Sediments

Debris distribution on the surface of the ice shelf was mapped from Terra ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) imagery and aerial photography interpretation (Fig. 1a). Field observations of sediment–landform assemblages and their relationship to the ice shelf were made in January 2004 to indicate debris provenance, debris-transport history and the nature of debris modification during transport by the ice shelf. Sedimentary facies were described according to texture, particle size and clast-shape analyzes. Poorly sorted sediments were classified in the field using the Reference Hambrey, Glasser and MiddletonHambrey and Glasser (2003) modification of the Reference MoncrieffMoncrieff (1989) classification. Clast morphology was analyzed for samples of 50 clasts, including clast roundness on a modified Reference PowersPowers (1953) scale and the measurement of a, b and c axes for clast shape. These data were analyzed using the approach of Reference Benn and BallantyneBenn and Ballantyne (1994), in which the RA index (% of angular and very angular clasts) is plotted against the C40 index (% of clasts with c/a axial ratio ≤ 0.4) on a covariant plot. This method has been shown to provide good discrimination between glacial facies in high-latitude environments (Reference Bennett, Hambrey and HuddartBennett and others, 1997). Clast lithological analyzis is based on field identification of hand specimens.

Radar data

Subsurface sediment–structure relationships and ice depths were analyzed using GPR data collected during November and December 2003, using a Sensors and Software Inc. pulseEKKO 100 GPR system with 100MHz antennae. Data were collected using the ‘stop and collect’ constant-offset technique, in which antennae were fixed 1 m apart and data collected using a 0.25 m step size. Antennae were orientated at right angles to the direction of data collection, and each trace is the average of 32 stacks. Post-processing and migration of the profiles was carried out using Win EKKO software supplied by Sensors and Software Inc. The final profiles were plotted with Kingdom Suite by Seismic Micro-Technology Inc. A high-pass filter was applied to remove the decaying transient component at the start of the trace (i.e. dewow), and a bandpass filter was applied to remove noise. Automatic gain control was also applied. A radio-wave velocity (RWV) of 0.1783 m ns^–1, derived from central midpoint surveys, was used to convert two-way travel time to depth. The GPR measurements were all made at locations in the ablation zone with exposed ice at the surface, so no correction was made for the presence of firn that may cause spatially variable RWVs in the accumulation area. Thickness measurements with the pulseEKKO system are estimated to be accurate to ±0.25 m.

Point measurements of ice thickness were also made at locations 7 and 17 (Fig. 1) with a monopulse radar transmitter based on the design of Reference Narod and ClarkeNarod and Clarke (1994), and manufactured by Icefield Instruments Inc. This was operated at a centre frequency of 10 MHz, with the antennae parallel to each other (H-plane) and separated by a horizontal distance of 10 m. The received traces were recorded with an airwave-triggered Tektronix 720A digital oscilloscope connected to a palmtop computer. Accuracy of this measurement is estimated to be ±1.7 m. The location and topography of these points and the GPR profiles were calculated from post-processed global positioning system (GPS) measurements.

Ablation

Surface ablation was measured in the field by repeat measurements of PVC poles inserted 1 m into the ice shelf at eight exposed ice locations across the study area (Fig. 1a). The poles were established in November 2003 with a Kovacs ice drill, and remeasured in December 2003 and January 2004. The measurement period includes all but three positive degree-days during the 2003/04 summer season, as measured by an automatic weather station located on the ice shelf approximately 250 m southeast of Bratina Island. To calculate water equivalents, ablation measurements are multiplied by 0.917, the average density (g cm^–3) of solid ice.

Velocity

Surface velocities were measured by recording repeat positions of 18 PVC poles (Fig. 1b and c) with a survey- grade Trimble 4700 GPS receiver in summer 2003/04 and summer 2004/05 (i.e. over a period of 296–358 days). The GPS receiver was run for at least 30 min at each stake on each visit, and the data were differentially corrected using a temporary GPS base station on Bratina Island and the permanent base station at McMurdo (MCM4). Errors are estimated at <30 mm in the horizontal direction. Velocities were also calculated over shorter periods during the summer, but are not displayed here as there was no significant temporal variability in surface motion, and the errors are reduced with the longer period between measurements.

