Sea ice deformation

Ice thickness and concentration are strongly determined by differential ice velocity. Consider two neighboring plates (floes) of ice. If the velocity difference between the neighboring plates is such that they tend to move apart (diverge), a lead is created. During winter, new ice may form in the lead. If the motion changes, such as by a change in the winds so that the neighboring plates move toward each other (converge), the lead closes. Any new ice that was formed in the lead must rearrange itself to occupy a smaller area. The typical processes for this are rafting, where one plate of ice rides atop the other, and ridging, where the ice is crushed into pieces that pile into ridges rising up to several meters above and sometimes many meters below the surrounding ice (keels). Deformed ice is hence thicker than undeformed ice (especially FYI), and s.i /■•■>■■• r-".:;

Figure 7.7 Mean pattern of sea ice drift in the Arctic for summer, based primarily on data from the IABP, the North Pole program and other sources with overlay of sea level pressure from NCEP/NCAR (ice drift field courtesy of I. Rigor, Polar Science Center, University of Washington, Seattle, WA, sea level pressure field by the authors).

Figure 7.7 Mean pattern of sea ice drift in the Arctic for summer, based primarily on data from the IABP, the North Pole program and other sources with overlay of sea level pressure from NCEP/NCAR (ice drift field courtesy of I. Rigor, Polar Science Center, University of Washington, Seattle, WA, sea level pressure field by the authors).

thus more likely to survive a melt season (and become MYI). The large ice thicknesses north of the Canadian Arctic Archipelago are associated with ridging and rafting. The basic idea is that the sea ice accommodates divergence by increasing the area of open water rather than thinning. It accommodates convergent motion by reducing the area of open water and by rafting and ridging (Thorndike, 1986).

A standard data product from the IABP is ice velocity derivatives, from which divergence and shear (which together represent the total deformation) can be calculated along with vorticity (the rotational part of the flow). Similar products can be made from satellite-derived ice motion fields. Care must be exercised in interpreting such data because the true ice motion field is granular, with velocity discontinuities along the floe boundaries or larger ice aggregations. Put differently, the field is inherently non-differentiable. If we calculate from IABP or SSM/I velocity fields that an area of

Figure 7.8 Mean pattern of sea ice drift in the Arctic for January 1989 based on SSM/I retrievals with overlay of sea level pressure from NCEP/NCAR (ice drift field courtesy of C. Fowler, University of Colorado, Boulder, CO, sea level pressure field by the authors).

the pack ice is converging, this is interpreted as an aggregate closing of the pack ice over some length scale much larger (tens to hundreds of km) than the typical distance between floes. If the ice concentration is <100%, the aggregate closing is seen as a regional increase in ice concentration. If the concentration is 100%, the aggregate closing implies the development of ridges and keels, increasing the ice thickness. Locally, however, the motion could be non-divergent. Divergent flow calculated over some length scale is interpreted similarly. The aggregate opening is seen as a decrease in the regional ice concentration, even though the motion could be locally non-divergent or convergent. Similarly, in an area of large-scale shear only, there can be local opening, closing and the development of ridges and keels. By the same token, the large-scale vorticity field masks local deviations from the general rotational flow. Estimates of ice deformation are also prone to error as they are sensitive to small errors in the velocity

Imagenes Excavacion Terreno
Figure 7.9 Mean pattern of sea ice drift in the Arctic for January 1991 based on SSM/I retrievals with overlay of mean sea level pressure from NCEP/NCAR (ice drift field courtesy of C. Fowler, University of Colorado, Boulder, CO, sea level pressure field by the authors).

field. For a detailed discussion of ice deformation, the reader should consult Thorndike (1986).

Using results from Thorndike and Colony's (1982) linear regression model, one can obtain estimates of the large-scale divergence and convergence from the geostrophic wind at sea level and the turning angle of the ice velocity with respect to the geostrophic wind. Since the turning angle is to the right of the geostrophic wind, cyclonic ice motion is associated with divergence, and anticyclonic motion is associated with convergence. As discussed, the turning angle is larger in summer, when the ice motion approaches "free drift" and is smaller in winter, when internal ice forces are strong. Serreze et al. (1989) applied the Thorndike and Colony approach to estimate ice divergence in the Canada Basin during summer. They found that due to frequent episodes of cyclonic ice motion, forced by the frequent migration of atmospheric lows into the region, the

Figure 7.10 Time series of sea ice divergence (in percent per day) in the vicinity of the SHEBA station calculated at four different spatial scales (average sizes) centered on the station, based on the RADARSAT RGPS (from Stern and Moritz, 2002, by permission of AGU).

calculated divergence in the interior ice pack can be large. For example, during the 30-day period August 13 to September 11, 1980, when the ice motion was strongly cyclonic, aggregate divergence rates of up to 0.75% per day were calculated. Over the 30-day period, this amounts to a total opening of over 20%. Estimates based in the IPAB data were somewhat lower (0.50% per day). This is counter to the situation for most of the year, when ice motion in this area tends to be anticyclonic (the Beaufort Gyre) and hence slightly convergent.

As part of the SHEBA field study, Stern and Moritz (2002) evaluated time series of ice divergence at several different scales centered over the main SHEBA camp (Figure 7.10). They based these estimates on the RADARSAT (synthetic aperture radar, or SAR) Geophysical Processor System, using the same basic feature-tracking approach just described with respect to SSM/I. However, the ice motion fields from RADARSAT are at a much finer (5 km) spatial resolution. While the basic time series structure as well as the magnitudes of divergence are similar at the different scales, (all fairly large, ranging from 50 x 50 km to 200 x 200 km), there are differences. Divergences/convergences exceeding 1% per day are not uncommon. The extremes are around 4-5% per day.

Stern and Moritz (2002) discuss the annual cycle with reference to Figure 7.10. In autumn and early winter, as the pack ice is growing and becoming stronger, the divergence is mostly positive (new leads form and new ice grows). There is a cumulative divergence of about 25% between 1 November and 25 December. January contains large divergence and convergence events. Winds are moderate through the month, varying from northerly to easterly. The ice drifts generally westward and leads form with a generally northwest-southeast orientation. At the end of January, wind speeds pick up quickly, and the direction changes to slightly south of easterly. This causes large convergences. Leads formed earlier in the month close and there are episodes of ice ridging. From February though July, divergent and convergent events are small, but the ice undergoes a gradual convergence of about 15% as the SHEBA station drifts within the Beaufort Gyre. The large divergence at the end of July is associated with a storm. After this time, the nature of the deformation becomes more random, consistent with low summer ice strength associated with "free-drift". After about 11 September, the ice enters a period when, in response to seasonal cooling, it begins to gain strength and eventually redevelops the more winter-like property of "plates and cracks". The ice pack becomes well consolidated by 4 October.

0 0

Post a comment