ArticlePDF Available

Massive subsurface ice formed by refreezing of ice-shelf melt ponds

Authors:

Abstract and Figures

Surface melt ponds form intermittently on several Antarctic ice shelves. Although implicated in ice-shelf break up, the consequences of such ponding for ice formation and ice-shelf structure have not been evaluated. Here we report the discovery of a massive subsurface ice layer, at least 16 km across, several kilometres long and tens of metres deep, located in an area of intense melting and intermittent ponding on Larsen C Ice Shelf, Antarctica. We combine borehole optical televiewer logging and radar measurements with remote sensing and firn modelling to investigate the layer, found to be ∼10 C warmer and ∼170 kg m¯³ denser than anticipated in the absence of ponding and hitherto used in models of ice-shelf fracture and flow. Surface ponding and ice layers such as the one we report are likely to form on a wider range of Antarctic ice shelves in response to climatic warming in forthcoming decades.
Content may be subject to copyright.
ARTICLE
Received 11 Aug 2015 |Accepted 10 May 2016 |Published 10 Jun 2016
Massive subsurface ice formed by refreezing
of ice-shelf melt ponds
Bryn Hubbard1, Adrian Luckman2, David W. Ashmore1, Suzanne Bevan2, Bernd Kulessa2,
Peter Kuipers Munneke2, Morgane Philippe3, Daniela Jansen4, Adam Booth5, Heidi Sevestre6,
Jean-Louis Tison3, Martin O’Leary2& Ian Rutt2
Surface melt ponds form intermittently on several Antarctic ice shelves. Although implicated
in ice-shelf break up, the consequences of such ponding for ice formation and ice-shelf
structure have not been evaluated. Here we report the discovery of a massive subsurface ice
layer, at least 16 km across, several kilometres long and tens of metres deep, located in an
area of intense melting and intermittent ponding on Larsen C Ice Shelf, Antarctica. We
combine borehole optical televiewer logging and radar measurements with remote sensing
and firn modelling to investigate the layer, found to be B10 °C warmer and B170 kg m 3
denser than anticipated in the absence of ponding and hitherto used in models of ice-shelf
fracture and flow. Surface ponding and ice layers such as the one we report are likely to form
on a wider range of Antarctic ice shelves in response to climatic warming in forthcoming
decades.
DOI: 10.1038/ncomms11897 OPEN
1Centre for Glaciology, Department of Geography and Earth Sciences, Aberystwyth University, Aberystwyth SY23 3DB, UK. 2Glaciology Group, Department
of Geography, Swansea University, Swansea SA2 8PP, UK. 3Laboratoire de Glaciologie, De
´partement Ge
´osciences, Environnement et Socie
´te
´, Universite
´Libre
de Bruxelles, Bruxelles 1050, Belgium. 4Alfred Wegener Institut, Helmholtz-Zentrum fu
¨r Polar- und Meeresforschung D-27568, Bremerhaven, Germany.
5School of Earth and Environment, University of Leeds, Leeds LS2 9JT, UK. 6Department of Arctic Geology, University Centre in Svalbard, N-9171
Longyearbyen, Norway. Correspondence and requests for materials should be addressed to B.H. (email: byh@aber.ac.uk).
NATURE COMMUNICATIONS | 7:11897 | DOI: 10.1038/ncomms11897 | www.nature.com/naturecommunications 1
The B49,000 km2Larsen C Ice Shelf (LCIS) is considered
susceptible to future collapse because of its exposure to
intense surface melting1across the northern sector of the
Antarctic Peninsula (Fig. 1). Satellite and airborne radar data on
the LCIS indicate low firn-air content2, promoting the formation
of melt ponds and the potential for hydrofracturing3,4. Analysis of
remotely sensed images of Cabinet Inlet, located in the northwest
sector of the LCIS, reveals the presence during some summer
months of surface melt ponds that generally form in flow-parallel
troughs that are some tens of kilometres long and hundreds of
metres wide (Fig. 1). These ponds form in Cabinet Inlet as a result
of melting by fo
¨hn winds that blow from the Graham Land
mountains and eastwards through the shelf’s northernmost-
fringing inlets5–7. Although appearing to generate substantial
surface melt, these winds do not persist for more than a few days
even through the summer, and ponds appear intermittently on
satellite images (Fig. 1). The additional stress and hydrofracturing
potential of such ponded surface water has been proposed as a
means of shelf destabilization8–10 and has been implicated in
the collapse of Larsen B Ice Shelf in 200211,12. However,
the implications of such ponding for the formation of new ice
and its influence on ice-shelf structure have not been reported.
In this study, we report the presence of a massive ice layer, at
least 16 km across, several kilometres long and tens of metres
deep, present beneath an area of intermittent pond formation on
LCIS, Antarctica. We combine field data with firn-modelling and
remotely sensed data to investigate the layer’s properties and
formation. The layer is found to be composed of two units: an
upper, solid ice unit formed largely from the continual refreezing
of ponded water; and a lower, infiltration ice unit formed
largely from the refreezing of meltwater that has percolated into
very dense firn. The layer is found to be B10 °C warmer and
B170 kg m 3denser than that which would have been present
in the absence of the influence of intense surface melting and
pond formation. The implications of the layer’s presence for
ice-shelf thickness estimates, flow and stability are explored.
Results
Borehole drilling and logging. In early austral summer
2014/2015, we drilled a B100-m long borehole into the flank of a
Cabinet Inlet trough indicated by satellite imagery to have
repeatedly hosted melt ponds over the past 15 years, but not since
2008/2009. Although the shelf was snow-covered at the time of
drilling, inspection of the wall of a 2-m deep pit revealed the
presence of both numerous ice layers within the snowpack and an
unusually thick ice layer at a depth of 2.0 m that prevented
continued excavation. The borehole was logged to a depth of
B97 m by optical televiewer (OPTV)13, providing a geometrically
accurate image of the complete borehole wall at a vertical and
lateral resolution of B1 mm. The resulting OPTV log (Fig. 2a)
contrasts starkly with those retrieved from other accumulating
b
a
Cabinet Inlet
Churchill
Peninsula
Cole
Peninsula
50 km 10 km
N
E
S
W
123
Figure 1 | Cabinet Inlet basemap and surface ponding. Location map of Cabinet Inlet study site on Larsen C Ice Shelf based on MODIS data from 3rd
December 2014. (a) Main figure location in Antarctica (red box). (b) Landsat expansion of Cabinet Inlet from 31st December 2001, showing surface
ponding (dark patches) and the locations of the borehole (yellow dot) and 200-MHz GPR transects (red lines) presented in Fig. 4.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11897
2NATURE COMMUNICATIONS | 7:11897 | DOI: 10.1038/ncomms11897 | www.nature.com/naturecommunications
ice-shelf and ice-sheet locations (for example, Fig. 2b), including
the interior of the Greenland Ice Sheet14 and an East Antarctic ice
shelf and ice rise15,16. In each of these other cases, OPTV log
luminosity decreases gradually with depth as surface snow
metamorphoses through firn to dense ice over depths of several
tens of metres. In contrast, our Cabinet Inlet OPTV log (Fig. 2a)
shows a sharp contact between high-luminosity surface snow or
firn and low-luminosity ice at a depth of only 2.9 m below the
ice-shelf surface. This ice extends to the 97-m deep base of
the log. Converting the luminosity of this OPTV log to
density reveals a mean density for this entire ice layer of
870 kg m 3. Figure 2a also reveals a transition in ice type at a
depth of B45 m, with the overlying layer (named Unit 1)
hosting more frequent horizontal layers and being more dense
(mean ¼888 kg m 3) than the underlying ice layer (Unit 2;
mean ¼854 kg m 3), which is principally composed of bubbly
host ice containing coarse, irregularly dipping bubble-free ice
layers. The proximity of this generally massive ice layer to the
shelf surface, and the sharpness of its contact with the overlying
snowpack, preclude formation by the usual process of
compaction–metamorphism. Instead, we interpret this Unit 1
ice as having formed as a consequence of refreezing, following
periods of intense surface melting and intermittent pond
formation. The characteristics revealed by the OPTV image are
consistent with this unit (2.9 to B45 m) being largely bubble-free
pond ice formed from the refreezing of surface water ponded
during extended periods of intense, presumably summer, melting.
