摘要
在水文及其他行业中广义极值分布(GEV)是一种较为常用的分布线型,包括极值Ⅰ型、极值Ⅱ型和极值Ⅲ型分布。在介绍了该分布线性矩与其参数关系后,给出了有效的可以考虑历史洪水信息的线性矩公式,并对由线性矩推求分布密度函数中k值作了改进。通过统计试验计算分析了线性矩在该分布中的统计性能及历史洪水公式的适用性,并与绝对值准则和平方和准则两种不同优化适线法、矩法作了比较。统计试验表明,改进k值计算后的线性矩法略优于传统线性矩法。从总体上讲,改进k值计算后的线性矩法优于平方和准则适线法及矩法,其参数及设计值的不偏性很好,与绝对值准则适线法总体上相当。
General extreme value distribution （GEV） is commonly used in hydrology and other fields, including extreme value type Ⅰ, extreme value type Ⅱ and extreme value type Ⅲ. After introducing L-moments of GEV distribution and its relationship with distribution parameters, a suitable formula of sample L-moments in consideration of the historical flood information was proposed, and the calculation of parameter k in probability density function was improved. The statistical performance of L-moments for estimation of population parameters and suitability of the historical flood formula proposed were analysed by Monte-Carlo method, then comparison with the moment method and curve-fitting methods with absolute criteria and square criteria was done. The results of Monte-Carlo experiments show that L-moment method with improved k value is better than the original one, both of them have very good un-bias of parameters and quantiles estimation. In general speaking, the L-moment method with improved k value is similar to the curve-fitting method with absolute criteria in statistical performance. However, it is better than the curvefitting method with square criteria and moment method.
出处
《水文》
CSCD
北大核心
2008年第3期8-13,共6页
Journal of China Hydrology
关键词
广义极值分布
线性矩
改进
历史洪水
优化适线法
general extreme value distribution （GEV）
L-moment
improvement
historical flood
curve-fitting method
Monte-Carlo
