Modeling deep ocean convection: Large eddy simulation in comparison with laboratory experiments

Abstract

A large-eddy simulation model (LES) has been applied to study deep convective processes in a stratified ocean driven by the energetic cooling at the ocean surface. Closely related to a recent laboratory experiment, the numerical experiment deals with the inverted problem of the growth of a convective mixed layer driven by a localized source of bottom heating in a rotating, stably stratified fluid. In general, good agreement is found between numerical and laboratory results,. After onset of the heating a well-mixed layer forms above the heated circular surface. Although small-scale turbulence quantities like rms velocities and length scale can be best described by the nonrotating turbulent velocity and length scales, they are also found to differ significantly from a nonrotating control run, which indicates that rotation affects but does not control the turbulence. Due to the horizontal radial temperature gradient between the mixed layer and the ambient fluid a rim current develops around the periphery of the heated surface. Its near-surface maximum can be well described by a simple thermal wind law. The strong counterrotating current also observed in the laboratory at greater heights above the surface is found to be mainly driven by surface friction and should not be observed in the ocean. As time progresses, the rim current becomes unstable, eventually generating a field of baroclinic eddies that stop the mixed layer growth by causing some horizontal exchange between the convective layer and its cooler surrounding. The wavelength of the instabilities slowly increases with time and is clearly related to the local Rossby radius.