An unfrozen unlithified sediment bedglaciological implications

A large proportion of the beds that directly underlay the soles of mid-latitude ice sheets of the last glacial period were composed of thick unlithified sediment sequences rather than rock. Soft beds have been the subject of much research in the past 25 yr since their significance was first recognized. They have proved to be phenomenologically rich, and as a consequence are still a very active source of research and debate. It is thus important to register what is known, what has been speculated and what the principal remaining problems are in determining their role in the traction at the bed of a glacier and in the generation of till. (The term till is frequently misused in the glaciological literature to refer to actively deforming subsole material. The term should be restricted to a deposited sediment, not to sediment in transport. Deforming subsole material is here referred to as deforming sediment, the deforming horizon or the deforming subsole nappe.)

1 On every occasion that boreholes through or excavations beneath modern glaciers have found unlithified sediment immediately below the glacier sole and measurements have been made that are able to detect deformation, they have done so. In the majority of cases, they report the thickness of the deforming horizon to be no greater than about 0.3-0.65 m (Boulton & Hindmarsh, 1987, 0.38-0.45m; Blake, 1992, 0.3 m; Humphrey et al., 1993, 0.65 m; Iverson et al., 1994, 0.35 m;

Table 2.1

Glacier

Debris volume

Debris zone

Potential melt-out

Source

concentration (%)

thickness (m)

till thickness (m)

East Antarctic margin

0-12

15

0-1.8

Yevteyev (1959)

Antarctic Byrd core

7

4.8

0.34

Gow et al. (1979)

Camp Century, Greenland

0.1

15.7

0.016

Herron & Langway (1979)

Nordenskioldbreen Spitsbergen

40

0.4

0.16

Boulton (1970)

Barnes ice cap, Baffin

6-10

8

0.048-0.08

Barnett & Holdsworth (1974)

Breidamerkurjokull, Iceland

50

0.15-0.3

0.075-0.15

Boulton et al. (1974)

Breidamerkurjokull, Iceland

8-10

0.05-0.2

0.004-0.02

Boulton (1979)

Matanuska, Alaska dispersed facies

0.04-8.4

0.2-8

>0.0008

Lawson (1979a)

Matanuska, Alaska stratified facies

0.02-74

3-15

>0.006

Lawson (1979a)

Glacier d'Argentiere, France

43

0.02-0.04

0.009-0.017

Boulton et al. (1979)

Myrdalsjokull, Iceland

15-31

2-5

0.3-1.55

Humlum (1981)

Bondhusbreen, Norway

0.39

5

1.95

Hagen et al. (1983)

Watts, Baffin Island

14-57

0.8-2.9

>0.4

Dowdeswell & Sharp (1986)

Based on data compiled by Kirkbride (1995)

Based on data compiled by Kirkbride (1995)

Figure 2.5 Examples of subglacial shear deformation. (a) Shear fold at the base of a thick till unit at Whitevale, Toronto, in which underlying sandy sediment have been folded into the till as a consequence of a local stress concentration that caused a fold that has subsequently been attenuated by simple shear strain. (b) Accumulation of a series of individual fold units (1-5) in a till near Sangaste, Estonia. The individual folds represent cumulative deposition of successive deforming horizons whose integrity and internal pattern of strain is reflected by individual folded units. Discontinuous shear planes occur between individual units. There is a thin zone of homogenized sediment at the base of the till, but below this, original sedimentary bedding has been relatively little disturbed.

Figure 2.5 Examples of subglacial shear deformation. (a) Shear fold at the base of a thick till unit at Whitevale, Toronto, in which underlying sandy sediment have been folded into the till as a consequence of a local stress concentration that caused a fold that has subsequently been attenuated by simple shear strain. (b) Accumulation of a series of individual fold units (1-5) in a till near Sangaste, Estonia. The individual folds represent cumulative deposition of successive deforming horizons whose integrity and internal pattern of strain is reflected by individual folded units. Discontinuous shear planes occur between individual units. There is a thin zone of homogenized sediment at the base of the till, but below this, original sedimentary bedding has been relatively little disturbed.

