Thin ice thickness distributions estimated from advanced very high resolution radiometer (AVHRR) and RADARSAT Geophysical Processor System (RGPS) data sets were compared over the Beaufort Sea and the Canada Basin for the period December 1996 to February 1997. The comparisons show a compelling agreement. High correlations were found in cases where thin ice grew in large, wide leads extending several hundred kilometers. At these large scales, estimates from AVHRR images and RGPS showed similar amounts of thin ice in leads. However, when major surface deformation occurred on small scales (100 m to 10 km), the finer spatial resolution (100 m) of RADARSAT images enabled the RGPS algorithm to derive more thin ice than that of AVHRR. Under such conditions the correlation between the two dropped, and a small negative bias (about 1%) was observed in the estimates from AVHRR. This bias, mostly concentrated at the very thin end of the thickness distribution, caused a further deficit in the AVHRR-derived thin ice growth, roughly 0.2 cm/d. Although the AVHRR and RGPS algorithms treat snowfall differently in the ice thickness calculations, both snow assumptions appear reasonable. However, RGPS may underestimate the thin ice production because of the 3-day sampling interval. With a better understanding of the sources of uncertainty and improved satellite estimates, a combination of these two satellite data could offer a wider coverage of thin ice thickness observations in the Arctic Basin.

Aided by submarine observations of ice thickness for model evaluation, we investigate the effects of assimilating buoy motion data and satellite SSM/I (85 Ghz) ice motion data on simulation of Arctic sea ice. The sea-ice model is a thickness and enthalpy distribution model and is coupled to an ocean model. Ice motion data are assimilated by means of optimal interpolation. Assimilating motion data, particularly from drifting buoys, significantly improves the modeled ice motion, reducing the error to 0.04 m s^-1 from 0.07 m s^-1 and increasing the correlation with observations to 0.90 from 0.66. Without data assimilation, the modeled ice moves too slowly with excessive stoppage. Assimilation leads to more robust ice motion with substantially reduced stoppage, which in turn leads to strengthened ice outflow at Fram Strait and enhanced ice deformation everywhere. Enhanced deformation doubles the production of ridged ice to an Arctic Ocean average of 0.77 m yr^-1, and raises the amount of ridged ice to half the total ice volume per unit area of 2.58 m. Assimilation also significantly alters the spatial distribution of ice mass and brings the modeled ice thickness into better agreement with the thickness observed in four recent submarine cruises, reducing the error to 0.66 m from 0.76 m, and increasing the correlation with observations to 0.65 from 0.45. Buoy data are most effective in reducing model errors because of their small measurement error. SSM/I data, because of their more complete spatial coverage, are helpful in regions with few buoys, particularly in coastal areas. Assimilating both SSM/I and buoy data combines their individual advantages and brings about the best overall model performance in simulating both ice motion and ice thickness.

Observations of sea-ice draft from submarine cruises in much of the Arctic Ocean show that the ice cover was unusually thin in the mid-1990s. Here we limit our examination to digitally recorded draft data from eight cruises spanning the years 1987 to 1997 and find a decrease of about 1 m over the 11-year span. Comparisons of our modeled draft with observed draft show good agreement in the temporal change. Comparing average draft over entire cruises, the RMS discrepancy between modeled and observed draft is 0.3 m and the correlation is 0.98. Agreement in the spatial patterns of draft is somewhat lower; the RMS discrepancy of 50-km averages of draft is 0.7 m and the correlation is 0.73. We review reports of interannual variations of ice thickness or volume from other model studies. All models agree that thickness decreased by between 0.6 and 0.9 m from 1987 to 1996. Our model shows a modest recovery in thickness from 1996 to 1999. For the 1950s, 1960s, and 1970s, models tend to disagree on the size and to a lesser extent the timing or phase of interannual variations.

Temperatures and albedos derived from satellite imagery are combined with a thermodynamic ice model to estimate thin ice thickness distributions over the Beaufort and the northern Greenland seas. The study shows that thin ice (thinner than 1 m) occupied over half the area in the seasonal ice zones in November and December of 1990 but dropped significantly in April 1991; the Beaufort Shelf showed the largest seasonal change. The aggregate properties of surface salt flux and compressive ice strength, both strongly dependent on the thin end of the ice thickness distribution, were estimated with data from advanced very high resolution radiometer to reveal the spatial and temporal variations over a large scale. The salt flux from growing thin ice was 1–2 orders of magnitude larger on the Beaufort and Greenland Shelves than in the deep basins. On the shelves, flux from thin ice accounted for over 90% of the total surface salt budget. These satellite-derived estimates provided detailed spatial information on salt flux, which can be of great use in studies of surface patterns of salinity forcing and shelf-basin interaction. Also revealed by the satellite data was the wide range in values of compressive strength, which affects how freely the ice cover can deform. Strengths were low in early winter and in seasonal ice zones and nearly doubled in spring. Both thin ice fraction and compressive ice strength estimated from satellite imagery were in good agreement with those simulated by a coupled ice-ocean model.