几种Copulas模拟不同历时降雨量的影响分析  

Impact analysis for rainfall depthsimulation of different durations through several Copulas

作　　者：许月萍[1] 童杨斌[1] 富强[1] 朱蓉[1] 

机构地区：[1]浙江大学建筑工程学院,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2009年第6期1107-1111,共5页Journal of Zhejiang University：Engineering Science

基　　金：武汉大学水资源与水电工科学国家重点实验室开放基金资助项目(2007B032);浙江省自然科学基金资助项目(Y507071);教育部博士点新教师基金资助项目(200803351029)

摘　　要：为减小水文频率分析中的不确定性,采用4种不同的Copulas联结函数来模拟不同历时降雨量的相关关系,并得出边际分布分别为广义极限和广义逻辑斯特的两元联合分布.模拟结果表明,Farlie-Gumbel-Morgenstern和Gaussian Copulas能较好地模拟变量之间的相关关系,而Gumbel和Clayton Copulas则相对较差.通过计算Copula的条件分布可以得到不同历时降雨量组合的概率.根据Copula联结函数来推求不同重现期的设计暴雨,可以同时考虑不同历时降雨量的相关性,该方法科学合理,为水文频率分析方法提供了新的思路.To reduce uncertainty in hydrological frequency analysis, four different Copulas were used to model the bivariate distribution functions for rainfall depths of various durations with generalized extremevalue distribution and generalized logistic distribution as marginals. The simulation showed that the Farlie- Gumbel-Morgenstern and Gaussian Copulas can model the dependence between variables reasonably, while Gumbel and Clayton Copulas did less well. The conditional probabilities of rainfall depth combinations for different durations can be obtained. The Copula method to calculate design rainstorm for different return periods is reasonable and provides a new alternative to traditional frequency analysis.

关 键 词：COPULA 设计暴雨 广义极限分布 广义逻辑斯特分布 降雨量 不同历时降雨 

分 类 号：TV122.5[水利工程—水文学及水资源]