A hybrid coupled model (HCM) is used to explore the underlying dynamics governing tropical Atlantic variability (TAV) and the dynamic regime that may be most relevant to TAV. By coupling an empirical atmospheric feedback model to an ocean GCM, the authors have conducted a detailed investigation on the potential importance of an unstable ocean-atmosphere interaction between wind-induced heat flux and sea surface temperature (SST) in driving decadal climate variability in the tropical Atlantic basin. The investigation consists of a systematic parameter sensitivity study of the hybrid coupled model. It is shown that in a strong coupling regime the local air-sea feedbacks can support a self-sustained decadal oscillation that exhibits strong cross-equatorial SST gradient and meridional wind variability. An upper-ocean heat budget analysis suggests that the oscillation results from an imbalance between the positive and negative feedbacks in the model. The dominant negative feedback that counteracts the positive feedback between surface heat flux and SST appears to be the advection of heat by ocean currents. The major imbalance in the model occurs in the north tropical Atlantic between 5 degrees and 15 degreesN, caused by a phase delay between the surface heat flux forcing and horizontal heat advection. It is suggested that this may be one of the crucial regions of ocean-atmosphere interactions for TAV. Based on the HCM results, a simple 1D model is derived to further elucidate key coupled dynamics. The model assumes that air-sea coupling takes place in a limited area within the deep Tropics of the Atlantic sector and the change of upper-ocean heat transport is regulated by the advection of anomalous temperatures by the mean meridional current and equatorial upwelling. The analysis shows that the simple model captures many of the salient features of the decadal SST cycle in the HCM, suggesting that the decadal oscillations simulated by the HCM are primarily controlled by the coupled dynamics local to the deep Tropics. The parameter sensitivity study further suggests that in reality the local air-sea coupling in the tropical Atlantic is most likely to be too weak to maintain a self-sustained oscillation, and stochastic forcing may be necessary to excite the coupled variability. Using a realistic representation of external "noise" derived from a 145-yr simulation of the National Center for Atmospheric Research atmospheric GCM (CCM3) forced with the observed SST annual cycle, the effect of stochastic forcing on TAV when the coupled system resides in a stable dynamical regime is examined. It is found that the local air-sea feedback and the North Atlantic oscillation-(NAO) dominated "noise" forcing are both required to simulate a realistic TAV. In the absence of the local air-sea feedback, the "noise" forcing can produce substantial SST anomalies in the subtropical Atlantic up to about 15 degreesN, particularly off the coast of North Africa. The local air-sea feedback appears to be particularly important for generating the covarying pattern of interhemipheric SST gradient and cross-equatorial atmospheric flow within the deep Tropics. However, too-strong local coupling can lead to an exaggerated tropical response. It is therefore conjectured that TAV may best fit into a weakly coupled scenario in which at minimum the air-sea feedback plays a role in enhancing the persistence of the cross-equatorial gradient of SST and the circulation anomalies, while the NAO provides an important source of external forcing to excite the coupled variability in the Tropics. Furthermore, it is argued that the "noise" forcing can significantly weaken the correlation between the SST variability on either side of the equator, thus hiding any underlying weak "dipole" structure in the SST.
