Dissipation of midocean mesoscale eddies: Inferences from satellite altimetry data, in-situ data, and idealized models

The energy cycles of midocean mesoscale eddies are examined by comparing numerical simulations of forced-dissipated two-layer geostrophic turbulence with satellite altimetry data and in-situ data. Spectral fluxes computed from altimetry data reveal that at larger horizontal scales an inverse cascade of kinetic energy towards larger horizontal scales takes place, while at smaller scales, a forward cascade towards smaller scales takes place. The existence of upscale and downscale cascades suggests that frictions acting on large scales, and frictions acting on small scales, are both felt by mesoscale eddies. The forward cascade further suggests that at least some of the thermocline dissipation seen in microstructure data is due to geostrophic eddies. Realistic vertical structures of eddy kinetic energy can be attained in the model when moderately strong large-scale friction acts on the bottom layer only. This is true whether the bottom drag is quadratic or linear in the flow. Scaling arguments similar to those done in tidal studies suggest that quadratic drag associated with bottom boundary layers may be insufficient to dissipate the 1 TW of power put into the general circulation and eddy fields by winds. Thus, as with tides, the main source of bottom friction may be the breaking of internal waves generated over rough topography. When our turbulence models are run at high resolution and only a spectral filter is used for small-scale dissipation, very little energy dissipation takes place in the upper layer, and the forward cascades at small horizontal scales are much smaller than those in altimetry data. Inclusion of an eddy viscosity of order 50 m2 s-1 in the model generates the forward cascades at small scales, as seen in satellite observations, and energy dissipation in the upper layer, consistent with in-situ microstructure observations. Eddy viscosities of this order are in agreement with values inferred from in-situ current meter data at the POLYMODE Local Dynamics Experiment (see Polzin, this session), and from a nearly global analysis of altimetric data (see Scott, this session). The viscosities may represent interactions between mesoscale eddies and internal waves. Our results have implications for the ocean energy budget, for the strength of the meridional overturning circulation, which is sensitive to the spatial distribution of dissipation, and for ocean general circulation models, which often close their energy budgets with eddy viscosities much larger than those observed, rather than realistically small eddy viscosities coupled with a physically motivated topographic wave drag.