中国、北半球和全球的气温突变分析及其趋势预测研究

本文采用均值差异假设检验研究了中国、北半球和全球气温历史序列的突变现象。分析表明，中国气温从本世纪以来，在４０年代末扣年代初曾出现一次由暖到冷的突变。北半球和全球均曾在１９世纪末和本世纪２０年代发生了突变现象。功率谱分析表明，气温的突变指数曲线具有明显的周期性。一系列比较研究证明，按照分析出的突变点将气温序列分段建模，无论数值误差还是变化趋势，效果均优于整段序列的模型。所以，对未来气温变化趋势作预测，应首先搞清楚未来会处在怎样的气候阶段中，会不会出现突变。研究表明，本文叙述的均生函数累加延拓的时序建模方案，对气温序列有很好的拟合和预测效果。

Detection of Abrupt Changes and Trend Prediction of the Air Temperature in China, the Northern Hemisphere and the Globe

Abrupt changes of yearly air temperature series in China, the Northern Hemisphere and the Globe are detected with a statistical test in this paper. It turns out that an abrupt change from a warm period to a cold one of the temperature in China occurred during the end of the 1940s and the beginning of 1950s, and tWo abrupt changes of the temperature in the Northern Hemisphere and the Globe also happened both in the 1890s and the 1920s. Meanwhile, it is found that the abrupt change indexes of the three temperature series have significant periodicity. The results of predictions for independent observations show that the models upon several periods, which are divided from the temperature series by the points of the abrupt changes, are far better than the model upon the whole temperature series, having examined the quantity, the change trend of the fitting and prediction. Therefore, when we predict the temperature.in the future the first priority is to detect the climate period, to which the future climate will be transfered, and the abrupt change, which will be occurred or not. Our work indicates the modeling scheme based on integrating extension series of the mean generating function is feasible for the simulation and prediction of the temperature series.