AN IMPROVED PROCEDURE FOR EL-NINO FORECASTING - IMPLICATIONS FOR PREDICTABILITY

A coupled ocean-atmosphere data assimilation procedure yields improved forecasts of El Nino for the 1980s compared with previous forecasting procedures. As in earlier forecasts with the same model, no oceanic d ata were used, and only wind information was assimilated. The improvem ent is attributed to the explicit consideration of air-sea interaction in the initialization. These results suggest that El Nino is more pre dictable than previously estimated, but that predictability may vary o n decadal or longer time scales. This procedure also eliminates the we ll-known spring barrier to El Nino prediction, which implies that it m ay not be intrinsic to the real climate system.