中文摘要：

选取河西地区13个气象站点1960年-2011年逐年日降水资料，计算河西地区极端降水阂值，分析了河西地区极端降水的时空变化特征，并对未来变化特征进行了预测。结果表明：①各量级的降水日变化不尽相同，其中以小雪日的增加和微雨日的减少最为显著；②各极端降水指数均呈增加趋势，在20世纪60年代增加幅度大；③在0．05的置信度下，极端降水指数在1966年发生突变现象，突变之后极端降水指数向增多趋势转变；④极端降水指数在不同的时间序列存在长短不同的周期振荡，同时在不同时间尺度上所反映的极端降水指数的偏多和偏少结构不一样；⑤在空间分布上，极端降水指数的变化趋势存在较大的区域性差异，由河西西北向东南呈增加趋势，表明极端降水的分布是与所处的气候带有关的，季风可能是引起其变化的主要原因；⑥Hurst指数表明未来降水日数与过去趋势一致，但以小雪日（增加）和微雨日（减少）的持续强度最显著；极端降水指数中，R95极端降水量和最大5天连续降水量未来变化趋势和过去一致，但持续强度不大，R99极端降水量和最大1天降水量未来变化趋势不明显。

英文摘要：

Located in semiarid northwestern China, and far from any ocean, Hexi belongs to the typical continental climate zone. All kinds of weather conditions can be frequently experienced there, especially extreme precipitation. Socio-economic development, human health and the natural environment have become vulnerable to extreme precipitation events in Hexi. Based on daily precipitation data from 1960 to 2009 at 13 meteorological stations across Hexi, regression analysis, inverse distance weighted, Mann-Kendall, sliding average, correlation analysis, Morlet wavelet analysis and R/S analysis were employed to analyze spatial and temporal variability, and future patterns, of rainy days and extreme precipitation events. We found that drizzle days have decreased and light rain days have increased. Very wet day precipitation, extremely wet day precipitation, annual maximum with 1-day and 5-day precipitation all increased at rates of 3.12, 1.39, 0.505 and 0.868, respectively. Annual total precipitation and precipitation extremes are correlated. Extreme precipitation indices show abrupt change at the 0.05 level. Mutations of extreme precipitation events took place in 1966, and changes increased in tendency after 1966. Different vibration periods were displayed very clearly in extreme precipitation indices and reflects different structures which are more or less in different time scales. Differences in spatial distribution still exist, especially in the southeast where the range of change is largest; the monsoon was the major reason for this change. The Hurst index shows that future light snow days （increasing） and future drizzle days （decreasing） are an obvious trend; future extreme precipitation events showed no obvious future trend.