The Deep Western What?
The Deep Western Boundary Current (DWBC) runs along the western boundary of the Atlantic Ocean from the Subpolar North Atlantic to the South American region. In its western subpolar origins, it is composed of the densest water masses formed by deep convection in the western and eastern Supolar North Atlantic (North Atlantic Deep Water, NADW). For this reason, and due to inferences from theoretical models, the DWBC was classically considered as the principle southward conduit of the NADW, and of the Atlantic Meridional Overturning Circulation (AMOC).
Boundary Current "Leakiness"
However, evidence mounted in recent decades that the DWBC "leaks" material in abundance into the ocean interior around two large underwater capes before entering the subtropical region. The leakiness further leads to "interior pathways", which are alternative (to DWBC) deep southward pathways of the NADW, traveling through the ocean interior well separated from the western boundary. This realization significantly shifted the community understanding of deep flow in the ocean as part of the AMOC, a circulation pattern of significant climatic significance.
The mechanisms and dynamics of leakiness from the DWBC were not completely clear however. An hypothesis was raised in the literature of DWBC leakiness by interaction with eddies spwaned from a second boundary current, the North Atlantic Current (NAC) passing northward near the DWBC. To address the question of the DWBC leakiness mechanism, we conducted two separate studies: 1. An idealized investigation of boundary current instability along curved boundaries (Solodoch et al , 2016; JFM); 2. Diagnosis of DWBC leakiness within a high resolution realistic numerical model and in observations (Solodoch et al 2020; JPO) , as detailed below.
Baorclinic Instability of curved flow around bathymetry
In the first study, we investigated the instability of boundary currents travelling along curved boundaries, similar (although in an idealized setting) to the capes that the DWBC circumnavigates in its leaky area. Such instability can lead to eddy shedding and excahge of material with the interior. We computed the instability tendency in various curved flow settings, and compaerd it with the tendency for rectilinear flow settings. Somewhat surprisingly, we found (Solodoch et al , 2016; JFM) that the curvature of the flow and of the boundary may either reduce or increase instability.
Leakiness by inertial separation
To proceed, in the second study, we shifted our attention to high-resolution numerical modeling using the numerical model ROMS. We built a realistic ROMS configuration of our area of interest in the North Atlantic, and analyzed DWBC leakiness in the model. For comparison we also analyzed several observational datasets , including float datasets (Argo floats and Export Pathways (ExPath) floats; the latter were previously used to demonstrate the leakiness phenomenon). We seeded Lagrangian "particles" in our models in large numbers, validated their trajectories against the 60 ExPath floats trajectories', and then studied the statistics of the leakiness paths and origins using the particles. The Lagriangian analysis was accompanied by Eulerian analysis, including statistics of offshore transport, cluster analysis of typical flow patterns, an energetic budget, and a potential vorticity budget.
Our findings (Solodoch et al 2020; JPO) show that a main dynamical reason for the leakiness is inertial separation due to the sharpness of the underwater capes. Furthermore, and consistently, we found that the leakiness is not intermittent but happens much by a continuous splitting of the DWBC into components leaving to offshore in several "hotspot" locations of topographic variation. The contribution of eddies and variability to the leakiness process is however significant, and the interaction of the steady and eddying leakiness components is quite subtle.
The findings suggest that the leakiness process depends primarily on the abrupt variations of seafloor topography in this area, and hence the leakiness may be a resilient phenomenon under various climatic conditions. Additionally, scaling results offer a guide as to the numerical resolution necessary to capture the separation process. Finaly, the dynamics of inertial separation are not unique to the studied region, and may apply to other areas in the ocean where boundary current separation has been observed.
Approximate time-mean streamfunction (right panels) on two different potential isopycnal surfaces. The depths of each surface is shown on the left panel in the same row. Within both density surfaces, streamlines leave the Deep Western Boundary Current and separate into the ocean interior, i.e., away from the western boundary. The locations of separation of these streamlines is consistent with leakiness of floats in previous observational campaigns and with the numerical float experiments discussed above. The leakiness occurs near the bends in the Flemish Cap underwater cape shown here (1,3&4 km isobaths in black contours), and on its southern face where isobath steepness increases. The leakiness is thus manifested to a large degree as a stready separation process, and aditional analyses supports an interpretation of an inertial separation process.