Hostname: page-component-cc8bf7c57-hbs24 Total loading time: 0 Render date: 2024-12-09T17:51:41.580Z Has data issue: false hasContentIssue false

Englacial drainage structures in an East Antarctic outlet glacier

Published online by Cambridge University Press:  23 December 2019

Thomas Schaap*
Affiliation:
Discipline of Earth Science, University of Tasmania, Hobart, Tasmania7001, Australia
Michael J. Roach
Affiliation:
Discipline of Earth Science, University of Tasmania, Hobart, Tasmania7001, Australia
Leo E. Peters
Affiliation:
Seismic Research Centre, The University of the West Indies, St. Augustine, Trinidad and Tobago, West Indies
Sue Cook
Affiliation:
Institute for Marine and Antarctic Studies, University of Tasmania, Private Bag 129, Hobart7001, Australia
Bernd Kulessa
Affiliation:
Glaciology Group, College of Science, Swansea University, Singleton Park, SwanseaSA2 8PP, UK School of Technology, Environments and Design, University of Tasmania, Hobart, Tasmania7001, Australia
Christian Schoof
Affiliation:
Department of Earth and Ocean Sciences, University of British Columbia, Vancouver, British Columbia, Canada
*
Author for correspondence: Thomas Schaap, E-mail: thomas.schaap@utas.edu.au
Rights & Permissions [Opens in a new window]

Abstract

Ground-penetrating radar data acquired in the 2016/17 austral summer on Sørsdal Glacier, East Antarctica, provide evidence for meltwater lenses within porous surface ice that are conceptually similar to firn aquifers observed on the Greenland Ice Sheet and the Arctic and Alpine glaciers. These englacial water bodies are associated with a dry relict surface basin and consistent with perennial drainage into an interconnected englacial drainage system, which may explain a large englacial outburst flood observed in satellite imagery in the early 2016/17 melt season. Our observations indicate the rarely-documented presence of an englacial hydrological system in Antarctica, with implications for the storage and routing of surface meltwater. Future work should ascertain the spatial prevalence of such systems around the Antarctic coastline, and identify the degree of surface runoff redistribution and storage in the near surface, to quantify their impact on surface mass balance.

Type
Letter
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
Copyright © The Author(s) 2019

Introduction

Englacial channels represent current and relict flow pathways for meltwater through the body of a glacier (Gulley and others, Reference Gulley, Benn, Screaton and Martin2009). They can provide pathways for surface water to reach the bed of a glacier and can affect its dynamics by altering the subglacial hydrological system (Fountain and Walder, Reference Fountain and Walder1998; Zwally and others, Reference Zwally2002; Schoof, Reference Schoof2010).

Englacial channels have been observed to form by three main mechanisms: (a) ‘cut-and-closure’, (b) exploitation of permeable structures and (c) hydrofracture of crevasses (Gulley and others, Reference Gulley, Benn, Screaton and Martin2009). The cut-and-closure formation mechanism described by Gulley and others (Reference Gulley, Benn, Screaton and Martin2009) occurs where supraglacial streams incise into the ice through melting, and the upper portion of the resulting canyon closes through ice creep. Systems which form by this mechanism are characterised by meandering channels with a low gradient and occasional step-pools. This mechanism requires the downcutting rate of the stream to be higher than the ablation rate of the surrounding ice, with this mechanism being most common in regions of low surface ablation and high stream flow rates (i.e. high surface slope and large catchment areas), and has typically been observed in glaciers at high latitude (Gulley and others, Reference Gulley, Benn, Screaton and Martin2009; Vatne and Irvine-Fynn, Reference Vatne and Irvine-Fynn2016; Benn and others, Reference Benn2017).

Shreve (Reference Shreve1972) proposed a different model involving pre-existing permeable structures, such as veins in the ice, which are exploited and enlarged. However, the primary porosity and hence hydraulic conductivity of glacier ice is extremely low. Secondary porosity created by relatively permeable macro-structures, such as debris layers or fractures, is therefore a more plausible mechanism. These permeable pathways are expanded by the transfer of heat from meltwater passage into conduits. The morphology of these conduits depends on their origin, and conduits which encounter crevasses can have pronounced morphological undulations (Gulley and others, Reference Gulley, Benn, Screaton and Martin2009).

The hydro-fracture model provides an efficient drainage pathway through the entire ice thickness using pre-existing surface crevasses. The depth of surface crevasses is normally limited by the closing action of ice creep. However, if a surface crevasse is water-filled, the additional pressure counteracts the ice overburden pressure and allows the crevasse to deepen. If the surface of the water body contains a sufficient volume, the crack will quickly deepen through the entire ice thickness (Weertman, Reference Weertman1973; van der Veen, Reference van der Veen2007). This mechanism allows the rapid drainage of surface melt lakes in Greenland (Das and others, Reference Das2008; Doyle and others, Reference Doyle2013; Jones and others, Reference Jones, Kulessa, Doyle, Dow and Hubbard2013) and on Antarctic ice shelves (Scambos and others, Reference Scambos2009). Hydro-fracture produces shafts with a plunge angle that is dependent on the local stress regime (Gulley and others, Reference Gulley, Benn, Screaton and Martin2009) and can also lead to a permanent connection between an ice-sheet surface and bed in the form of a moulin (Alley and others, Reference Alley, Dupont, Parizek and Anandakrishnan2005).

