Except for areas of landfast ice near the coast, the Arctic ice pack is in near constant motion. The mean annual circulation has two major features, a clockwise motion in the Canada Basin known as the Beaufort Gyre, and a motion of ice from the Siberian coast, across the pole and through Fram Strait, known as the Transpolar Drift Stream (Fig. 23.3). Most of the ice that leaves the Arctic on an annual basis exits through the Fram Strait (between Spitzbergen and Greenland). This flux is primarily of thick, multiyear ice. The long-term mean ice circulation is determined by roughly equal contributions of the surface wind field and surface ocean currents. However, the ice motion exhibits pronounced variability, which is primarily wind driven. Depending on the wind field and internal stresses in the ice cover (floe-to-floe interactions), the ice motion may be divergent, decreasing ice concentration, or convergent, increasing ice concentration. If ice concentration is already 100%, convergence will raft ice floes atop one another (part of the sea ice deformation process). Divergence leads to more open water. In autumn and winter, the open-water areas will quickly freeze, forming areas of new, thinner first-year ice. In summer, the exposure of dark open-water areas associated with divergence promotes strong absorption of solar radiation, increasing melt. Understanding the observed loss of Arctic Sea ice requires that we address such factors.
The NAO/AO link seems to involve several related processes. The first two can be understood with the aid of Plate 23.2, adapted from the study of Rigor et al. (2002). The top panel shows observed trends in summer sea-ice concentration (based on passive microwave records) along with observed trends in sea-ice motion for the previous winter (based on IABP data), both computed over the period 1979-1998. Following earlier discussion, the trends in summer ice extent have been largest along the Siberian and Alaskan coasts. Ice concentration trends largely mimic the trends in extent. The trends in winter ice motion indicate a change to a more cyclonic (counterclockwise) pattern. Care should be exercised in the interpretation—although the actual winter ice motion remained generally anticyclonic (broadly like that in Fig. 23.3), the trend has been toward more cyclonic conditions. This manifests itself as a shrinking of the Beaufort Gyre, and an increase in the size of the Transpolar Drift Stream during high NAO/AO conditions. The bottom panel shows the components of summer ice concentration and winter ice motion for each year regressed on the winter AO index. The key point is that the patterns are very similar—most of the change in summer ice concentration and winter ice motion relates to the trend in the winter AO.
The relevant processes are as follows. Changes in the winter wind field associated with the general upward trend in the winter AO changed the sea-ice motion field. This change features enhanced transport of ice away from the Siberian and Alaskan coasts and more ice divergence (or less convergence). Both processes lead to openings in the ice cover, which quickly refreeze to form thin, first-year ice. By spring, there is an anomalous coverage of this thin ice. With more thin ice, less energy is needed to completely melt the ice in summer. Unless there is a compensating motion of ice into the region, there will be less ice at summer's end. A second effect is that the thinner ice also leads to an earlier melt onset, further favouring ice losses. Except in summer, sea ice insulates the cold atmosphere from the relatively warm underlying ocean. With thinner ice in spring, the insulating effect is reduced. There are strong heat fluxes to the atmosphere, manifested as a positive SAT trend. The warmer air temperatures, together with the fact that the thinner, first-year ice melts at a lower temperature, fosters early melt (salt depresses the freezing point, and first-year ice is saltier than multiyear ice). Hence, the SAT rise in spring favouring early melt is not so much a direct advective effect of the NAO/AO, but rather a result of wind-induced changes in the ice cover.
