利用1981-2010年雷暴观测资料， 采用EOF分解和主值函数分析等方法， 对中国年平均雷暴日的时空分布特征、 年际变化及异常变化特征进行了分析。结果表明, 年平均雷暴日的地理分布可大致分为4个区域， 分别为东南高发区、 西南高发区、 东北次高区和西北低发区。年平均雷暴日的时间分布表现为夏季多、 冬季少； 一天之中雷暴出现的时间集中在下午到晚上。雷暴日数的年际变化呈现出1980年代和2000年代2个相对多发期和1990年代相对少发期， 其中2000年代雷暴事件的相对多发表现为下午和夜间雷暴事件的增加。近30年中国北方地区的雷暴整体呈现出减少趋势， 而南方则是先减后增， 其距平场的年代际变化较为明显。EOF分解后距平场第一向量的方差贡献达到32.4%， 在空间上表现为南北相异型， 其余各向量也不同程度地表现出区域性异常， 前12个向量累积方差贡献达到80.9%。此外， ENSO事件对中国雷暴尤其是南方地区影响较为明显， 两广和云贵地区呈现出相反的距平变化。

The features of spatial and temporal distribution, annual variability and abnormal changes of the thunderstorm are made with EOF and main function analysis  using the thunder observation data from 1981 to 2010. The results show that the distribution of annual thunderstorm day in China is divided into 4 sub-regions. The highest one is in southeast of China, the other highest in southwest of China, the higher in northeast and north China, the lowest in northwest of China. The number of annual thunderstorm shows much less in winter than in summer, meanwhile thunderstorm appears more in the time from the afternoon to evening. The annual thunderstorm days show two periods of high value in 1980s and 2000s and a low period in 1990s and high value of the period in 2000s is because of the increases of thunderstorm events in the afternoon and evening. In the past 30 years, the thunderstorm in northern China shows an overall reduction while changed from decrease into increase in southern China, the interdecadal variation of anomaly field is more obvious. The variance contribution of the first principal component accounts for about 32.4% and the spatial characteristic appears the different-shaped between north and south. The rest vector fields also show abnormal regional changes in different levels. The percentage of the first 12 principal component's cumulative variance contribution is 80.9%. In addition, the ENSO events have certain impact on thunderstorms in China especially in the southern region of China, the manifestation between Guangdong, Guangxi and Yunnan, Guizhou show the opposite anomaly performance.