Are there optimum sites for global paleotemperature reconstruction?

Abstract

In order to place the relatively short instrumental record of global mean annual temperature in perspective, long proxy records are needed. Since it is unlikely there will ever be a globally extensive network of well-calibrated proxy records, it is necessary to identify sites which capture most of the variance of the larger scale average, to be used as representative time series for global means.

This problem is approached by examining different spatially extensive data sets (instrumental and GCM-derived) to determine the pattern of spatial correlation with the larger-scale mean series over different time-averages. A 1,000 year simulation (from the GFDL coupled atmosphere-ocean GCM) is used to examine the usefulness of various proxy indicator sites as predictors of global mean annual temperature series, for both inter-annual, and inter-decadal variations. Realistic networks of sites are used, representing locations where paleoclimatic records from corals, high latitude tree-rings, polar ice cores, mid and low latitude ice cores and mid-latitude historical and varved sediments have been obtained. From this limited set of sites, optimum networks are identified for both inter-annual and inter-decadal reconstructions. A combination of low latitude oceanic sites and interior continental locations seems to provide the best network for the reconstruction of global mean annual temperature.