Black island medial moraine area

Sediment–landform assemblage 1: the Black Island medial moraine (BIMM)

Description: The BIMM comprises a linear chain of large (~5 m high) ice-cored debris cones and ridges separated by intervening meltwater ponds. Individual debris cones and ridges are texturally variable but are typically composed of sandy gravel, sandy pebble-gravel or sandy cobble-gravel (Fig. 2). Gravel-sized clasts in these facies are predominantly sub-angular and sub-rounded with occasional angular and rounded clasts, of mixed lithology (although dominated by locally derived fine-grained volcanics) and do not bear striations or facets (Figs 3 and 4). A clast-rich sandy diamicton occurs within the middle of the moraine, forming a ridge ~200 m long and ~3 m wide (Fig. 2b). Gravel-sized clasts in the diamicton are angular, sub-angular and subrounded, of mixed lithology with occasional (6%) faceted clasts (Fig. 3). Striated boulders were also observed in the diamicton. The remains of numerous marine fauna are found in this moraine, including fish (Trematomus hansoni), bryozoan (Hornera sp.) sponges (Cinachyra antarctica), starfish (Echinodermata) and shells (Antarctic barnacle and Antarctic scallop A. colbecki). Field observations indicate that sediment in and around the BIMM sits directly on clean debris-free ice on the ice-shelf surface (Fig. 2c and d).

Fig. 2. GPR profile from across BIMM, with interpretation and photographs of associated lithofacies in the BIMM sediment–landform assemblages. The GPR profile is presented as both unmigrated and migrated data. Photographs of sediments show: (a) Sandy gravel on ~5 m high ice-cored debris cone. Figure on snowpatch indicates scale. (b) Clast-rich sandy diamicton forming ~1 m high debris ridge. Ice axe gives scale. (c) Thin (~50 mm) cover of sand and granule gravel overlying bare ice. Pen gives scale. (d) Patches of well-sorted sand on the ice-shelf surface immediately adjacent to BIMM. The sand was deposited by a large surface meltwater stream flowing to the left of the photograph. Image is ~2 m across.

Fig. 3. Clast roundness histograms (modified Powers roundness) for lithofacies on the BIMM, Bratina Island and the surface of the McMurdo Ice Shelf. Roundness categories are VA – very angular, A – angular, SA – sub-angular, SR – sub-rounded, R – rounded and WR – well rounded. Lithology categories are: FGV – fine-grained volcanics, G – granite, TMM – Transantarctic Mountains metasediments, VL – vesicular lava, VQ – vein quartz. Approximate locations of samples are noted on Figures 2 and 6.

Fig. 4. Covariant plot of the RA index (% of angular and very angular clasts) against the C40 index (% of clasts with c/a axial ratio ≤ 0.4) for lithofacies on and around the McMurdo Ice Shelf. Each symbol represents a sample of 50 clasts. The facies envelopes are based on data presented by Reference Benn and BallantyneBenn and Ballantyne (1994) and Reference Bennett, Hambrey and HuddartBennett and others (1997). The moraine complex envelope contains 44 samples of clasts in diamicton emplaced on land by the 1948 surge or rapid advance of the Kongsvegen/Kronebreen tidewater glacier in Svalbard (Reference Bennett, Hambrey, Huddart, Glasser and CrawfordBennett and others, 1999).

Interpretation: The degree of sorting of sediments in the BIMM indicates that this material has undergone active glacial transport. The presence of high percentages of sub-angular and sub-rounded clasts, together with striated and faceted clasts in the diamicton indicates material originally modified by crushing and abrasion during transport at the ice/bed interface (Reference BoultonBoulton, 1978) or along the side-walls of tributary valleys and nunataks (Reference Evans and CofaighEvans and Cofaigh, 2003). The presence of marine faunal remains in the moraine also indicates a basal origin.

Our preferred interpretation of the origin of this feature is that it represents basally derived supraglacial material forming a medial moraine. The orientation of the medial moraine differs slightly from the orientation of measured surface velocities, suggesting that this feature may be a former flow-unit boundary which is now being slowly reorientated under the contemporary flow regime. This material probably appears on the ice surface as a result of the high surface ablation rates on the ice shelf (see below) and possibly also as a result of upward flowlines in the flow confluence around Black Island (cf. Reference Gomez and SmallGomez and Small, 1985). This interpretation fits with the published ¹⁴C dates of Reference Kellogg, Kellogg, Stuiver and BentleyKellogg and others (1990), which indicate that marine shells in the debris increase in age with distance from Black Island. Reference Kellogg, Kellogg, Stuiver and BentleyKellogg and others (1990) argued that the progression in age of the shells along the BIMM traces the retreat of the ice- shelf grounding line through Holocene times, although the progression of dates could equally be interpreted as dating the time when fauna were entrained from shore debris on Black Island.

