Eddies and the general circulation of numerical model gyres: An energetic perspective

Abstract

Energy budgets for the long‐time‐averaged circulation of mesoscale resolution numerical ocean circulation model (EGCM) experiments are the only widely used analysis methods that permit investigation into the eddy‐mean flow interaction of these model oceanic systems. All published energy results from the closed basin EGCM experiments are presented here in nondimensional form to facilitate comparison of the various types of behavior. The ubiquitous basin‐averaged energy analysis method is necessarily sensitive to changes in model forcing and dissipative processes, and it is shown that several natural energy measures of eddy‐mean flow interaction are strongly correlated with forcing and dissipation choices. In particular, it is found that by these energy measures, eddies are relatively unimportant in the mean flow when the horizontal momentum subgrid scale process corresponds to constant coefficient viscosity, but they are much more important when this process is modeled by a constant coefficient biharmonic operator. Energy budgets over subrogions of the full domain should offer increased insight into model mechanisms, but the use of quasi‐geostrophic dynamics can severely limit the number of independent energy pathways in the flow. Owing to these constraints the published regional two‐layer quasi‐geostrophic energy budgets offer information beyond that of the basin budgets. It is suggested that the assumptions of these quasi‐geostrophic models also limit the generality of inferences about the role of vertical processes based on these calculations. In none of the energy budgets reported is there significant direct driving of a mean flow by Reynolds stress work, and in only one case does mean buoyancy work act to try to increase mean flow kinetic energy. From these budgets it appears that if eddies are energetically important in the model flows, their only unambiguous role is as a pathway to dissipation. Further more detailed regional analysis is necessary to see if this result continues to hold when smaller horizontal subregions of the domain are analyzed. Such analysis activity, as well as more attention to alternative means of investigating eddy‐mean flow interaction, appears to be needed.