Englacial channels have been directly observed and geophysically detected in cold-based (Vatne, Reference Vatne2001), temperate and polythermal (Stuart and others, Reference Stuart, Murray, Gamble, Hayes and Hodson2003; Catania and others, Reference Catania, Neumann and Price2008; Benn and others, Reference Benn2017) glaciers from Greenland, Svalbard and the Himalaya. Englacial meltwater accumulations have also been described in East Antarctica (Lenaerts and others, Reference Lenaerts2016). Ground-penetrating radar (GPR) detection of englacial channels has been noted as far south as King George Island, near the Antarctic Peninsula (Kim and others, Reference Kim, Lee, Hong, Hong and Shon2010; da Rosa and others, Reference da Rosa2014). Few direct and indirect observations of englacial hydrological systems have been reported from Antarctica (e.g. Langley and others, Reference Langley, Leeson, Stokes and Jamieson2016), and GPR observations of near-surface drainage systems in Antarctica have not yet been reported.

Surface meltwater is visible in satellite imagery around much of Antarctica (Kingslake and others, Reference Kingslake, Ely, Das and Bell2017). This meltwater is typically contained in surface melt ponds, or drains supraglacially in streams over the ice-sheet surface. Surface melt ponds have been associated with rapid ice-shelf retreat (e.g. Scambos and others, Reference Scambos2009), with the rapid simultaneous drainage of multiple surface melt ponds being suggested as the trigger for collapse (Banwell and others, Reference Banwell, MacAyeal and Sergienko2013; Macayeal and Sergienko, Reference Macayeal and Sergienko2013). Surface water drainage has also been put forward as an explanation for surface structures known as ‘ice dolines’ (Bindschadler and others, Reference Bindschadler, Scambos, Rott, Skvarca and Vornberger2002). Ice dolines are large, steep-sided depressions that have been observed on the surface of ice shelves, and have been suggested to form as englacial water bodies drain, allowing the surface of the ice to collapse (Mellor, Reference Mellor1960).

Here we present GPR observations that indicate the presence of a near-surface englacial drainage system in the Sørsdal Glacier, Princess Elizabeth Land, East Antarctica, and use forward models of electromagnetic wave propagation to characterise the englacial channel contents.

Study area

Sørsdal Glacier is a major outlet of the East Antarctic Ice Sheet, flowing adjacent to the Vestfold Hills in Princess Elizabeth Land (Fig. 1). The floating section of the glacier is ~20 km long and a maximum of 10 km wide. Sørsdal Glacier experiences significant surface melt in the austral summer, resulting in the seasonal development of surface meltwater ponds.

Fig. 1. Maps showing the location of Sørsdal Glacier and Channel Lake with GPR transects (solid black lines labelled 1 to 5). Base imagery captured 29th March 2017 by Sentinel 2 satellite.

This study focuses on a single surface feature in the ablation zone of Sørsdal Glacier, which we have informally named ‘Channel Lake’. In February 2017, when fieldwork for this study was conducted, Channel Lake was characterised by an elongate, mostly dry surface basin. The surface consisted of solid ice with meltwater puddles scattered across. No snow or firn was observed from the surface or in shallow (<1 m) boreholes drilled with a hand auger. Raised hummocky features were seen on the glacier surface, appearing as a down-glacier continuation of Channel Lake (Fig. 2a). The basin itself was bordered by steep, crevassed ridges, similar to those observed in ice dolines (Bindschadler and others, Reference Bindschadler, Scambos, Rott, Skvarca and Vornberger2002) (Fig. 2b). Channel Lake was located ~22 km up-glacier of the calving front and ~1 km up-glacier of the grounding line. Seismic reflection data (gathered supplementary to this study) from Channel Lake showed the ice was ~1400 m thick. The Channel Lake basin formed a topographic depression ~10 m below the surrounding glacier surface at its deepest point and extended ~4 km along-flow (including associated downstream structures) and ~200 m across. The average speed of the ice in the Channel Lake area is 290 m a−1 calculated using velocity data from MEaSUREs v2 (Rignot and others, Reference Rignot, Mouginot and Scheuchl2017).

Fig. 2. (a) Photograph of Channel Lake taken from a helicopter, looking northeast. Channel Lake and its associated downstream structures are together ~4 km in length; (b) photograph of Channel Lake basin looking east from the southern ridge. Here, the bounding ridges are up to 10 m high and are ~100 m apart. Both photographs were taken during summer 2016/17.

