Reconstruction of near-global annual precipitation using correlations with sea surface temperature and sea level pressure

Abstract

An indirect precipitation analysis method is described which allows analysis of large spatial-scale and multidecadal variations over land and oceans beginning 1900. The method uses covariance between precipitation and analyses of sea level pressure (SLP) and sea surface temperature (SST). Both SLP and SST analyses can be produced using in situ data for the 20th century. Here a canonical correlation analysis is developed to specify annual precipitation anomalies from annual anomalies of SLP and SST on a 5° spatial grid. Covariance relationships are computed using 26 years of satellite-based precipitation data beginning 1979 and are used to analyze annual average precipitation anomalies for the full period. This indirect analysis indicates global variations consistent with the satellite-based analysis for the recent period. Cross-validation testing shows most skill in the tropics where variations are largest, with decreasing skill at higher latitudes, and large-scale averages have much more skill than at individual locations. For the full period over oceans the analysis indicates increasing precipitation with increasing temperature over the 20th century. That oceanic change is correlated with the change from climate models, but the analysis change is more than twice as strong as the change indicated by the models. Over land the analysis is consistent with gauge observations over the 20th century, which are independent observations before 1979. This study shows that indirect precipitation analyses can show many climate-scale variations that cannot be resolved in studies based on direct analysis of precipitation data. Copyright 2009 by the American Geophysical Union.