基于贝叶斯理论的水文频率分析方法研究  

Hydrological frequency analysis based on Bayesian theory

作　　者：梁忠民[1] 戴荣[1] 雷杨[1] 余钟波[1,2] 

机构地区：[1]水文水资源与水利工程科学国家重点实验室,河海大学,南京210098 [2]Department of Geoscience, University of Nevada, Las Vegas, NV 89154,USA

出　　处：《水力发电学报》2009年第4期22-26,共5页Journal of Hydroelectric Engineering

基　　金：国家自然科学基金(50779013);高等学校博士学科点专项科研基金资助课题(20070294018);长江学者和创新团队发展计划资助(IRT0717)

摘　　要：探讨了贝叶斯理论在水文频率分析计算中的应用。根据贝叶斯公式耦合先验和样本信息,采用Markov chain Monte Carlo(MCMC)抽样技术估计参数的后验分布,并通过参数的随机样本构造设计值的抽样分布,根据设计值的抽样分布可以推求设计值的点估计和区间估计。与传统水文频率分析方法相比,基于贝叶斯理论的分析方法不仅能提供设计值的各种估计,同时能够对估计的不确定性进行定量评价,为水文频率分析计算提供更丰富的信息。The application of Bayesian theory to hydrological frequency analysis is investigated in this study.Proior and sample information are coupled under the framework of Bayesian formula,the Markov Chain Monte Carlo（MCMC） sampling approach is used to estimate the posterior distributions of parameters of a probability distribution function.Thus the sampling distribution of a quantile or the hydrological design value is obtained through a large number of parameter sets generated from these parameter posterior distributions,upon which an estimation of the quantile or confidence limits can be acquired.Comparing with the traditional hydrological frequency analysis method,the Bayesian method could provide not only various kinds of estimators of a quantile,but also quantitative evaluation on the uncertainty of these estimators.Therefore,the Bayesian method brings more useful information to hydrological frequency analysis.

关 键 词：水文学及水资源 水文频率分析 贝叶斯理论 MARKOV CHAIN MONTE Carlo 

分 类 号：P333.6[天文地球—水文科学]