EXPERIMENTAL PREDICATIONS OF CLIMATIC VARIABILITY FOR LEAD TIMES OF 12 MONTHS

Interannual variability of present-day climate has been clearly linked to the occurrence of large-scale sea-surface temperature anomalies, p articularly in the Pacific Ocean. Prediction of such anomalies many se asons in advance by using relatively simple, coupled oceanic-atmospher ic models provides a basis for predicting the atmospheric climate over the same period. This is achieved by inserting the sea-surface temper ature anomalies predicted by the coupled model at appropriate time int ervals into a global climatic model and integrating forward in time. P rediction experiments for 1991 and 1992 have been conducted on this ba sis using sea-surface temperature anomalies predicted for the Pacific Ocean' Global and regional rainfall outcomes are presented here, altho ugh the predictions are not expected to be valid for the whole globe. Detailed results are given for Australia and southern Africa, which ex perienced severe drought conditions in 1991 and 1992. Useful skill, wi th statistical significance for predictions when substantial rainfall anomalies occurred, was obtained for these regions, indicating the pot ential utility of the method. Considerable scope exists for improvemen ts in the technique presented here, particularly in the global climati c model used, so that increasing accuracy should result as the techniq ue is developed. Operational implementation of this method is essentia lly straightforward.