Glacier Surge After Ice Shelf Collapse

See allHide authors and affiliations

Science  07 Mar 2003:
Vol. 299, Issue 5612, pp. 1560-1562
DOI: 10.1126/science.1077987


The possibility that the West Antarctic Ice Sheet will collapse as a consequence of ice shelf disintegration has been debated for many years. This matter is of concern because such an event would imply a sudden increase in sea level. Evidence is presented here showing drastic dynamic perturbations on former tributary glaciers that fed sections of the Larsen Ice Shelf on the Antarctic Peninsula before its collapse in 1995. Satellite images and airborne surveys allowed unambiguous identification of active surging phases of Boydell, Sjögren, Edgeworth, Bombardier, and Drygalski glaciers. This discovery calls for a reconsideration of former hypotheses about the stabilizing role of ice shelves.

Ice shelves are floating ice sheets attached either to land or to a grounded ice sheet. The transition zone between floating and grounded ice—the grounding line—is a complex zone where a major change in ice flow conditions takes place (1). Grounding lines are of great importance in Antarctica because they are natural gates through which most ice is exported off the continent. The mechanical transition taking place at grounding lines has promoted discussions about their stability as well as their possible influence on the stability of ice sheets, with special reference to the marine-based West Antarctic Ice Sheet (WAIS) (2–5). In the past, the main hypothesis was that ice shelves buttress the inland ice, and their removal could trigger the surge of ice streams, which drain WAIS, causing severe depletion of continental ice levels and therefore a sudden increase in global sea level. However, modern theoretical models disregard the buttressing effects of ice shelves, suggesting that their disintegration should not affect the inland ice (6–8). In view of this disagreement, the collapse of ice shelves on the Antarctic Peninsula (9–11) represents a unique opportunity to study the effects of ice shelves on grounded ice and provides the necessary ground truth to verify previous and present ideas about the matter (12).

Vaughan (13) reported that glaciers that formerly nourished the Wordie Ice Shelf (Fig. 1) did not suffer noticeable perturbation after its disintegration. He concluded that the ice shelf backpressure (14) did not play an important role in the force balance of the nourishing glaciers, although at least one, the Fleming glacier, was found to have an important sliding velocity component (15). On the other hand, Rott et al. (16) found that large outlet glaciers formerly nourishing the northernmost sections of the Larsen Ice Shelf (LIS) [that is, Larsen A and Prince Gustav Channel (Fig. 1)], which disintegrated in early 1995, suffered strong acceleration and retreat beyond their grounding lines after the collapse of LIS. For instance, Drygalski glacier lost 24 km2 of its front, coincident with up to a threefold increase in ice velocity between 1995 and 1999. Sjögren and Boydell glaciers have undergone a similar ice loss, the former doubling its velocity during the same period.

Figure 1

Map of the Antarctic Peninsula, showing locations cited in the text.

In late 2001 and early 2002, we conducted an airborne reconnaissance and Global Positioning System (GPS) mapping survey over the northeastern Antarctic Peninsula. During the October 2001 survey, we observed the presence of distinctive “ice terraces” on both margins of Drygalski glacier (Fig. 2). A few months later, in February-March 2002, the existence of similar features was confirmed on the margins of Sjögren and Boydell glaciers (Fig. 3). Ice terraces are slices of ice attached to the rock margins, elevated ∼20 to 40 m above the actual glacier surface. They clearly resemble the blocks of “stranded ice” left by the passage of a surge front on some glaciers (17). The spatial location of terraces and their primary layering are indicative of their glacial origin. Furthermore, the lack of chaotic morphology and icefalls in the surrounding area discards the possibility that these ice terraces were deposited by avalanches. Our observations indicate that these ice features were once part of the glacier margin. Their existence is indicative of a sudden lowering of glacier surfaces, suggesting important dynamic unstabilities.

Figure 2

Drygalski glacier. (A) Low-altitude aerial photograph taken on 13 February 2002, showing the ice terraces discovered on the northern margin of Sentinel Nunatak in October 2001. The view is to the west. (B) Aerial photograph showing a close-up view toward the ice terraces.

Figure 3

(A and B) Ice terraces on the margins of Sjögren glacier as photographed on 15 March 2002.