Here the fine-scale horizontal layering apparent in Fig. 2a likely
reflects the episodic nature of the process, which would involve
four general stages: (a) snow accumulation, (b) surface melting,
(c) meltwater infiltration into underlying snow, eventually
resulting in its saturation and pond formation, and (d) the
freezing of that meltwater layer to form pond ice, probably during
early austral autumn. In contrast, Unit 2 (approximately 45–97-m
depth) also contains layers of bubble-poor ice, but in this zone
they are typically decimetres thick, contorted and isolated within
host ice that is optically brighter (likely due to the presence of
reflective bubbles; Fig. 2a). We interpret this lower unit as ice
dominated by infiltration refreezing formed by meltwater
percolating into underlying firn. This still involves the influence
of intense surface melting but, in contrast to the Unit 1,
of generally insufficient magnitude to form a continuous
quasi-massive layer of largely bubble-free pond ice. These
physical characteristics and depth ranges of the two units we
identify are consistent with the upper (Unit 1) ‘pond ice’ forming
within the Cabinet Inlet region of pond formation and the lower
(Unit 2) ‘infiltration ice’ forming, and being inherited from,
up-flow of the region of pond formation (Fig. 1).
Firn modelling and satellite image analysis. We evaluate this
hypothesis for the recent past through a one-dimensional firn
densification and hydrology model17, driven by surface mass
fluxes and temperature data from the RACMO2.3 regional
climate model for Cabinet Inlet18. Model results (Fig. 3a,b)
indicate the occurrence of intermittent, but substantial, surface
melt events at the borehole location, consistent with Unit 1
forming from the refreezing of surface meltwater. As well as
predicting the presence of such a layer, the model predicts
(Fig. 3b) that the layer’s upper surface was, in summer
2014–2015, expected to be 2.9-m below the snow surface, and
that the overlying snowpack contains a substantial ice layer
between B1.9 and 2.2 m, agreeing with our direct observations
from snow-pit digging, the OPTV log (Fig. 2a) and ground-
penetrating radar (GPR) data (Fig. 4) addressed below.
Further, this analysis is supported by a time series of moderate
resolution imaging spectroradiometer (MODIS) satellite images
(Fig. 3c), indicating that melt ponds formed annually between
2001 and 2009, while none appears in any of the images available
between early 2009 and the time of fieldwork in late 2014. This
reconstruction matches very closely the RACMO-based firn
densification reconstruction (Fig. 3b) with ‘pond years’,
coinciding with periods of intense near-surface firn
densification (for example, 2005–2007).
Ground-penetrating radar. An approximation of the lateral
extent of the massive subsurface ice layer we report is provided
by the area known from satellite images to host melt ponds:
B60-km across-flow and B20-km along-flow (Fig. 1). However,
the precise degree to which this zone is underlain by massive
NE S WN
–20 –10 0
DIR OPTVLCIS OPTV Temp. (°C)
NE S WN
Depth
(m)
LCIS density
(kg m–3)
DIR density
(kg m–3)
0
500
1,000
250
750
0
500
1,000
250
750
cba
0
10
20
30
40
50
60
70
80
90
Figure 2 | Cabinet Inlet borehole data. (a) OPTV log of the LCIS borehole.
This raw log is of the complete borehole wall, unrolled to progress
north—east—south—west—north from left to right. The log is classified
into two units, described and interpreted in the text: Unit 1 extends from a
depth of B2.9–45 m, and Unit 2 from a depth of B45 m to the base.
(b) OPTV log of a typical ice-shelf borehole unaffected by significant
melting (Derwael Ice Rise, Roi Baudouin Ice Shelf, Antarctica) showing
(a) the normal reduction in luminosity with depth, interpreted as
progressive densification from snow, through firn to ice, and (b) the usual
presence of regular horizontal planes closing up with depth, interpreted as
annual layering. Density profiles derived from OPTV luminosity (described
in Methods) are superimposed as red lines on both OPTV logs. (c) Englacial
temperatures (a) measured by borehole thermistor string (solid dots),
(b) that would normally be used in an ice-shelf model for this site (open
circles), and (c) predicted by our firn model at a depth of 11 m (cross).
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11897 ARTICLE
NATURE COMMUNICATIONS | 7:11897 | DOI: 10.1038/ncomms11897 | www.nature.com/naturecommunications 3
pond and/or infiltration ice, and the depth of such layers where
they are present beyond our borehole, cannot be determined from
temporally discrete satellite images. To evaluate this we carried
out GPR profiling at 200 MHz along three transects focused on
the borehole location (Fig. 1). The resulting radargrams (Fig. 4)
reveal the presence of numerous near-surface reflectors and one
substantial reflector at a depth that varies between B1 and B3m
along the transects. Very little radar energy above the background
noise level was received from below this reflection. This absence
of signal return is consistent with the presence of a refrozen ice
layer that, although characterized by minor density stratification,
is physically and chemically uniform compared with firn layering
unmodified by meltwater. This main reflector is B2.9-m deep at
the borehole location, coincident with field-based digging, drilling
and OPTV logging, as well as with firn modelling (above).
We therefore infer that this strong GPR reflection represents the
upper surface of a spatially extensive ice layer that is both thick
and relatively homogeneous. Since meltwater ponds are confined
to surface troughs that persist in MODIS images for decades
within this region of Cabinet Inlet, it is unlikely that this ice layer
is ubiquitously composed of Unit 1 pond ice. Away from the
troughs, this near-surface reflector therefore probably indicates
the uppermost surface of an ice layer that is more similar to
Unit 2 infiltration ice, that is, still influenced by intense surface
melting and subsurface refreezing, but in these areas not actually
forming standing ponds. In the absence of further direct inves-
tigation, such as by ice coring or borehole analysis, it is not yet
possible to determine at high resolution the lateral variability of
the two units we identify. Nonetheless, our OPTV and GPR data
together indicate that a widespread ice layer extends at least
across all of our B16-km flow-orthogonal and B6-km flow-
parallel GPR study area. Our borehole OPTV log also indicates
that Units 1 and 2 combined extend to a depth of at least 97 m at
the location of our borehole. While it is highly likely that this
thickness—defined by the depth of the base of Unit 2—varies
spatially across Cabinet Inlet, it is not currently possible to specify
this distribution without further borehole and/or GPR data.
Discussion
The presence of the refrozen ice layer we report above affects the
physical properties of the LCIS in at least two ways. First,
the layer’s density is substantially higher than that of the snow and
firn that would otherwise have formed over this depth range by
standard compaction–metamorphism. For example, a recent LCIS
model19 used a density of B700 kg m 3for the shelf’s uppermost
B100 m, based on inverting seismic data recorded in the shelf’s
southern sector. In contrast, our OPTV-derived densities indicate a
measured density of B870 kg m 3over this depth range in the
Cabinet Inlet area, 24% higher. Such a density enhancement
influences calculations of the shelf’s thickness based on the surface
elevation data20. In this case, using an ice density of 917kg m 3,a
sea water density of 1,026 kg m 3, a surface elevation of 63.5 m
(Fig. 4) and the assumed firn density of 700 kg m 3for the
uppermost 97m of the shelf yields a total shelf thickness of 382m.
Repeating the calculation with our OPTV-reconstructed density of
870 kg m3for the uppermost 97 m yields a total shelf thickness of
551 m. Although substantially thicker than that calculated from the
‘standard firn’ model, a thickness of 551 m is close to that of 564 m
reconstructed for the area by the Bedmap2 consortium21.This
close correspondence reflects the fact that the (altimetry-based)
Bedmap reconstructions include a correction to account for
enhanced densification, resulting from active surface melting on ice
shelves22. Our OPTV-based density reconstruction also yields a
firn-air content (the column-length equivalent accounted for by
material with a density lower than that of bubble-free glacial ice) of
5.0 m. While this value is substantially lower than the 23.0m that
would result from firn of density 700 kg m 3, it is far closer to the
spatially distributed range of 0–4m predicted for the area based on
recent combined analyses of remotely sensed surface elevation and
shelf thickness fields9,10. However, direct comparisons such as
these are somewhat confounded by our firn-air content of 5.0m
being based on a single-point measurement (recorded, as noted
above, on the limb of an elongate trough hosting an ephemeral
surface pond), whereas the reconstructions based on remotely
sensed data integrate data over a coarser spatial field.