Boulton et al., 2001a, generally <0.5 m; Iverson et al., 2003, >0.4m). An exception to this has been the finding of Truffer et al. (2000) who demonstrated that a décollement surface in the deforming bed must lie at a depth of greater than 2 m below the glacier sole.

2 It is common to find shear and drag fold structures reflecting longitudinal shear deformation in sediments deposited beneath former glaciers and ice sheets (Hart, 1995b; Benn & Evans, 1998; Boulton & Dobbie, 1998). Some demonstrate subglacial shear deformation to depths of several metres (see Fig. 2.9). Such deep folding tends, however, to be localized, and may simply reflect local stress concentrations and blocking of shear movement that locally causes deformation to descend to greater than normal depths. It is more common to find structures in which fold packages are much thinner, with thicknesses similar to those of measured active deforming horizons (Fig. 2.5a). It is frequently found that folds in tills comprise a large number of such highly attenuated fold packages and boudins (Benn & Evans, 1996) that appear to have accumulated sequentially one above the other (Fig. 2.5b) rather than representing the 'freezing' of a single deforming horizon.

3 The roughness of the glacier bed is of fundamental importance to the décollement process. The effective roughness of a sediment bed is quite different from that of a rock bed. Whether décollement occurs by ice sliding over its surface or by internal deformation, the roughness on the surface of a shear plane is primarily at the millimetric or submillimetric scale of the grains and occasional metric scale of large clasts (if present) rather than at the 10-100 m roughness scale of smoothly eroded rock beds. As a consequence, the dominant mode of décollement is by ice sliding through regelation (Weertman, 1957) around individual grains, or by deformation within the sediment through grain against grain movement. Plastic flow of ice, that dominates on scales >10 cm to

1 m, will be relatively unimportant, in contrast to bedrock surfaces where it dominates. However, direct studies of the glacier sole in temperate valley glaciers (Kamb & LaChapelle, 1964; Boulton et al., 1979) show, in most of the few cases studied, that the sole consists of a debris-rich horizon in the basal few centimetres, with the glacier sole forming a frozen sediment carpet. It seems in these cases that any sliding between the glacier sole and its bed does not occur at an ice-sediment contact but at a sediment-sediment contact. As in other granular sediments, water pressure will be a fundamental determinant of failure, either at the glacier sole or in the underlying sediments.

4 Fischer & Clarke (1997a) have demonstrated stick-slip behaviour at the base of a glacier in which slip occurs at the glacier sole during periods of the highest water pressures, with décollement being transferred down into the sediment bed as water pressures fall (see also Iverson et al., 2003). Figure 2.6 shows the patterns of cumulative subglacial shear strain in 6-h increments recorded by strain markers (that are also water pressure transducers) in a subglacial sediment (Boulton et al., 2001a). Strain is concentrated at the glacier sole (between transducers at 0 and 0.1) during water pressure peaks and at lower levels (between 0.1 and 0.3 m, or 0.3 and 0.5 m) during periods of lower water pressure. Figure 2.7 suggests how this might arise. Piotrowski & Tulaczyk (1999) and Piotrowski (this volume, Chapter 9) have suggested that there may also be a spatial variation in the partitioning between basal sliding and sediment deformation. Boulton (1987) suggested that sediment deformation would be minimized and friction would be maximized against the up-glacier parts of drumlins, which would be 'sticky spots' (Whillans, 1987) at the glacier bed, with easy deformation in interdrumlin zones.