Bratina Island area

Sediment–landform assemblage 3: Bratina Island moraines

Description: Wrapped around the lower slopes of Bratina Island moraine is a group of large sub-horizontal curvilinear moraines (Fig. 5). The moraines are continuous over distances of >2 km. Nearest the contemporary McMurdo Ice Shelf they are 2–10 m in height, with a single sharp crest. Higher on the island the ridges are lower (~5m in height) and more subdued. Individual ridges vary in their composition, but the majority are composed of sandy gravel. Gravel-sized clasts in the sandy gravel are mainly subrounded with occasional sub-angular and rounded clasts, of mixed lithology and do not bear striations or facets (Figs 3 and 4). The moraines are notable for their high proportion of exotic pink and white granite clasts (typically 30–40% in samples of 50) which give these deposits a distinctive colour. Shells and shell fragments are also common. Below the elevation of the moraines, Bratina Island is separated from the contemporary McMurdo Ice Shelf by a tide crack varying in width from ~2 to 15 m, along which is exposed a clast-poor to clast-rich muddy diamicton containing subrounded and sub-angular clasts of mixed lithology (Fig. 6). No striated and faceted clasts occur in this lithofacies, which is generally capped by a thin (millimetre-scale) white salt crust.

Fig. 5. Oblique aerial photograph of Bratina Island (B.I.) showing ice-cored ridges and cones of debris on the contemporary ice-shelf surface, ice-shelf moraines and clean ice areas.

Fig. 6. GPR profile from Bratina Island out onto ice shelf, interpretation and photographs of associated lithofacies and sediment–landform assemblages. The GPR profile is presented as both unmigrated and migrated data. Sediments show: (a) Bratina Island moraines. Figure (top right) gives scale. (b) Tide crack between the ice shelf and Bratina Island. Figure (top centre) gives scale. (c) Aerial view of ice-cored ridges and cones on the surface of the ice shelf.

Interpretation: Sub-horizontal single- or multiple-crested ridges of variable composition (ranging from coarse angular debris to muddy diamicton) have been described from the George VI Sound on the Antarctic Peninsula, where they were interpreted as ice-shelf moraines (Reference Sugden and ClappertonSugden and Clapperton, 1981). The Bratina Island moraines are interpreted here as accumulations of sediment formed where the contemporary (but vertically more extensive) McMurdo Ice Shelf abuts the land. The sharp-crested ridges probably retain an ice core, whereas the subdued ridges represent a later stage in the degradation of the ice cores. The Bratina Island moraines are very close in elevation (within 15–20 m) to the surface height of the contemporary ice shelf, suggesting that they were formed relatively recently (probably within the last century). The diamicton exposed in the tide cracks is interpreted as material elevated by upwardly directed flowlines, in response either to the high surface ablation rates or to longitudinal shortening where the ice shelf meets the topographic barrier created by Bratina Island.

Sediment–landform assemblage 4: ice-cored ridges and cones

Description: The ice-cored ridges and cones surrounding Bratina Island are individual mounds of sandy gravel, gravelly sand, sandy pebble-gravel or diamicton (Figs 6 and 7), often containing scattered fossil marine shells and contemporary faunal remains. Individual mounds are separated by large (100 m scale) meltwater ponds. Active failures and exposures in the walls of the mounds reveal bare ice beneath and show that the ice-shelf surface sediments are generally thin (~0.1 m), attaining a maximum thickness of 0.5 m. Debris on the ice-shelf surface rests directly on clean debris-free ice, and sandwiching of sediment between ice layers is common (Fig. 7a and b). Gravel-sized clasts are predominantly sub-angular and subrounded with occasional angular and rounded clasts, of mixed lithology (although dominated by fine-grained volcanics), and do not bear striations or facets (Figs 3 and 4).

