摘要
不同时段长度的径流极值事件之间存在相关性,极值事件的共同发生具有重要的现实意义。基于极值理论,利用广义极值分布对不同时段长度的径流年极大值序列进行分布拟合,并以此为边缘分布分别建立Gumbel Copula、Clayton Copula和t-Student Copula两变量联合分布模型。结果表明:t-Student Copula上尾和下尾均存在相关性,能更好地给出真实数据厚尾分布的几何特征,捕捉到尾部的变化;两变量联合分布法得到的设计值较大,更偏于安全;利用Copula函数来描述不同时段长度的极端径流事件之间的相关关系,可以更加灵活可靠地推求设计径流过程和分析径流事件的条件频率。
There are dependence structures between extreme events of flood at different periods of time and it has important practical signifi-cance if extreme events occur at the same time.Based on the extreme value theory,this paper fit GEV distributions to the maximum value ofannual runoff series at different periods of time and fit Gumbel Copula、Clayton Copula and ellipse t-Student Copula to the fitted marginal dis-tributions. The results show that there are correlations in both the upper tail and the lower tail of t-Student Copula,which can depict the geo-metric features of the thick tail distribution of the real data and capture the changes of the tail dependence.The design value calculated by Copu-la function is bigger and more security.By using Copula method to depict the dependence between extreme events of flood at different periods oftime,we can calculate the design runoff processes and analyze the condition probability of the flood events much more flexible and reliable.
出处
《人民黄河》
CAS
北大核心
2017年第4期29-32,37,共5页
Yellow River
基金
国家自然科学基金资助项目(41371047)
中国科学院战略性先导科技专项(XDA05110102)
水文水资源与水利工程科学国家重点实验室专项经费项目(1069-514031112)
国家重点研发计划项目(2016YFC0402704)