Second, as well as altering its density, the ice column will be
warmed by latent heat released by the freezing of ponded and
percolating meltwater. This effect is quantified by a thermistor
4
3
2
1
0
M:A ratio
Depth (m)
a
b
c
Year
0
2
4
6
8
10
Density (kg m–3)
900
800
700
600
500
400
No ponds
Ponds
2014
2012
2010
2000
2008
2006
2004
2002
1990
1998
1996
1994
1992
Figure 3 | Firn model and remotely sensed outputs for Cabinet Inlet for
the period 1st January 1990 to the time of fieldwork on 1 st December
2014. (a) Ratio of melt to accumulation (M:A) predicted by the firn model
for each 12-month period (1st March–28th February). The horizontal
dashed line marks an M:A ratio of 1. (b) Predicted firn density plotted
against depth. The predicted massive shallow ice layer (dark blue) is never
more than B8 m below the shelf surface and extends to within B1 m of the
surface in years 1993–2001 and 2005–2009, both containing individual
years of high M:A ratio. (c) Classification of time series of summer MODIS
satellite images (available since 2001) of Cabinet Inlet according to the
presence (red bars) or absence (grey bars) of surface ponds.
50
–2,330,000
Easting (m)
–2,320,000
1,160,000
1,170,000
Northing (m)
60
70
Ele. (m a.s.l.)
1
2
3
Figure 4 | Cabinet Inlet radargrams. 200 MHz GPR profiles along the
three transects identified in Fig. 1, overlaid on a Landsat image of the region
from 31st December 2001 when surface melt ponds were present. The
borehole location is marked by the yellow dot.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11897
4NATURE COMMUNICATIONS | 7:11897 | DOI: 10.1038/ncomms11897 | www.nature.com/naturecommunications
string installed into the borehole following OPTV logging
(Fig. 2c). The mean annual temperature measured at a depth of
11 m was 5.9 °C, while the mean annual surface temperature at
the site (which normally defines that at a depth of 10–15 m in the
absence of significant refreezing) is 16.9 °C. Further, the
entire ice profile recorded by our thermistor string (Fig. 2c) is
warmer than would be expected without factoring in refreezing
of ponded meltwater. This effect is confirmed by our firn
model, which under-predicts englacial temperatures despite
accounting for heat released by the refreezing of percolating
meltwater. In this case, modelled pore-water refreezing
contributes B1.5 °C to the firn, yielding a predicted temperature
at 11 m of 15.4 °C (Fig. 2c). This is still almost 10 °C colder
than our measured firn temperature at that depth, demonstrating
that the refreezing of ponded meltwater (not included in the
model) provides significant, hitherto unconsidered, englacial
heating in this region. As well as being denser, this ice is
therefore also warmer than that which would otherwise be used in
numerical models of the flow of the LCIS19.
Replacing on the order of 10 10 0.1 km of standard firn
with relatively warm and dense ice will exert some influence on
ice shelf flow and stability. However, evaluating this influence at
the ice-shelf scale is not straightforward. In the first instance, the
warmer ice layer we identify will be less viscous than colder ice,
and this may go some way to accounting for the anomalously
high-rate factor that has in the past been necessary for numerical
models of the flow of the LCIS, and in particular its northern
sector, to match empirical data23,24. A consequence of such a
temperature-induced acceleration would be a general reduction in
back stress within confined embayments, such as Cabinet Inlet,
potentially increasing shear stresses along the flow unit’s
lateral margins. This process is consistent with observations
that the northern sector of Larsen B Ice Shelf experienced an
increase in lateral rifting before its break up in 2002 (ref. 25). In
contrast, the influence of the layer’s presence on brittle
deformation may be in the opposite direction, with enhanced
ductile flow accommodating strain and the solid ice we report
being more resistant to tensile fracture than lower-density and
finer-grained firn. In such cases, the flow-parallel alignment of
elongate solid ice bodies such as that we report herein might serve
to resist the flow-orthogonal crevassing that commonly develops
in response to longitudinal tensile stresses, approaching the
marine limit of ice shelves. Finally, the spatial extent of this layer
at the ice-shelf scale, and the way in which its deformation
interacts with that of other material units, such as suture ice, and
basal channels and crevasses, will also influence the way in which
the layer’s presence affects overall shelf stability. In particular,
longitudinal troughs located on the surface of ice shelves have
been related to spatially coincident basal channels26, which have
themselves been associated with shelf instability27 through local
thinning and crevassing28,29. Although it is not yet known
whether the surface troughs on LCIS investigated herein are
associated with such basal channels, the recent finding that
material density is locally enhanced below similar surface troughs
on the Roi Baudouin Ice Shelf, Antarctica30, is consistent with
the enhanced melting and massive ice formation reported herein.
In the light of these complexities, identifying and exploring the
net influence of pond ice formation on ice-shelf stability can
only be achieved with confidence by including the full spatial
extent and physical properties of the refrozen ice layer into
a shelf-wide flow and fracture model, which may be regarded as a
future research priority.
The surface ponding responsible for the subsurface ice layer we
report herein is currently restricted to warmer regions of
Antarctica’s fringing ice shelves and in particular, but not
exclusively, to the northern and western sectors of the Antarctic
Peninsula. However, regional warming is predicted to spread
southwards and intensify substantially over forthcoming
decades5,31, so ponding is expected also to become more
widespread. Similarly, massive layers of warm, dense ice are
therefore highly likely to form within ice shelves present across a
substantial area of Antarctica, perhaps the entire continent if
warming continues into the 22nd century5, with important
consequences for ice-shelf flow, hydrofracture and stability.
Methods
Borehole drilling and logging.The borehole was drilled by pressurized hot water
and logged by OPTV13. The resulting OPTV log, analysed by WellCAD and BIFAT
software32, provides a geometrically accurate image, with a pixel size of B1mm,of
the material composition of the complete borehole wall, as well as the structural
geometry of layers and inclusions intersecting it33. Since the OPTV log records an
image of reflected light, it also provides a proxy for the density of compacted snow,
firn and ice due to the progressive decrease in reflectivity, as voids close and bubbles
are occluded and collapsed13,15. Hubbard et al.15 exploited this relationship and
identified an exponential relationship between OPTV luminosity (L) and material
density (D) on the basis of samples recovered from a core retrieved from a borehole
logged by OPTV on the Roi Baudouin Ice Shelf, East Antarctica. However, due to
equipment loss, the light-emitting diode brightness of the OPTV probe used in the
current study was different from that of Hubbard et al.15, and a new calibration was
undertaken. This was based on the correlation of 40 core samples recovered from a
logged borehole, again located on the Roi Baudouin Ice Shelf (Fig. 5). The new
calibration yields a best-fit regression equation of D¼950–40.1 e(0.0101L)(R2¼0.82),
with root mean square values of the residuals of 40.4, 35.2 and 21.7 kg m 3for the
density ranges 600–700, 700–800 and 800–900kg m 3, respectively. While these
values provide an error range for absolute densities derived from OPTV luminosity,
the relative changes in density reported herein, that is, along a single borehole log,
reduce to the precision of the method. This is approximated by the density range
(910 kg m 3) divided by the luminosity range (256), or B3.5kg m 3.
Borehole temperatures (Fig. 2c) were recorded by negative temperature
coefficient thermistors, recorded across a Wheatstone half-bridge by Campbell
Scientific micro-loggers. Resistances were converted to temperature using a
polynomial34 fitted to the manufacturer’s calibration curve refined via a second-stage
calibration in a distilled water/ice bath. Once recalibrated, sensors were replaced into
a new water/ice bath to determine temperature error, yielding a root mean square
error of ±0.03 °C. Borehole temperatures were logged for at least 100 days, and the
undisturbed ice temperature was calculated from an exponential function fitted to
the cooling curve35. By this time, all temperatures recorded below the near-surface
zone of seasonal thermal disturbance had stabilized to values that varied (over
timescales of days to weeks) by o0.05°C. Thus, the englacial temperatures we report
herein were not influenced by the hot-water drilling process.