Channel Lake is similar to features that have previously been associated with surface lake drainage in East Antarctica (Langley and others, Reference Langley, Leeson, Stokes and Jamieson2016). In satellite imagery taken from the 2014 and 2015 austral summers, Channel Lake can be seen as a supraglacial lake filled with water (Fig. 3). The area of the water-filled lake was ~0.6 km2 in January 2014, although there is high uncertainty in this estimate due to difficulty in determining lake boundaries given that it was partially ice-covered. This water seems to have disappeared between January 2015 and January 2016, with the January 2016 image showing a dry basin similar to that encountered during our fieldwork in the 2016/17 austral summer. Furthermore, a satellite image from August 2016 shows a long, dark-coloured patch extending down-glacier of Channel Lake (Fig. 3). This dark region may be the manifestation of a refrozen accumulation of meltwater at the surface or near-surface. The drainage of Channel Lake and the emergence of water at the surface some distance down-glacier alludes to the possibility of a near-surface englacial drainage system which transported the water from Channel Lake to its re-emergence downstream. Here we investigate whether the features observed in satellite imagery can be associated with and hydrologically connected to englacial drainage pathways.

Fig. 3. Satellite imagery (Landsat 8 OLI/TIRS C1 Level 1) showing the evolution of Channel Lake on Sørsdal Glacier between 2014 and 2016.

Methods

GPR acquisition

GPR data were collected in February 2017 across five transects: four across the basin feature of Channel Lake, and one downstream on flat ice (Fig. 1). Each transect was repeated with 800 and 250 MHz MALA Ramac antennas connected to a MALA X3Mc control system. Table 1 describes the sampling parameters for each antenna.

Table 1. Sampling parameters for GPR data on Sørsdal Glacier

The GPR data were acquired by manually towing the antennas across the ice surface. The GPR transects were oriented orthogonally to the long axis of Channel Lake, with an approximate north–south orientation, and were numbered 1–5 from east to west; up-glacier to down-glacier (Fig. 1). The surface topography presented a significant challenge in the eastern transects (Lines 1–3) due to steep embankments littered with cracks and crevasses that made it difficult to maintain the antennas parallel to the surface and directly along the profiles at times leading to a minor loss of data on some profiles. The 800 MHz data on Line 5 suffered the most, where a degraded connection to the antenna severely degraded the quality of the data to the point of being unintelligible. Location data were gathered by handheld GPS devices.

GPR processing

We opted for a minimalistic processing flow to minimise the distortion of signal attributes. A dewow filter was applied to each profile to remove inherent electronic noise, and a manual time-varying gain function amplified low-amplitude features at depth. This gain was only applied to help identify features, and was subsequently removed to preserve original signal characteristics for analysis. The profiles were then time-migrated using a well-known phase-shift algorithm (Stolt, Reference Stolt1978) to collapse diffractions and correct for dipping reflectors. In Figure 4, the radargrams along Lines 1–3 are then displayed with a static correction of relative changes in topography, based on in-field GPS observations with an estimated average vertical precision of ±0.5 m. The surface topography of Lines 4 and 5 was flat without noticeable undulations.

Fig. 4. GPR radargrams (locations shown in Fig. 1). Examples of each geometric class are annotated. Note the differences in scale and surface topography.

Forward modelling

We consider a system of ice, air and water, with relative dielectric constants (ɛ r), electrical conductivities (σ), electromagnetic velocities and magnetic permeabilities (μ r) taken from Plewes and Hubbard (Reference Plewes and Hubbard2001) (Table 2). In situ observations from shallow augering around Channel Lake only identified the presence of ice, air and water, with no evidence of firn or snow.

Table 2. Input properties of the modelled materials, taken from Plewes and Hubbard (Reference Plewes and Hubbard2001)

The construction and interpretation of the forward models followed a similar methodology to that described by Stuart and others (Reference Stuart, Murray, Gamble, Hayes and Hodson2003). Models were constructed in a 2-D space; however, they were only varied on the depth axis and not on the horizontal, and the boundary conditions were set to absorb all electromagnetic energy. This was done to facilitate a controlled forward model environment, avoiding the signal arising from non-horizontal reflectors which would otherwise interfere with the reflection signal from the interface being analysed. Three main types of model were tested, simulating different materials within a pocket in ice: air, water and air on top of water. For each model, a series of depths and thicknesses were tested for the pockets, and differing water/air thicknesses were tested in the case of the air-on-water model. Zero-offset traces were calculated by finite differencing of electromagnetic energy propagating vertically downwards from a 250 MHz Ricker wavelet source at the transmitter location, with a time step of 0.02 ns and a time window of 400 ns. The output of this process was a radargram which was analysed in a manner similar to real data, with the exception that the forward-modelled data did not require any additional signal processing.

