The last glacial cycle abrupt glaciological responses to climate

Research on marine cores from the 1970s onward have shown that the isotopic changes in foraminiferal S18O have variations that match those predicted from calculations of the variations in the Earth's orbit (Hays et al., 1976). In particular, peaks in spectral analysis with periodicities of ca. 22, 41 and 100 kyr match variations in obliquity, precession and eccentricity, and it was widely argued that changes in solar insolation, at high northern latitudes, forced changes in the global ice-sheet inventory (Budd & Smith, 1981).

Broecker & Van Donk (1979) drew attention to the 'saw toothed' nature of the global ice volume changes, with the glacial intervals being abruptly terminated after a slow and progressive build-up of the global ice volume. The abrupt terminations of the glacial cycles suggest that causes other than the predictable changes in orbitally driven solar insolation provide positive feedback during the deglaciation processes. During glacial stages many continental ice-sheet margins fronted tidewater (Fig. 21.6c) and one explanation for the abrupt terminations is based on the collapse of the Northern Hemisphere's marine-based ice sheets (Denton et al., 1986; Andrews, 1991).

When we consider glacier-climate relationships (Fig. 21.1) on time-scales of 104yr we are now dealing with the growth and destruction of large ice sheets (Drewry & Morris, 1992) (Fig. 21.6c). The interaction between the height and albedo of large ice sheets and the climate system is important, as is the impact of freshwater inputs to the ocean and the effect on the THC. In addition on these time-scales, changes in the elevation of ice sheets and relative sea level will occur at the margins of ice sheets as the relaxation time for the process of glacial isostatic adjustment is of the order of 2-3kyr for the LIS (Dyke & Peltier, 2000) (Fig. 21.6c). The massive Northern Hemisphere ice sheets over northern North America (LIS) and Fennoscandia were centred spatially over the large interior seas of Foxe Basin/Hudson Bay, and the Baltic/Gulf of Bothnia, hence are 'marine-based' (Denton & Hughes, 1981). Owing to the weight of the ice sheets and subsequent glacial isostatic depression of the crust, a large fraction of the beds of these ice sheets was well below sea level during glacial maxima (Peltier & Andrews, 1976; Peltier, 1994).

The modern analogue for these ice sheets is the West Antarctic Ice Sheet; theoretical arguments for the sensitivity of this ice sheet to future, and past, climate changes have been advanced (Hughes, 1977; Mercer, 1979; MacAyeal, 1992a; Anderson, 1999). It came as a surprise, however, that the evidence for massive and abrupt changes in ice-sheet discharge (from studies of marine cores in the North Atlantic) was associated with periodic collapse of the LIS that are now referred to as Heinrich or H-events (Andrews, 1998; Bond et al., 1992; Broecker et al., 1992; Broecker, 1994; Heinrich, 1988). In the original paper, however, Heinrich (1988) envisioned a link with insolation forcing, which was the ruling paradigm at that time, and only with the 1992 publications (Andrews & Tedesco, 1992; Bond et al., 1992; Broecker et al., 1992) did the nature of the glaciological response take centre stage. Thus, in the 1990s the focus on ice-sheet-climate shifted to take into account evidence for abrupt changes in the cryosphere-climate system at much higher periodicities than solar insolation changes related to the Earth's orbit around the sun (Fig. 21.9a & b).

Over the past decade or more the literature on 'glacial' events dated between ca. 10 and 50 cal.kyr is dominated by the results from the study of marine sediments. The evidence on land for glacial H- or D-O events is not particularly compelling. The marine studies highlighted a number of dramatic glacial responses, largely viewed from the perspective of changes in the ice rafted component (IRD) of deep-sea sediments, and the evidence for significant additions of meltwater to the surface ocean. During major glacial cycles the ice sheets thickened and extended outwards. In many areas this has resulted in ice sheets extending beyond the coastline onto the continental shelf, often to the shelf break (Piper et al., 1991; Vorren et al., 1998) (Fig. 21.6c). During the advance phase of the glacial cycle, the crust will be undergoing glacial isostatic depression and the relative sea level at the seaward margin of the ice sheet will, probably, be rising (this depends on the balance between global ice-sheet growth, which extracts water from the ocean, and the regional rate of ice-sheet growth, which results in isostatic depression (Peltier & Andrews, 1976; Lambeck, 1990; Peltier, 1996; Lambeck et al., 2000)). Because of the density differences between ice and marine water (900 versus ca. 1028kgm-3) there is a buoyancy force working to lift the ice sheet from its bed. This works to reduce friction at the bed and to cause an acceleration of glacier flow, which could not be matched by an increase in accumulation, hence leading toward a collapse and rapid retreat of ice margins on glaciated continental margins. This scenario is also enhanced by the fact that deep troughs which lead toward fjords cross most glaciated continental shelves (Holtedahl, 1958).As ice retreats, glacial isostatic recovery would cause uplift at the sea floor and a marine regression. Thus there are hints here of a self-regulating cycle of glacial response where an increase in the regional mass balance results in an ice advance onto the continental margin, but thereafter the termination of the advance and the retreat might be caused by rapid calving driven by changes in water depth (Fig. 21.6c). This is a fundamentally different set of responses from those climate-mass-balance interactions that control the activity of a margin on land.

