The ocean mesoscale in a warming world: insights from multiscale modeling

Abstract

State-of-the-art climate models and computing infrastructure are now able to resolve mesoscale ocean eddy activity in many contexts. However, in computationally intensive model applications, such as the Coupled Model Intercomparison Project (CMIP) or simulations of the high latitudes, grid resolutions largely remain eddy-parameterizing due to resource constraints. These missing mesoscale processes are understood to be crucial drivers of ocean circulation and climate and may become still more relevant in the context of anthropogenic climate change. To overcome the computational limitations of traditional models, multiscale modeling strategies have been developed which can distribute grid resolution and resources based on resolution requirements and research goals. Here, several strategies for resolving the mesoscale using multiscale methods are described and the results of their implementation with the Finite volumE Sea ice Ocean Model (FESOM) are reported. In the first application, FESOM participates in CMIP6 with the strategy of concentrating computational resources on the major eddy-rich regions of the ocean. The resulting simulations are able to reproduce between 51 and 82% of observed eddy kinetic energy (EKE) in each region and project substantial climate change impacts on mesoscale activity for the first time at such a scale. The results include a poleward shift of eddy activity in most western boundary currents; EKE intensification in the Antarctic Circumpolar Current (ACC), Brazil and Malvinas Currents, and Kuroshio Current; EKE decline in the Gulf Stream; and intensification of Agulhas leakage. In a second application, FESOM is used to concentrate computational resources in the Southern Ocean and cost-reducing modeling strategies are used to enable fully eddy-resolving climate change projections with the regionally focused grid. The simulations faithfully reproduce EKE in the Southern Ocean and project intensified eddy activity in line with the CMIP6 analysis. The climate change signal is difficult to reliably discern from natural variability after 1 °C of warming, but becomes clear after 4 °C. Finally, the high-resolution Southern Ocean simulations are used to investigate high-latitude eddy activity where ice cover and low eddy size make observations and traditional modeling methods difficult. Detailed, near circumpolar mesoscale activity is detected and related to gyre circulation, the Antarctic Slope Current, and bathymetry. There is a strong seasonal cycle which suppresses winter eddy activity at the surface and selectively dampens cyclonic eddies. After prolonged anthropogenic warming, broad intensification of eddy activity occurs alongside regional decline, ACC eddy activity encroaches further into the high latitudes, and the seasonal cycle is diminished. Collectively, this work demonstrates the effectiveness of multiscale modeling in reducing the cost of resolving mesoscale ocean activity, facilitating the study of eddy activity and its interactions with the broader climate in previously unachievable contexts.