Estimation of sea ice parameters using an assimilated sea ice model with a variable drag formulation

In this work, sea ice concentrations in the regions of Baffin Bay, Hudson Bay, the Labrador Sea and the Gulf of St. Lawrence were estimated using a standalone regional configuration of Los Alamos Sea ice Model, CICE. A sensitivity analysis of the mixed layer depth criteria that affects the predicted sea ice formation/ablation has been carried out to tune the model. Sea ice concentration from AMSR2 OSISAF, sea ice edge from SSMIS/ASCAT OSISAF, and sea surface temperature from OISST AVHRR-only products were then assimilated into the tuned model using a combined optimal interpolation method. Extending previous studies, which used a constant drag formulation in their assimilation schemes, we use a variable drag formulation to represent the friction associated with effective sea ice surface roughness at the ice-atmosphere and ice-ocean interfaces. The variable drag coefficient is estimated from properties of the sea ice cover such as the ice concentration, the vertical extent and area of the ridges, the freeboard, the floe draft, the size of floes and melt ponds. The new features of the assimilated model include tracers to estimate ridged ice, level ice, melt ponds, freeboard, sail height and keel depth. The estimates of ice concentration from the assimilated models are then compared with the ice concentration product from the Canadian Ice Service. The results show that the updates provided to the tracers through the assimilation of sea ice products are reasonable and the model estimates of ice concentration from the assimilated models are better than the non-assimilated model estimates. Finally, the estimates of updated sea ice parameters such as ridged ice area, volume and ice thickness are compared with the estimates from non-assimilated models and the literature to verify the results.