
江苏省研究生培养创新工程(CXLX13-239)