中文摘要：

基于长江流域147个气象站1960-2005年最大值降水序列（AM）与超门限峰值降水序列（POT），选取4大类20种分布函数，采用极大似然法和线性矩法估算了参数，经柯尔莫洛夫-斯米尔诺夫检验，确定了降水极值的最优概率模型。对AM与POT两套极端强降水序列的频率分析均表明，Wakeby分布函数能够较好的拟合长江流域降水极值的概率分布。同时指出了降水极值的拟合存在的不确定性。

英文摘要：

20 distribution functions which come from 4 categories for the AM and POT extreme precipitation series at 147 stations over the Yangtze River Basin, have been applied to obtain the robust probability model which fits for each station by KS goodness of fit test. Research results show that Wakeby distribution can simulate the probability distribution of precipitation extremes calculated both from AM and POT series quite well. But there are still some uncertainties for the estimated design value of extreme precipitation.