Firn modelling.The firn model used is IMAU-FDM v1.0, which takes into account
firn compaction, meltwater percolation and refreezing17,18. At the surface, the firn
model is forced with mass fluxes (snowfall, snowmelt, rain, sublimation, and
snowdrift sublimation and erosion) and surface temperature from the regional
500
600
700
800
900
20 40 60 80 100 120 140 160 180 200 220
OPTV luminosity (0 – 255)
D = 950 – 40.1 e
(0.0101L)
R2 = 0.82
Density (kg m–3)
Figure 5 | Optical televiewer density–luminosity calibration. Material
density, measured gravimetrically on core samples, plotted against
equivalent OPTV luminosity, both measured along a borehole located on
Roi Baudouin Ice Shelf, East Antarctica.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11897 ARTICLE
NATURE COMMUNICATIONS | 7:11897 | DOI: 10.1038/ncomms11897 | www.nature.com/naturecommunications 5
climate model RACMO2.3, run at a 5.5-km horizontal resolution in a domain over
the Antarctic Peninsula.
Satellite image analysis.MODIS Terra and Aqua level 1 (250 m) data were
ordered via the level 1 and atmosphere archive and distribution system, and used to
check for the presence of melt ponds in Cabinet Inlet. A total of 577 cloud-free
band 2 (848 nm) images between 1st December and 30th April from 2000 (Aqua)
or 2002 (Terra) to 2015 were classified to generate the time series shown in Fig. 3c.
Ground-penetrating radar.GPR data (Fig. 4) were collected using a Sensors and
Software Pulse Ekko Pro system operated in common offset mode with a 0.8-ns
sample interval and a 4,000-ns sample window. The console was mounted on the
skidoo and the 200MHz antennae towed 15 m behind on a plastic sledge at
B10 km h 1with an eight stack trace recorded every 3m. A Leica VIVA GS10
GNSS rover unit was connected directly to the GPR console; the base station was
located at the drill site. Surface elevation was referenced to the Earth Gravitational
Model 1996 geoid, taken to represent sea level, and no correction was made for
dynamic topography. GPR data were processed in Reflex-W. Processing steps
included de-wow of 25-ns filter length, spherical divergence compensation, spectral
whitening between 70 and 200MHz, and a two-dimensional mean-averaging filter.
The static correction was based on Leica Geo Office post-processed GNSS data,
resulting in a vertical accuracy of ±0.5 m. Radar wave propagation velocities used to
convert travel times to the depths shown in Fig. 4 were 0.2m ns1for the upper-
most 3 m (based on a local 500 MHz common midpoint gather from the uppermost
reflector of Unit 1) and a typical value for glacial ice36 of 0.17 m ns 1below that.
Data availability.The data that support the findings of this study are available
from http://www.projectmidas.org/data/hubbard2016/ and from the corresponding
author upon request.
References
1. Holland, P. R. et al. Oceanic and atmospheric forcing of Larsen C Ice-Shelf
thinning. Cryosphere 9, 1005–1024 (2015).
2. Holland, P. R. et al. The air content of Larsen Ice Shelf. Geophys. Res. Lett. 38,
L10503 (2011).
3. Luckman, A. et al. Basal crevasses in Larsen C Ice Shelf and implications for
their global abundance. Cryosphere 6, 113–123 (2012).
4. McGrath, D. et al. Basal crevasses on the Larsen C Ice Shelf, Antarctica:
Implications for meltwater ponding and hydrofracture. Geophys. Res. Lett. 39,
L16504 (2012).
5. Luckman, A. et al. Surface melt and ponding on Larsen C Ice Shelf and the
impact of fo
¨hn winds. Antarctic Sci. 26, 625–635 (2014).
6. Kuipers Munneke, P., van den Broeke, M. R., King, J. C., Gray, T. & Reijmer, C.
H. Near-surface climate and surface energy budget of Larsen C ice shelf,
Antarctic Peninsula. Cryosphere 6, 353–363 (2012).
7. Cape, M. R. et al. Foehn winds link climate-driven warming to ice shelf
evolution in Antarctica. J. Geophys. Res. 120, 11037–11057 (2015).
8. van den Broeke, M. Strong surface melting preceded collapse of Antarctic
Peninsula ice shelf. Geophys. Res. Lett. 32, L12815 (2005).
9. Scambos, T. et al. Ice shelf disintegration by plate bending and hydro-fracture:
satellite observations and model results of the 2008 Wilkins ice shelf break-ups.
Earth Planet Sci. Lett. 280, 51–60 (2009).
10. Kuipers Munneke, P., Ligtenberg, S. R. M., Van Den Broeke, M. R. & Vaughan,
D. G. Firn air depletion as a precursor of Antarctic ice-shelf collapse. J. Glaciol.
60, 205–214 (2014).
11. Glasser, N. F. & Scambos, T. A. A structural glaciological analysis of the 2002
Larsen B ice-shelf collapse. J. Glaciol. 54, 3–16 (2008).
12. Banwell, A. F., MacAyeal, D. R. & Sergienko, O. V. Breakup of the Larsen B Ice
Shelf triggered by chain reaction drainage of supraglacial lakes. Geophys. Res.
Lett. 40, 5872–5876 (2013).
13. Hubbard, B., Roberson, S., Samyn, D. & Merton-Lyn, D. Digital optical
televiewing of ice boreholes. J. Glaciol. 54, 1–8 (2008).
14. Hubbard, B. & Malone, T. Optical-televiewer-based logging of the uppermost 630
m of the NEEM deep ice borehole, Greenland. Ann. Glaciol. 54, 83–89 (2013).
15. Hubbard, B. et al. Ice-shelf density reconstructed from optical-televiewer
borehole logging. Geophys. Res. Lett. 2013, GL058023 (2013).
16. Hubbard, B. et al. Optical-televiewer-based identification and characterization
of material facies associated with an Antarctic ice-shelf rift. Ann. Glaciol. 53,
137–146 (2012).
17. Ligtenberg, S. R. M., Helsen, M. M. & van den Broeke, M. R. An improved
semi-empirical model for the densification of Antarctic firn. Cryosphere 5,
809–819 (2011).
18. van Wessem, J. M. et al. The modelled surface mass balance of the Antarctic
Peninsula at 5.5 km horizontal resolution. Cryosphere 10, 271–285 (2016).
19. Jansen, D. et al. Present stability of the Larsen C ice shelf, Antarctic Peninsula.
J. Glaciol. 56, 593–600 (2010).
20. Bamber, J. & Bentley, C. A comparison of satellite-altimetry and ice-thickness
measurements of the Ross Ice Shelf, Antarctica. Ann. Glaciol. 20, 357–364 (1994).
21. Fretwell, P. et al. Bedmap2: improved ice bed, surface and thickness datasets for
Antarctica. Cryosphere 7, 375–393 (2013).
22. Griggs, J. A. & Bamber, J. L. Ice shelf thickness over Larsen C, Antarctica,
derived from satellite altimetry. Geophys. Res. Lett. 36, L19501 (2009).
23. Khazendar, A., Rignot, E. & Larour, E. Acceleration and spatial rheology of
Larsen C Ice Shelf, Antarctic Peninsula. Geophys. Res. Lett. 38, L09502 (2011).
24. Borstad, C. P., Rignot, E., Mouginot, J. & Schodlok, M. P. Creep deformation
and buttressing capacity of damaged ice shelves: theory and application to
Larsen C ice shelf. Cryosphere 7, 1931–1947 (2013).
25. Rack, W. & Rott, H. Pattern of retreat and disintegration of Larsen B Ice Shelf,
Antarctic Peninsula. Ann. Glaciol. 39, 505–510 (2004).
26. Stanton, T. P. et al. Channelized ice melting in the ocean boundary layer
beneath Pine Island Glacier, Antarctica. Science 341, 1236–1239 (2013).
27. Alley, K. E., Scambos, T. A., Siegfried, M. R. & Fricker, H. A. Impacts of warm
water on Antarctic ice shelf stability through basal channel formation. Nat.
Geosci. 9, 290–293 (2016).
28. Vaughan, D. G. et al. Subglacial melt channels and fracture in the floating part
of Pine Island Glacier, Antarctica. J. Geophys. Res. 117, F03012 (2012).
