中国格点化日降水极值统计模型及阈值的选取

采用年最大值法（AM）及超阈值峰量法（POT）分别构建基于0.5°×0.5°网格的全国地面日降水极值序列，建立基于广义极值分布（GEV）和广义帕累托分布（GPD）的降水极值统计模型，通过K-S检验评估模型拟合效果，研究全国日降水极值的统计规律及其空间分布特征，提出适用于不同地区极端日降水的极值分布模型与阈值选取标准，结果表明：(1)POT序列比AM序列更符合降水极值序列的要求；(2)为便于比较并提高模型拟合效果，POT序列的阈值由百分位数法确定效果较好；(3)阈值方案优选结果在空间分布上与中国干湿区域的划分有很好的相关性，在湿润地区宜将第90～94百分位数作为阈值，在半湿润和半干旱地区宜将第94～97百分位数作为阈值，在干旱地区则使用第97～99百分位数较为合适。

Statistical Model and Threshold Value Selection of Gridded Daily Precipitation Extremes in China

Based on the national daily precipitation 0.5°× 0.5° gridded dataset, annual maximum (AM) samples and peaks over threshold (POT) samples were selected. The generalized extreme value distribution (GEV) and the generalized Pareto distribution (GPD) were employed to establish statistical models of precipitation extremes respectively. The goodness of fit of each model was evaluated by Kolmogorov-Smirnov test. The statistical analysis was performed. Extreme value distribution model of precipitation and threshold value selection criteria applicable to different areas were proposed. The results show that: (1) The simulated results of POT samples are superior to those of AM samples; (2) The method of sample percentile for determining threshold value is better than the others; (3) The geographical distribution pattern of optimization results is similar to the distribution of dry and wet regions in China. The 90?94 percentile is the fittest to determine threshold value in humid regions. The 94?97 percentile is better in semi-arid and sub-humid regions. The 97?99 percentile is the most suitable in arid regions.