The initial evidence for abrupt changes in climate, but not a priori glacial response, during MIS 2 and 3 (or ca. 13 to 50cal.

j it"

W-S "iflti



10.00 15.00 ACCUM


Figure 21.10 GISP2 data—accumulation (cmyr-1) versus S18O at 500-yr intervals for the past 50,000 yr. The figures represent the estimated age (cal.kyr) for a data point. The line is the best fit for the correlation between the two variables.

kyr) was highlighted by the data from the Greenland Ice Sheet (Fig. 21.9a), which indicated very abrupt changes in the isotope (climate) data (Johnsen et al., 1992)—the question is whether these were reflected in a glacial response? Data from marine sites close to the margins of the Greenland Ice Sheet (Voelker, 1999,2000; van Kreveld et al., 2000) and the eastern LIS (Andrews & Barber, 2002) indicated that in the interval 12-50cal. kyr there were coeval changes in the marine environments (Fig. 21.9c & d) on either side of Greenland (Fig. 21.3). The climate for the region during this interval (Fig. 21.10) (Meese et al., 1994) indicates that colder temperatures occurred with lower accumulation (r2 = 0.77). The points in the upper right quadrant of Fig. 21.10 are the Holocene values and the graph shows that there is an overall relationship between accumulation and S18O (which is generally considered to be linearly related to temperature (Dans-gaard et al., 1969)). This linkage persists in the time domain, with interstadials being times of higher net accumulation on the summit of the ice sheet. Of concern in this chapter are the interrelationships between the climate-ocean variables and the feedbacks between ice sheets and glaciers (Figs 21.1 & 21.2) (Sakai & Peltier, 1997).

The data from the Greenland Ice Sheet reflect changes in climate at around 3 km in the atmosphere; the marine data show a series of parallel changes in a variety of proxies (Fig. 21.9), but how do these link to actual changes in the extent and position of ice margins during the D-O or H-events? Heinrich events represent large-scale collapse of the LIS with a 5-7 kyr periodicity, and possibly linked to events around other ice sheets (Bond & Lotti, 1995). Dansgaard-Oeschger events represent climate cycles present in ice-core and marine records (van Kreveld et al., 2000; Andrews & Barber, 2002), with a series of them forming a saw-toothed pattern that leads to an H-event (Moros et al., 2002); these have been termed 'Bond cycles' (Broecker, 1994). Note, however, that the glaciologically massive H-events are not dramatically different in their isotopic composition from the D-O stadial events (Fig. 21.9a). Furthermore, as noted by several authors (Mayewski et al., 1997), a ca. 1500-yr cycle is embedded in the Greenland ice-core data; hence Bond et al. (1999) argued that the IRD events within the Holocene are a continuation of this fundamental cycle (e.g. Fig. 21.8B). This periodicity is seen in the band-pass filter on the GISP2 S18O data (Fig. 21.9a) and on the multi-tapered (MTM) spectra (Mann & Lees, 1996) from 100-yr resolution GISP2 data (not shown).

