Coupled Variability and Predictability in a Stochastic Climate Model of the Tropical Atlantic

Abstract

The coupled variability and predictability of the tropical Atlantic ocean?atmosphere system were analyzed within the framework of a linear stochastic climate model. Despite the existence of a meridional dipole as the leading mode, tropical Atlantic variability (TAV) is dominated by equatorial features and the subtropical variability is largely uncorrelated between the northern and southern Atlantic. This suggests that atmospheric stochastic forcing plays a dominant role in defining the spatial patterns of TAV, whereas the active air?sea feedbacks mainly enhance variability at interannual and decadal time scales, causing the spectra distinctive from the red spectrum. Under the stochastic forcing, the useful predictive skill for sea surface temperature measured by normalized error variance is limited to 2 months on average, which is 1 month longer than the predictive skill of damped persistence, indicating that the contribution of ocean dynamics and air?sea feedbacks is moderate in the tropical Atlantic. To achieve maximum predictability, processes such as ocean dynamics, thermodynamical and dynamical air?sea feedbacks, and the delicate mode?mode interactions should be correctly resolved in the coupled models. Therefore, predicting TAV poses more challenge than predicting El Niño in the tropical Pacific.