Abstract/Summary

The Arctic sea ice cover is in retreat. Accurate representation of the marginal ice zone (MIZ), the region of the sea ice cover that separates open ocean from the pack ice, is important to capture this retreat in models. The MIZ is associated with complex interactions of the atmosphere, sea ice, and oceans, and a highly heterogenous sea ice cover. Several important sea ice properties and processes that determine the evolution of the MIZ, including lateral melting, momentum exchange, and sea ice rheology, are dependent on floe size. Climate models have historically treated floe size as a fixed parameter, if at all. Observations have shown that floes adopt sizes from scales of metres to kilometres. Here I investigate two alternative models of the floe size distribution (FSD). The first approach assumes the FSD follows a power law with a fixed exponent and the second approach is a prognostic floe size-thickness distribution model where the shape of the FSD freely evolves. These models are used to understand how variable floe size in the MIZ changes the seasonal retreat of the Arctic sea ice cover, both through the impact on lateral melt volume and on momentum exchange coefficients. I discuss the advantages and disadvantages of each approach, including an assessment of whether either model improves sea ice model performance compared to observations. I find a high sensitivity to poorly constrained FSD parameters and parameterisations, highlighting the need to better characterise the FSD with observations. I show that winter floe formation and growth processes strongly influence FSD impacts on the sea ice over the melt season. I also demonstrate the need to incorporate brittle fracture in FSD models. I conclude that simple representations of floe size are sufficient to project future sea ice trends, but FSD models are important to capture the spatial distribution of the sea ice.