中文摘要：

以位于珠江三角洲感潮河网地区的洪潮遭遇概率问题为实例,对当前应用较多的各种二变量联合概率模型进行了比较研究,包括混合Gumbel分布、Gumbel-Logistic分布、以及阿基米德族Copula函数中的Gumbel-Hougard Copula模型、Clayton Copula模型、Genest-Ghoudi Copula模型等。结果表明,不同的二变量概率分布模型对于变量之间的相关性都有不同程度的限制,实际应用时需要首先根据水文变量的相关性特点选择适宜的模型。Copula函数模型则不限定边际分布型式,比混合Gumbel模型和Gumbel-Logistic模型具有更强的灵活性。多种拟合优度检验方法对各模型的检验结果表明,虽然各模型都能通过K-S检验,但其拟合优度有显著不同,实际应用时需进行拟合优度检验以择优。

英文摘要：

Taken the case of problem of flood and stormtide encounter in estuary area of the Pearl River delta,several bivariate hydrological probability models are compared and analysed,which include mixed Gumbel model,Gumbel-Logistic model,and the Copula method,such as Gumbel-Hougard Copula,Clayton Copula,and Genest-Ghoudi Copula,etc,which belong to Archimedean Copulas.The results showed that the different bivariate probability models have different degrees limitation to the correlation between variables,and the fitful models should be chosen correctly by variables correlativity.Compared with mixed Gumbel model and Gumbel-Logistic model,the Copula models have no limitation to marginal distribution,and the marginal distribution can be any types of probability distribution function.Several goodness-of-fit test methods are applied to select the best models based on Kolmogorov-Smirnov test,RMSE,BIAS,and AIC criterion.For each model,goodness-of-fit are different obviously,and the best model should be determined according to the characteristics of observed hydrological datum.