基于Copula函数的滦河流域旱涝组合事件概率特征分析

Study on the characteristic of multiply events of drought and flood probability in Luanhe River Basin based on Copula

以滦河流域多伦、承德、青龙和乐亭4个代表性气象站点1957—2010年日降水数据为基础资料,选取广义帕累托分布(Generalized pareto Distribution,GP)拟合各个气象站点春、夏、秋三季的降水距平百分率序列,并利用L-矩法估计参数,在此基础上采用二维Copula函数(Clayton和Gumbel-Hougaard分布函数)拟合春-夏、夏-秋降水距平百分率序列,依据降水距平百分率划分旱涝的标准进一步计算出各个站点旱涝交替和连旱连涝这两类旱涝组合事件发生的概率,结果表明:(1)Clayton和Gumbel-Hougaard分布函数均能较好地拟合季节间的降水距平序列;(2)青龙和乐亭较容易发生旱涝组合事件,春夏以持续干旱为主,夏秋以旱涝交替为主,夏秋季节间旱涝组合事件发生频率较高。

Based on the daily precipitation data of 1957-2010 at Duolun, Chengde, Qinglong and Leting Meteorological stations in the Luanhe River Basin, the sequences of percentage of precipitation anomalies （Spring/Summer/Autumn） were fitted by GP model. By the L-moment method, the parameters of GP were estimated. Two kinds of Copula functions （Clayton and Gumbel-Hougaard） were used to fit the spring-sum-mer and summer-autumn percentage of precipitation anomalies sequences and the probability of multiple events of drought and flood were calculated. The results show that （1） Both the Clayton and Gumbel-Hou-gaard can better fit the seasonal percentage of precipitation anomalies sequences; （2） The multiple events of drought and flood are most likely to take place in Qinlong and Leting, the frequency of long drought is high during spring and summer, the frequency of alternating occurrences of drought and flood is high dur-ing summer and autumn and the summer-autumn multiple events of drought and flood are easily to ex-plored.