Air–sea interactions in the tropical Atlantic: a view based on lagged rotated maximum covariance analysis.

Abstract

The dominant air–sea feedbacks that are at play in the tropical Atlantic are revisited, using the 1958–2002 NCEP reanalysis. To separate between different modes of variability and distinguish between cause and effect, a lagged rotated maximum covariance analysis (MCA) of monthly sea surface temperature (SST), wind, and surface heat flux anomalies is performed. The dominant mode is the ENSO-like zonal equatorial SST mode, which has its maximum amplitude in boreal summer and is a strongly coupled ocean–atmosphere mode sustained by a positive feedback between wind and SST. The turbulent heat flux feedback is negative, except west of 25°W where it is positive, but countered by a negative radiative feedback associated with the meridional displacement of the ITCZ. As the maximum covariance patterns change little between lead and lag conditions, the in-phase covariability between SST and the atmosphere can be used to infer the atmospheric response to the SST anomaly. The second climate mode involves an SST anomaly in the tropical North Atlantic, which is primarily generated by the surface heat flux and, in boreal winter, wind changes off the coast of Africa. After it has been generated, the SST anomaly is sustained in the deep Tropics by the positive wind–evaporation–SST feedback linked to the wind response to the SST. However, north of about 10°N where the SST anomaly is largest, the wind response is weak and the heat flux feedback is negative, thus damping the SST anomaly. As the in-phase maximum covariance patterns primarily reflect the atmospheric forcing of the SST, simultaneous correlations cannot be used to describe the atmospheric response to the SST anomaly, except in the deep Tropics. Using instead the maximum covariance patterns when SST leads the atmosphere reconciles the results of recent atmospheric general circulation model experiments with the observations.