摘要
基于copula函数论述了两变量的联合概率分布方法。此方法的主要优点是边缘分布可由不同的分布函数构成,变量间可具相关性。以粤东汕尾海域极值波高与相应风速为研究实例,经分析获得以下结果：（1）优选的极值波高和风速可分别由P-III型和GEV分布表示;（2）拟合优度检验指标表明二者的最优连接函数为Archimedean copula类的Gumbel-Hougaard copula;（3）与联合分布比较,重现期介于5—200年之间的波高边缘分布设计值的相对差值大约介于3.1%~8.1%之间,风速设计值相对差值大约介于2.8%~6.4%之间;（4）特定风速设计频率条件下,波高与风速的遭遇概率随波高设计频率的减小而减小,特定波高设计频率随风速条件频率的减小,二者的遭遇概率随之增大。
This article introduced the method of bivariate joint probability distribution based on the copula function.A major advantage of this method is that marginal distributions of individual variables can be of any form and the variables can be correlated.Some of conclusions were reached by using extreme wave height and wind speed as an example collected in Shanwei sea area as the following：（1） Optimized marginal distributions of wave height and wind velocity can be represented by the Pearson pattern three and generalized extreme value distribution,respectively;（2） Gumbel–Hougaard Copula that belongs to Archimedean copula family was the optimal copula selected by the goodness-of-fit test;（3） The relative differences of the special frequency design values between the marginal distribution of wave height and the joint distribution fall in between 3.1% and 8.1% for the return periods between 5 and 200 years,and the relative differences of wind velocity fall in between 2.8%~6.4%;（4） The encountering probabilities of wave height given wind velocity decrease along with decreasing the frequency of wave height;whereas the encountering probabilitiy increase while the specific wave height frequency is along with decrease of wind velocity frequency.
出处
《海洋通报》
CAS
CSCD
北大核心
2011年第2期158-163,共6页
Marine Science Bulletin
基金
国家自然科学基金资助项目(40576041)
关键词
极值波高和风速
边缘分布
联合概率分布
COPULA函数
条件分布
extreme wave height and wind velocity
marginal probability distribution
joint probability distribution
copula function
conditional probability distribution
