Seasonal climate forecasts over tropical Africa and South America are hampered by the lack of skillful predictions of sea surface temperatures (SST) in the tropical Atlantic. Thermodynamic interactions between the atmosphere and the surface mixed-layer of the ocean are believed to play an important role over the tropical and subtropical Atlantic, and may yield a source of seasonal predictability.

This mechanism is investigated using the latest version of the UCLA atmospheric General Circulation Model, which includes a state of the art stratus parametrization, coupled to a simple, uniformly 50m deep mixed layer in the Atlantic basin. A monthly climatology of SST is prescribed outside of the basin and poleward of 50N or 50S in the Atlantic. Land-surface parameters are hold at their climatological values. Over 20 years of simulation are analyzed to identify the modes of variability arising from the air-sea interactions internal to Atlantic, and the results are compared with a control experiment which uses prescribed climatological SST everywhere.

The coupling is found to give rise to a 'dipolar' leading mode of rainfall variability straddling the equator, not present in the control run. This structure exhibits a red spectrum and, as expected, is associated with SST anomalies on either side of the equator. Further investigation of the SST and coupled SST-SLP variability shows a well-defined structure in the tropical South Atlantic, which resembles the equatorial zonal mode, despite the lack of ocean dynamics. The associated SLP pattern is similar to the leading mode in the control run but with a highly reddened spectrum and an enhanced quasi-biennial component. The latter has also been found in the observations. Local air-sea interactions in the tropical and subtropical South Atlantic are found to play an important role in the quasi-biennial component, and these will be discussed in detail in the presentation.