论文部分内容阅读
以太湖流域为例,选取26个代表站点,基于BMM和POT法选取日极大值降水序列,分别用GEV、Gumbel分布和GPD分布进行拟合,研究各站点极值分布拟合情况及空间区域各站点极值分布的选择情况,并综合对比分析了BMM和POT法拟合优劣以及相同重现期下空间极值分布的异同。结果表明,基于BMM法的各站点极值分布拟合情况不同,但AIC、BIC准则保持一致,空间分布上东南大部分区域适合用GEV分布来拟合降水极值,西北大部分区域适合用Gumbel分布拟合;基于POT法的各站点拟合情况各异,但普遍较BMM法的分布拟合要好,各站点阈值选择在70~90 mm之间;两类方法的重现期极值空间分布基本一致,POT法较BMM法可避免干旱年份对降水极值分布的影响。
Taking Taihu Lake Basin as an example, 26 representative sites were selected, and daily maximum precipitation sequences were selected based on BMM and POT methods. The distributions of GEV, Gumbel and GPD were fitted respectively to study the fitting of extreme value distribution and spatial region The selection of the extreme value distribution of each site, and a comprehensive comparison and analysis of the similarities and differences between the fitting advantages and disadvantages of BMM and POT method and the extreme value distribution of space under the same recurrence period. The results show that the fitting of extreme value distribution is different for each site based on BMM method, but the AIC and BIC criterion are consistent. Most of the southeast area is suitable for fitting the precipitation extreme value in most of the southeast area, and Gumbel The distribution fitting of each site based on the POT method is different, but the distribution fitting is generally better than the BMM method, and the threshold selection of each site is between 70 and 90 mm. The extreme spatial distribution Basically, the POT method can avoid the influence of drought year on the distribution of precipitation extreme value compared with BMM method.