Prediction of Time-Mean Atmospheric Circulation and Rainfall: influence of Pacific Sea Surface Temperature Anomaly

Abstract

A global general circulation model was integrated for 60 days with the observed initial conditions at 0000 UTC on 15, 16 and 17 December 1982 and climatological boundary conditions of sea surface temperature (SST), soil moisture, snow cover, sea ice and albedo. These integrations were repeated after the observed SST anomalies over the Pacific Ocean during the winter of 1982?83 were added to the climatological SST. Model forecasts for 10-, 30- and 60-day averages of circulation and rainfall were examined. A comparison of the model-predicted, time-averaged circulation and rainfall anomaly with the corresponding observations show that the inclusion of the sea surface temperature (SST) anomalies over the Pacific Ocean produces spectacular improvements in the forecasts of tropical circulation and rainfall. There is also clear improvement in the forecasts of extratropical circulation, especially after the model drift is removed from the forecasts. The model drift was calculated as the moan difference between the model forecasts with climatological boundary conditions for the five different and independent initial conditions (at 0000 UTC on 1 January 1979, 1980, 1981, 1982 and 1984) and the corresponding verifying observations. Due to an insufficient number of forecasts for defining the model drift, only a zonally symmetric model drift was removed from the forecasts. This study, like several other similar studies, clearly demonstrates that if SST anomalies could be predicted in advance, it would be possible to make substantial improvements in long range prediction over the tropics and also, to a limited extent, over the midlatitudes.