Northern Hemisphere Ice-Sheet Influences Global Climate Change

Australia: The Land Where Time Began

Northern Hemisphere Ice-Sheet Influences Global Climate Change

Active Interaction of large ice sheets with the rest of the climate system takes place by amplifying the pacing, and potentially driving climate change over several time scales. Ocean surface temperatures, ocean circulation, continental water balance, vegetation, and the albedo of the land surface result from direct and indirect influences of ice sheets on climate, and these changes in turn cause additional changes in the climate system and help in the synchronisation of global climate change. Clark et al. suggest the missing link in understanding the interactions between the climate and the ice that are integral to the transition that occurred in the Middle Pleistocene may be the effect of the underlying geological substrate on ice sheet dynamics; the 100 ky climate cycle; high amplitude, millennial scale variability of climate; and the low aspect ratio ice sheets of the Last Glacial Maximum (LGM).

Large sheets first developed in the Northern Hemisphere about 2.54 Ma (Shackleton, Berger & Peltier, 1990; Mix et al., 1995), following a long-term global cooling trend that continued through much of the Cainozoic. The Milankovitch theory of glaciation that is driven by orbital changes is supported by statistical analyses of palaeoclimate data, that showed that the ice sheets of the Northern Hemisphere have waxed and waned with the same periods (100 ky, 41 ky and 23 ky) as the orbital parameters, which are eccentricity, obliquity and precession, that control the season distribution of insolation at high northern latitudes (Hays, Imbrie & Shackleton, 1984). These orbital periodicities are also shown by other features of the climate system, though may lag insolation forcing of climate change at high northern latitudes by much longer (5 to 15 ky, depending on the period), than is expected (Imbrie et al., 1992; Imbrie et al., 1993). Ice sheets may have been responsible for amplifying and transmitting changes that correspond to orbital periodicities in seasonality elsewhere through the climate system with a phase lag which corresponds to their long time constant (Imbrie et al., 1992; Imbrie et al., 1993), as ice sheets are 1 of the few components of the climate system that have a time constant of this length. According to this hypothesis, interactions among ice sheets in the northern hemisphere and other features of the climate system therefore translate high latitude insolation forcing into a global climate signal that has dominant orbital-scale glacial cycles (10⁴ to 10⁵ years) (Imbrie et al., 1992; Imbrie et al., 1993) in which millennial scale variations (10³ to 10⁴ years) are embedded (Bond et al., 1993).

Several questions regarding interactions between the ice sheet and the climate remain, in spite of the success of the Milankovitch theory in explaining many aspects of temporal and spatial variability of the climate change of the late Cainozoic. There are a number of questions that needed to be answered:

1) What are the mechanisms that almost synchronise the climates of the Northern Hemisphere and the Southern Hemisphere at orbital time scales in spite of asynchronous insolation forcing?

2) What is the origin of the transition invariability of the global ice volume in the Middle Pleistocene about 1.2 Ma, from 41 ka cycles that were dominant to 100 ky cycles, while there is a lack of change in insolation forcing?

3) What is the origin of the 100 ky cycle in the absence of any substantive insolation forcing at this period?

4) Which mechanisms are responsible for suborbital, climate variability at millennial scale?

5) Which processes of the LGM 21 ka to be surprisingly thin and therefore have a different influence on climate than would have been the case had the ice sheets been thicker?

In this study Clark et al. address these issues by first discussing the mechanisms by which ice sheets are able to influence global climate and cause climate change that is near-synchronous in the polar hemispheres. The evidence that the dynamics of modern and former ice sheets are strongly influenced by geological and topographic characteristics of the substrate beneath the ice sheet is reviewed. Clark et al. proposed that the effect of the substrate underlying the ice sheets of the Northern Hemisphere is to modulate the response of the ice sheet to forcing of insolation. These modulated responses are then transmitted as a global climate signal through the effects on the climate of ice sheets and may explain several of the key issues that surround the evolution and behaviour of the climate system over the past 2.5 Ma.

