游程理论和Copula函数在二维干旱变量联合分布中的应用

Application of the theory of run and Copula function to the joint distribution of two- dimension drought variables

采用3种不同的截取水平作为游程理论的干旱识别方式,从都江堰站51年逐月降水距平百分率指数序列中识别干旱事件,然后建立干旱历时与干旱烈度的两变量Copula函数模型,并进行概率分析和重现期计算。结果表明:1)干旱统计结果与截取水平的相关性强,具有一定的可预测性。2)Gumbel Copula函数为最优的联合分布函数。3)不同截取水平的概率分析结果具有明显的差异,且差异程度波动大,而重现期计算结果差异小。4)干旱风险评估以重现期分析为主、概率特征分析为辅更合理。表明游程理论和Copula函数的联合应用,可合理评估区域干旱风险。

Based on the precipitation anomaly percentage index coming from the monthly date of Dujiangyan meteorological station in 51 years,the drought events were identified by using the three different cutting levels of the theory of run,the joint distributions of drought duration and severity were constructed by using Copula function,and probabilistic characteristics and return period were analyzed. The results showed： 1） Drought statistical results was greatly related to cutting levels,and it was predictability. 2） Gumbel Copula function fitted the drought duration and severity best under different cutting levels. 3） The result of probability analysis showed great difference,and the degree of difference varied,while the return period had small difference. 4） Drought risk assessment was more reasonable when using the analysis of return period supplemented with the analysis of probabilistic characteristics. It was indicated that the combination of the theory of run and Copula function in regional drought risk assessment was reasonable.