29. Sergienko, O. V. Basal channels on ice shelves. J. Geophys. Res. 118, 1342–1355
(2013).
30. Drews, R. et al. Constraining variable density of ice shelves using wide-angle
radar measurements. Cryosphere 10, 811–823 (2016).
31. Ligtenberg, S. R. M., van de Berg, W. J., van den Broeke, M. R., Rae, J. G. L. &
van Meijgaard, E. Future surface mass balance of the Antarctic ice sheet and its
influence on sea level change, simulated by a regional atmospheric climate
model. Clim. Dyn. 41, 867–884 (2013).
32. Malone, T., Hubbard, B., Merton-Lyn, D., Worthington, P. & Zwiggelaar, R.
Borehole and Ice Feature Annotation Tool (BIFAT): a program for the
automatic and manual annotation of glacier borehole images. Comput. Geosci.
51, 381–389 (2013).
33. Roberson, S. & Hubbard, B. Application of borehole optical televiewing to
investigating the 3-D structure of glaciers: implications for the formation of
longitudinal debris ridges, midre Lovenbreen, Svalbard. J. Glaciol. 56, 143–156
(2010).
34. Steinhart, J. S. & Hart, S. R. Calibration curves for thermistors. Deep Sea Res.
Oceanogr. Abstr. 15, 497–503 (1968).
35. Humphrey, N. & Echelmeyer, K. Hot-water drilling and borehole closure in
cold ice. J. Glaciol. 36, 287–298 (1990).
36. Plewes, L. A. & Hubbard, B. A review of the use of radio-echo sounding in
glaciology. Progress Phys. Geogr. 25, 203–236 (2001).
Acknowledgements
This research was funded by the Natural Environment Research Council, grants NE/
L005409/1 and NE/L006707/1, and Aberystwyth University’s Capital Equipment fund.
The Leica VIVA GS10 GNSS system and Sensors and Software Pulse Ekko Pro radar
transmitter were loaned by the Natural Environment Research Council Geophysical
Equipment Facility, loan number 1028. We thank the British Antarctic Survey for
logistical support, and in particular the project’s field assistants Ashly Fusiarski and
Nicholas Gillett.
Author contributions
A.L. and B.H. led the overall project. Fieldwork was carried out by D.A., S.B., B.H., B.K.,
H.S., A.B. and A.L. D.J., M.O.L. and I.R. contributed to ice-shelf modelling, guiding
the field programme and interpretations. S.B. and A.L. led the remote-sensing
contributions. P.K.M. carried out the firn modelling. M.P., J.-L.T. and B.H. contributed
to the OPTV calibration. All authors contributed to writing the manuscript, which
was led by B.H.
Additional information
Competing financial interests: The authors declare no competing financial interests.
Reprints and permission information is available online at http://npg.nature.com/
reprintsandpermissions/
How to cite this article: Hubbard, B. et al. Massive subsurface ice formed by
refreezing of ice-shelf melt ponds. Nat. Commun. 7:11897 doi: 10.1038/ncomms11897
(2016).
This work is licensed under a Creative Commons Attribution 4.0
International License. The images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise
in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material.
To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11897
6NATURE COMMUNICATIONS | 7:11897 | DOI: 10.1038/ncomms11897 | www.nature.com/naturecommunications
... Luckman et al. (2014) and Bevan et al. (2018) used satellite measurements to show that the annual melt duration on Larsen C is highest in the north, where temperatures are closer to the melting point, and in inlets close to the mountains, where foehn winds are most intense and frequent (Elvidge et al., 2015(Elvidge et al., , 2020Turton et al., 2018). These foehn-induced eastwest gradients in melt are also seen in borehole and firn measurements Holland et al., 2011;Hubbard et al., 2016). ...
... The simulated spatial pattern of meltwater production ( Figure 3) and number of melt days ( Figure S4 in Supporting Information S1) is consistent with satellite observations of the annual number of days that surface melting occurs (e.g., Bevan et al., 2018;Luckman et al., 2014), with a clear northsouth gradient across the ice shelf, and more melting observed in inlets. The hindcast also simulates peak mean meltwater production during the high melt years identified in Bevan et al. (2018Bevan et al. ( ), for example, the 2006Bevan et al. ( /2007Bevan et al. ( , 2015Bevan et al. ( /2016Bevan et al. ( , and 2016Bevan et al. ( /2017 melt seasons, when ice shelf averaged cumulative annual melt of 187 mm w.e., 157 mm w.e. and 161 mm w.e. over 56 days, 48 and 54 days, respectively, is modeled. The spatial patterns of surface melt shown in Bevan et al. (their Figure 6) are quite closely reproduced in Figure S4 in Supporting Information S1, which shows the number of melt days per year. ...
... The simulated spatial pattern of meltwater production ( Figure 3) and number of melt days ( Figure S4 in Supporting Information S1) is consistent with satellite observations of the annual number of days that surface melting occurs (e.g., Bevan et al., 2018;Luckman et al., 2014), with a clear northsouth gradient across the ice shelf, and more melting observed in inlets. The hindcast also simulates peak mean meltwater production during the high melt years identified in Bevan et al. (2018Bevan et al. ( ), for example, the 2006Bevan et al. ( /2007Bevan et al. ( , 2015Bevan et al. ( /2016Bevan et al. ( , and 2016Bevan et al. ( /2017 melt seasons, when ice shelf averaged cumulative annual melt of 187 mm w.e., 157 mm w.e. and 161 mm w.e. over 56 days, 48 and 54 days, respectively, is modeled. The spatial patterns of surface melt shown in Bevan et al. (their Figure 6) are quite closely reproduced in Figure S4 in Supporting Information S1, which shows the number of melt days per year. ...
Article
Full-text available
Following collapses of the neighboring Larsen A and B ice shelves, Larsen C has become a focus of increased attention. Determining how the prevailing meteorological conditions influence its surface melt regime is of paramount importance for understanding the dominant processes causing melt and ultimately for predicting its future. To this end, a new, high‐resolution (4 km grid spacing) Met Office Unified Model (MetUM) hindcast of atmospheric conditions and surface melt processes over the central Antarctic Peninsula is introduced. The hindcast is capable of simulating observed near‐surface meteorology and surface melt conditions over Larsen C. In contrast with previous model simulations, the MetUM captures the observed east‐west gradient in surface melting associated with foehn winds, as well as the interannual variability in melt shown in previous observational studies. As exemplars, we focus on two case studies—the months preceding the collapse of the Larsen B ice shelf in March 2002 and the high foehn, high melt period of March‐May 2016—to test the hindcast's ability to reproduce the atmospheric effects that contributed to considerable melting during those periods. The results suggest that the MetUM hindcast is a useful tool with which to explore the dominant causes of surface melting on Larsen C.
... Of concern is that SGLs are expected to become more extensive on Antarctic ice shelves, due to increases in surface melt extent and intensity in response to future atmospheric warming [11][12][13][14] . Regular, prolonged surface melt reduces the meltwater retention capacity of ice shelves by saturating their firn layer and reducing firn air content (FAC) 11,[15][16][17][18] . Excess meltwater that cannot be stored in the firn runs off to form SGLs on the snow or ice surface, filling topographic hollows, including rifts and crevasses 16 . ...
... Depleted FAC, refrozen subsurface meltwater and high surface runoff volumes have been linked to SGL formation on East Antarctic ice shelves 22,27,45 and on the Antarctic Peninsula 17,47 and can render ice shelves more susceptible to hydrofracture 14 . Meltwater percolating through the firn that encounters shallower impermeable ice lenses (formed by refrozen meltwater) can percolate less far into the ice before fully saturating the snowpack 17,40,48 . Continued surface melting over successive melt seasons gradually depletes FAC when more pore space is lost by melt and refreezing during densification than is replenished by snowfall 44 . ...
... Grounding line from ref. 77 and coastline from ref. 78 . exert a localised warming effect on ice temperatures through the release of latent heat 16,17 . When firn saturation prevents meltwater percolation and refreezing within the firn, the firn is flooded and excess surface runoff can form SGLs 18,38,49 . ...