Influence on climate of ice sheets

Ice sheets are among the largest topographic features on the Earth, which is why they influence the climate, and they are responsible for some of the largest reginal anomalies in albedo and radiation balance, as well as representing the largest reservoir of freshwater that is readily exchangeable on Earth. The variations in fluxes from ice sheets are especially large, because they shrink at the faster rate of surface melting, or even the faster rate of ice sheet dynamics (surging), even though they grow at the usually slow rate of snowfall. Ice sheets reorganise continental drainage by damming rivers and reversing the flow of rivers through the isostatic depression of bedrock beneath the ice, thereby forming lakes that fill over a number of years to centuries, though they may drain at an order or orders of magnitude faster when the ice dams fail (Walder & Costa, 1992).

It is suggested by experiments with climate models that there are several mechanisms by which climate may be influenced by Northern Hemisphere ice sheets. Features of the ice sheet-climate interactions that are common to a number of simulations, such as:

· the southwards displacement of the winter jet stream by high ice sheets, cooling that is substantial over and downwind of the ice sheets,

· Reorganisation and straightening of storm tracks along the southern margin of the Laurentide Ice Sheet and across the North Atlantic region,

· And generation of large anticyclones at the surface of the ice sheets (Manabe & Broccoli, 1985; Pollard & Thompson, 1997; Kutzbach et al., 1998; Ganopolski et al., 1998).

The height of the ice sheet determines the circulation effects, whereas the generalised temperature effect is primarily dependent on the area of the ice sheet (Rind, 1996). These effects are transmitted through the atmosphere downwind to the adjacent North Atlantic Ocean, where they cause a reduction of the sea surface temperatures (SSTs) and expansion of the sea ice (Manabe & Broccoli, 1985; Ganopolski et al., 1998).

It is indicated by several dynamical ocean models that the strength of the thermohaline circulation in the North Atlantic Ocean, which it transfers substantial amounts of heat to high latitudes in the Northern Hemisphere, is sensitive to the freshwater budget at the formation sites of the North Atlantic Deep Water (NADW) (Maier-Raimer & Mikolajewicz, 1988; Stoker, Wright & Broecker, 1992; Rahmstorf, 1995). A major regulator of the formation North Atlantic Deep Water and associated heat transport (Ganopolski et al., 1998; Broecker, Bond & Klas, 1994; Weaver, 1999), because they affect the freshwater budget of the North Atlantic directly by the release of meltwater and icebergs and indirectly through atmospheric controls on precipitation and evaporation over the North Atlantic. According to Clark et al. the inherent symmetry in the rates off ice sheet processes – slow buildup but rapid decay and slow filling of lakes at the margin of the ice but rapid drainage in outburst flood can cause orders of magnitude changes in freshwater fluxes.

It is suggested that the North Atlantic is coupled tightly to the ice sheets of the Northern Hemisphere by the transmission of influences of ice sheets in the Northern Hemisphere to the SSTs and formation of NADW in the North Atlantic. (Ruddiman, 1987). It is likely that changes in the formation of NADW through other mechanisms that are not yet determined, though it appears that ice sheets are amplified by these processes (Bond et al., 1999). In any event, forcing of the ice sheets is a mechanism that is well understood and this may explain many of the variations of NADW in the past (Mix & Fairbanks, 1992; Raymo, Ruddiman & Shackleton, 1990; Keigwin et al., 1991). Substantial changes in the North Atlantic system may occur largely in response to freshwater delivered from ice sheets, accompanied by ice sheet size changes that are only modest, is a corollary of this argument.

The release of freshwater from ice sheets to sites where NADW forms is caused by climate forcing, or its amplification, may be transmitted to distant regions through the atmosphere and ocean (Stocker, Wright & Broecker, 1992; Rahmstorf, 1995; Rind et al., 1997; Hostetler et al., 1999). A long way from the North Atlantic, however, climate anomalies on orbital time scales are much more prominent than those arising only from changes in the North Atlantic SSTs at millennial time scales, as a result of large differences between CO₂ and ice sheets during glacial as opposed to interglacial times, though only minimally associated with meltwater forcing of North Atlantic SSTs and NADW formation at millennial time scales (Alley & Clark, 1999; Hostetler & Bartlein, 1999).

