Circulation anomaly mechanisms in the tropical Atlantic sector during the northeast Brazil rainy season: Results from the GISS general circulation model

Abstract

The role of the sea surface temperature (SST) pattern in the tropical Atlantic on the overlying atmosphere is studied from a 7‐year run of the general circulation model (GCM) of the Goddard Institute for Space Studies (GISS) (period 1980–1986) and in comparison to upper air analyses of the European Centre for Medium‐Range Weather Forecasts (ECMWF), surface ship observations of the SST and wind fields (COADS), and rainfall measurements in Northeast Brazil (Nordeste). Observations for March to April show that reduced northward SST decrease in the tropical Atlantic is accompanied by reduced northward decrease of lower‐tropospheric thickness, which accounts for an inverse change in the meridional pressure gradient and reduced northerly wind component; this in turn entails a northward displaced Intertropical Convergence Zone and drought in the Nordeste. The GCM simulation reproduces the thermal and hydrostatic forcing of SST on the meridional gradients of lower‐tropospheric thickness and pressure in the proper sense, albeit more weakly than observed. The modeled meridional wind component follows the meridional pressure gradient in the proper sense, but mean and range are smaller than observed, and the modeled interannual variability is not altogether consistent with observational evidence. The modeled March to April precipitation in the Nordeste shows little relation to the observed rainfall, although it is weakly related with the observed interhemispheric SST contrast and with the modeled near‐surface meridional wind component. Both modeled and observed March to April Nordeste rainfall are negatively related with SST anomalies in the equatorial Pacific in March to April as well as in the immediately preceding January. Both in January and in March to April the modeled soil moisture exhibits a strong dependence on the modeled precipitation, but neither resembles in interannual variability nor in absolute amounts of the observed rainfall. Overall the GISS GCM seems to reproduce reasonably well the response of lower‐tropospheric thickness and near‐surface pressure to SST, while the simulation of wind and rainfall is somewhat deficient.