The amplitude of signals reflected from interfaces between different media is determined by the amplitude reflection coefficient (R). These values may be used to infer the interfacing materials which produce reflections in GPR data. Five separate possible interfaces were considered in this study: ice over water, ice over air, air over ice, air over water and water over ice. Table 3 lists the calculated amplitude reflection coefficients of these interfaces as calculated using Eqn (1):

(1)$$R = \displaystyle{{\sqrt {\varepsilon _1} -\sqrt {\varepsilon _2}} \over {\sqrt {\varepsilon _1} + \sqrt {\varepsilon _2}}} \;, $$

Table 3. Amplitude reflection coefficients for the various interfaces calculated from Eqn (1) and the material properties in Table 2

ɛ 1 and ɛ 2 are the dielectric constants of the overlying and underlying media respectively, assuming μ r ≈ 1 for each medium (Ward and Hohmann, Reference Ward, Hohmann and Nabighian1987).

From these values, we can expect strong reflection signals from interfaces involving water and moderate reflections from interfaces of air. In practice, deducing the exact materials from reflection signals in glacier GPR data is challenging without precise knowledge of source wavelet properties or in situ observations to correlate with (e.g. ice cores intersecting features found in GPR data). The most powerful diagnostic for our purpose of identifying the nature of the englacial reflections is therefore the polarity of the reflections' first peaks, with the amplitude (often referred to as ‘brightness’ in GPR data) of the reflections only assessed in a relative sense. The presence of water is therefore indicated by bright reflectors with a negative peak polarity.

Results

The GPR transects across Channel Lake showed multiple irregular subsurface features in every transect (Fig. 4). The highest spatial density of subsurface features is apparent underneath Channel Lake, with fewer features apparent to the north and south. The features occur mostly within 5–10 m of the surface and have horizontal extents from <1 m to over 50 m. Some radargrams also revealed horizontally extensive reflectors within the top 2 m. Within any given transect, some features are discrete, while others appear either interconnected or too close together to adequately resolve (Fig. 4).

We classify the features according to their geometries, identifying three main categories that are labelled A, B and C in Figure 4. Type A features are characterised by hyperbolae that are abundant in the unmigrated data and were observed on most profiles in groups beneath the margins of Channel Lake (e.g. Fig. 5). Type B features appear as bright, flat and laterally extensive reflections, and are particularly abundant in Lines 3 and 4 at shallow depths of <5 m (e.g. Fig. 6). Type C features encompass a broad range of bright, undulating, horizontally extensive reflections, and are observed in all radargrams in this study at depths typically exceeding 5 m (e.g. Fig. 7).

Fig. 5. Section of non-static-corrected 800 MHz GPR data from Line 1 that shows the characteristic diffractions of Type A features (left), and the migration effect, which attempts to collapse these diffractions to points. We infer that these point features represent englacial crevasses.

Fig. 6. Section of 800 MHz GPR data from Line 4 illustrating a typical Type B feature as a bright, shallow, horizontally extensive feature. Red line represents the extracted trace, which is compared against a forward model trace from an ice-water interface.

Fig. 7. (a) Radargrams and extracted traces showing the signal characteristics of Type C features, with ice–water forward model for comparison. (b) Radargrams and traces from two Type B features compared against an ice–air–water forward model trace. Note that the lower reflector, interpreted as the top of a water body in both features, is apparently undulating in the radargrams. This is likely a distorted view of a flat surface.

The forward models simulated single englacial pockets at varying depths, ranging from completely water-filled to completely air-filled, and combinations of the two. Consistent with our expectations, we simulated stronger reflections from ice–water or air–water interfaces than ice–air interfaces and confirmed the polarity of the reflections' first peaks being robustly diagnostic of the interface type (Table 3). No clear reflections were simulated for water–ice interfaces, in agreement with Stuart and others' (Reference Stuart, Murray, Gamble, Hayes and Hodson2003) inferences for englacial channels in a cold-ice Svalbard glacier, which implies that volume estimates of water-bearing englacial voids cannot readily be inferred from our GPR data. The vertical resolution of GPR data (i.e. the vertical distance between which two signals can be separable) is theoretically limited to a quarter of its pulse wavelength (Reynolds, Reference Reynolds2011), and the skin depth of the signal is a function of its frequency and the conductivity of the medium. The estimates presented by Table 4 show that most reflection features in this study lie within the skin depth of ice for both antenna frequencies, and while the thickness of water bodies may be indeterminable, our 800 MHz should resolve an air pocket thicker than ~0.1 m. However, these values are only theoretical calculations based on the idealistic electrical parameters described in Table 2. It is likely that the vertical resolution of each system is in fact coarser, although this is difficult to quantify without in situ testing. Additionally, the skin depth in ice may vary significantly depending on the water content.