The reduced formation of NADW, however, does not contribute much to synchronisation of interhemispheric climate change. At times of glacial maxima shallower and southwards displacement of the formation of NADW that is seen in some models leads to cooling in high northern latitudes through expanded sea ice in the North Atlantic (Ganopolski et al., 1998), though the rate of formation of the NADW and its outflow to the Southern Ocean are reduced only slightly from those of modern times (Ganopolski et al., 1998; Weaver et al., 1998). A near collapse of the formation of NADW, such as happens when it is perturbed by a large pulse of freshwater (Stocker, Wright & Broecker, 1992; Rahmstorf, 1995; Weaver, 1999), in contrast, causes warming in parts of the Southern Hemisphere either by a reduction of cross-equatorial flow of Atlantic surface waters, which leaves heat in the South Atlantic (Stocker, Wright & Broecker, 1992; Mix, Ruddiman & McIntyre, 1992), or by stimulating drift to the south to supply the formation of deepwater in the south (Weaver, 1999; Schiller, Mikolajewicz & Voss).

Additional feedbacks that transmit the ice sheet signal globally and contribute to synchronising the hemispheres, are provided by many atmospheric and oceanic responses to changes that are ice-induced. Colder temperatures over Eurasia, snow cover increases, and vegetation type changes increases albedo and aridity and weaken the African and Asian monsoons (Kutzbach et al., 1998; Prell & Kutzbach, 1996; deMenocal & Rind, 1993), which thereby reduces the export of tropical water vapour and affects heat exchange across the equator. Glacial surface temperatures were lower than at present over much of the globe, with the largest differences occurring above ice sheets and regions of more extensive sea ice in both hemispheres (Manabe & Broccoli, 1985; Pollard & Thompson, 1997; Kutzbach et al., 1998). Enhanced polar cooling that is associated with ice albedo and other feedbacks increases the equator-to-pole temperature gradient, and this causes wind strength increases (Ganopolski et al., 1998; deMenocal & Rind, 1993; Overpeck et al., 1989). The tropics are cooled by stronger winds, by upwelling of colder waters, or entrainment of extratropical waters, which further cools the tropics and extratropics by water vapour feedbacks in the atmosphere (Ganopolski et al., 1998; Bush & Philander, 1998; Ágústsdόttir et al., 1999).

It is suggested by model results that lower concentrations of CO₂ are required to explain the magnitude and symmetry of global cooling that is observed during global glaciations (Pollard & Thompson, 1997; Weaver et al., 1998; Broccoli & Manabe, 1987). The identification of why atmospheric concentrations of CO₂ have changed is problematic, however, as is the establishment of temporal relation to global ice volume changes. Records of deep sea sediment changes in δ¹³C values suggest that the CO₂ concentration change leads to sea level (global ice volume) (Shackleton & Pisias, 1985), though the integrity of the δ¹³C record as a measure of the atmospheric concentrations of CO₂ is not certain (Curry & Crowley, 1997). It is similarly suggested by the interpretation of δ¹⁸O values of atmospheric O₂ (δ¹⁸O_atm) in ice core records as a proxy of sea level, that changes in the levels of atmospheric CO₂ lead ice volume (Sowers et al., 1991; Petit et al., 1999). Other factors may, however, influence δ¹⁸O_atm values (Sowers et al., 1991; Broecker & Henderson, 1998), and it has proven to be difficult to put the ice core chronology on the same time scale as the deep sea δ¹⁸O record of global ice volume (Broecker & Henderson, 1998; Raymo & Horowitz, 1996). The only well-dated records that currently best link sea level to atmospheric concentrations of CO₂ for the last deglaciation, and these suggested that the initial rise in atmospheric concentrations of CO₂ lags sea level rise by 0 to 4 ky. It is likely that multiple controls on atmospheric CO₂ concentrations are likely to have controlled concentrations of CO₂, and ice sheets are not likely to have controlled them completely. There are several plausible processes, however, by which ice-induced changes in sea level, temperature, windiness, dust and other factors could influence atmospheric concentrations of CO₂ (Broecker & Henderson, 1998; Berger, 1999), which would provide a strong feedback on the growth and decay of ice sheets (Pagani et al., 1999; Pearson & Palmer, 1999).