5 The effective rheological behaviour of sediments deforming beneath a glacier is a matter of considerable debate. Boulton & Jones (1979) assumed a Coulomb failure criterion for mate-

Julian days

Julian days

Distance - metres

Figure 2.6 (a) Six-hourly patterns of longitudinal shear strain measured beneath Breidamerkurjokull, Iceland (Boulton et al., 2001a). Strains are greatest on days 252, 254 and 255, which are also days of relatively high water pressure. Several patterns of strain occur. Most strain appears to be by basal sliding at 12.00 hours on days 252, 254 and 255 (periods of high and increasing water pressures), whereas at 18.00 hours on each of those days most strain appears to occur between 0.1 and 0.3m depth. Significant strain occurs between 0.5 and 1.0 m on days 254 and 255. (b) Progressive net cumulative strain from days 250 to 261. Although detailed short-term patterns vary, as shown in (a), the net effect is a simple pattern, with about half the strain being taken up by sediment deformation and half by basal sliding. At x and y, for example, almost all net strain is by basal sliding.

Distance - metres

Figure 2.6 (a) Six-hourly patterns of longitudinal shear strain measured beneath Breidamerkurjokull, Iceland (Boulton et al., 2001a). Strains are greatest on days 252, 254 and 255, which are also days of relatively high water pressure. Several patterns of strain occur. Most strain appears to be by basal sliding at 12.00 hours on days 252, 254 and 255 (periods of high and increasing water pressures), whereas at 18.00 hours on each of those days most strain appears to occur between 0.1 and 0.3m depth. Significant strain occurs between 0.5 and 1.0 m on days 254 and 255. (b) Progressive net cumulative strain from days 250 to 261. Although detailed short-term patterns vary, as shown in (a), the net effect is a simple pattern, with about half the strain being taken up by sediment deformation and half by basal sliding. At x and y, for example, almost all net strain is by basal sliding.

Stick-slip and water pressure cycle

Water pressure falls. Ice/ bed coupling Increases. Less sliding, more deformation. Water pressure falls further in dilating horizon.

Dilated horizon consolidates, further ice/bed inter-locking. Dilatant shear zone descends.

ooOOO nOOoQ

Dilated horizon consolidates, further ice/bed inter-locking. Dilatant shear zone descends.

Water pressure falls to below critical level for failure. Longitudinal stress builds up, later to be released by sliding.

Sliding

Sediment deformation -

No deformation

Figure 2.7 Suggested explanation of the stick-slip process apparent in Fig. 2.6.

rial in this setting. Boulton & Hindmarsh (1987) found that a non-linearly viscous law or a Bingham solid law would fit seven data points relating effective pressure to shear stress calculated from the average gravitational driving stress. A number of subsequent field experiments (Kamb, 1991; Hooke et al., 1997; Tulaczyk et al., 2000a) and laboratory experiments (Iverson et al., 1998) have demonstrated that deforming sediments (in the former cases) and till (in the latter) show plastic behaviour that can be described using a Coulomb failure criterion. The anomaly is that such a failure criterion, if applied, for example, to the setting shown in Fig. 2.4, where effective pressure is least immediately beneath the glacier sole, predicts failure in a thin shear zone that the experiments of Hooyer & Iverson (2000b) suggest should not be more than 20 mm in thickness, immediately beneath the glacier sole. However, typical measured thicknesses of deforming horizons (1 above) are more than an order of magnitude greater, and the vertical strain profile (Fig. 2.6b) is one characteristic of a viscous material. Clast interlocking (Tulaczyk, 1999) would create a shear zone of 10-15 times clast diameter, but even in most tills interlocking is only commonly likely between millimetric grains. Several suggestions have been made to reconcile these data:

• Hindmarsh (1997) has suggested that a plastic rheology may appear viscous at large scales, although failing to suggest the process by which small-scale plasticity is transformed into large-scale viscosity.

• Boulton & Dobbie (1998) and Iverson & Iverson (2001) have suggested that short-term water pressure fluctuations such as those shown in Fig. 2.4 could produce vertical variations in the location of Coulomb failure so that they aggregate to a time-integrated deformation profile of a viscous form as shown in Fig. 2.6b, as well as the stick-slip behaviour shown in Fig. 2.6a. Iverson & Iverson (2001) have simplified such a cumulative deformation to a law of the form:

where e is the strain rate, A is a constant, P is ice pressure, S is sediment strength, m is the coefficient of internal friction, N is effective pressure and t is shear stress. Figure 2.7 also shows how localized failure and dilation could displace the location of failure without any external changes in water pressures.