Heinrich events must represent major changes in the dynamics of the LIS but whether these massive collapses of the ice sheet, which may be associated with rapid changes in sea levels (Andrews, 1998; Chappell, 2002), are driven by internal glacio-logical dynamics, 'climate' at the bed-ice interface (Fig. 21.1), or by surface climate, is debatable (Alley et al., 1999; Clarke et al., 1999; van Kreveld et al., 2000). The 'binge-purge' model for the origins of H-events (MacAyeal, 1993a,b) is similar to some theories for the origins of surging glaciers (Canadian Journal of Earth Sciences, 1969). In these cases the links with atmospheric or ocean climate are not direct and the events are driven by changes in the conditions at the bed-ice interface, particularly temperature and water. Not included within the 'binge-purge' model is the possibility that large subglacial lakes may have existed under the LIS and H-events might be associated with 'outburst' floods (Shoemaker, 1992a,b; Hesse & Khodabakhsh, 1998). Because Hand D-O oscillations are associated with ice-sheet behaviour largely associated with marine (tidewater) margins, many of the caveats associated with the relationship (or lack thereof) between glacier response and climate (e.g. Figs 21.2B & 21.6b) at LIA and Holocene time-scales (Mann, 1986) apply on these longer time-scales. In particular, the presence on the Canadian shelf of a large 600 m deep basin seaward of the Hudson Strait sill (Andrews & MacLean, 2003) may do much to explain the massive collapse of the LIS called for in H-events.

As noted above, much of our information on abrupt glaciolog-ical variability in ice sheets comes from marine cores (Jennings et al., 1996; Stoner et al., 1996; Scourse et al., 2000; Groussett et al., 2001; Hemming et al., 2002b), so are there coeval responses of land-based margins (Mooers, 1997; Kirby & Andrews, 1999; Rashid et al., 2003)? Margins that descend to sea level or lake level (tidewater and calving margins) are intuitively more prone to be unstable (Thomas, 1977, 1979a; Broecker, 1994). However, one of the dramatic revolutions in the past several years has been the recognition of fast-flowing ice streams and margins that are situated on soft sediments, which deform rapidly (Clark & Walder, 1994; O Cofaigh & Evans, 2001; Dowdeswell et al., 2004a). This was partly predicted in the 1970s when Matthews (1974) documented very low gradients on marginal lobes of the LIS across the Canadian Prairies. These low gradients were compatible with shear stresses of 4-10 kPa, an order of magnitude lower than from most glaciers (Paterson, 1981); similar shear stresses were also measured on moraine segments from the highly lobate margins of the southwest and northwest LIS (Beget, 1987; Clark, 1994; Clark et al., 1996b). Soft sediments also probably lay under the large ice streams that drained the LIS, and especially Hudson Strait (Andrews & MacLean, 2003). However, D-O events are characteristic of MIS 3 (Johnsen et al., 1992) (Fig. 21.9a & b) and, at

Age ka BP



1600 1800 2000 2200 2400

Age ka BP

Figure 21.11 (A) S18O records on benthic foraminifera for the past 0-1 Myr, and (B) 1.5 to 2.5 Myr (see text for source).

present, it appears that the LIS did not extend onto the Canadian Prairies during this period of time (Andrews, 1987b, Fulton, 1989; Dyke et al., 2002). Thus although the southwest margin of the LIS was very dynamic during MIS 2 (Lowell et al., 1995) it appears (= no evidence?) that this sector of the ice sheet was not involved in the D-O interstadial-stadial responses coeval with those in the Greenland ice-core record (Fig. 21.9a) or seen in the North Atlantic sediment records (van Kreveld et al., 2000; Moros et al., 2002). It is probably no accident that H- and D-O events are recorded offshore from the margins of ice sheets terminating in the sea, hence involving glacier dynamics associated with changes in water depth driven by a combination of glacial isostatic forcing and exchange of freshwater between oceans and ice sheets.

On D-O time-scales there is actually a significant disconnect between the spatial continuity of ocean and climate evidence and the glacial record. For example, although the margins of the Younger Dryas (YD) (H-0, Fig. 21.9a) ice sheets are relatively well-mapped in Fennoscandia (FS) and Great Britain, there is no comparable moraine system around the margin of the Greenland Ice Sheet (it might be the Milne Land stadial? (Funder, 1989; Kelly et al., 1999)). Furthermore, although the margins of the YD time period are reasonably well constrained for the LIS (Dyke, 2004) there does not appear to be the same continuity of moraine expression as noted for the FS ice sheet, although some moraine sequences have been linked to the YD. There are indeed coeval moraine intervals to some of the younger H-events (Lowell et al., 1995) but there is not a 1:1 match, and the link between the marine-based H-events and the history of the terrestrial LIS ice extent over the past 50,000 yr is not easy to decipher (Stoner et al., 1996; Kirby & Andrews, 1999; Rashid et al., 2003;). Exactly what the terrestrial glacier/ice sheet responses were to D-O events is uncertain in terms of the creation of a geological record, such as an end moraine or a till stratigraphy.

0 0

Post a comment