非定长相关时间序列集内平均序列的摆动异常及其解释
ON THE ABNORMAL SWING OF THE AVERAGED TIME SERIES PRODUCED FROM A NUMBER OF CORRELATED TIME SERIES
-
摘要: 文中对多个时间序列的平均序列中常常出现的异常摆动问题进行了探讨.结果表明,给定时间序列集中序列个数的多少会造成其所组成的平均序列的方差的变化,即序列个数越多,平均序列方差越小的可能性就越大,由此定义了方差增长率α.当给定方差增长率限定值α0并使其满足α≤α0时,可以得到最小序列个数m0,且α0越大或序列间平均相关系数越大,所需要的时间序列数越少.这一结果可应用于气候变化研究中有关平均时间序列建立的研究中,亦可运用到其他相关研究领域.Abstract: In many climatological researches,the averaged time series can be produced from a number of time series to enhance common signals.The abnormal swing always exists in the averaged series,so that the changes of the former part of the series,with less samples,are larger than those of the latter part,with more samples.This phenomenon could be found either in tree ring chronologies produced from tree ring series with different lengths or in the averaged time series of some climatic parameters produced from records of meteorological station network.
下载: