Reduced space optimal analyses of historical SST and SLP data

Abstract

We have developed a statistical technique for the objective analysis of observed historical climate fields, which contains comparatively precise data and good coverage in the last few decades, and poor observational coverage prior. This method combines the classical approach of least-squares optimal estimation with the novelty of space reduction and is specifically designed to reconstruct the recoverable portion of the signal from the observations. Applications of this method to reconstructions of near-global monthly fields of sea surface temperatures (SST) and sea level pressure (SLP) from 1850s to present along with verified error bars are publicly available: from http://ingrid.ldeo.columbia.edu/SOURCES/.KAPLAN/, click on RSA_MOHSST5 for SST analyses and RSA_COADS_SLP1 for SLP analyses.

The limitations of the technique concerning long-term and small-scale variability, assumption of stationarity of means and covariances and incompleteness of coverage and the ways to address them, as well as prospects of multivariate analyses (like SST and SLP togther, or SST and sea ice concentration) are going to be addressed.