Tropical Atlantic biases and their relation to surface wind stress and terrestrial precipitation

Most coupled general circulation models (GCMs) perform poorly in the tropical Atlantic in terms of climatological seasonal cycle and interannual variability. The reasons for this poor performance are investigated in a suite of sensitivity experiments with the Geophysical Fluid Dynamics Laboratory (GFDL) coupled GCM. The experiments show that a significant portion of the equatorial SST biases in the model is due to weaker than observed equatorial easterlies during boreal spring. Due to these weak easterlies, the tilt of the equatorial thermocline is reduced, with shoaling in the west and deepening in the east. The erroneously deep thermocline in the east prevents cold tongue formation in the following season despite vigorous upwelling, thus inhibiting the Bjerknes feedback. It is further shown that the surface wind errors are due, in part, to deficient precipitation over equatorial South America and excessive precipitation over equatorial Africa, which already exist in the uncoupled atmospheric GCM. Additional tests indicate that the precipitation biases are highly sensitive to land surface conditions such as albedo and soil moisture. This suggests that improving the representation of land surface processes in GCMs offers a way of improving their performance in the tropical Atlantic. The weaker than observed equatorial easterlies also contribute remotely, via equatorial and coastal Kelvin waves, to the severe warm SST biases along the southwest African coast. However, the strength of the subtropical anticyclone and along-shore winds also play an important role.