Article
Full-text available
Antarctic supraglacial lakes (SGLs) have been linked to ice shelf collapse and the subsequent acceleration of inland ice flow, but observations of SGLs remain relatively scarce and their interannual variability is largely unknown. This makes it difficult to assess whether some ice shelves are close to thresholds of stability under climate warming. Here, we present the first observations of SGLs across the entire East Antarctic Ice Sheet over multiple melt seasons (2014–2020). Interannual variability in SGL volume is >200% on some ice shelves, but patterns are highly asynchronous. More extensive, deeper SGLs correlate with higher summer (December-January-February) air temperatures, but comparisons with modelled melt and runoff are complex. However, we find that modelled January melt and the ratio of November firn air content to summer melt are important predictors of SGL volume on some potentially vulnerable ice shelves, suggesting large increases in SGLs should be expected under future atmospheric warming.
... When snow and ice melts on the surface of ice shelves, producing meltwater, this meltwater percolates downwards or flows across the surface, producing surface channels, ponds and firn aquifers. Drainage systems on Antarctic ice shelves' are observed to be large and persistent in time (Mellor and McKinnon, 1960;Reynolds, 1981;Phillips, 1998;Banwell et al., 2013;Hubbard et al., 2016;Langley et al., 2016;Lenaerts et al., 2016b;Buzzard et al., 2018;Stokes et al., 2019;Arthur et al., 2020). Meltwater forms in areas where the surface energy balance brings ice to the melting temperature (Schlatter, 1972). ...
... Surface meltwater drainage systems have been regularly observed in summer on Antarctic ice shelves since exploration of these regions began (Mellor and McKinnon, 1960;. Meltwater flows laterally, creates channels and fills surface depressions (Winther et al., 1996;Lenaerts et al., 2016a;Dell et al., 2020), and in some cases drains englacially, into or through the ice shelf (e.g., Dow et al., 2018;Stokes et al., 2019;Dunmire et al., 2020;Schaap et al., 2020;Warner et al., 2021) or into the firn (e.g., Hubbard et al., 2016;Montgomery et al., 2020). The change in local load from water movement and ponding on the shelf can cause flexural stresses large enough to generate fractures (MacAyeal et al., 2015). ...
... It is unclear if this variability can be explained by or is linked to seasonal meltwater production, or whether more complex glaciohydrological coupling is at play. For example, large refrozen ice masses have been found advected down-stream of lakes on Larsen C Ice Shelf (Hubbard et al., 2016) and Amery Ice Shelf (Phillips, 1998), reducing firn permeability over large areas. If this encourages subsequent surface meltwater transport, as has been observed in Greenland (MacFerrin et al., 2019), complex, multi-year coupling between meltwater production, drainage, and refreezing is possible. ...
Thesis
In this thesis, I have used remote sensing and modeling techniques to investigate Antarctic ice shelf surface hydrology with the purpose of answering three key questions: 1) How do surface drainage systems evolve over a typical summertime melt season, over several consecutive melt seasons, and over several decades? 2) What controls the expansion of surface hydrology networks? and 3) Will surface drainage expand into areas vulnerable to hydrofracture and important for buttressing when meltwater volume increases in a warmer, future climate?
... This can induce a positive feedback whereby previously melted areas are more likely to experience further melting, due to the increased absorption of short-wave radiation associated with low-albedo surfaces (Kingslake et al., 2017). It is possible that this feedback is further enhanced by the presence of ice slabs and lenses, which can from beneath areas of intermittent pond formation (Hubbard et al., 2016). These dense layers of ice inhibit meltwater percolation and can be several degrees warmer than ice that has not undergone lateral heat fluctuations that result from the melting and refreezing of ice (Hubbard et al., 2016). ...
... It is possible that this feedback is further enhanced by the presence of ice slabs and lenses, which can from beneath areas of intermittent pond formation (Hubbard et al., 2016). These dense layers of ice inhibit meltwater percolation and can be several degrees warmer than ice that has not undergone lateral heat fluctuations that result from the melting and refreezing of ice (Hubbard et al., 2016). Such ice slabs have been shown to have important implications for lake development over multiple melt seasons, based on modelling of the Larsen C ice shelf (Buzzard et al., 2018). ...
... However, given the vast amount of ice that is discharged through the AIS, it is crucial that we continue to develop our understanding of how varying levels of surface meltwater can influence hydrological and ice dynamic processes in the region. Repeated cycles of melting and refreezing at the ice surface releases latent heat, weakening the ice structure and making it more prone to future climatic perturbations (Hubbard et al., 2016). Changes in temperature or precipitation patterns, in addition to predicted ocean warming, could also influence the vulnerability of the ice shelf to melt-induced fracture. ...
Article
Full-text available
Surface meltwater is widespread around the Antarctic Ice Sheet margin and has the potential to influence ice shelf stability, ice flow and ice–albedo feedbacks. Our understanding of the seasonal and multi-year evolution of Antarctic surface meltwater is limited. Attempts to generate robust meltwater cover time series have largely been constrained by computational expense or limited ice surface visibility associated with mapping from optical satellite imagery. Here, we add a novel method for calculating visibility metrics to an existing meltwater detection method within Google Earth Engine. This enables us to quantify uncertainty induced by cloud cover and variable image data coverage, allowing time series of surface meltwater area to be automatically generated over large spatial and temporal scales. We demonstrate our method on the Amery Ice Shelf region of East Antarctica, analysing 4164 Landsat 7 and 8 optical images between 2005 and 2020. Results show high interannual variability in surface meltwater cover, with mapped cumulative lake area totals ranging from 384 to 3898 km2 per melt season. By incorporating image visibility assessments, however, we estimate that cumulative total lake areas are on average 42 % higher than minimum mapped values. We show that modelled melt predictions from a regional climate model provide a good indication of lake cover in the Amery region and that annual lake coverage is typically highest in years with a negative austral summer SAM index. Our results demonstrate that our method could be scaled up to generate a multi-year time series record of surface water extent from optical imagery at a continent-wide scale.
... The daily melt anomalies were computed by comparison with this climatology and compared with AR occurrences. Data obtained before 1993 must be considered with caution, in particular with summer 1987-88 29 Finally, to complement our analysis on the relationship between melt pond observations and AR activity obtained using the MODIS image interpretation by Hubbard et al. 2016 32 , we also studied the Landsat8 image database 77 (Moussavi et al., 2020), used by Banwell et al., 2021 29 . Information on lake extent on the Larsen ice shelves were retrieved and merged over periods of 16 days (corresponding to the Landsat revisit time). ...
... High runoff values, produced by melting and rainfall when the rn is already saturated with liquid water, are detrimental for ice-shelf stability 31 . Indeed, these high melt and runoff rates lead to melt pond formation as demonstrated by the signi cant correlation between melt pond observations from MODIS and summer AR occurrences 32 (Supplementary Fig. 7). ...
Preprint
Full-text available
The disintegration of the ice shelves along the Antarctic Peninsula have spurred much discussion on the various processes leading to their eventual dramatic collapse, but without a consensus on an atmospheric forcing that could connect these processes. Here, using an atmospheric river (AR) detection algorithm along with a regional climate model and satellite observations, we show that particularly intense ARs have a ~40% probability of inducing extreme events of temperature, surface melt, sea-ice disintegration, or large swells; all processes proven to induce ice-shelf destabilization. This was observed during the collapses of the Larsen A, B, and overall, 60% of calving events triggered by ARs from 2000-2020. The loss of the buttressing effect from these ice shelves leads to further continental ice loss and subsequent sea-level rise. Understanding how ARs connect various disparate processes cited in ice-shelf collapse theories is essential for identifying other at-risk ice shelves like the Larsen C.
... When firn pore spaces become saturated, slush is formed and this may be particularly likely where firn overlies former blue ice areas or refrozen lakes, or where refreezing of infiltrated water has formed extensive ice layers at depth within the firn. Melting and refreezing of slush promotes firn air content depletion, thereby increasing its density and increasing an ice shelf's vulnerability to ponding (Kuipers Munneke and others, 2014;Hubbard and others, 2016;Alley and others, 2018). Ponded water has been shown to drive ice-shelf collapse events through hydrofracture others, 2003, 2004;Banwell and others, 2013;Banwell and MacAyeal, 2015;Robel and Banwell, 2019) and therefore several studies have mapped the changing extent of ponded water on ice shelves (e.g. ...
