Direct information on the spatial-temporal distribution of ice thickness is still rather limited, based largely on under-ice submarine transects and moored under-ice buoys, which measure ice draft with upward-looking sonar. However, the laser altimeter system on NASA's ICESat mission is starting to provide satellite-derived thickness estimates based on sea ice freeboard (Kwok et al., 2004). Draft is the part of the ice (roughly 90%) that projects below the water surface. Freeboard is the part that projects above the water surface. The ice cover includes areas of level ice, hummocked ice, pressure ridges with keels that may extend to depths of 20 m or more, as well as openings. Mean ice thickness is typically considered from the thickness of ice that is present averaged along with open water (zero ice thickness). Figure 7.5 provides a reasonable estimate of the mean winter ice thickness based on submarine sonar data. Immediately obvious is that the thickest ice is along the northern coasts of Canada and Greenland, where it ranges from 6 to 8 m. As discussed, this is a result of the mean Beaufort Gyre ice circulation having a component of on-shore motion, resulting in ridging. Thinner ice in the Eurasian shelf seas arises from the continual transport of ice away from this region toward Fram Strait, forming open water areas where new, thin ice can grow in autumn and winter. It hence follows that the Eurasian shelf seas are the primary areas of new ice production in the Arctic. This region is sometimes referred to as the "ice factory" of the Arctic Ocean.
Given that mean thickness is somewhat of a statistical abstraction, it is arguably better to characterize the thickness probability distribution. For example, in the PIZ, pressure ridges with keels 9 m deep recur once every 50-200 m of submarine track (Bourke and McLaren, 1992). The total ice thickness, including the above-water ice freeboard (and its snow cover), is approximately 1.115 times the average draft for level ice, and 1.130 times the draft for ice with ridges (Bourke and Paquette, 1989). Wadhams (1992) obtained similar results.
The annual cycle of thickness is still poorly known, particularly on a regional basis. Bourke and McLaren (1992) indicate that mean drafts at the same locations are 0.51.0 m lower in summer than winter (Bourke and Garrett, 1987). In the central Beaufort Sea, submarine data collected in 1958 and 1970 showed mean thicknesses of only 1.7 m in summer (McLaren, 1989). There appears to be considerable interannual variability in mean ice thickness. For example, Rothrock et al. (1999) compared sea ice draft data collected during submarine cruises between 1993 and 1997 with earlier records (1958-76). These comparisons indicate that the mean ice draft at the end of the summer melt period decreased by 1.3 m in most of the deepwater portions of the Arctic Ocean (see also the earlier paper of Wadhams, 1990). While subsequent work (Holloway and Sou, 2002) points to a significant problem of undersampling, the follow-on studies of Rothrock et al. (2003) and Rothrock and Zhang (2005) support the idea of recent thinning. We will return to these issues in Chapter 11.
Was this article helpful?