基于Copula理论与方法,以广州1951~2010年的日降水为例,以最大日降水量为基准,构建最大日降水量(W1)与历时3日(W3)降水量,最大日降水量(W1)与历时7日(W7)降水量两个组合的联合概率分布模式。经择优检验建立了边缘分布为广义极值和P-III型的Gumbel-Hougaard Copula两变量联合分布。随之,推算了两个组合降水的同现重现期和设计暴雨值。最后,依据条件分布计算了在大于或小于年最大日降水量特定设计暴雨条件下超过历时3日或7日降水设计值的风险率。
Based on the Copula theory and method,the joint probability distribution models between annual maximum daily rainstorm and three-day or seven-day rainstorm amounts were build with the maximum daily rainstorm as a reference using the daily rainfall data in Guangzhou City from 1951 to 2010 as a case study.Firstly,the Gumbel-Hougaard Copula function selected by the goodness-of-fit test was used to build the joint distributions which are the marginal distributions of generalized extreme value and Pearson type III;Then,the co-occurrence return periods and the design rainstorm values were calculated;Finally,the risk probabilities of exceed successive three-day or seven-day rainstorm amounts were calculated based on conditional distribution.