Statistical prediction of sea-surface temperature over the tropical Atlantic

摘要

A statistical system to predict sea-surface temperature anomalies (SSTAs) over the tropical oceans, with emphasis on the tropical Atlantic, is described. Canonical correlation analysis is used to identify critical sequences of predictor patterns, which tend to evolve into subsequent patterns and which can be used to form a forecast. The results indicate that SST fields over the equatorial Pacific and tropical Atlantic can be a potential predictor of the SSTAs over the tropical Atlantic 3-4 months in advance. The spatial structures of the SSTAs over the tropical Atlantic for the period March-May are well captured by the predictions done with initial conditions from September to February. Model performance is better over the northern tropical Atlantic than over the southern tropical Atlantic, where persistence is hardly beaten. Results of this work can contribute to improve seasonal climate predictions of rainfall anomalies over the northeast Brazil region.