Table 4. Theoretical values of vertical resolution and skin depth for ice, water, and air for both 250 MHz and 800 MHz GPR antennas

Discussion

Type A features were mainly found underneath the crevassed ridges at the edges of the Channel Lake basin, and therefore likely represent small englacial crevasses or fractures. Type B reflections are flat, lie within 5 m of the surface and often extend over 20 m along transects. The reflection signal characteristics of Type B features are difficult to distinguish as a result of interference with noise at early two-way travel times; however, their high brightness and comparison to the model traces suggest these are indicative of water (Fig. 6). These features are consistent with phreatic surfaces of perched water bodies filling pore spaces or cavities in subsurface ice. Indeed, our in situ observations during other sensor installations consistently revealed the widespread presence of water or air-filled cavities 30–90 cm beneath the Channel Lake basin. The GPR observations of Type B features share some characteristics with firn aquifers observed in locations such as the Greenland Ice Sheet (Forster and others, Reference Forster2013), Svalbard (Christianson and others, Reference Christianson, Kohler, Alley, Nuth and van Pelt2015) and the European Alps (Kulessa and others, Reference Kulessa, Booth, Hobbs and Hubbard2008; Kulessa and others, Reference Kulessa, Grobbe, Revil, Zhu and Slobin press), which are able to accommodate substantial meltwater flow and storage. However, our observed features are highly spatially constrained to the Channel Lake basin, and the lack of firn observed at the site requires a different explanation for the presence of subsurface cavities. One possibility is that subsurface porosity was created by a lake drainage event, as previously floating ice descended onto the underlying surface. An alternative explanation could be that the water was created directly by subsurface melting, a process which has been observed to occur in Antarctica when surface air temperatures are below zero and penetration of solar radiation allows melting of the ice at depth (Liston and others, Reference Liston, Winther, Bruland, Elvehøy and Sand1999). Despite the more spatially constrained nature of our observations, surface meltwater ponds have been observed extensively around the coast of Antarctica (Kingslake and others, Reference Kingslake, Ely, Das and Bell2017) and this type of subsurface meltwater lens could have an important effect on the movement and storage of meltwater in Antarctica.

Comparing the forward modelling results to our GPR observations, most of the Type C englacial reflections in Sørsdal Glacier are bright and characterised by negative polarities (Fig. 7a), and are therefore consistent with the presence of ice–water interfaces – potentially indicating water-filled englacial voids at depths that typically exceed 5 m (Fig. 4). Exceptionally, two large Type C features in Lines 1 and 4 (Fig. 7b) are characterised by a dimmer reflection with positive polarity above a brighter reflection with negative polarity. These observations are consistent with the presence of englacial voids containing pockets of air above water layers, the surfaces of which are likely to be flat in reality but appear to be undulating in our migrated images. These undulations are artefacts likely caused by off-line effects within the cone-shaped radar beam transmitted by our system, generated as our profiles cross voids at oblique angles (Arcone and Yankielun, Reference Arcone and Yankielun2000; Benn and others, Reference Benn2017), and because small-scale differences in radar velocities are not accounted for in our 2-D migration algorithm. Notwithstanding the undulations, we can infer that the travel time difference between the dimmer and brighter reflectors is consistent with air pockets thicker than 1 m at the centre location (marked by red lines in Fig. 7b). As expected, we also cannot delineate the base of the void beneath the water layer.

Our GPR observations on Sørsdal Glacier detected numerous Type C englacial features, which we can infer to be either water-filled or containing a water layer below an air pocket. It is possible that many of the features that appear as water-filled in our data do in fact contain an air pocket above the water layer, but that this air pocket is so thin that we cannot distinguish it in our GPR data. The geometries of these Type C features, and the morphology of the Channel Lake itself point to a number of possible formation mechanisms.

Although we cannot assert the exact formation mechanism, we propose that the Channel Lake basin was probably formed by the drainage of a previous surface lake for three main reasons. First, the Channel Lake basin was a dry depression when we visited it in February 2017, but was observed to be water-filled in satellite imagery from previous years. During the time of our survey campaign, large volumes of water were in fact observed to pool in lakes on the glacier surface elsewhere. Second, the oval shape of the basin with crevassed margins is similar to depressions observed elsewhere in East Antarctica that were caused by surface lake drainage (Langley and others, Reference Langley, Leeson, Stokes and Jamieson2016). Third, there are geomorphological similarities to previously-observed ice dolines, which form characteristically elongate and distinctive basins in otherwise flat surface ice (Bindschadler and others, Reference Bindschadler, Scambos, Rott, Skvarca and Vornberger2002) although the edges of the Channel Lake are less sharply-defined.

