南水北调中线工程代表站降水极值的概率分布特征研究  

Probability Distribution Characteristics of Extreme Precipitation at Representative Stations of the Middle Route of South-to-North Water Diversion Project

作　　者：竹磊磊 王纪军[1,2] 李凤秀[1,2] 周丹丹[3] 李芳 ZHU Leilei;WANG Jijun;LI Fengxiu;ZHOU Dandan;LI Fang(China Meteorological Administration·Henan Key Laboratory of Agrometeorological Support and Applied Technique,Zhengzhou 450003,China;Henan Climate Center,Zhengzhou 450003,China;Xuchang Meteorological Bureau,Xuchang 461000,Henan China)

机构地区：[1]中国气象局河南省农业气象保障与应用技术重点开放实验室,郑州450003 [2]河南省气候中心,郑州450003 [3]许昌市气象局,河南许昌461000

出　　处：《河南科学》2023年第9期1300-1308,共9页Henan Science

基　　金：中国气象局·河南省农业气象保障与应用技术重点开放研究室开放研究基金(AMF201509);中国气象局气候变化专项(CCSF201911);中国气象局·河南省农业气象保障与应用技术重点实验室应用技术研究基金项目(KM202245)。

摘　　要：基于南水北调中线工程受水区4个代表气象站1961—2021年年最大日降水量和年最大连续降水量,建立Pearson-Ⅲ型分布(P-Ⅲ)、广义极值分布(GEV)和Wakeby分布(WAK)的一维概率分布模型;根据从以上3个模型中选出的最优分布模型构建各站的Frank、Clayton和Gumbel Copula二维联合概率分布模型,并根据最优一维、二维模型分析重现期问题.结果表明:①南阳站、石家庄站和北京站强降水有由多日向单日集中的趋势.②GEV和WAK较P-Ⅲ分布在拟合斜率迅速减小的分布曲线时效果更好,WAK的效果又优于GEV.③郑州站和石家庄站20年一遇以上的降水值大于南阳站和北京站.④各站两个降水极值序列的Spearman和Kendall秩相关系数均较高,其中石家庄站最高,北京站次之,郑州站最低.⑤Gumbel Copula函数构建的二维联合分布最优;各站观测值的“或”(“且”)联合重现期大部分在10年(30年)以下,所有单要素重现期组合下的“或”(“且”)联合重现期,各站间的排序与两个秩相关系数各站间的排序一致(相反).Based on annual maximum daily precipitation(AMDP)and annual maximum continuous precipitation(AMCP)of four representative meteorological stations in the water-receiving area of the Middle Route of South-to-North Water Diversion Project during 1961-2021,the one-dimensional probability distribution models of Pearson-Ⅲ(P-Ⅲ)distribution,generalized extreme value(GEV)distribution and Wakeby(WAK)distribution are established for fitting the distributions of AMDP and AMCP.According to the optimal distribution model selected from the above three models,the two-dimensional joint probability distribution models of Frank Copula,Gumbel Copula and Clayton Copula are established for each station,and the return period problem is analyzed according to the optimal one-dimensional and two-dimensional models.The results show that:①The heavy rain at Nanyang station,Shijiazhuang station and Beijing station show a tendency from multi-day to single-day concentration.②GEV distribution and WAK distribution models are better than P-Ⅲdistribution model in fitting the distribution curves with rapidly decreasing slope,and the effect of WAK distribution model is better than that of GEV distribution model.③The extreme precipitation of more than once in 20 years of Zhengzhou and Shijiazhuang stations is greater than that of Nanyang and Beijing stations.④The Spearman and Kendall rank correlation coefficients between two extreme precipitation series of all stations are high,among which Shijiazhuang station is the highest,Beijing station is the second,and Zhengzhou station is the lowest.⑤the Gumbel Copula two-dimensional joint distribution function is more suitable than Frank and Clayton Copula functions in this study.Most of the“OR”(“AND”)joint return periods observed at each station are less than 10(30)years,and under all combinations of single element return periods the orders of“OR”(“AND”)joint return periods between stations are consistent with those spearman and Kendall rank correlation.

关 键 词：降水极值 Pearson-Ⅲ型分布 广义极值分布 Wakeby分布 COPULA函数 

分 类 号：P468[天文地球—大气科学及气象学]