洪水频率计算中密度估计法及其稳健性的研究

Density Estimation Method and Its Robustness in Computation of Flood Frequency/Magnitude

本文引进密度函数的核估计结合再抽样的方法,研究洪水频率的分析计算;并用统计试验方法,对P-Ⅱ-CFM法,LN3-PWM法及密度估计法(DEM),分析比较了五组参数,三组样本特征等总共90种方案.统计结果表明:在这三种方法中,稳健性最好的是非参数的密度估计法,P-Ⅱ-CFM法次之,LN3-PWM法最差.另外,由于DEM法推求出的设计值无偏性及有效性都较好,而且它还跳出了传统的推求设计值要先假定分布线型的框框,避开了水文设计值计算上困惑多年的线型问题.因此,是一种有应用和研究价值的方法.

The Density Estimation Method combined with Resampling Method is used to study flood frequercy computation. The Pearson Type-Ⅲ (p- Ⅲ) Curve-Fitting Method, the Three Parameter Log-Normal (LN3) Probability Weighted Moments Method and the Density Estimation Method (DEM) are compared and analysed in regard to flood series frown P- Ⅲ distribution and LN3 distribution by using statistical testing method. There are ninety schemes which are compared and analysed toward five-set parameters and three-set sample characterstics. The statistical results indicate that the DEM method is the most robust estimation method, and the LN3-PWM method is the least robust of these three methods. Besides the DEM's bias and root mean square are smaller than those of the methods whose original distribution does not correspond with the assumed distribution, thus it is worth applying to flood frequency analysis.