研究pair-copula在干旱特性联合概率中的应用.以渭河流域咸阳站降雨资料为例,采用游程理论,选取干旱历时、干旱烈度和烈度峰值为干旱特性变量,应用Pearson线性相关系数、Spearman相关系数和Kendall秩相关系数进行相依性度量.采用4种常用的copula函数构造了12种pair-copulas,以RMSE、AIC、BIC为准则选择最优的pair-copula.运用Rosenblatt变换的Bootstrap法进行copula拟合度检验,推导3变量的联合概率分布.与3维对称、非对称阿基米德copulas和椭圆copula比较,表明paircopula可以描述多变量水文概率分布.%This paper studied how to apply pair-copula in drought characteristics probability. Taking the rainfalls at the Xianyang station on the Weihe River as a study case, drought duration, severity and peak were selected as the research variables based on the run theory, using Pearson linear correlation coefficient rn, the Spearman correlation coefficient pn and the Kendall rank correlation coefficient Tn to measure the dependency, then using 4 kinds of commonly used copula functions to construct 12 kinds of pair-copulas, taking RMSE, AIC, BIC as criterions to choose the most superior pair-copula, using Rosenblatt-Bootstrap transformation for goodness-of-fit test, obtaining the joint probability expression of three variables, after comparison with three dimensional symmetrical and asymmetrical Archimedean copulas and one Meta-elliptical copula, to indicate that pair-copula can describe the multivariable hydrology probability distribution.
展开▼
