ALBERT

Description: A numerical model based on the shallow-water equations is developed to represent the flow of Antarctic Bottom Water (AABW) in the Brazil Basin (southwest Atlantic Ocean). The aim is twofold. First, an attempt is made to identify in a model that includes both simplified dynamics and realistic bathymetry (at 1/6° resolution) the impacts of the elevated diapycnal mixing rates near the Mid-Atlantic Ridge (MAR) documented by dissipation data of the Deep Basin Experiment (DBE). To this end, different assumptions regarding the distribution of the velocity across the AABW layer interface (w) are considered. Second, the extent to which the shallow-water model can replicate observations relative to AABW circulation in the basin, in particular the trajectory and velocity of neutrally buoyant floats released in the AABW during the DBE, is examined. The model flows are characterized by small Rossby numbers, except in the northward-flowing western boundary current where kinetic energy is largely concentrated. To interpret the flows, model streamlines are compared with isopleths of linear potential vorticity f/h0 of the shallow-water theory (f is the planetary vorticity and h0 is the layer thickness in the absence of motion). The f/h0 contours are oriented northwest–southeast in the western part of the basin and southwest–northeast in the eastern part, reflecting the bowl-shaped topography of the Southern Hemisphere basin. With a spatially uniform (positive) w, the ubiquitous vortex stretching produces a flow to the southeast, consistent with the Stommel–Arons theory. This flow occurs in most of the basin interior, even in the east where f/h0 contours converge to the northeastern end of the basin. With strongly positive w near the ridge and zero or slightly negative w elsewhere, the flow follows more closely f/h0 contours in the western interior and intersects them near the ridge. The confinement of the diapycnal mass flux near the MAR drastically reduces the southward flow in the interior or even reverses its direction, leading to a circulation quite distinct from that of the Stommel– Arons theory. The model results compare favorably to some (but not all) hydrographic estimates of AABW circulation patterns and rates. On the other hand, the model streamlines and velocities show important differences with, respectively, the trajectory and the velocity of the floats launched in the AABW layer. The prescription of vanishing w in the interior does not systematically improve the fit of the model streamlines to the float trajectories, and the model velocities simulated with spatially uniform w or spatially variable w are on average smaller by one order of magnitude than the float velocities. A variety of mechanisms, which are not included in the numerical experiments, may explain the differences between the model results and the float data.