不同Copula函数在洪水峰量联合分布中的应用比较  

Comparison of the Application of Different Copula Functions in Joint Distribution of Flood Peak and Volume

作　　者：杜懿[1,2] 麻荣永 DU Yi;MA Rongyong(College of Civil Engineering and Architecture,Guangxl University,Nanning 530004,Guangxi,China;Guangxi Key Laboratory of Disaster Prevention and Engineering Safety,Nanning 530004,Guangxi,China)

机构地区：[1]广西大学土木建筑工程学院,广西南宁530004 [2]广西防灾减灾与工程安全重点实验室,广西南宁530004

出　　处：《水力发电》2018年第12期24-26,58,共4页Water Power

基　　金：国家自然科学基金资助项目(51369005);广西防灾减灾与工程安全重点实验室系统性研究项目(2013ZDX04)

摘　　要：为推求适合郁江南宁水文站洪水峰量联合分布的最优Copula连接函数,将二元正态Copula函数与t-Copula函数引入到水文多变量频率分析计算中,并与应用广泛的Archimedean Copula函数族进行比较。结果表明,所选用的5种Copula函数均适用于本次研究,其中经验频率与理论频率的拟合相关系数分别为1. 016 2、1. 016 2、1. 034 9、0. 968 5和1. 014 1,平方欧式距离分别为0. 011 7、0. 010 9、0. 018 6、0. 017 9和0. 020 3。比较发现,二元t-Copula函数表现最优,最能体现郁江南宁水文站洪水峰量联合分布的特性规律。In order to get the best Copula connection function in the joint distribution of flood peak and volume in Nanning Hydrological Station of Yujiang River,the new normal Copula function and t-Copula function are introduced into the calculation of hydrological multivariate frequency analysis,and compared with the Archimedean Copula function.The results show that all five Copula functions are suitable for this research,the fitting correlation coefficients of empirical frequency and theoretical frequency are 1.016 2,1.016 2,1.034 9,0.968 5 and 1.014 1 respectively,and the square Euclidian distance are 0.011 7,0.010 9,0.018 6,0.017 9 and 0.020 3 respectively.The comparisons show that the t-Copula function is the most suitable,which can best reflect the characteristic of the joint distribution of flood peak and volume in Nanning Hydrological Station of Yujiang River.

关 键 词：水文变量 频率分析 联合分布 Coupla函数 

分 类 号：P333[天文地球—水文科学]