ENSO PREDICTION EXPERIMENTS USING A SIMPLE OCEAN-ATMOSPHERE MODEL

A coupled model consisting of a reduced gravity ocean and an empirical atmosphere is developed to investigate predictability of ENSO. The oc ean model has no seasonal cycle, and is forced only by zonal wind stre ss anomalies. The empirical atmosphere is constructed by using statist ical techniques to relate observed wind stress anomalies directly to t he sea surface temperature anomalies simulated by the model. The focus of this study is on detailed analysis of individual predictions, as w ell as on statistical scores from an ensemble of predictions. The mode l can in general successfully predict an event up to about one year in advance, but the model has little ability to predict low amplitude va riability that is not related to ENSO. Predictions based on POP analys is of the ocean model data are also made for several ENSO events, with skill comparable to that of the dynamical coupled model.