STUDYING THE POSSIBILITY OF PALEOCLIMATE RECONSTRUCTION WITH MATHEMATICAL STATISTICS METHODS

E. V. Dmitriev and A. I. Chavro

The verification of general circulation models requires reliable information about climate change over the past 1000 years. However, sufficiently long instrumental observational data sets do not exist so far. Therefore, it is an important problem to reconstruct the low-frequency climate variability from different proxy indicators. Recently, one of the most frequently used methods for the interpretation of proxy data has been the relatively new method suggested by Michael Mann. The motivation for using this method is the claim that it is more robust against factors such as non-climatic variability in the proxy data [16]. In this paper, we describe a comparison of this method with the methods based on multivariate linear regression, which were used earlier mainly to reconstruct low-frequency variations of global geophysical parameters from proxy data. The comparison is in agreement with the conclusions of theoretical analysis of the Mann method [1].