Ice-Sheet Dynamics

In order to understand what controls the evolution and behaviour of ice sheets it is necessary to understand the influence they have had on climate over the past 2.5 My. Given the importance of ice sheets in the climate system, what controls their evolution and behaviour is necessary to understand their influence over the past 2.5 My. The same need applies equally to questions of the future stability of the West Antarctic Ice Sheet (WAIS) (Bentley, 1998). A strong connection between the dynamics of ice sheets and the geology they rest on (Boulton & Jones, 1979; Alley et al., 1986), has been revealed by studies of modern and former ice sheets, that the substrate can modulate the behaviour of ice sheets and through the influence of ice sheets on the climate system, change climate (MacAyeal, 1992; ______, 1993; Clark, 1994; Anandakrishnan et al., 1998; Clark & Pollard, 1998).

The influence of ice sheets on climate is determined by the dynamics of ice sheets that affects their size, response to climate change, and the release of freshwater from them to the oceans. The movement of glacial ice is accomplished by some combination of internal deformation of the ice, basal sliding, and deformation of subglacial sediment (Paterson, 1994). If the basal temperature of the ice sheet is lower than the melting point, the ice is coupled to the underlying substrate, and almost all motion occurs by internal deformation of the ice. If the temperature is at the pressure melting point, ice motion by basal sliding is facilitated by water that is produced, and by sediment deformation in places where there is unconsolidated sediment. Compared to a frozen-bed glacier, a glacier that has an unfrozen bed has a lower aspect ratio, higher balance velocity, and a response time that is shorter, as well as other mechanisms that can generate instability in the ice (MacAyeal, 1992; ______, 1993).

The physics of basal sliding and deformation of subglacial sediment remain poorly understood (Paterson, 1994), unlike the constitutive law for internal ice deformation that is relatively well tested. Sliding occurs over hard bedrock (hard beds) and over unconsolidated sediments, though it is in general favoured when the glacial ice is above the soft beds that are low friction where there is low bed relief and the basal water pressure is high. Deformation of subglacial sediment occurs when soft beds that are saturated with water deform under the shear stress that is applied by the overlying ice sheet. It is not known what the appropriate constitutive law for the deformation of subglacial sediment is, and rheologies that have been proposed range from slightly nonlinear (Boulton & Hindmarsh, 1987) to perfectly plastic (Kamb, 1991). Reconciliation of these contrasting observations by proposing a viscous behaviour at the large scale results from multiple, distributed, small scale failure events [see also (Iverson et al., 1998)].

According to Clark et al. they favour a general hypothesis in which basal motion is partitioned variously between sliding and deformation of subglacial sediment, depending on the temporal and spatial variations in subglacial hydrologic conditions and the properties of sediment (Iverson et al., 1999), though the specific processes by which the soft beds influence basal motion has not yet been resolved. The ice may be decoupled from its bed, thereby increasing sliding at the expense of deformation of the sediment, under conditions resulting in basal water pressure that is sufficiently high. A substantial challenge to the modelling of the long term behaviour and evolution of ice sheets is represented the formulation of rules describing the complex spatial and temporal variability that governs basal motion – factors such basal thermal regime, subglacial hydrology, ice bed coupling, sediment rheology, and continuity (MacAyeal, 1992; Marshall et al., in press).

It is suggested by observations beneath the WAIS that basal motion is linked strongly to the geology of the substrate. Basal motion is responsible for the presence of ice streams that are flowing rapidly in the WAIS (velocities of 10² to 10³ m/year), and this is confirmed on either side by ice sheet flow that is slow-moving (10⁰ to 10¹ m/yr), and it accounts for nearly all discharge from the WAIS (Hughes, 1975), and also for the low-aspect ratio of the ice sheet (Alley et al., 1986). It is shown by geophysical observations (Blankenship et al., 1997) and drilling (Engelhardt et al., 1990) that several of the ice streams in the WAIS that drain into the Ross Sea overlie soft beds that have basal water pressures that almost cause floatation of the ice. It is suggested by the head of one of these ice streams coinciding with the upstream edge of a sedimentary basin that the presence of sedimentary basins determines the presence of ice streaming.

