Vorticity Balance in Coarse-Resolution Global Ocean Simulations

Abstract

The vorticity budget of the vertically integrated circulation from two global ocean simulations is analyzed using a horizontal spacing of 2° ? 2° in longitude/latitude. The two simulations differ in their initial hydrographic conditions and surface wind and buoyancy forcing. The constrained simulation obtains optimal initial condition and surface forcing through assimilating observational data using the model's adjoint, whereas the unconstrained simulation uses Levitus climatological conditions for initialization and is driven by NCEP?NCAR reanalysis forcing, plus restoring to the monthly surface temperature and salinity climatological conditions. The goal is to examine the dynamics that sets the time-mean circulation and to understand the differences between the constrained and unconstrained simulations. It is found that, similar to eddy-permitting simulations, the bottom pressure torque (BPT) in coarse-resolution models plays an important role in the western boundary currents and in the Southern Ocean, and largely balances the difference between wind stress curl and ?V for the depth-integrated flow. BPT is a controlling factor of the interior abyssal flow. The geostrophic vorticity relation holds in the interior basins in intermediate and deep layers and breaks down in the upper ocean toward the surface. In the upper layer of the interior basins, the model simulations show statistically significant deviation from the Sverdrup balance. In the subtropical gyre regions, the deviation from Sverdrup balance is confined to zonal bands that are balanced by the curls of lateral friction and nonlinear advection. The differences between the constrained and unconstrained simulations are significant in vorticity terms. The adjustment to Levitus hydrographic climatological data as the model's initial condition causes the most significant changes in BPT, which is the main reason for changes in abyssal flow. The analysis also points to needs for further improvement of models and controlling the influence of data errors in ocean state estimation.