Article
Full-text available
Surface meltwater is becoming increasingly widespread on Antarctic ice shelves. It is stored within surface ponds and streams, or within firn pore spaces, which may saturate to form slush. Slush can reduce firn air content, increasing an ice-shelf's vulnerability to break-up. To date, no study has mapped the changing extent of slush across ice shelves. Here, we use Google Earth Engine and Landsat 8 images from six ice shelves to generate training classes using a k -means clustering algorithm, which are used to train a random forest classifier to identify both slush and ponded water. Validation using expert elicitation gives accuracies of 84% and 82% for the ponded water and slush classes, respectively. Errors result from subjectivity in identifying the ponded water/slush boundary, and from inclusion of cloud and shadows. We apply our classifier to the Roi Baudouin Ice Shelf for the entire 2013–20 Landsat 8 record. On average, 64% of all surface meltwater is classified as slush and 36% as ponded water. Total meltwater areal extent is greatest between late January and mid-February. This highlights the importance of mapping slush when studying surface meltwater on ice shelves. Future research will apply the classifier across all Antarctic ice shelves.
... The short-term surface melt events associated with HWs on ice shelves may not significantly contribute to the annual Antarctic mass losses (volume change) because in Antarctica most of the annual meltwater from snowmelt refreezes 33 . Snow properties change in case of melting much more than in case of meltwater refreezing. ...
Article
Full-text available
Summer temperatures are often above freezing along the Antarctic coastline, which makes ice shelves and coastal snowpacks vulnerable to warming events (understood as periods of consecutive days with warmer than usual conditions). Here, we project changes in the frequency, duration and amplitude of summertime warming events expected until end of century according to two emission scenarios. By using both global and regional climate models, we found that these events are expected to be more frequent and last longer, continent-wide. By end of century, the number of warming events is projected to double in most of West Antarctica and to triple in the vast interior of East Antarctica, even under a moderate-emission scenario. We also found that the expected rise of warming events in coastal areas surrounding the continent will likely lead to enhanced surface melt, which may pose a risk for the future stability of several Antarctic ice shelves.
Article
Full-text available
The disintegration of the ice shelves along the Antarctic Peninsula have spurred much discussion on the various processes leading to their eventual dramatic collapse, but without a consensus on an atmospheric forcing that could connect these processes. Here, using an atmospheric river detection algorithm along with a regional climate model and satellite observations, we show that the most intense atmospheric rivers induce extremes in temperature, surface melt, sea-ice disintegration, or large swells that destabilize the ice shelves with 40% probability. This was observed during the collapses of the Larsen A and B ice shelves during the summers of 1995 and 2002 respectively. Overall, 60% of calving events from 2000–2020 were triggered by atmospheric rivers. The loss of the buttressing effect from these ice shelves leads to further continental ice loss and subsequent sea-level rise. Under future warming projections, the Larsen C ice shelf will be at-risk from the same processes. The most intense atmospheric rivers to hit the Antarctic Peninsula induce extremes in temperature, surface melt, sea ice disintegration or swell that destabilize the ice shelves with 40% probability, suggest analyses of observations and regional climate model simulations.
Article
Ice shelf collapse reduces buttressing and enables grounded glaciers to contribute more rapidly to sea-level rise in a warming climate. The abrupt collapses of the Larsen A (1995) and B (2002) ice shelves on the Antarctic Peninsula (AP) occurred, at least for Larsen B, when long-period ocean swells damaged the calving front and the ice shelf was inundated with melt lakes that led to large-scale hydrofracture cascades. During collapse, field and satellite observations indicate föhn winds were present on both ice shelves. Here we use a regional climate model and machine learning analyses to evaluate the contributory roles of föhn winds and associated melt events prior to and during the collapses for ice shelves on the AP. Föhn winds caused about 25 % ± 3 % of the total annual melt in just 9 d on Larsen A prior to and during collapse and were present during the Larsen B collapse, which helped form extensive melt lakes. At the same time, the off-coast wind direction created by föhn winds helped melt and physically push sea ice away from the ice shelf calving fronts that allowed long-period ocean swells to reach and damage the front, which has been theorized to have ultimately triggered collapse. Collapsed ice shelves experienced enhanced surface melt driven by föhn winds over a large spatial extent and near the calving front, whereas SCAR inlet and the Larsen C ice shelves are affected less by föhn-wind-induced melt and do not experience large-scale melt ponds. These results suggest SCAR inlet and the Larsen C ice shelves may be less likely to experience rapid collapse due to föhn-driven melt so long as surface temperatures and föhn occurrence remain within historical bounds.
Article
Full-text available
We present Bedmap2, a new suite of gridded products describing surface elevation, ice-thickness and the seafloor and subglacial bed elevation of the Antarctic south of 60 • S. We derived these products using data from a variety of sources, including many substantial surveys completed since the original Bedmap compilation (Bedmap1) in 2001. In particular, the Bedmap2 ice thickness grid is made from 25 million measurements, over two orders of magnitude more than were used in Bedmap1. In most parts of Antarc-tica the subglacial landscape is visible in much greater detail than was previously available and the improved data-coverage has in many areas revealed the full scale of mountain ranges, valleys, basins and troughs, only fragments of which were previously indicated in local surveys. The derived statistics for Bedmap2 show that the volume of ice contained in the Antarctic ice sheet (27 million km 3) and its potential contribution to sea-level rise (58 m) are similar to those of Bedmap1, but the mean thickness of the ice sheet is 4.6 % greater, the mean depth of the bed beneath the grounded ice sheet is 72 m lower and the area of ice sheet grounded on bed below sea level is increased by 10 %. The Bedmap2 compilation highlights several areas beneath the ice sheet where the bed elevation is substantially lower than the deepest bed indicated by Bedmap1. These products, along with grids of data coverage and uncertainty, provide new opportunities for detailed modelling of the past and future evolution of the Antarctic ice sheets.
Article
Full-text available
Around the perimeter of Antarctica, much of the ice sheet discharges to the ocean through floating ice shelves. The buttressing provided by ice shelves is critical for modulating the flux of ice into the ocean, and the presently observed thinning of ice shelves is believed to be reducing their buttressing capacity and contributing to the acceleration and thinning of the grounded ice sheet. However, relatively little attention has been paid to the role that fractures play in the flow and stability of ice shelves and their capacity to buttress the flow of grounded ice. Here, we develop an analytical framework for describing the role that fractures play in the creep deformation and buttressing capacity of ice shelves. We apply principles of continuum damage mechanics to derive a new analytical relation for the creep of an ice shelf as a function of ice thickness, temperature, material properties, resistive backstress and damage. By combining this analytical theory with an inverse method solution for the spatial rheology of an ice shelf, both backstress and damage can be calculated. We demonstrate the applicability of this new theory using satellite remote sensing and Operation IceBridge data for the Larsen C ice shelf, finding damage associated with known crevasses and rifts. We find that increasing thickness of mélange between rift flanks correlates with decreasing damage, with some rifts deforming coherently with the ice shelf as if completely healed. We quantify the stabilizing backstress caused by ice rises and lateral confinement, finding high backstress associated with two ice rises that likely stabilize the ice front in its current configuration. Though overall the ice shelf appears stable at present, the ice in contact with the Bawden ice rise is weakened by fractures, and additional damage or thinning in this area could portend significant change for the shelf. Using this new approach, field and remote sensing data can be utilized to monitor the structural integrity of ice shelves, their ability to buttress the flow of ice at the grounding line, and thus their indirect contribution to ice sheet mass balance and global sea level.
Article
Full-text available
The catastrophic collapses of Larsen A and B ice shelves on the eastern Antarctic Peninsula have caused their tributary glaciers to accelerate, contributing to sea-level rise and freshening the Antarctic Bottom Water formed nearby. The surface of Larsen C Ice Shelf (LCIS), the largest ice shelf on the peninsula, is lowering. This could be caused by unbalanced ocean melting (ice loss) or enhanced firn melting and compaction (englacial air loss). Using a novel method to analyse eight radar surveys, this study derives separate estimates of ice and air thickness changes during a 15-year period. The uncertainties are considerable, but the primary estimate is that the surveyed lowering (0.066 ± 0.017 m yr−1) is caused by both ice loss (0.28 ± 0.18 m yr−1) and firn-air loss (0.037 ± 0.026 m yr−1). The ice loss is much larger than the air loss, but both contribute approximately equally to the lowering because the ice is floating. The ice loss could be explained by high basal melting and/or ice divergence, and the air loss by low surface accumulation or high surface melting and/or compaction. The primary estimate therefore requires that at least two forcings caused the surveyed lowering. Mechanisms are discussed by which LCIS stability could be compromised in the future. The most rapid pathways to collapse are offered by the ungrounding of LCIS from Bawden Ice Rise or ice-front retreat past a "compressive arch" in strain rates. Recent evidence suggests that either mechanism could pose an imminent risk.
