Discovery of Till Deposition at the Grounding Line of Whillans Ice Stream

See allHide authors and affiliations

Science  30 Mar 2007:
Vol. 315, Issue 5820, pp. 1835-1838
DOI: 10.1126/science.1138393


We report on the discovery of a grounding-line sedimentary wedge (“till delta”) deposited by Whillans Ice Stream, West Antarctica. Our observation is that grounding-line deposition serves to thicken the ice and stabilize the position of the grounding line. The ice thickness at the grounding line is greater than that of floating ice in hydrostatic equilibrium. Thus, the grounding line will tend to remain in the same location despite changes in sea level (until sea level rises enough to overcome the excess thickness that is due to the wedge). Further, our observation demonstrates the occurrence of rapid subglacial erosion, sediment transport by distributed subglacial till deformation, and grounding-line sedimentation, which have important implications for ice dynamics, numerical modeling of ice flow, and interpretation of the sedimentation record.

Subglacial sediment deformation and transport is an integral part of the glacialsystem, affecting or controlling both short-term (1) and long-term (2) glacier fluctuations through changes in the basal speed of the glaciers. Extensive data sets on sediment transport are available for many modern mountain glaciers (3), allowing assessment of the effects of sediment on ice dynamics. However, knowledge of sediment fluxes from modern ice sheets remains much sketchier. We demonstrate that rapid sedimentation has been occurring at the grounding line of a modern Antarctic ice stream, and, in a companion paper (4), we show that this sedimentation plays an important role in ice-sheet stability. The large sediment fluxes implied by our observations point to distributed deformation over a considerable thickness of subglacial till and thus to a deformation rate that increases with a low power of the basal shear stress.

We conducted radar surveys from the floating Ross Ice Shelf onto the grounded ice of Whillans Ice Stream as part of a larger study of the behavior of the West Antarctic Ice Sheet. We used a pulsed-radar and a resistively loaded dipole antenna, with a center frequency of 2 MHz as the source, and a similar antenna and a storage oscilloscope for receiving the return energy [see (5) for details on the instrument]. Figure 1 is a radargram along the line “ab” (whose location is shown on the inset map of the region); ice flow is from right (kilometer 25) to left (kilometer 0) (with a smaller component of flow out of the page). The main features apparent in the radargram are the strong basal echo and, between kilometer 25 and kilometer 17, a strong second echo indicating the base of a subglacial wedge.

Fig. 1.

Radargram along line ab from grounded ice (right) to floating ice (left). Abscissa is distance in kilometers from the farthest downstream position; ordinate is depth below the WGS-84 geoid. Flow is from right to left on the radargram (with a component out of the page) and from northeast to southwest on the map (inset) showing the study region. The map of Antarctica is oriented with the Greenwich meridian directed upward, and the location of Whillans Ice Stream is indicated by the white dot. On the right of the radargram, a single radar trace (from the location marked with triangles on the radargram) is plotted to illustrate the Ricker-wavelet–like form of the ice-bottom reflection and the phase change of the reflection at the wedge bottom.

Numerous basal crevasses are apparent in the radargram (hyperbolic arrivals associated with the line diffractors of the crevasse shoulders), downflow of the grounding line at kilometer 12. We determine the grounding-line location from the drop in ice surface elevation zs (10 m over a short distance of 3 km) and the coincident slight decrease in ice thickness hi (Fig. 2); the ice-thickness change is insufficient to explain most of the surface drop. This “surface ramp” marks a change in bed slope and a marked increase in basal crevasses. In addition, Global Positioning System (GPS) and strand-crack observations (6) are consistent with our interpretation of the grounding line.

Fig. 2.

Surface elevation zs (upper curve) and bed elevation zb (lower curve) along line ab. Note that the two elevation scales are different and that there is a discontinuity between them. Abscissa is the same as in Fig. 1.

We infer that the subglacial wedge continues downstream of kilometer 17 to the grounding line at kilometer 12. The abrupt termination of radar returns from the base of the subglacial wedge (at kilometer 17) is likely a consequence of seawater infiltration beneath the ice from the upglacier-most basal crevasse at kilometer 17. As sketched in Fig. 3, the observed basal crevasses on our line may be extensions of features opened by tidal flexure of floating ice where the grounding line is embayed nearby; relatively dense seawater injected along such a crevasse during opening would subsequently drain downglacier along the ice-wedge interface above the fresher pore waters of the wedge. The skin depth for seawater is on the order of 10–3 m (7), so even a thin layer would effectively mask deeper reflectors (8). Thus, we suggest that the subglacial wedge is present between kilometer 17 and kilometer 12.

Fig. 3.

Cartoon illustrating the subglacial wedge. (Left) Map view of the region showing the radar line crossing the grounding line from grounded ice to floating ice. (Right) Cross section along the radar line. The wedge is well-imaged for part of its length, but ends abruptly when basal crevasses begin to appear. We suggest that the wedge continues beyond the apparent termination but can only speculate on its extent (illustrated by question marks).

