Remarkable control of western boundary currents by Eddy Killing, a mechanical air-sea coupling process

Western boundary currents (WBCs) are critical to Earth's climate. In the last decade, mesoscale air-sea interactions emerged as an important factor of WBC dynamics. Recently, coupled models including the feedback of surface oceanic currents to the atmosphere confirmed the existence of a physical process called eddy killing, which may correct long-lasting biases in the representation of WBCs by providing an unambiguous energy sink mechanism. Using ocean-atmosphere coupled simulations of the Gulf Stream and the Agulhas Current, we show that eddy killing reduces the eddy-mean flow interaction (both forward and inverse cascades) and leads to more realistic solutions. Model and data fluxes are in good agreement when the same coarse grid is used for their computation, although in this case they are underestimated. We conclude that the uncoupled approach is no longer suitable for continued ocean model improvement and discuss new formulations that should better account for air-sea interactions. Plain Language Summary Western boundary currents (WBCs), such as the Gulf Stream and the Agulhas Current play a crucial role in global ocean circulation and in determining and stabilizing the Earth's climate. In the last decade, mesoscale air-sea interactions emerged as important in WBC dynamics. Recently, coupled models including the feedback of surface oceanic currents to the atmosphere revealed a process called eddy killing, which potentially corrects long-lasting biases in the representation of WBCs. In this study, using ocean-atmosphere coupled simulations of the Gulf Stream and Agulhas Current, we show that eddy killing reduces the interactions between eddies and mean flow. The influence of the eddies on the mean flow can be measured by the cascade of energy, and, in particular, the inverse cascade of energy. The reduction of inverse energy flux by eddy killing leads to realistic solutions and, in particular, to the observed stabilization of WBCs. Model and data fluxes are in good agreement when the same coarse grid is used for their computation, although in this case they are underestimated. We conclude that uncoupled models are no longer suitable for continuing our model improvement of ocean dynamics and discuss new formulations that should better account for air-sea interactions