Importance of mesoscale advection in setting AMOC pathways and timescales

The Atlantic Meridional Overturning Circulation (AMOC) plays a key role in climate due to uptake and redistribution of heat and carbon anomalies. This redistribution takes place along the main pathways that link the high-latitude North Atlantic with the mid-latitudes and the Southern Ocean. This study aims to establish the relative importance of the material transport induced by mesoscale currents (“eddy-induced mixing”) and large-scale circulation in setting the AMOC pathways and timescales, using a suite of high-resolution simulations of oceanic tracers in an offline tracer model. The offline formulation uses previously calculated velocities and mixing intensity to simulate tracer distributions, which leads to a dramatic reduction in the required computational time and allows for running extended high-resolution simulations with multiple tracers.

AMOC pathways and timescales are studied using the Boundary Impulse Response (BIR) tracer, which in this study is defined as a response to an impulse at either the North Atlantic surface (NA BIR) or the southern lateral (SO BIR) boundaries of the Atlantic domain. While the NA BIR tracer quantifies ventilation of the Atlantic by the sinking branch of AMOC, the SO BIR tracer describes propagation of a signal from the Southern Ocean. In order to isolate the importance of the eddy-induced mixing in these processes, we contrast two offline simulations: the control case with the full advection and a sensitivity run in which the mesoscale advection is removed. Eddy-induced mixing is shown to remove the BIR tracers from the mean pathways, which facilitates ventilation and signal propagation into the low- and mid-latitudes. We then investigate the extent to which these effects of eddies can be represented by eddy diffusion. Using a separate set of idealized tracers, we estimate a spatially inhomogeneous tensor of eddy-induced lateral diffusivities and use it to replace the eddy-induced mixing in our offline simulations. The remaining biases in tracer distribution and the importance of spatial variability in eddy-induced diffusivities are discussed in the context of eddy parameterization in non eddy-resolving models.