Converting the radar basal-echo time to hi [using a velocity ci = 168 m μs–1 (9)] and subtracting hi from zs (determined by differential GPS measurements), we calculate bed elevation zb (Fig. 2). The bed rises gradually along the wedge and the grounding zone before dropping abruptly (7 m over 2 km, from kilometer 14 to kilometer 12) at the grounding line. The bed drop is coincident with the rapid drop in zs (“surface ramp”) of 10 m.

To characterize the properties of the wedge, we measure both the polarity of the echo and the return power Pr (in dB) [defined as Math, where t is time, f(t) is the echo amplitude, and (t2t1) is the duration of the echo, which is very similar to a three-humped Ricker wavelet (10)] for its top and bottom.

The basal echo is of reversed polarity along the whole line, which indicates that the permittivity of the subglacial material is greater than that of ice (for ice, we use a permittivity value of ϵi = 3.2). The echo from the base of wedge, however, is of normal polarity, which shows that the permittivity of the material beneath the wedge is lower than that of the wedge.

The reflection off the base of the wedge is about 12 dB less than that off the top of the wedge (Fig. 4). The difference in bed return power ΔPr varies almost linearly with wedge thickness. Fitting a straight line to the data results in ΔPr = –6.7Δt – 7.2, where Δt is the two-way travel time through the wedge. Assuming that the power loss at the base of the wedge is independent of distance along the wedge, we find that the attenuation within the wedge is –6.7 ± 4 dB μs–1 and that the power loss at the base is –7.2 ± 2 dB (where the uncertainties correspond to one SD).

Fig. 4.

(Top) ΔPr between the top and bottom of the subglacial wedge plotted against Δt. A linear fit through the points is shown. (Bottom) Plot of the residuals between the data and the fit.

We reject the hypothesis that the wedge consists of free water because the hydrologic potential H = ρwgzb + ρig(zszb) decreases notably along the wedge (where ρi and ρw are ice and water densities, respectively, and g is gravitational acceleration). For a water body that is this large [i.e., many ice thicknesses along flow and at least a few ice thicknesses across flow (11)], the hydrologic potential would be constant. In addition, both salt water and ground-water have high electrical conductivity that would lead to a much greater attenuation than that observed. We reject the hypothesis that the wedge is basal accretion ice because of the large absolute reflection coefficient at the top of the wedge (12), which is only slightly lower than the reflection from the ice shelf. Finally, we reject the possibility that this feature is an artifact such as sideswipe (13), because the reflection polarity is opposite to that of the primary bed reflection; sideswipe would have the same polarity as the bed.

In Table 1, we estimate the attenuation for sediments with varying porosity. We include fresh water and seawater for reference. We also calculate the reflection coefficient for the assumed wedge fill over a rock with ϵ′ = 6.6, where ϵ′ is the real part of the electrical permittivity [the “unfrozen bedrock” case of Peters et al. (14)].

Table 1.

Calculated attenuations [exp(–2πΔtf tan δ) in dB μs–1 for radar frequency f = 2 MHz, where tan δ is the complex part of the electrical permittivity and is related to absorption] and reflectivity at the base of different wedge fill materials. fw, fresh water; gw, groundwater. The complex permittivity is given by ϵ = ϵ′(1 – i tan δ), where i is the square root of –1; we determine ϵ′ and tan δ from equations 24 and 25 of (29). This range encompasses reasonable glacial till values (indicated by the “a” superscript in the tan δ column). We assume that the material below the wedge is unfrozen bedrock with ϵ′ = 6.6.

Wedge fillPermittivity ϵ′tan δAttenuation (dB μs-1)Reflectivity (dB)
Unfrozen till (45% fw) 30 0.01 to 0.1a -1 to -5 -10
Unfrozen till (25% fw) 18 0.01 to 0.1a -1 to -5 -10
Unfrozen till (15% fw) 12 0.01 to 0.1a -1 to -5 -9
Unfrozen till (40% gw) 18 0.82 -89 -6
Fresh water 80 0.002 -1 -2.5
Groundwater 80 1.4 -152 -2.5
Seawater 77 11.3 103 -2.6

The measured attenuation data, ΔPr, are best explained by a wedge fill consisting of an unfrozen till layer with low-electrical-conductivity (almost fresh) water in the pores. We cannot distinguish between low- and high-porosity sediment with this technique, but borehole measurements farther upstream on this glacier have found that porosity ϕ ≈ 45% (15, 16). For the case of a sedimentary fill, the radar wave speed is Math for ϵ′ = 18 (unfrozen till; Table 1), and the maximum thickness of the wedge is 31 m (where c0 = 300 m μs–1 is the speed of light in a vacuum). The imaged volume of the wedge is on the order of 105 m3 per unit width, with perhaps a similar volume obscured by seawater down-glacier of kilometer 17, including material inferred beyond the grounding line. The grounding-line position has probably been stable near the present position for a millennium (17). Hence, deposition of the wedge has resulted from a sediment flux on the order of 102 m3 m–1 a–1.

