摘要
由极端降水事件所导致的灾害对自然和社会系统的影响十分严重。以岷江流域为研究区域,探索不同概率分布对该流域不同地区降水极值的适用性,并揭示其统计特征。基于流域内15个气象站1962—2012年逐日降水数据,分别采用年极大抽样和百分位法从中提取年极大降水序列(AM)和超门限峰值降水序列(POT)。选取广义极值分布(GEV)和广义帕雷托分布(GPD),采用L-矩参数估计法得到两种分布函数的参数。根据K-S检验结果评价拟合效果并确定最优分布,计算不同重现期水平下的降水量。结果表明:1)GEV分布拟合AM序列时,有5个站点属于极值Ⅱ型分布,其余属于Weibull分布;拟合POT序列时,所有站点均属于极值Ⅱ型分布。2)AM序列的最优分布全为GEV分布;POT序列有13个站的最优分布为GPD分布,仅有2个站为GEV分布;GPD分布不适宜拟合雅安站的AM序列。3)流域降水极大值集中在岷江中下游地区,上游降水相对较少。4)在雨量较少的地区,两种分布推算的不同重现期降水量相差不超过6 mm;对于多雨地区,当重现期为50、100 a时,这种差值大于20mm。通过分析,认为两种分布皆可用于岷江流域降水极值的拟合,GEV分布更适用于拟合AM序列,GPD分布更适用于POT序列,但两种分布拟合各自最适宜的样本序列时难以比较。AM序列和POT序列最优分布推算的不同重现水平下的降水量存在差异,这种差异随重现期增大而增大,同时也与站点降水量相关。
Disasters caused by the extreme precipitation events have serious impacts on natural and social system. In this study, the statistical characteristics of the probability distributions of the precipitation extremes in different areas of the Minjiang River basin were explored and their applicability was discussed. Based on the annual maximum （AM） and peak over threshold （POT） of extreme precipi- tation series from 15 sites in the Minjiang River basin during 1962-2012, the generalized extreme value distribution and generalized Pareto distribution were used to investigate statistic characteristics. The parameters of GEV and GPD distribution functions were esti- mated by the L-moments method, and the precipitation extremes with different return period were calculated by the robust distribution which was determined with K-S test. The results showed that there were 5 sites that belong to extreme-Ⅱ distribution and the rest sites to Weibull distribution when the AM series were fitted by GEV distribution. All the sites belonged to extreme- Ⅱ distribution when the POT series were fitted by GEV distribution. Furthermore,the optimal distributions of 13 sites were GPD distribution for POT series and all of the robust distributions were GEV distribution for AM series. The maximal precipitation was concentrated in the middle and lower reaches of Minjiang river,while the upstream precipitation was relatively small. Finally,the difference of precipitation extremes with different return period did not exceed 6 mm in regions with little rainfall. In rainy regions,the difference was more than 20 mm when the return period was 50 or 100 years. It was believed that these two kinds of distributions are applicable to Minjiang River basin but GEV distribution is more suitable for AM series and GPD distribution is more suitable for POT series. The predpitation extremes of different return periods were different when calculated by the robust distribution of different series. The difference along with the return period was increased but associated with the precipitation of the sites.
出处
《工程科学与技术》
EI
CAS
CSCD
北大核心
2017年第1期78-85,共8页
Advanced Engineering Sciences
基金
国家自然科学基金资助项目(5150090103
41301021)
关键词
岷江流域
降水极值
重现期
概率分布
L-矩
Minjiang basin
precipitation extremes
return period
probability distribution
L-moments
