Reproducing Sea-Ice Deformation Distributions With Viscous-Plastic Sea-Ice Models

High resolution sea-ice dynamic models offer the potential to discriminate between sea-ice rheologies based on their ability to reproduce the satellite-derived deformation fields. Recent studies have shown that sea-ice viscous-plastic (VP) models do not reproduce the observed statistical properties of the strain rate distributions of the RADARSAT Geophysical Processor System (RGPS) deformation fields [1][2]. We use the elliptical VP rheology and we compute the probability density functions (PDFs) for simulated strain rate invariants (divergence and maximum shear stress) and compare against the deformations obtained with the 3-day gridded products from RGPS. We find that the large shear deformations are well reproduced by the elliptical VP model and the deformations do not follow a Gaussian distribution as reported in Girard et al. [1][2]. On the other hand, we do find an overestimation of the shear in the range of mid-magnitude deformations in all of our VP simulations tested with different spatial resolutions and numerical parameters. Runs with no internal stress (free-drift) or with constant viscosity coefficients (Newtonian fluid) also show this overestimation. We trace back this discrepancy to the elliptical yield curve aspect ratio (e = 2) having too little shear strength, hence not resisting enough the inherent shear in the wind forcing associated with synoptic weather systems. Experiments where we simply increase the shear resistance of the ice by modifying the ellipse ratio confirm the need for a rheology with an increased shear strength.

[1] Girard et al. (2009), Evaluation of high-resolution sea ice models [...], Journal of Geophysical Research, 114
[2] Girard et al. (2011), A new modeling framework for sea-ice mechanics [...], Annals of Glaciology, 57, 123-132