It is also suggested by geological evidence from the areas that were formerly covered by the Norther Hemisphere ice sheets of the last ice age also that there is a strong relation between the distribution of soft beds and basal flow that is enhanced (Boulton & Jones, 1979; Clark, 1994). There were sedimentary basins in central areas of the Fennoscandian and Laurentide Ice Sheets that at present are largely below sea level (Hudson Bay, Gulf of Bothnia), and when they were beneath the ice sheets the weight of the ice depressed them even further. Crystalline bedrock, which was in turn surrounded by sedimentary bedrock, surrounded these core areas. Typically, areas of bedrock are of low relief and are covered by unconsolidated sediments, that are relatively continuous and of low permeability, which suggests these areas of soft-bed were predisposed to ice flow that was fast when basal motion was activated. Contrasting with this, the sediment cover that was of higher relief and continuous were characteristic of areas of crystalline bedrock suggest stronger ice bed coupling and therefore reduced ice flow (Marshall et al., 1996).

Transition in the Middle Pleistocene

At the transition during the Middle Pleistocene, about 1.2 Ma, the dominant 41 ky ice volume variations changed to the dominant 100 ky variations under what was essentially the same orbital forcing (Pisias & Moore, 1981). Records of those features of the climate system that are driven by ice sheets also show the transition (deMenocal, 1995; Williams et al., 1997; Clemens, Murray & Prell, 1996; Ding et al., 1994), which suggests the mechanism that was responsible for the transition in the size of the ice sheets and their variability was ultimately responsible for a substantial change in the behaviour of the climate system.

Ice sheet-climate models that have been used to explore the cause of the transition in the Middle Pleistocene produce a transition as a nonlinear response to either a prescribed, long-term cooling trend that was associated with decreasing concentrations of atmospheric CO₂ (Oerlemans, 1984; Saltzman & Maasch, 1991; Berger et al., 1999; Paillard, 1998), or to a switch in model physics that was imposed suddenly (DeBlonde & Peltier, 1997). There is a lack of data that constrains a long-term cooling trend or a decrease in concentrations of atmospheric CO₂ over the past 3 My, but it is suggested by these models that such a trend is a possible cause of the transition.

It is indicated by geological records that the Laurentide Ice Sheet, which dominates the global ice volume signal, was more extensive in area before the transition than after it (Clark & Pollard, 1998). The record of the δ¹⁸O of global ice volume, in contrast, indicates a large increase in the volume of the ice after the transition. These records, that are apparently contradictory, can be reconciled invoking a change at the transition from ice sheets that are thinner (about 2 km) to thicker (about 3 km), which requires a change in basal flow condition. Hard-bedded areas were covered by a thick (10s of metres) soil that was deeply weathered (regolith) that built up in northern latitudes over 10s of millions of years prior to the growth of the ice sheets, at the initiation of the glaciation of the Northern Hemisphere. According to Clark et al. this soft bed can maintain relatively thin ice sheets of low volume, which respond linearly to the dominant (about 21and 41 ky) orbital factoring (Clark & Pollard, 1998). Glacial erosion of the regolith and the exposure of the crystalline bedrock that resulted, therefore, may have allowed the thickness of the ice sheet and the depression of the bedrock to become large enough to introduce mechanisms that are responsible for the dominant nonlinear, about 100 ky, response to orbital forcing over the past 1.2 My (Clark & Pollard, 1998).

The 100 ky Cycle

Changes of ice volume show a linear response only to obliquity and precession (Imbrie et al., 1993), though the main periodicities of the δ¹⁸O record of global ice volume (100 ky, 41 ky and 23 ky) are the same as those that dominate orbital insolation changes. The effect of eccentricity variations, about 100 ky and longer, on insolation, in contrast, is to modulate the amplitude of precession variations, about 23 ky and 19 ky, so the resulting 100 ky amplitude in variations in insolation forcing is much too small to explain the large response of ice volume at this period (Imbrie et al., 1993). It is suggested by this that either the response of ice sheets to the orbital forcing is nonlinear or that some climate oscillation that is internal is either phase locked to orbital forcing or its phase is independent (reviewed in (Imbrie et al., 1993) (Muller & MacDonald, 1997)). Whichever is the case, there is no longer direct response of ice sheets to orbital forcing, though through their influence on the climate system they become the primary mechanism that is responsible for driving the 100 ky climate cycle.

