气候突变的聚类分析

CLUSTER ANALYSIS OF CLIMATE JUMP

气候数据是时间序列，是一类有序样品．本文给出了一种新的有序样品谱系聚类算法，先对数据进行平滑拟合，然后用聚类分析的方法分析了北半球的气候资料，证实了1920年为气候突变点．同时也找到了较低的各个层次上的突变点．

The Climate data is a time series which can be considered as a series of ordered samples. In this paper, a new method-hierarchy cluster algorithm of ordered samples is presented. First,the samples are processed by smoothing fit.Then the cluster analysis has been used in the analysis of climate data. It is shown that the 1920th year is the jump point. Moreover, jump points of lower each hierarchy have been discovered.