重庆多历时极值降水的概率型优选及应用  

Probability distribution model optimization and application of multiple duration extreme precipitation in Chongqing

作　　者：王颖 刘晓冉[1] 程炳岩[1] 孙佳[1] WANG Ying;LIU Xiaoran;CHENG Bingyan;SUN Jia(Chongqing Climate Center,Chongqing 401147,China)

机构地区：[1]重庆市气候中心,重庆401147

出　　处：《气象科学》2020年第4期560-568,共9页Journal of the Meteorological Sciences

基　　金：公益性行业(气象)科研专项(GYHY201506014);2018年重庆市气象局智慧气候应用创新团队项目。

摘　　要：基于重庆34个国家气象站1981—2016年5 min,24 h至3 d的过程最大雨量共18个年极值降水序列,利用线性矩法计算6种常用概率分布函数的模型参数,通过备选最优模型筛选法客观选取各站各历时极值降水的最优概率模型,并将优选结果应用于重庆不同历时百年重现期降水的计算。结果表明:模型优选方法得到的重庆各站不同历时极值降水的最优线型略有差异,广义极值分布占比最大,三参数Weibull分布次之,三参数对数正态分布第三,皮尔逊Ⅲ型和Gumbel分布相当,指数分布最差。最优线型计算的重庆不同历时百年重现期降水的空间分布大值区由短历时的点状分布向长历时的片状分布转变,渝东北的大值中心受地形影响不断向北移动。基于线性矩法的概率模型参数估计及客观的线型优选过程具有较强的可操作性和适应性,可应用于其他工程气象参数的推算中。Based on 18 durations of annual maximum rainfall series of 34 national stations in Chongqing from 1981 to 2016,the L-moments are used to estimate the parameters of six common probability distributions.The optimal line,ensured objectively by alternative optimal model method,is applied to calculate the 100-year return period extreme precipitation with different durations.The result is as follows:There exist different optimal distributions for different duration of extreme rainfall in each station.The ratio of the largest to the smallest proportion is the Generalized of Extreme Value distribution(GEV),three-parameter Weibull distribution(Weibull3),three-parameter lognormal distribution(LN3),Gumbel distribution and Pearson type Ⅲ distribution(P-Ⅲ).As time goes on,the relatively large precipitation area that calculated by the best model has changed from a point-like distribution with a short duration to a sheet-like distribution with a long duration.Under the influence of topography,the large center in northeast Chongqing is constantly moving northwards.It's operable and adaptable for reckoning on some other engineering meteorological parameters by selecting one or more optimal probability distribution model,whose parameters is estimated by L-moments,through the objective selection process.

关 键 词：多历时极值降水 概率模型优选 线性矩估计 备选最优模型筛选法 

分 类 号：P426.62[天文地球—大气科学及气象学]