The wedge that we observe matches closely, in size, shape, and setting, with the numerous wedges on the floor of the Ross Sea beyond the Ross Ice Shelf (18, 19). Those wedges were deposited at the edge of the continental shelf at the Last glacial maximum and during retreat across the continental shelf. Those relict wedges are typically tens of kilometers long and tens of meters high. The wedges occur in troughs that were occupied by ice-age extensions of the modern ice streams, with a trough-mouth wedge and one to three retreat wedges along each trough. (No information is available on the possible occurrence of additional retreat wedges beneath the floating Ross Ice Shelf.) Available sampling indicates that the wedges in the Ross Sea are composed of diamicton, which can be distinguished from water-washed sub–ice-shelf sediments (20), and that “Groundingzone wedges contain the sediment that was transported within the subglacial deforming till layer” (19).

Megascale glacial lineations on the upglacier sides of the Ross Sea wedges show that ice streaming persisted during deposition and that the wedges exerted drag on the flowing ice. The wedges sit on the Ross Sea Unconformity that eroded during glacial advance and thus were deposited during the glacial maximum (shelf-edge) and retreat of the ice. Any wedges associated with the glacial advance or with earlier glacial cycles have been extensively modified or removed and no longer preserve their topographic form. Similarly, upglacier of our field site, prograding beds are observed that are likely to have originated from an earlier grounding-line wedge but are without preserved topographic expression (21).

Our modern observations, together with the observations of deglacial features, provide several insights to ice-sheet flow and the coupled sedimentary system. The modern grounding line, just beyond the crest of its wedge, occurs where the bed falls away rather than where the ice thins enough to allow its flotation over a nearly horizontal bed. The ice just upglacier of the grounding line is substantially thicker than that needed to allow flotation, owing to the restraint from friction with the wedge. We show in a companion paper (4) that this is a necessary result of sub–ice-shelf sedimentation and serves to stabilize the grounding-line position. This phenomenon also allows deposition of a discrete wedge rather than the spreading of deposits more uniformly over a broader area.

The implied sedimentation rate for the modern wedge requires long-term erosion rates beneath Whillans Ice Stream and its catchment of just over 0.1 mm a–1, substantially smaller than those for actively eroding mountain glaciers but still notable as compared to many subaerial situations (22, 23). Erosion is likely concentrated in restricted places where basal melting occurs over poorly consolidated sedimentary substrates. Erosion in these regions then would be sufficiently rapid to matter to ice-sheet evolution over multiple ice-age cycles.

The sediment flux implied by our observations likely points to transport dominated by distributed deformation over a considerable thickness (tens of centimeters or more) of subglacial till, averaged over the past millennium. Steady-state meltwater fluxes for the West Antarctic Ice Sheet are incapable of transporting such large quantities of sediment. Dilute debris likely exists in a basal ice layer ∼10 m thick, frozen onto the base of the ice stream (15) from a through-going water system. Debris concentrations are poorly known in this layer but are probably on the order of 3% or less, based on limiting values reported in (24) for presumably comparable ice beneath the neighboring Kamb Ice Stream. The grounding-zone melt rate in this area has been estimated as ∼0.05 m a–1 (25), based on modeling that assumed deeper water than what is locally present, which likely overestimates the heat transport and melt rate. The modeled melt rate would give an estimated sediment flux to the wedge (from the melting off of the debris in the ice) that is nearly an order of magnitude too small to explain our observed deposit.

Existence of subglacial till deformation is supported by two observations beneath West Antarctic ice streams (15, 26) that suggest a flux of sediment on the order of 150 m3 m–1 a–1, though with considerable scatter in the results. Our observation is consistent with these measurements. Finally, the depth of deformation in subglacial till may provide insight to the “flow law” relating strain rate to stress in the material. Sustained high stress is expected to collapse deformation to a surface (27). Considerations of measured profiles of sediment properties with depth indicate that the deformation surface would be at or very close to the base of the ice, where ice velocity would be largely decoupled from the stress supported on the basal sediment (Coulomb-plastic behavior) (27). However, processes perhaps linked to dilatancy and the non-steady response to tidal forcing can allow more-distributed deformation and strain rate increasing as a low power of stress (27, 28). Hence, the deposit that we observe and the older deposits on the Ross Sea floor point to subglacial till deformation distributed across a considerable depth and thus to a low stress exponent in the sediment flow law.

Our observations overall show that the ice-sheet flow and sedimentary processes indicated by very-short-term measurements under ice streams (15), and by geological observations of deglaciated regions (19), have continued over intermediate times. The sediment transport is intimately linked to the lubrication allowing ice-stream persistence. The sediment deposition serves to stabilize the grounding line, which has numerous implications (4).

Supporting Online Material

Fig. S1

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article