Fig. 7. Field photographs and graphic logs illustrating the nature of ice–debris relationships on the McMurdo Ice Shelf. (a) Exposure of ice-shelf ice with thin cover of gravel and showing evidence of contemporary downwasting. (b) Refrozen meltwater pond showing small-scale reworking of sediments on the ice-shelf surface.

Interpretation: There is no obvious source area for the large volumes of supraglacial material present on the contemporary ice-shelf surface. The clast roundness and lithology (including the presence of Transantarctic Mountain metasedimentary rocks) indicates that this material originated outside the current extent of the ice shelf and cannot be accounted for simply by rockfall onto the ice-shelf surface in the local area. We therefore suggest that this material was transported into the McMurdo Sound area by a larger and more dynamically active ice sheet or ice shelf. Individual ridges and cones may be inherited from pre-existing glaciological structures; differential ablation concentrates debris on the ice surface. The large-scale patterns of debris distribution, visible on satellite images and aerial photography (e.g. Figs 1 and 5), strongly suggest that the ridges and cones are attributable to ice-deformation processes, which are now being affected by ablation processes.

This debris may therefore represent material originally subglacial or englacial in origin, which has been elevated to the ice surface by ice-deformation processes (e.g. longitudinal and transverse shortening) where the ice flowed into this topographic embayment. Evidence for this statement includes the nature of the folds visible on the ice-shelf surface from aerial photography and satellite imagery, which shows tight to isoclinal folds, with large parasitic folds still attached to their limbs. Some smaller-scale folds (e.g. those within the Swirls) are no longer attached to their limbs and hence are rootless. Isolated chevron folds also occur outside the Swirls area. The simplest explanation for these folds is that they represent debris bands that were originally continuous, forming approximately linear features that have been deformed by ice flowing generally northward between Black Island and Brown Peninsula but at several different internal velocities. Clast roundness and clast surface feature data also indicate basal transport. This material is effectively stranded on the surface of the contemporary McMurdo Ice Shelf, becoming increasingly concentrated on the ice-shelf surface through time by surface ablation.

To summarize, the ice-cored ridges and cones on the ice-shelf surface are interpreted as downwasting debris originally transported into the area by a larger and more dynamically active ice sheet or ice shelf. Carbon-14 dates of biotic remains in this debris and δ¹⁸O values of surface ice indicate that some of the debris has been present on the ice surface since around 20 kyr BP (Reference Kellogg, Kellogg, Stuiver and BentleyKellogg and others, 1990). The presence of contemporary corals and sponges suggests that sediment is also being continually fed to the ice-shelf surface wherever the underlying seabed is shallow enough that anchor ice is forming and attaching to the base of the ice shelf. Thus, even under contemporary flow regimes, sediment is added to the ice shelf.

Ablation, velocity and ice thickness in the western McMurdo Ice Shelf

The results of the ablation measurements are shown in Table 2. Between 43 and 441mmw.e. of ablation was measured in this area of the McMurdo Ice Shelf during the 2003/04 summer field season. Ablation is highest in the area around Bratina Island. Ablation rates vary according to the albedo of the surrounding ice; for example location 16 is within a debris-covered moraine and has nearly three times more ablation than the adjacent location 17, which is positioned on debris-free ice. Measurement points near Bratina Island (locations 1–8), which are situated on or near to debris-covered ice, have the highest ablation rates. Surface melt was accommodated by surface streams, which were observed to drain into tide cracks (Bratina Island) or flow in a northerly direction towards the ice-shelf edge (BIMM area).

Table 2. Stake coordinates, surface velocity, ice depth and ablation data from the western part of the McMurdo Ice Shelf. Blank cells indicate data not available. See Figure 1 for stake locations

There is a general northward movement of ice towards the seaward margin of the ice shelf, with the lowest velocities measured around Bratina Island (Table 2). Velocities measured on this part of the ice shelf are far lower than velocities recorded on other Antarctic ice shelves such as the Larsen Ice Shelf (Reference Rack, Rott, Siegel and SkvarcaRack and others, 2000) and the Filchner– Ronne Ice Shelf (Reference Rignot and MacAyealRignot and MacAyeal, 1998). Ice thicknesses measured range from 6 to 22 m (Table 2).