• Fowler (2002) has drawn attention to a more fundamental problem: the unconstrained nature of the velocity field in perfectly plastic behaviour. The crucial issue remains therefore: how does a sediment bed generate resistance to glacier flow; what rheology and what flow law should be applied to sediment-floored ice sheets?

It is clear that the subglacial hydraulic regime, and its time dependence, are of fundamental importance to the behaviour of subglacial sediment beds. For a relatively fine-grained material, such as the clay-silt-sand matrix of a till, high porewater pressures that generate very low effective pressures are enough significantly to reduce interlocking and the strength of the till. Soft sediment beds are therefore fundamentally different from rock beds in that sustained high water pressures in them, resulting from poor drainage, can sustain a state in which interlocking is poor, the frictional resistance offered by the bed is perennially less than the yield strength of ice, and low shear stress flow can be sustained over long periods. Several of the active ice streams of the West Antarctic ice sheet appear to reflect this state (e.g. Alley

Figure 2.8 Patterns of head distribution along a flow-parallel section beneath a modelled ice sheet in which a silt lens (k about 10-8ms-1) and a clay lens (k about 10-9ms-1) overlie a thick sandy unit. There is a relatively small horizontal head gradient and insignificant vertical head gradient in the sand because of its high conductivity, but there are strong vertical head gradients in the clay and silt lenses as water drains through them from the glacier into the underlying sandy aquifer. At the ice-bed interface, effective pressures and friction will tend to be high along an ice-sand interface, lower at an ice-silt interface and even lower at an ice-clay interface, with strong areal variation in patterns of shear resistance and the nature of décollement at the ice-bed interface. Thick lines in the clay lens represent heads at 100 kPa intervals.

Figure 2.8 Patterns of head distribution along a flow-parallel section beneath a modelled ice sheet in which a silt lens (k about 10-8ms-1) and a clay lens (k about 10-9ms-1) overlie a thick sandy unit. There is a relatively small horizontal head gradient and insignificant vertical head gradient in the sand because of its high conductivity, but there are strong vertical head gradients in the clay and silt lenses as water drains through them from the glacier into the underlying sandy aquifer. At the ice-bed interface, effective pressures and friction will tend to be high along an ice-sand interface, lower at an ice-silt interface and even lower at an ice-clay interface, with strong areal variation in patterns of shear resistance and the nature of décollement at the ice-bed interface. Thick lines in the clay lens represent heads at 100 kPa intervals.

Distance from ice sheet terminus - km

Distance from ice sheet terminus - km et al., 1987; Whillans & Van der Veen, 1997) and some parts of Pleistocene ice sheets seem to have done so (Mathews, 1974; Boulton & Jones, 1979). Time-dependent variations in water pressure may be the cause of apparently viscous behaviour of subglacial sediments (see Hindmarsh, 1997). It may also be the cause of partitioning of strain between the ice-bed interface and the sediment as a consequence of its effect on the degree of interlocking between the glacier sole and underlying sediments (Iverson et al., 1995). It has been suggested by Piotrowski & Tulaczyk (1999) that a sediment-floored subglacial bed may be a patchwork of zones where slip is concentrated at the ice-bed interface and zones where a larger proportion of the forward movement of the glacier is accounted for by deformation in the sediment. Any such a patchwork is likely to change through time as a consequence of seasonal and diurnal changes, changes in the points of injection of surface water, and changes in the local hydraulic geometry of the bed that will change drainage pathways and effective pressures.

Figure 2.8 illustrates how the friction at the base of a glacier can vary as a consequence of varying geohydrological properties in the bed. The ice load and flux of water at the base of the glacier has been prescribed, as has the conductivity of subglacial beds and the dependence of permeability on effective pressure. The results indicate how fine-grained low-conductivity sediment masses overlying highly conductive strata can locally reduce effective pressures and frictional resistance on the bed.

Was this article helpful?

0 0

Post a comment