| description abstract | Although observations indicate that localization of deep convection results from oceanic preconditioning and only to a lesser extent from the gradients in the atmospheric forcing, most laboratory and numerical simulations of oceanic deep convection localize convection via the buoyancy forcing. It is as yet unclear to what extent the localized forcing simplification is representative of realistic preconditioned convection and whether conclusions drawn from analyzing the former scenario can be applied to the latter. Comparison between these two localized convection scenarios is the focus of this study. The analysis is conducted using a high-resolution, nonhydrostatic numerical model that assumes no variations in the latitudinal direction. Although this model cannot represent baroclinic instability, it represents well the violent mixing phase of chimney formation that precedes baroclinic instability. Because of the inherently different initial and boundary conditions, there is no single comparison to be made between the two scenarios; all of the comparisons, however, show that a localized forcing tends to enhance the frontal structures at the edge of the convected water mass. The authors explain these results by means of some simple analytical calculations that indicate that convection as a result of a localized forcing induces a continuous increase in the horizontal density gradient, while convection in a surface-intensified, cyclonic gyre results in a decrease in the horizontal gradient. The authors conclude that the two scenarios are not equivalent and discuss how some of the conclusions, valid for localized forcing, do not necessarily apply to the preconditioned scenario. In particular, the equilibrium state observed in localized forcing simulations, in which the lateral fluxes due to the baroclinic eddies balance the surface buoyancy loss, may not have an analog for the case of preconditioned convection. | |
