Western boundaries have been suggested as mesoscale eddy graveyards, using a diagnostic of the eddy kinetic energy (EKE) flux divergence based on sea surface height (η). The graveyard’s paradigm relies on the approximation of geostrophy — required by the use of η — and other approximations that support long baroclinic Rossby waves as the dominant contribution to the EKE flux divergence. However, a recent study showed an opposite paradigm in the Agulhas Current region using an unapproximated EKE flux divergence. Here, we assess the validity of the approximations used to derive the η-based EKE flux divergence using a regional numerical simulation of the Agulhas Current. The EKE flux divergence consists of the eddy pressure work (EPW) and the EKE advection (AEKE). We show that geostrophy is valid for inferring AEKE, but that all approximations are invalid for inferring EPW. A scale analysis shows that at mesoscale (L > O(30)km), EPW is dominated by coupled geostrophic-ageostrophic EKE flux and that Rossby waves effect is weak. There is also a hitherto neglected topographic contribution, which can be locally dominant. AEKE is dominated by the geostrophic EKE flux, which makes a substantial contribution (54%) to the net regional mesoscale EKE source represented by the EKE flux divergence. Other contributions, including topographic and ageostrophic effects, are also significant. Our results support the use of η to infer a qualitative estimate of the EKE flux divergence in the Agulhas Current region. However, they invalidate the approximations on mesoscale eddy dynamics that underlie the graveyard’s paradigm.

Western boundaries (WB) have been suggested to be hotspots of mesoscale eddy decay, using an eddy kinetic energy (EKE) flux divergence based on sea surface height (η). The η-based diagnostic requires approximations, including the use of geostrophic velocities. Here, we assess to what extent mesoscale EKE flux divergence can be inferred from η using a numerical simulation of the Agulhas Current. The EKE flux divergence is composed of two terms: the eddy-pressure work (linear component) and the advection of EKE (nonlinear component). Both are mainly positive in the WB region (net EKE sources). However, it is not reliably accounted by both η-based diagnostics. The η-based eddy-pressure work has a net contribution in the WB region of the opposite sign than the true one. Ageostrophic eddy-pressure work dominates the geostrophic one (corresponding to a β-contribution). It is explained by mesoscale eddies’s scale to fall below the scale of ζ/β (ζ: root mean square of normalized relative vorticity for mesoscale eddies; β: latitudinal variations of Coriolis parameter). The advection done by geostrophic EKE flux dominates the EKE flux divergence in the WB region. It results in the EKE flux divergence to be qualitatively estimable using η (up to 54 % of the net EKE source). Our results in the Agulhas Current show a mesoscale eddy dynamics in contrast with the decay’s paradigm at western boundaries. Further analysis in other western boundaries are required to complete our understanding of mesoscale eddies dynamics.