Interpretation: Ablation rates in this area reflect local variations in albedo, which are primarily a function of debris cover. Ice within debris-covered areas has a higher ablation rate than debris-free ice. In addition, the large amount of debris-covered ice around Bratina Island, and the albedo of Bratina Island itself may create a local microclimate. For example, a comparison of temperature data collected on debris-covered ice near Bratina Island with those recorded by an automatic weather station at Pegasus South (~24 km away near Pegasus Runway, positioned on permanently snow-covered ice shelf) revealed that temperatures were, on average, 1.7°C warmer near Bratina Island during the 2003/04 summer. As demonstrated at location 4, localized ice collapse and calving into surface melt ponds also plays a role in the net loss of ice in this area. Ablation measured on the McMurdo Ice Shelf is high when compared to the 109mmw.e. a^–1 calculated for Larsen Ice Shelf at Larsen AP meteorological station (66.9° S) by Reference Shepherd, Wingham, Payne and SkvarcaShepherd and others (2003).

The presence of large meltwater streams on the surface of the ice shelf supports these ablation data. Large meltwater streams attributed to high melt rates on ice shelves have previously only been described from the Antarctic Peninsula (e.g. Reference MercerMercer, 1978; Reference Rott, Rack, Nagler and SkvarcaRott and others, 1998; Reference Scambos, Hulbe, Fahnestock and BohlanderScambos and others, 2000).

A conceptual model for supraglacial ice-shelf transport

The distribution of landforms and sediments on the ice shelf and adjacent land masses can be used to propose a conceptual model of debris transport and deposition for the McMurdo Ice Shelf (Fig. 8). Land surfaces surrounding the ice shelf are mantled with deposits containing abundant erratic material, which relate to a vertically and horizontally expanded ice-sheet/ice-shelf system at the Last Glacial Maximum (LGM). Horizontal and vertical recession from this maximum is recorded by a series of ice-shelf moraines constructed against these adjacent land surfaces. Sediment on the contemporary ice-shelf surface is a mixture of relict and active debris-transport processes.

Fig. 8. Schematic cross-section through the ice shelf, indicating processes responsible for debris transport and deposition.

Spatially extensive areas of debris on the ice-shelf surface represent the debris load of a grounded ice-sheet/ice-shelf system at the LGM, now effectively stranded on the ice-shelf surface in an area of low contemporary ice-shelf velocity. The low surface velocities, coupled with high surface ablation rates, mean that this debris was not carried to the ice-shelf edge and lost to calving processes. This represents a process by which ice shelves can maintain and sequester a debris load for thousands of years, potentially providing a mechanism by which ice shelves could accumulate sufficient debris to contribute to Heinrich events (Reference Hulbe, MacAyeal, Denton, Kleman and LowellHulbe and others, 2004). Under contemporary conditions, this debris is locally redistributed by surface ablation (creating an ice- cored landscape characterized by large debris-rich mounds and cycles of ice-core degradation and freezing), ice-shelf flow (forming medial moraines) and meltwater streams (locally reworking supraglacial material and redistributing it across the ice-shelf surface).

Angular supraglacial debris, representing point-sourced rockfall onto the ice-shelf surface and passive supraglacial transport, is generally absent, and most of the surface debris displays evidence of active transport, in the form of subtle basal glacial or fluvial modification. Once released from the ice shelf, either by transport to the ice-shelf edge or by catastrophic break-up of the ice shelf, this debris is a potential source of coarse-grained sediment to the sea floor. Episodic release of this coarse-grained debris by ice shelves could therefore be misinterpreted in the geological record as evidence for palaeo-grounding-line advances, since coarse-grained sedimentation is most commonly associated with sedimentary processes at, or close to, the grounding line of Antarctic glaciers and ice shelves (Reference Anderson, Kennedy, Smith, Domack, Anderson and AshleyAnderson and others, 1991; Reference Domack, Jacobson, Shipp and AndersonDomack and others, 1999; Reference Evans and PudseyEvans and Pudsey, 2002). Our conceptual model provides a framework for supraglacial ice-shelf debris transport that can be tested against the sedimentary record on the Antarctic continental shelf, in other contemporary ice-shelf settings and in the geological record.

