Does increased spatial resolution improve the simulation of Arctic sea ice lows in NEMO4.2-SI3?

The Arctic total sea ice extent has rapidly declined since the beginning of satellite observations. This decline materialized into record sea ice lows in the summers of 2007 and 2012. Those sea ice lows exhibit an important spatial heterogeneity and are likely caused by different dynamic and thermodynamic drivers of atmospheric and oceanic origins. Using the global ocean–sea ice model NEMO4.2-SI³ in the same setup but at three different horizontal resolutions (namely, 1/12°, 1/4° and 1°), we thoroughly examine the most extreme sea ice states simulated in summer by the model from a mass balance perspective. This method allows to disentangle the dominating mechanisms leading to the sea ice lows, such as dynamic redistribution and compression of sea ice in 2007, or preconditioning and excess basal melt in 2012. It also highlights the importance of processes at the ice-ocean interface to drive the evolution of sea ice at all considered temporal scales. We then compare how increased spatial resolution, allowing for the simulation of finer-scale physical processes such as ocean eddies, impacts the modelled sea ice thickness and concentration distribution, as well as the different ice mass fluxes. A particular attention is being paid to the influence of ocean heat content anomalies, as increased horizontal resolution provides a more realistic simulation of heat inflow in the Beaufort Gyre through subsurface eddies of Pacific origin. This study highlights the benefits of increased spatial resolution for realistically simulating the Arctic sea ice cover and weighs them with the associated computational cost. The decomposition of the ice mass budget into its different thermodynamic and dynamic terms puts forward the often downplayed role of the ocean in determining the interannual variability of Arctic sea ice and provides a stepping stone for further studies.