Areas dominated by sheet flow, e.g. the southwestern shield-area parts of the Keewatin and Quebec domes, display 'classic' glacial landscapes comprising abundant glacial lineations, eskers aligned to lineations, ribbed moraine and a scarcity of sharp lateral contrasts in lineation length and development. Older ice-flow directions may be manifested by overridden large 'ghost' drumlins or older cross-cut striae.
Landscapes that probably are sites of former ice-stream webs or networks show fundamental differences from sheet-flow landscapes (Clark, 1999; Stokes & Clark, 2001) in that sharp lateral boundaries of lineation swarms are common, eskers are few and often not aligned to flow traces, and ribbed moraine is lacking or present only near former ice-stream heads. Older flow directions are mainly manifested as beheaded lineation patches outside ice-stream corridors (Plate 38.7). Topographical control on flow direction generally appears to have been strong in this type of landscape. In analogy with evidence from the present-day ice streams at Siple Coast in Antarctica, we interpret such landscapes to reflect networks of ice streams. Typical for these are that they display upstream tributaries into sheet-flow areas, frozen-based interstream ridges, head convergences and variability in length, width and velocity on short time-scales (Hodge & Doppelhammer, 1996; Bamber et al., 2000a; Gades et al., 2000).
Reconstruction of event sequences in such areas necessitates an approach significantly different from that used in sheet-flow areas. In sheet-flow situations, regional cross-cutting and overprinting of old lineation systems is seen as reflecting slow migration of dispersal centres, in response to climate-controlled mass-balance distribution changes on the ice-sheet surface. In ice-stream-dominated landscapes, on the other hand, disjunct and beheaded lineation patches probably reflect rapid onset of ice streaming at a particular location, and consequent drawdown in individual ice drainage basins. The time-scale for major directional changes may be in the order of a few hundred years. Inferences about overall ice configuration changes must in such areas be based on the collective evidence from several ice streams. For instance, the onset of the four major Finnish ice streams (Kleman et al., 1997; Boulton et al., 2001c) in Late-glacial time most probably caused a major westward shift in the ice-divide position of the Fennoscandian Ice Sheet. This conclusion would be rather uncertain if based only on the evidence from one ice-stream corridor.
Our investigations of palaeo-ice streams, hitherto, suggest that the term ice stream, although coherent and meaningful from a process point of view, in reality embraces a range of different fast-flow situations. This situation necessitates a subdivision into functionally different types of ice streams in the inversion context. We have observed and mapped the traces of a large number of palaeo-ice streams, varying widely in size and glaciological context, and with topographic guidance varying between strong and insignificant. In addition, we have uncovered evidence for small ephemeral ice streams in the Canadian Arctic during the deglaciation. They indicate flow patterns that are very different from older apparently semi-stable ice streams in the respective areas. We hold the view that a relevant classification should comprise at least three different types of ice streams.
1 Topographically governed ice streams are constrained by topography and fixed in space but variable in time (Marshall et al., 1996; Stokes, 2000; Kaplan et al., 2001). The Hudson Strait, Laurentian Channel and Norwegian Channel ice streams are prime candidates for this type.
2 Transient rigid-bed ice streams form without topographical or substratum control when thawed spots start to develop under a largely cold-based ice sheet, which then finds itself with a steeper profile than the reduced bed traction can sustain. The Dubawnt ice stream (Stokes, 2000) is the prime candidate for this type.
3 Ephemeral ice streams probably develop in response to a rapid calving and break-up of ice in adjacent marine areas. Prime candidates are the east-trending ice stream on Prince of Wales Island (Dyke et al., 1992) and the small Cap Krusenstern ice stream, the traces of which overprint large glacial lineations of the Amundsen Gulf ice stream of probable LGM age.
The development of appropriate inversion methods for areas dominated by ice streams is only in its infancy, although important progress has been achieved (Stokes & Clark, 2001).
Was this article helpful?