The 100 ky cycle is, in most cases asymmetric, with long, about 90 ky, growth phases that are fluctuating and rapid, about 10 ky terminations In most cases the 100 ky cycle is asymmetric, with long about 90 ky, fluctuating growth phases and are rapid, about 10 ky, terminations (Broecker & van Donk, 1970). The large 100 ky ice sheets required some instability to trigger deglaciation (Imbrie et al., 1993; Bond et al., 1993), which contrasts with the smaller ice sheets that prevailed prior to 1.2 Ma and responded linearly to insolation forcing. Precession and obliquity forcing are suggested by many model results that cause ice sheets to grow to some critical size beyond which they stop responding linearly to orbital forcing; deglaciation then occurs through nonlinear interactions between the ice sheets, oceans and atmosphere. Once some threshold is exceeded (Imbrie et al., 1993) that permits the triggering of deglaciation by the next summer insolation maximum in the Northern Hemisphere, these interactions can develop. It is suggested by the linkage of the 100 ky cycle to that of eccentricity, that eccentricity may play a role in the triggering of deglaciation through its modulation of the precession cycle (Hays, Imbrie & Shackleton, 1984). However, any model of the 100 ky cycle must explain why the ice sheets no longer respond in a completely linear manner to orbital forcing after the transition of the Middle Pleistocene, as well as for the mechanism or mechanisms for rapid deglaciation.

It is indicated by modelling results that that thin ice sheets are maintained by widespread soft beds, which respond linearly to insolation forcing, whereas the growth of thicker ice sheets that require mechanisms of deglaciation that are nonlinear are allowed by widespread hard beds (Clark & Pollard, 1998). However, the Laurentide and Fennoscandian Ice Sheets rested on extensive marginal areas of soft beds when at their maximum extents. The role soft beds may have played in causing terminations has not been explored by any models, though there are several existing models of the 100 ky cycle that require fast ice flow for deglaciation (Hyde & Peltier, 1985; Tarasov & Peltier, 1999), which suggests that soft beds may be involved by enabling fast motion.

Clark et al. evaluated the relation between the timing of the advance of the ice sheet onto the outer soft-bedded zones and 100 ky cycles by the identification of the point on the δ¹⁸O global ice volume (sea level) curve where the Fennoscandian and Laurentide Ice Sheets grow large enough to grow onto marginal regions that are soft bedded. It is suggested by this relation that both ice sheets were caused by orbital forcing to grow to a large size on intermediate hard-bedded regions, which was possibly modulated by an inner core of soft beds (____, 1993). However, the ice sheets advanced only onto the outer zone of soft beds late in the 100 ky glaciation cycle (Mangerud, 1993), after which they were followed by major terminations (I, II, IV, V, and VII). It is consistent with those models of the 100 ky cycle that invoke runaway deglaciations only following the ice sheets attaining a threshold thickness and volume, as it is indicated by this relation that only the largest ice sheets advanced onto soft beds. It is proposed by Clark et al. that growth of 100 ky ice sheets onto the outer soft beds combines with other key feedback processes such as changes in sea level (Imbrie et al., 1993) and glacial isostasy (Peltier, 1998) to cause the abrupt terminations of 100 ky cycles. Clark et al. also suggest soft beds may have been deeply frozen in many areas on the first advance of the ice sheets onto them, though in some areas, such as lake basins that were already in existence may have remained unfrozen from the outset. Once the ice sheets had advanced over the soft beds that were frozen the ice sheets would have been able to maintain steep profiles and high surface elevations due to the long time scales of the response of the permafrost to the insulation that was provided by the overlying ice sheet (order of 10³ to 10⁴ years) (Marshall, 1996). Geothermal heat flow beneath the ice sheets would lead to subsequent thawing of permafrost which would enable rapid discharge of ice to low (warmer) elevations and to the adjacent oceans and lakes, where rapid ablation would occur (Pollard & Thompson, 1997; MacAyeal, 1992; ____, 1993). This may result in a West Antarctic type ice sheet that was dominated by ice steams, which had a time response that was reduced substantially and therefore with the ability to be lowered further and to more rapid deglaciation during the next insolation rise in the Northern Hemisphere that brings warmer summers and retreat of ice.

