Arctic landfast ice simulation with brittle rheology and probabilistic grounding

Landfast ice, i.e., sea ice that is mechanically immobilized for several weeks along the coasts, significantly influences the underlying ocean by controlling the occurrence of coastal polynyas and the formation of dense water within. However, it is usually poorly represented in numerical models. In the Arctic, the accurate simulation of landfast ice relies on parameterizing sea ice grounding in shallow water areas and on the sea ice rheology capability to form ice arches in regions with restricted geometry. In this study, we compare a brittle rheology (i.e., the Brittle Bingham-Maxwell or BBM one), newly implemented in the ocean-sea ice model NEMO-SI3, with a standard viscous-plastic rheology (i.e., the aEVP), which is widely used in sea ice models. The performance of the two rheologies in forming ice arches and landfast ice is evaluated at the scale of the Arctic at a 0.25° horizontal resolution. For the grounding parameterization, we apply a probabilistic grounding scheme based on the ice thickness distribution and investigate how leveraging subgrid-scale bathymetry statistics can enhance its performance.