双参数Copula函数在洪水联合分布中的应用研究

Analysis of Two-parameter Copula Function in Flood Joint Distribution

采用金沙江屏山站和岷江高场站的日径流资料,分别构造了两江洪峰流量以及金沙江峰、量之间的联合分布,比较了单参数与双参数Archimedean Copula函数的拟合效果,估计了两江洪峰遭遇风险的条件概率,探讨了双参数Archimedean Copula函数在洪水联合分布中的应用。结果表明:双参数Copula函数拟合效果明显优于单参数Copula函数,且同时具有上尾相关性和下尾相关性,在水文极值分析中是有广阔的应用前景;当金沙江发生千年一遇洪水时,岷江发生千年一遇洪水的概率为3.458%。

The daily runoff data from the Pingshan station on the Jinshajiang River and Gaoyang station on the Minjiang River were selected to construct the joint distribution of flood peaks and flood volumes.The goodness of fit between one-parameter and two-parameter Archimedean copulas were compared and the conditional probabilities of flood combination risk were derived to explore two-parameter Archimedean copula in bivariate flood frequency analysis.The results show that the goodness of fit of two-parameter copula is much better than that of one-parameter copulas and the two-parameter copula has properties of both upper and lower tail dependences so that it could be widely promoted in hydrological extreme events analysis.The case study results show that the conditional probability of the Minjiang River occurring 1000-year flood equals to 3.458% if 1000-year flood occurs at the Jinshajiang River.