Millennial time scales

Climate variations on a millennial time scale (10³ years) were not long enough and occurred too frequently to be explained by orbital forcing, though the ice sheets of the Northern Hemisphere clearly have a role that may, in many ways, parallel their role in climate change at orbital time scales. In particular, large, abrupt and millennial-scale climate changes are forced or amplified by ice sheets through the release of freshwater to the North Atlantic, which caused changes in the SSTs and the formation of NADW that were transmitted through the atmosphere and ocean with the signal being amplified and transmitted further regionally and globally by various feedbacks (Rind et al., 1997; Hostetler et al., 1999; Broecker, 1998; Alley et al., 1999).

Climate variability was dominated by 2 modes on a millennial scale during glaciations, Dansgaard-Oeschger (D/O) cycles, that had approximate spacing of 1,500 years, and Heinrich Events, which had by comparison a spacing that was long and variable (10³ to 10⁴ years) (Bond et al., 1993; Bond et al., 1999). The D/O oscillation, an oceanic process, was often triggered by changes in meltwater (Keigwin et al., 1991; Alley & Clark, 1999; Broecker et al., 1989) though possibly also oscillating freely as stochastic variability (Weaver, 1999) or in response to mechanisms of an El Niño-Southern Oscillation type (Cane & Clement, 1999). D/O climate change is centred on the North Atlantic as well as regions that have a strong atmospheric response to changes in the North Atlantic.

Surging of the Laurentide Ice Sheet through the Hudson Strait was involved in most Heinrich events, apparently triggered by D/O cooling (Bond et al., 1993; Bond et al., 1999). During a Heinrich event icebergs released to the North Atlantic caused a near shutdown of the formation of NADW (Keigwin & Lehman, 1994). Heinrich events were transmitted elsewhere through the ocean, as well as atmospheric transmission that were present in D/O oscillations (Weaver, 1999; Broecker, 1998; Alley et al., 1999).

A mechanism for millennium-scale ice sheet behaviour was provided by soft beds. The rapid advance and retreat of surge lobes over marginal areas of soft beds apparently regulated the routing of meltwater from the Laurentide Ice Sheet to the North Atlantic, and some abrupt climate changes were apparently triggered by changes in this routing (MacAyeal, 1992; Broecker et al., 1989; Barber et al., 1989; Clark et al., 1996). Instability of ice dynamics is involved in Heinrich events that are modelled readily by incorporating soft beds in the Hudson Bay and Hudson Strait (____, 1993; Marshall & Clarke, 1997).

Last Glacial Maximum – ice sheets

The LGM that occurred at 21 ka is a critical period for understanding climate dynamics, because of the palaeoclimate data that has provided boundary conditions for climate models and evaluating an evaluation of model performance (COHMAP Members, 1988). There are a number of important issues in the climate of the LGM that have remained unresolved (Bard, 1999), which includes the thickness of the ice sheets in the Northern Hemisphere. Geophysical (Earth) models that incorporate rebound of the crust and relative sea level change (Peltier, 1994; Lambeck, Smither & Johnston, 1998) that is associated with the rebound of the crust, reconstruct ice sheets that are 1,000 to 2,000 m thinner than those reconstructed by ice sheet models, which include only ice flow by internal ice deformation and basal sliding (Tarasov & Peltier, 1999; CLIMAP Project Members, 1981; Huybrechts & T’Siobbel, 1997). A novel proposal that uses an ice rheology that is not standard, and is 20 times as soft at low stresses as in traditional models would produce would reconstruct the thin ice sheets (Tarasov & Peltier, 1999; Peltier, 1998). However, questions are raised by ice-texture data about the applicability of this proposed rheology to ice that is deforming more rapidly that largely controls the form of the ice-sheet (Alley, 1992). Spreading of the ice shelf is described by the accepted rheology (Thomas, 1985), and this solution to the low aspect ratio problem does not yet account for the evidence for basal lubrication in marginal regions of former ice sheets (Boulto0n & Jones, 1979; Clark, 1994).

