基于Halphen分布和最大熵的洪水频率分析

设计洪水估算值对水利水电工程的设计规划至关重要,其关键是选择合适的分布线型和参数估计方法。目前采用单一的频率曲线进行水文分析计算,其设计结果具有较大的不确定性。引入Halphen分布函数进行洪水频率分析,并提出了基于最大熵原理的参数估计方法。利用丹江口水库1929~2014年的年最大日流量资料系列,分析检验Halphen分布的拟合效果,并与传统的分布进行比较。结果表明,Halphen函数具有灵活的形式和较好的尾部特征,适合于水文频率分析,其拟合结果基本优于其他的传统分布。

Flood Frequency Analysis Using Halphen Distribution and Maximum Entropy

Flood frequency analysis (FFA) provides an important basis for determining the construction size and operation strategy ofwater conservancy project. The main issue in FFA is how to select a suitable distribution and parameter estimation method. In thisstudy, the principle of maximum entropy method was employed to derive three types of Halphen distributions and estimate parameters.The annual maximum daily flow data series from 1929 to 2014 in the Danjiangkou Reservoir were selected to test the performancesof the proposed methods. The fitting ability of Halphen distributions was compared with those of common used distributionsin FFA. Results indicate that Halphen distributions have flexible forms and better tail properties and are suitable for flood frequency analysis.Compared with the widely used distributions in hydrology, the performance of Halphen distribution is comparable or even better.