Contemporary ice-dynamics, mass-balance and debris-entrainment mechanisms

Our data and observations from this area of the McMurdo Ice Shelf indicate that in view of its relatively thin ice, high ablation rates and slow velocities, the area is potentially unstable. In particular, low velocities in the ablation area may indicate that basal freezing is an important input to the mass balance of this part of the ice shelf. The spatial variability of basal freezing may be the controlling factor in mass balance and ice motion in this area, particularly around very slowly moving areas such as Bratina Island. This is in contrast to the dominance of surface inputs and outputs found in a typical glacial mass-balance system. The δ¹⁸O data collected on the ice surface in the ablation area of the ice shelf indicate that the ice here is of marine origin (Reference Kellogg, Kellogg, Stuiver and BentleyKellogg and others, 1990), whilst near the northern edge of the ice shelf, ice cores show the shelf to be composed of fresh-water ice, formed from surface meltwater refreezing at the base of the ice shelf after it has drained down tide cracks (Reference Gow, Weeks, Hendrickson and RowlandGow and others, 1965).

These observations suggest that the ice shelf is sustained by freezing-on of marine and fresh water under the ice shelf. Basal ice formation has been observed at other ice shelves. The bottommost 6 m of a 416 m core from the Ross Ice Shelf (Reference Zotikov, Zagorodnov and RaikovskyZotikov and others, 1980), 45 m of a 315 m core from the Amery Ice Shelf (Reference MorganMorgan, 1972), and 62 m of a 215 m core from site B13 and 167 m of a 320 m core from site B15 on the Filchner–Ronne Ice Shelf (Reference OerterOerter and others, 1992, Reference Oerter, Eicken, Kipfstuhl, Miller and Graf1994) are comprised of ice formed by basal freezing. Model calculations indicate that, in favourable conditions, up to 4 m of marine ice can form per year at the base of ice shelves (Reference Smedsrud and JenkinsSmedsrud and Jenkins, 2004).

Sub-ice-shelf accumulation is balanced by ablation on the ice-shelf surface, which leads to the concentration of debris on its upper surface. Our observation that surface debris sits directly on the ice shelf and has no connection to underlying ice structures confirms this. Measurements of basal melting and freezing rates in this area are essential to ascertain the potential future stability of this part of the McMurdo Ice Shelf.

The BIMM may represent debris accumulated along a former flow-unit boundary which, under the contemporary flow regime, is now being redistributed across the ice-shelf surface by meltwater streams and ponds. Large areas of surface debris in the Swirls area between Brown Peninsula and Black Island (Fig. 1) have no source area on the ice shelf and must represent deformation of pre-existing surface material that cannot be related to contemporary flow. On the basis of ¹⁴ C dates, Reference Kellogg, Kellogg, Stuiver and BentleyKellogg and others (1990) suggested that this debris was deposited by a grounded ice sheet that flowed westward into the area during the LGM. They argued that since the debris continues onto land, this part of the ice shelf has been stagnant since the LGM. Reference Kellogg, Kellogg, Stuiver and BentleyKellogg and others (1990) therefore suggested that the ice shelf in the Swirls area has only very recently become ungrounded.

Implications for models of ice-sheet/ice-shelf interactions in the Late Pleistocene and Holocene

Our field data can be used to test the two existing models for ice-sheet/ice-shelf interactions in the Ross Sea–McMurdo Sound area in the Late Pleistocene and Holocene. The first of these hypotheses is that at the LGM the area was inundated by a grounded ice sheet that filled the entire Ross Sea embayment (Reference Denton, Bockheim, Wilson and StuiverDenton and others, 1989; Reference Kellogg, Kellogg, Stuiver and BentleyKellogg and others, 1990). In this model, ice flow was from east to west across McMurdo Sound. The second model also invokes grounded ice but suggests that three coexisting ice lobes flowed into the area (Reference WilsonWilson, 2000). These lobes were an expanded and grounded Koettlitz Glacier lobe, a Ross Ice Shelf lobe and a smaller medial lobe fed from grounded ice in the Ross Sea that flowed between Brown Peninsula and Black Island. Our interpretation of the debris left stranded on the ice-shelf surface in the Swirls area is most easily accommodated by the model of Reference WilsonWilson (2000), in which this debris represents the deformed remnants of a set of terminal moraines marking the extent of a tongue of ice that flowed between Brown Peninsula and Black Island. This suggests that the ice shelf has been a permanent feature through the Late Pleistocene and Holocene, although it is likely to have thinned during this time.