Article
Full-text available
Basal crevasses extend upwards from the base of ice bodies and can penetrate more than halfway through the ice column under conditions found commonly on ice shelves. As a result, they may locally modify the exchange of mass and energy between ice shelf and ocean, and by altering the shelf's mechanical properties could play a fundamental role in ice shelf stability. Although early studies revealed that such features may be abundant on Antarctic ice shelves, their geometrical properties and spatial distribution has gained little attention. We investigate basal crevasses in Larsen C Ice Shelf using field radar survey, remote sensing and numerical modelling. We demonstrate that a group of features visible in MODIS imagery are the surface expressions of basal crevasses in the form of surface troughs, and find that basal crevasses can be generated as a result of stresses well downstream of the grounding line. We show that linear elastic fracture mechanics modelling is a good predictor of basal crevasse penetration height where stresses are predominantly tensile, and that measured surface trough depth does not always reflect this height, probably because of snow accumulation in the trough, marine ice accretion in the crevasse, or stress bridging from the surrounding ice. We conclude that all features visible in MODIS imagery of ice shelves and previously labelled simply as "crevasses", where they are not full thickness rifts, must be basal crevasse troughs, highlighting a fundamental structural property of many ice shelves that may have been previously overlooked.
Article
Full-text available
Data collected by two automatic weather stations (AWS) on the Larsen C ice shelf, Antarctica, between 22 January 2009 and 1 February 2011 are analyzed and used as input for a model that computes the surface energy budget (SEB), which includes melt energy. The two AWSs are separated by about 70 km in the north–south direction, and both the near-surface meteorology and the SEB show similarities, although small differences in all components (most notably the melt flux) can be seen. The impact of subsurface absorption of shortwave radiation on melt and snow temperature is significant, and discussed. In winter, longwave cooling of the surface is entirely compensated by a downward turbulent transport of sensible heat. In summer, the positive net radiative flux is compensated by melt, and quite frequently by upward turbulent diffusion of heat and moisture, leading to sublimation and weak convection over the ice shelf. The month of November 2010 is highlighted, when strong westerly flow over the Antarctic Peninsula led to a dry and warm föhn wind over the ice shelf, resulting in warm and sunny conditions. Under these conditions the increase in shortwave and sensible heat fluxes is larger than the decrease of net longwave and latent heat fluxes, providing energy for significant melt.
Article
Full-text available
We present Bedmap2, a new suite of gridded products describing surface elevation, ice-thickness and the seafloor and subglacial bed elevation of the Antarctic south of 60° S. We derived these products using data from a variety of sources, including many substantial surveys completed since the original Bedmap compilation (Bedmap1) in 2001. In particular, the Bedmap2 ice thickness grid is made from 25 million measurements, over two orders of magnitude more than were used in Bedmap1. In most parts of Antarctica the subglacial landscape is visible in much greater detail than was previously available and the improved data-coverage has in many areas revealed the full scale of mountain ranges, valleys, basins and troughs, only fragments of which were previously indicated in local surveys. The derived statistics for Bedmap2 show that the volume of ice contained in the Antarctic ice sheet (27 million km<sup>3</sup>) and its potential contribution to sea-level rise (58 m) are similar to those of Bedmap1, but the mean thickness of the ice sheet is 4.6% greater, the mean depth of the bed beneath the grounded ice sheet is 72 m lower and the area of ice sheet grounded on bed below sea level is increased by 10%. The Bedmap2 compilation highlights several areas beneath the ice sheet where the bed elevation is substantially lower than the deepest bed indicated by Bedmap1. These products, along with grids of data coverage and uncertainty, provide new opportunities for detailed modelling of the past and future evolution of the Antarctic ice sheets.
Article
Full-text available
The thickness of ice shelves, a basic parameter for mass balance estimates, is typically inferred using hydrostatic equilibrium, for which knowledge of the depth-averaged density is essential. The densification from snow to ice depends on a number of local factors (e.g., temperature and surface mass balance) causing spatial and temporal variations in density–depth profiles. However, direct measurements of firn density are sparse, requiring substantial logistical effort. Here, we infer density from radio-wave propagation speed using ground-based wide-angle radar data sets (10 MHz) collected at five sites on Roi Baudouin Ice Shelf (RBIS), Dronning Maud Land, Antarctica. We reconstruct depth to internal reflectors, local ice thickness, and firn-air content using a novel algorithm that includes traveltime inversion and ray tracing with a prescribed shape of the depth–density relationship. For the particular case of an ice-shelf channel, where ice thickness and surface slope change substantially over a few kilometers, the radar data suggest that firn inside the channel is about 5 % denser than outside the channel. Although this density difference is at the detection limit of the radar, it is consistent with a similar density anomaly reconstructed from optical televiewing, which reveals that the firn inside the channel is 4.7 % denser than that outside the channel. Hydrostatic ice thickness calculations used for determining basal melt rates should account for the denser firn in ice-shelf channels. The radar method presented here is robust and can easily be adapted to different radar frequencies and data-acquisition geometries.
Article
The launch of ERS-l provides coverage, by satellite altimetry, of a large part of the Ross Ice Shelf ineluding areas of input from Byrd Glacier and Ice Streams D and E. Five 35 d repeats of fast-delivery data, comprising approximately 100000 height estimates, have been used to produce a Digital Elevation Model (DEM) of the Ross Icc Shelf north of 81.5° S. Careful filtering of the altimeter data, which removed about 30% of the measurements, ensured that only valid values were used. The data were grldded to produce a DEM with a cell size of 10km. Repeatability of the data was assessed from an analysis of crossing points of ascending and descending tracks. The rms cross-over difference for the ice shelf was 0.94 m. Using the five repeat tracks gave a random error of 0.30 m for an averaged height measurement. Regionally correlated errors in the orbit and geoid add a systematic long wavelength bias of approximately 2m to the final elevation estimate. Two of the latest geoid models, OSU91-A and JGMI, were compared with the available in situ data and hYdrostatic models based on ice and water densities. The altimetry was compared with ice-thickness data from Ross Icc Shelf Geophysical and Glaciological Survey (RIGGS) stations and Scott Polar Research Institute radio-echo-sounding surveys undertaken in the 1970s. Differences between the DEM and heights calculated from ice thicknesses and a standard density -depth equation lie, in general, within the combined measurement errors. There are, however, several areas where this is not the case. Prominent north-south stripes of different ice thickness shown on a RIGGS map apparently do not exist. Low elevations are associated with high-density ice draining from East Antarctic outlet glaciers. The grounding line of Icc Streams D and E and an ice plain behind it are clearly demarcated by the break in surface slope. Grounded ice north of Steershead is also observed
Article
Antarctica's ice shelves provide resistance to the flow of grounded ice towards the ocean. If this resistance is decreased as a result of ice shelf thinning or disintegration, acceleration of grounded ice can occur, increasing rates of sea-level rise. Loss of ice shelf mass is accelerating, especially in West Antarctica, where warm seawater is reaching ocean cavities beneath ice shelves. Here we use satellite imagery, airborne ice-penetrating radar and satellite laser altimetry spanning the period from 2002 to 2014 to map extensive basal channels in the ice shelves surrounding Antarctica. The highest density of basal channels is found in West Antarctic ice shelves. Within the channels, warm water flows northwards, eroding the ice shelf base and driving channel evolution on annual to decadal timescales. Our observations show that basal channels are associated with the development of new zones of crevassing, suggesting that these channels may cause ice fracture. We conclude that basal channels can form and grow quickly as a result of warm ocean water intrusion, and that they can structurally weaken ice shelves, potentially leading to rapid ice shelf loss in some areas.