We also propose that the Channel Lake basin was drained through the englacial pathways imaged by our GPR surveys. Based on similar observations, Stuart and others (Reference Stuart, Murray, Gamble, Hayes and Hodson2003) inferred the widespread presence of ponds of water in relict englacial pathways in the cold austre Brøggerbreen in Svalbard, which covered between 14 and 90% of the channel height. These inferences are directly compatible with our own observations at Sørdal Glacier, and it is therefore likely that any surface ablation during our survey periods was drained into and potentially through the englacial pathways we have imaged. Although Type C features are common in our data, the large spacing between our survey profiles (Fig. 1) makes it challenging to infer the potential presence of interconnected voids with any degree of certainty (Fig. 8). Future campaigns aiming to further characterise near-surface englacial features at Channel Lake and other sites may aim to provide a more detailed analysis, using closer spacing to delineate and potentially determine interconnectivity of features. Otherwise further studies may seek to determine the spatial distribution of englacial features across the glacier and the rest of the Antarctic coast.

Fig. 8. Conceptualised image of hydrological features identified in our GPR data, classified by type (see text). Map locations of features shown at the top relative to Channel Lake, cross-sectional depth distribution of features shown irrespective of the horizontal location along any given profile line at bottom.

Conclusions

Our GPR data acquired along sub-parallel profiles on Sørsdal Glacier (Fig. 1) provide compelling evidence for the presence of englacial voids filled with water, or air pockets above water, at depths typically exceeding 5 m but <20 m (Type C features, Figs. 4 and 7). The data are furthermore consistent with lenses of meltwater held within porous ice at depths of <5 m, and perched on impermeable glacier ice, that are conceptually similar to the firn aquifers observed on the Greenland Ice Sheet, as well as the Arctic and Alpine glaciers. These englacial water bodies are observed adjacent to and more than 2 km downstream of a dry relict surface pond, which is consistent with pond drainage into an interconnected englacial drainage system. Indeed, entirely consistent with these inferences, a large early melt season outburst flood from this implied system was identified in 2016/17 satellite imagery.

Our observations imply that meltwater can be stored and possibly flow in shallow hydrological systems in East Antarctica, thus affecting the volume and timing of meltwater contribution to local surface mass balance, akin to processes inferred previously for the Greenland Ice Sheet (e.g. Forster and others, Reference Forster2013). Future work should therefore ascertain the spatial prevalence of such hydrological systems along the East Antarctic coastline, identify to what degree the meltwater is mobile and redistributed or refrozen and on what timescales, and quantify any impacts on surface mass-balance contributions.

Acknowledgements

Data collection was supported through the Australian Government's Australian Antarctic Science Grant Program under project AAS 4342. This work was supported by the Australian Government's Business Cooperative Research Centres Programme through the Antarctic Climate and Ecosystems Cooperative Research Centre (ACE CRC) and the Australian Research Council's Special Research Initiative for the Antarctic Gateway Partnership (project ID SR140300001). Landsat 8 imagery courtesy of the US Geological Survey. Thanks to Hugh Tassel and Will McAdam for providing advice and assistance in data processing.