The Rigor et al. (2002) model does not explain all aspects of the recent ice trend, such as the extreme minima of 2002 and 2003. As noted, the AO, although showing a positive trend since about 1970, and very positive from about 1989-1995, has in recent years retreated toward a more neutral state (Fig. 23.2). Serreze et al. (2003) highlight the very strongly cyclonic circulation of the atmosphere during the summer of 2002. A manifestation of the sea ice momentum balance is that ice drift tends to be to the right of the surface geostrophic wind (the wind parallel to contours of sea-level pressure, representing a balance between the pressure gradient and Coriolis forces). As a consequence, when the atmospheric circulation (hence sea-ice circulation) is cyclonic, the ice cover will tend to diverge, resulting in more open water. Although by itself, ice divergence in summer will spread the existing ice over a larger area, Serreze et al. (2003) argued that enhanced absorption of solar energy in the resulting dark open-water areas between floes would lead to much stronger melt, reducing ice extent. For each month from June through to August 2002, the sea-level pressure fields featured mean closed lows over the Arctic Ocean, an extremely unusual, stormy pattern. Further analysis shows the summer of 2003 as also characterized by an unusually cyclonic pattern.
The subsequent analysis of Rigor & Wallace (2004) gives a different view. Using a simple sea-ice model, they show that these recent extreme sea-ice minima may have been induced dynamically by changes in surface wind field over the Arctic associated with the previous very positive AO state from about 1989 to 1995. Put differently, the sea-ice system seems to have a memory of the previous high AO state. Plate 23.3 shows the estimated age (in years) of sea ice for 1986 and 2001 based on the model. The obvious difference between the two years is the much younger age of ice in 2001. In the 1980s, winds associated with primarily negative AO conditions favoured a larger Beaufort Gyre, which trapped sea ice in the central Arctic Ocean for over 10 years. New ice that was formed along the coast was advected away from the coast towards Fram Strait. For example, in the top panel we mark areas of sea ice that formed in the coastal Beaufort and Chukchi seas in the autumn of 1981 and 1984 which are drifting towards Fram Strait in the Transpolar Drift Stream. By comparison, sea ice returning to the Alaskan coast along the eastern flank of the Beaufort Gyre has been drifting in the central Arctic Ocean for many years. From 1989 to 1991, the average age (thickness) of sea ice in the Arctic Ocean decreased rapidly with the step to high NAO/AO conditions. The areal extent of older, thicker ice declined from 80% of the Arctic Ocean to 30%. Although some of this loss may represent melt, it seems that the old ice was primarily exported out of the Arctic via Fram Strait. Several investigators have documented an upward trend in either ice volume or area fluxes through the Fram Strait from the late 1970s through much of the 1990s (e.g. Kwok & Rothrock, 1999). The overall effect is that sea ice stayed in the Arctic Basin for a shorter period, meaning less time for the ice to ridge and thicken.
The AO then declined from its high positive state. During the summers of 2002 and 2003, the anomalously younger, thinner ice was advected into Alaskan coastal waters where very pronounced ice melt occurred. For example, in the bottom panel of Plate 23.3, we show the recirculation of the much younger ice formed in the Beaufort and Chukchi seas in the autumns of 1993,1997 and 1999 back towards the Alaskan coast instead of leaving the Arctic through Fram Strait. This argument helps to explain why the past several years have seen large ice reductions, despite the decline in the winter AO.
Rigor & Wallace (2004) show that, statistically, ice divergence associated with the cyclonic summer pattern noted by Serreze et al. (2003) would actually increase, rather than decrease, ice extent (the ice would be spread over a larger area). By this argument, the more dominant role of the cyclonic summer circulation was probably its contribution to the unusually diffuse ice cover in the
Beaufort Sea seen in September of both years (Plate 23.1). Another remarkable aspect of September 2002 and 2003 (again see Plate 23.1) was the virtual absence of sea ice in the Greenland Sea (off the east coast of Greenland). Serreze et al. (2003) argued that the cyclonic summer circulation led to a wind field that limited ice transport through Fram Strait and into the Greenland Sea. In the Rigor & Wallace (2004) framework, however, the Fram Strait flux was probably composed of unusually thin ice, such that ice entering the Greenland Sea rapidly melted away. A reasonable hypothesis is that both processes were at work.
Was this article helpful?