It is suggested by ice sheet models that introduce the effects of soft beds (92-94) successfully reproducing the low-aspect ratio Laurentide Ice Sheet reconstructed by Earth models that use relative sea level data, those soft beds provide a reasonable mechanism to explain the shape and volume of the ice sheet that is consistent with observations of relative sea level change and other geodynamic considerations. It is shown by sensitivity studies, however, that thick ice cover of Hudson Bay at the LGM as is constructed in Earth models is only possible if soft beds in that region are deactivated, whereas soft beds that underlay the outer periphery of the ice sheet are active (93,934). Moreover, that Laurentide Ice Sheet was caused to change from an asymmetrical, multidomed, low elevation geometry towards a high for of high elevation that was symmetrically domed by a progressive reduction in the effect of these outer soft beds on ice flow. The ice sheets of the LGM may have been thicker and higher than indicated by reconstructions in current Earth models (Mitrovica & Davis, 1996), as there are uncertainties concerning whether isostatic equilibrium had been achieved by that time. The deactivation of soft-bedded areas would have been made possible by ice sheets that were thicker and higher at the LGM, which may have occurred by increasing the fraction of the bed that was frozen (Fisher, Reeh & Langley, 1985; Licciardi et al., 1998, 1998). Reactivating large areas of soft beds, as has been discussed above, may have precipitated thinner ice sheets near the LGM as well as rapid deglaciation that caused the 100 ky cycle.

Discussion

It is demonstrated by climate model simulations and climate records demonstrate how important ice sheets are in modulating the climate variability of the Late Cainozoic directly, and through topographic and ice albedo forcing and indirectly through changes in sea level and the discharge of fresh water. Ice sheets have contributed to (near) synchronisation of interhemispheric climate change. Regional to hemispheric or broader atmospheric responses and, where transmitted through the deep ocean, an antiphase response on and downwind of the South Atlantic, were caused by smaller, faster, changes in ice sheet changes. Clark et al. suggest that interpretation of climate records should be viewed as the superposition of climate variability at these different time scales, particularly during transitions from glacial to interglacial when the changes that are occurring at millennial and orbital time scales are large (Alley & Clark, 1999).

It is demonstrated by long palaeoclimate records that several features of the climate system at low and southern latitudes respond to insolation forcing in northern latitudes, though at an earlier phase than the response to ice sheets in the Northern Hemisphere (Imbrie et al, 1992; Imbrie et al., 1993; Pisias & Mix, 1997; Harris & Mix, 1999). An important question remains as to what extent these early responses may influence ice sheets. Similarly, many parts of the climate system that respond to changes of ice volume also respond to insolation forcing on a regional scale (Clemens, Murray & Prell, 1996; Genthon, 1987; Colman et al., 1995; Morley & Heusser, 1997), and important feedbacks to the growth and decay of ice sheets may also be provided by these regional responses.

The influence of soft beds on the dynamics of ice sheets has been found to be an important concept in understanding the behaviour of the WAIS (Bentley, 1997; Oppenheimer, 1998; Alley et al., 1986; Anandakrishnan et al., 1994; 1998; Bell et al., 1998). It is suggested by geological records that the former Northern Hemisphere ice sheets were also influenced by soft beds, and it is proposed by Clark et al. that ice sheet behaviour that is geologically modulated may explain several issues in climate dynamics of the Late Cainozoic that are long standing. However there remains a number of critical issues that need to be resolved before a full understanding can be attained of this relation between the climate, atmosphere and ocean, ranging from a better understanding of the way in which soft beds interact with ice sheets, to further increase in understanding of how ice sheets, atmosphere and ocean interactions in the long term is influence by the behaviour of ice sheets.

Australian National Parks