References

Alley, RB, Dupont, TK, Parizek, BR and Anandakrishnan, S (2005) Access of surface meltwater to beds of sub-freezing glaciers: preliminary insights. Annals of Glaciology 40, 814.CrossRefGoogle Scholar
Arcone, SA and Yankielun, NE (2000) 1.4 GHz radar penetration and evidence of drainage structures in temperate ice: Black Rapids Glacier, Alaska, U.S.A. Journal of Glaciology 46(154), 477490. doi: 10.3189/172756500781833133.CrossRefGoogle Scholar
Banwell, AF, MacAyeal, DR and Sergienko, OV (2013) Breakup of the Larsen B Ice Shelf triggered by chain reaction drainage of supraglacial lakes. Geophysical Research Letters 40(22), 58725876.CrossRefGoogle Scholar
Benn, D and 5 others (2017) Structure and evolution of the drainage system of a Himalayan debris-covered glacier, and its relationship with patterns of mass loss. Cryosphere 11(5), 22472264. doi: 10.5194/tc-11-2247-2017.CrossRefGoogle Scholar
Bindschadler, R, Scambos, TA, Rott, H, Skvarca, P and Vornberger, P (2002) Ice dolines on Larsen Ice Shelf, Antarctica. Annals of Glaciology 34, 283290.CrossRefGoogle Scholar
Catania, GA, Neumann, TA and Price, SF (2008) Characterizing englacial drainage in the ablation zone of the Greenland ice sheet. Journal of Glaciology 54(187), 567578.CrossRefGoogle Scholar
Christianson, K, Kohler, J, Alley, RB, Nuth, C and van Pelt, WJJ (2015) Dynamic perennial firn aquifer on an Arctic glacier. Geophysical Research Letters 42(5), 14181426. doi: 10.1002/2014gl062806.CrossRefGoogle Scholar
da Rosa, KK and 5 others (2014) Stratigraphy of Wanda Glacier, King George Island, Antarctica, using ground penetrating radar. Revista Brasileira de Geofisica 32(1), 2130.CrossRefGoogle Scholar
Das, SB and 6 others (2008) Fracture propagation to the base of the Greenland Ice Sheet during supraglacial lake drainage. Science 320(5877), 778781.CrossRefGoogle ScholarPubMed
Doyle, SH and 9 others (2013) Ice tectonic deformation during the rapid in situ drainage of a supraglacial lake on the Greenland Ice Sheet. The Cryosphere 7(1), 129140. doi: 10.5194/tc-7-129-2013.CrossRefGoogle Scholar
Forster, RR and 12 others (2013) Extensive liquid meltwater storage in firn within the Greenland ice sheet. Nature Geoscience 7, 9598. doi: 10.1038/ngeo2043.CrossRefGoogle Scholar
Fountain, AG and Walder, JS (1998) Water flow through temperate glaciers. Reviews of Geophysics 36(3), 299328. doi: 10.1029/97RG03579.CrossRefGoogle Scholar
Gulley, JD, Benn, DI, Screaton, E and Martin, J (2009) Mechanisms of englacial conduit formation and their implications for subglacial recharge. Quaternary Science Reviews 28(19-20), 19841999. doi: 10.1016/j.quascirev.2009.04.002.CrossRefGoogle Scholar
Jones, GA, Kulessa, B, Doyle, SH, Dow, CF and Hubbard, A (2013) An automated approach to the location of icequakes using seismic waveform amplitudes. Annals of Glaciology 54(64), 19. doi: 10.3189/2013AoG64A074.CrossRefGoogle Scholar
Kim, KY, Lee, J, Hong, MH, Hong, JK and Shon, H (2010) Helicopter-borne and ground-towed radar surveys of the Fourcade Glacier on King George Island, Antarctica. Exploration Geophysics 41(1), 5160. doi: 10.1071/EG09052.CrossRefGoogle Scholar
Kingslake, J, Ely, JC, Das, I and Bell, RE (2017) Widespread movement of meltwater onto and across Antarctic ice shelves. Nature 544(7650), 349352.CrossRefGoogle ScholarPubMed
Kulessa, B, Booth, AD, Hobbs, A and Hubbard, AL (2008) Automated monitoring of subglacial hydrological processes with ground-penetrating radar (GPR) at high temporal resolution: scope and potential pitfalls. Geophysical Research Letters 35(24), L24502. doi: 10.1029/2008GL035855.CrossRefGoogle Scholar
Kulessa, B (in press s) Seismoelectric characterisation of ice sheets and glaciers. In Grobbe, NRevil, AZhu, Z and Slob, E ed. Seismoelectric Exploration: Theory, Experiments and Applications. AGU Books, Washington, DC, USA.Google Scholar
Langley, ES, Leeson, AA, Stokes, CR and Jamieson, SSR (2016) Seasonal evolution of supraglacial lakes on an East Antarctic outlet glacier. Geophysical Research Letters 43(16), 85638571. doi: 10.1002/2016GL069511.CrossRefGoogle Scholar
Lenaerts, JTM and 12 others (2016) Meltwater produced by wind–albedo interaction stored in an East Antarctic ice shelf. Nature Climate Change 7, 58. doi: 10.1038/nclimate3180.CrossRefGoogle Scholar
Liston, G, Winther, J, Bruland, O, Elvehøy, H and Sand, K (1999) Below-surface ice melt on the coastal Antarctic ice sheet. Journal of Glaciology 45(150), 273285. doi: 10.3189/S0022143000001775.CrossRefGoogle Scholar
Macayeal, DR and Sergienko, OV (2013) The flexural dynamics of melting ice shelves. Annals of Glaciology 54(63), 110. doi: 10.3189/2013AoG63A256.CrossRefGoogle Scholar
Mellor, M (1960) Antarctic ice terminology: ice dolines. Polar Record 10(64), 92. doi: 10.1017/S0032247400050786.CrossRefGoogle Scholar
Plewes, LA and Hubbard, B (2001) A review of the use of radio-echo sounding in glaciology. Progress in Physical Geography 25(2), 203236.CrossRefGoogle Scholar
Reynolds, JM (2011) An Introduction to Applied and Environmental Geophysics, 2nd Edn.John Wiley & Sons, Chichester, UK.Google Scholar
Rignot, E, Mouginot, J and Scheuchl, B (2017) MEaSUREs InSAR-Based Antarctica Ice Velocity Map, Version 2. Boulder, Colorado USA. NASA National Snow and Ice Data Center Distributed Active Archive Center. doi: https://doi.org/10.5067/D7GK8F5J8M8R.CrossRefGoogle Scholar
Scambos, T and 7 others (2009) Ice shelf disintegration by plate bending and hydro-fracture: satellite observations and model results of the 2008 Wilkins ice shelf break-ups. Earth and Planetary Science Letters 280(1–4), 5160.CrossRefGoogle Scholar
Schoof, C (2010) Ice-sheet acceleration driven by melt supply variability. Nature 468(7325), 803806. doi: 10.1038/nature09618.CrossRefGoogle ScholarPubMed
Shreve, RL (1972) Movement of water in glaciers. Journal of Glaciology 11(62), 205214. doi: 10.3189/S002214300002219X.CrossRefGoogle Scholar
Stolt, RH (1978) Migration by Fourier transform. Geophysics 43(1), 2348. doi: 10.1190/1.1440826.CrossRefGoogle Scholar
Stuart, G, Murray, T, Gamble, N, Hayes, K and Hodson, A (2003) Characterization of englacial channels by ground-penetrating radar: an example from austre Brøggerbreen, Svalbard. Journal of Geophysical Research: Solid Earth 108(11), EPM 7-1EPM 7-13.CrossRefGoogle Scholar
van der Veen, CJ (2007) Fracture propagation as means of rapidly transferring surface meltwater to the base of glaciers. Geophysical Research Letters 34(1), L01501. doi: 10.1029/2006GL028385.CrossRefGoogle Scholar
Vatne, G (2001) Geometry of englacial water conduits, Austre Brøggerbreen, Svalbard. Norsk Geografisk Tidsskrift 55(2), 8593. doi: 10.1080/713786833.CrossRefGoogle Scholar
Vatne, G and Irvine-Fynn, TDL (2016) Morphological dynamics of an englacial channel. Hydrology and Earth System Sciences 20(7), 29472964. doi: 10.5194/hess-20-2947-2016.CrossRefGoogle Scholar
Ward, SH and Hohmann, GW (1987) Electromagnetic theory for geophysical applications. In Nabighian, M ed. Electromagnetic Methods in Applied Geophysics. Society of Exploration Geophysicists, Tulsa, OK, USA, 131311. doi: 10.1190/1.9781560802631.ch4.Google Scholar
Weertman, J (1973) Can a water-filled crevasse reach the bottom surface of a glacier. IASH publ 95, 139145.Google Scholar
Zwally, HJ and 5 others (2002) Surface melt-induced acceleration of Greenland ice-sheet flow. Science 297(5579), 218222. doi: 10.1126/science.1072708.CrossRefGoogle ScholarPubMed
Figure 0

