Abstract
Statistical comparisons between numerical sea ice models and an observed large‐scale strain array in the western Weddell Sea during 1992 are used to evaluate the performance of three of the more generally utilized sea ice rheology formulations. Results show that sea ice velocity is reproduced with relatively high accuracy (90% coherence, >80% normalized cross correlation) in models having high‐quality atmospheric forcing fields (e.g., the European Centre for Medium‐Range Weather Forecasts). On the other hand, temporal and spatial variability of the velocity field, as exemplified by progressive vector plots and ice deformation, respectively, are reproduced less accurately (coherence and normalized cross correlation <50%). In terms of model sensitivity, this means that deformation and temporal variability are more discriminating in terms of elucidating specifics about the constitutive relation and mechanical properties of sea ice on a large scale. For example, inclusion of both compressive and shear stresses is important in attaining a proper probability distribution of deformation relative to observations. Additional analysis shows that adjustments to specific model parameters improve the model results for either drift or select deformation components, but no best solution could be found, given the models examined here. Results suggest that inclusion of more physically based processes, such as subdaily tidal and inertial oscillations, reconsideration of the boundary layer formulation, and consideration of anisotropy, may be necessary to include in next‐generation sea ice models, especially those that are intended for coupling with high‐resolution (eddy resolving) ocean models.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have