考虑不定量历史洪水的水文频率参数贝叶斯估计  

Parameter estimation in hydrologic frequency analysis using non-quantitative historical flood and Bayesian approach

作　　者：尚晓三[1,2] 王栋[1] 

机构地区：[1]南京大学地球科学与工程学院水科学系,南京210023 [2]安徽省水利水资源重点实验室,安徽蚌埠233000

出　　处：《水力发电学报》2015年第10期35-41,共7页Journal of Hydroelectric Engineering

摘　　要：历史洪水能够有效提高洪水频率分析结果精度和可靠性。在实际应用中,部分历史洪水难以定量,常用的确定性方法对此类资料难以得到有效利用,影响到洪水频率分析结果的可靠性。本次研究将最大似然估计和叶斯方法耦合起来,进行考虑不定量历史洪水信息的洪水频率分析,分别以模拟和实测序列加以验证,相比传统确定性方法,该方法不仅可有效利用不定量历史洪水信息,还可提高洪水频率分析成果的精度。Consideration of historical flood could improve the accuracy and reliability of flood frequency analysis. In practice, however, some historical floods cannot be quantitated and are difficult to use in a common deterministic methods, thus lowering the reliability of parameter estimation and design flood in the flood frequency analysis. This paper focuses on the coupling of a Bayesian method a maximum likelihood estimator in the frequency analysis using non-quantitative historical floods. This coupling method and its application are demonstrated in a case study of synthetic and measured flood series. Compared to several traditional deterministic method, the coupling method not merely makes it possible to make use of non-quantitative historical floods, but can improve estimation accuracy.

关 键 词：贝叶斯方法 不定量历史洪水:洪水频率分析 最大似然估计 

分 类 号：P333.6[天文地球—水文科学]