Fig. 1. Maps showing the location of Sørsdal Glacier and Channel Lake with GPR transects (solid black lines labelled 1 to 5). Base imagery captured 29th March 2017 by Sentinel 2 satellite.

Figure 1

Fig. 2. (a) Photograph of Channel Lake taken from a helicopter, looking northeast. Channel Lake and its associated downstream structures are together ~4 km in length; (b) photograph of Channel Lake basin looking east from the southern ridge. Here, the bounding ridges are up to 10 m high and are ~100 m apart. Both photographs were taken during summer 2016/17.

Figure 2

Fig. 3. Satellite imagery (Landsat 8 OLI/TIRS C1 Level 1) showing the evolution of Channel Lake on Sørsdal Glacier between 2014 and 2016.

Figure 3

Table 1. Sampling parameters for GPR data on Sørsdal Glacier

Figure 4

Fig. 4. GPR radargrams (locations shown in Fig. 1). Examples of each geometric class are annotated. Note the differences in scale and surface topography.

Figure 5

Table 2. Input properties of the modelled materials, taken from Plewes and Hubbard (2001)

Figure 6

Table 3. Amplitude reflection coefficients for the various interfaces calculated from Eqn (1) and the material properties in Table 2

Figure 7

Fig. 5. Section of non-static-corrected 800 MHz GPR data from Line 1 that shows the characteristic diffractions of Type A features (left), and the migration effect, which attempts to collapse these diffractions to points. We infer that these point features represent englacial crevasses.

Figure 8

Fig. 6. Section of 800 MHz GPR data from Line 4 illustrating a typical Type B feature as a bright, shallow, horizontally extensive feature. Red line represents the extracted trace, which is compared against a forward model trace from an ice-water interface.

Figure 9

Fig. 7. (a) Radargrams and extracted traces showing the signal characteristics of Type C features, with ice–water forward model for comparison. (b) Radargrams and traces from two Type B features compared against an ice–air–water forward model trace. Note that the lower reflector, interpreted as the top of a water body in both features, is apparently undulating in the radargrams. This is likely a distorted view of a flat surface.

Figure 10

Table 4. Theoretical values of vertical resolution and skin depth for ice, water, and air for both 250 MHz and 800 MHz GPR antennas

Figure 11

Fig. 8. Conceptualised image of hydrological features identified in our GPR data, classified by type (see text). Map locations of features shown at the top relative to Channel Lake, cross-sectional depth distribution of features shown irrespective of the horizontal location along any given profile line at bottom.