Eddy forcing of the mean circulation in the western North Atlantic

Abstract

This thesis addresses several aspects of the problem of determining the effect of the low-frequency eddy variability on the mean circulation of the Western North Atlantic. A framework for this study is first established by scale analysis of the eddy and mean terms in the mean momentum, vorticity, and heat balances in three regions of the Western North Atlantic -- the northern recirculation, the southern recirculation, and the mid-ocean. The data from the last decade of field experiments suggest somewhat different conclusions from the earlier analysis of Harrison (1980). In the momentum balance we confirm that the eddy terms are negligible compared to the lowest order mean geostrophic balance. The eddy term may be an 0(1) term in the vorticity balance only in the northern recirculation region where the mean flow is anisotropic. In the mean heat balance, if the mean temperature advection is scaled using the thermal wind relation, then the eddy heat flux is negligible in the mid-ocean, but it may be important in the recirculation areas. For all the balances the eddy terms are comparable to or an order of magnitude larger than the mean advective terms. We conclude from the scale analysis that the eddy field is most likely to be important in the Gulf Stream recirculation region. These balances are subsequently examined in more detail using data from the Local Dynamics Experiment (LDE). Several inconsistencies are first shown in McWilliams' (1983) model for the mean dynamical balances in the LDE. The sampling uncertainties do not allow us to draw conclusions about the long-term dynamical balances. However, it is shown that if we assume that the linear vorticity balance holds between the surface and the thermocline for a finite record, then the vertical velocity induced by the eddy heat flux divergence is non-zero. The local effect of the mesoscale eddy field on the mean potential vorticity distribution of the Gulf Stream recirculation region is determined from the quasigeostrophic eddy potential vorticity flux. This flux is calculated by finite difference of current and temperature time series from the Local Dynamics Experiment. This long-term array of moorings is the only experimental data from which the complete eddy flux can be calculated. The total eddy flux is dominated by the term due to the time variation in the thickness of isopycnal layers. This thickness flux is an order of magnitude larger than the relative vorticity flux. The total flux is statistically significant and directed 217° T to the southwest with a magnitude of 1.57 x 10 -5 cm/2s. The direction of the eddy flux with respect to the mean large scale potential vorticity gradient from hydrographic data indicates that eddies in this region tend to reduce the mean potential vorticity gradient. The results are qualitatively consistent with numerical model results and with other data from the Gulf Stream recirculation region. We find that the strength of the eddy transfer in the enstrophy cascade is comparable to the source terms in the mean enstrophy balance. The Austauch coefficient for potential vorticity mixing is estimated to be 0(107cm2/sec). An order of magnitude estimate of the enstrophy dissipation due only to the internal wave field shows that other processes must be important in enstrophy dissipation. The measured eddy potential vorticity fluxes are compared to the linear stability model of Gill, Green, and Simmons (1974). An earlier study (Hogg, 1984) has shown agreement between the empirical orthogonal modes of the data and the predicted wavenumbers, growth rates, and phase speeds of the most unstable waves. However, we show substantial disagreement in a comparison of the higher moments the eddy heat and potential vorticity fluxes. Because the critical layer of the model is located near the surface, the model predicts that most of the eddy potential vorticity and eddy heat flux should occur within about 300 meters of the surface. The data show much greater deep eddy heat flux than predicted by the model. It is suggested that the unstable modes in the ocean have a longer vertical scale because of the reduction in the buoyancy frequency near the surface. The evidence for in situ instability is also examined in the decay region of the Gulf Stream from an array of current and temperature recorders. Although there is vertical phase propagation in the empirical orthogonal modes for some of the variables at some of the moorings, there is not much evidence for a strong ongoing process of wave generation.