The melt cycle during two Arctic summers is simulated with a dynamic-thermodynamic sea ice model. The regional, seasonal and interannual variability of the surface albedo is compared directly with corresponding estimates derived by Robinson et al. (1986) from satellite imagery. Despite the large seasonal variability (~0.8 to ~0.2) of the observationally derived regional albedos, the largest error in the simulate albedo for any region or month is approximately 0.10. The model also captures much of the interannual variability of the regional albedos, which were observed to be lower in 1977 than in 1979 because of an earlier onset of melt. The component of the model's thermodynamic forcing that explains the most of the interannual variability of regional albedo (and ice thickness) is the air temperature. On the oher hand, the various sensitivity tests show that the model is at least as sensitive to the specification of monthly snowfall as to the parameterization of the snow and ice albedos. ¿ American Geophysical Union 1987 |