In addition to ice terraces, the clearest evidence of strong dynamic perturbations comes from the comparison of two high-resolution optical satellite images. A Landsat 7 Enhanced Thematic Mapper plus (ETM+) image taken on 21 February 2000 and an Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) image taken on 26 September 2001 were co-registered. Image analysis allowed the identification of active surging phases (18) on some of the six major outlet glaciers affected by LIS removal. Figure 4 shows the surge of Sjögren and Boydell glaciers in September 2001. Both glaciers have advanced about 1.25 km, accounting for a net area gain of 7.9 km2 during the 1.6-year time interval. The detection of typical surge waves, the opening of marginal crevasses by enhanced shearing, and the formation of a looped moraine confirm the passage of a surge front along both glacier tongues. A similar active surging phase is visible on Bombardier and Edgeworth glaciers (Fig. 5), with a maximum net advance of 1.65 km and an area increase of 6.5 km2. Furthermore, the Landsat image of 21 February 2000 (Fig. 5A) revealed a feature at the boundary between Bombardier and Dinsmoor glaciers interpreted as a remnant looped moraine (Fig. 5A). This indicates the possibility that these glaciers surged shortly before February 2000. It is important to note that no evidence could be found for surging of Dinsmoor glacier during the same period. Further evidence of dynamic changes comes from analyzing the surface crevasse patterns. We observed that typical transverse crevasse systems turned into a pattern of complex ice pinnacles, suggesting a dramatic increase in longitudinal stretching.

Figure 4

Surge of Sjögren and Boydell glaciers. (A) Landsat 7 ETM+ image, band 8 from 21 February 2000. (B) ASTER image, band 3 from 26 September 2001. The dashed line indicates the position of already-separated ice fronts as surveyed by means of airborne GPS on 15 March 2002.

Figure 5

Surge of Bombardier and Edgeworth glaciers. (A) Landsat 7 ETM+ image, band 8 from 21 February 2000. (B) ASTER image, band 3 from 26 September 2001. The dashed line indicates the position of the ice fronts as mapped on 15 March 2002. Both images show many icebergs and brash ice, indicating intense calving activity.

Mapping of ice front positions carried out in March 2002 by means of an airborne GPS survey allowed us to confirm a strong retreat at some glacier fronts after their surge. The comparison with positions derived from the September 2001 ASTER image indicates that Sjögren and Boydell glaciers lost 12.1 km2 after retreating past their former confluence, whereas Dinsmoor, Bombardier, and Edgeworth glaciers lost 6.5 km2 during the same period. The behavior of Drygalski glacier is less clear on the basis of available data, allowing us to detect only a minor retreat from February 2000 to March 2002. Both the satellite images and aerial surveys reveal the existence of dense brash ice persistent in the embayments of these glaciers, suggesting the continuation of intense calving activity. Furthermore, it should be noted that unlike the fast-flowing tributaries, the slowly flowing ice piedmont and many smaller glaciers located within the region have suffered only minor changes after the disintegration of the ice shelf.

Finally, an analysis of ice velocities on Sjögren glacier made with sequential satellite images, applying the analog feature-tracking method, confirms that acceleration continued from 1999 to 2001. The displacement of several surface crevasses measured between February 2000 and September 2001 yielded speeds of 1.8 to 2.4 m day−1 within the same sector where Rott et al.(16) measured 1.0 m day−1 in 1999. This represents a doubling of ice velocity in only 2 years. Using the same method, average velocities of 2.4 m day−1 were detected on a section of Edgeworth glacier, about 4 km upstream from the glacier front in September 2001, although this is not indicative of further acceleration since 1999.

The evidence presented here unambiguously shows that five of the six major tributaries that formerly nourished the disintegrated portions of LIS have recently experienced important dynamic perturbations. This includes not only the detected acceleration and retreat (16) but also an active surging. A thorough explanation of surges and their relation to ice shelf collapse is not yet possible, because the basal dynamics and thermal properties of these glaciers are unknown. Furthermore, the fact that only the major fast-flowing tributaries surged, not the slowly moving ice piedmonts and smaller glaciers, clearly demonstrates the importance of basal and thermal conditions in the control of glacier dynamics. This observation supports the hypothesis that the existence of conditions allowing fast sliding under the major tributaries would imply that the backstress of LIS played an important role in their force balance. Several factors may have contributed to generating the necessary conditions for fast sliding, such as the regional climatic warming (11) producing enhanced melting and percolation to the base of glaciers (19) and/or the occurrence of deformable beds. We suggest that in the absence of other sources of backstress, the removal of the ice shelf could have been sufficient to trigger the surges.

It should be emphasized that the grounded ice on the northeastern Antarctic Peninsula is rapidly retreating and therefore substantially contributing to the global rise in sea level. The risk increases when the possible surging response of the Hektoria-Green-Evans and Crane glaciers is considered; these glaciers formerly nourished the section of Larsen B Ice Shelf that disintegrated in early 2002. Further research is necessary to establish possible analogies between the behavior of these fast-flowing glaciers and the ice streams draining WAIS. In view of the controversial theoretical discussions about the stability of WAIS, the evidence presented here calls attention to early hypotheses about the buttressing effects of ice shelves and the inherent instabilities of ice stream–ice shelf coupling.

  • * To whom correspondence should be addressed. E-mail: glacio{at}


View Abstract

Navigate This Article