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

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.