TCThe CryosphereTCThe Cryosphere1994-0424Copernicus PublicationsGöttingen, Germany10.5194/tc-11-2411-2017Observationally constrained surface mass balance of Larsen C ice shelf, AntarcticaKuipers MunnekePeterp.kuipersmunneke@uu.nlhttps://orcid.org/0000-0001-5555-3831McGrathDanielhttps://orcid.org/0000-0002-9462-6842MedleyBrookehttps://orcid.org/0000-0002-9838-3665LuckmanAdrianhttps://orcid.org/0000-0002-9618-5905BevanSuzannehttps://orcid.org/0000-0003-2649-2982KulessaBerndhttps://orcid.org/0000-0002-4830-4949JansenDanielaBoothAdamhttps://orcid.org/0000-0002-8166-9608SmeetsPaulHubbardBrynhttps://orcid.org/0000-0002-3565-3875AshmoreDavidVan den BroekeMichielhttps://orcid.org/0000-0003-4662-7565SevestreHeidiSteffenKonradShepherdAndrewGourmelenNoelInstitute for Marine and Atmospheric research, Utrecht University, Utrecht, the NetherlandsGeosciences Department, Colorado State University, Fort Collins, CO, USAU.S. Geological Survey, Alaska Science Center, Anchorage, AK, USACryospheric Sciences Laboratory, NASA Goddard Space Flight Center, Greenbelt, MD, USAGeography Department, College of Science, Swansea University, Swansea, UKAlfred Wegener Institute Helmholtz-Centre for Polar and Marine Research, Bremerhaven, GermanySchool of Earth and Environment, University of Leeds, Leeds, UKCentre for Glaciology, Department of Geography and Earth Sciences, Aberystwyth University, Aberystwyth, UKDepartment of Geography and Sustainable Development, University of St Andrews, St Andrews, UKSwiss Federal Research Institute WSL, Birmensdorf, SwitzerlandSchool of Geosciences, University of Edinburgh, Edinburgh, UKPeter Kuipers Munneke (p.kuipersmunneke@uu.nl)1November20171162411242621March201724March201729August20178September2017This work is licensed under the Creative Commons Attribution 3.0 Unported License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/3.0/This article is available from https://tc.copernicus.org/articles/11/2411/2017/tc-11-2411-2017.htmlThe full text article is available as a PDF file from https://tc.copernicus.org/articles/11/2411/2017/tc-11-2411-2017.pdf
The surface mass balance (SMB) of the Larsen C ice shelf (LCIS), Antarctica,
is poorly constrained due to a dearth of in situ observations. Combining
several geophysical techniques, we reconstruct spatial and temporal patterns
of SMB over the LCIS. Continuous time series of snow height (2.5–6 years) at
five locations allow for multi-year estimates of seasonal and annual SMB over
the LCIS. There is high interannual variability in SMB as well as spatial
variability: in the north, SMB is 0.40 ± 0.06 to
0.41 ± 0.04 m w.e. year-1, while farther south, SMB is up to
0.50 ± 0.05 m w.e. year-1. This difference between north and
south is corroborated by winter snow accumulation derived from an airborne
radar survey from 2009, which showed an average snow thickness of
0.34 m w.e. north of 66∘ S, and 0.40 m w.e. south of
68∘ S. Analysis of ground-penetrating radar from several field
campaigns allows for a longer-term perspective of spatial variations in SMB:
a particularly strong and coherent reflection horizon below 25–44 m of
water-equivalent ice and firn is observed in radargrams collected across the
shelf. We propose that this horizon was formed synchronously across the ice
shelf. Combining snow height observations, ground and airborne radar, and SMB
output from a regional climate model yields a gridded estimate of SMB over
the LCIS. It confirms that SMB increases from north to south, overprinted by
a gradient of increasing SMB to the west, modulated in the west by
föhn-induced sublimation. Previous observations show a strong decrease in
firn air content toward the west, which we attribute to spatial patterns of
melt, refreezing, and densification rather than SMB.
Introduction
About 74 % of the grounded ice sheet of Antarctica drains into the Southern
Ocean through floating ice shelves . By buttressing the
grounded ice sheet e.g., ice shelves strongly
modulate the flux of ice into the ocean, thereby exerting an important
control over the contribution of mass variations of the Antarctic Ice Sheet
to global sea level.
In recent decades, the collapse of ice shelves along the Antarctic Peninsula
was immediately followed by a sustained velocity increase in
the glaciers previously feeding these ice shelves, by a factor 3–4 in
documented cases . It is likely that the
current (and growing) mass imbalance of the Antarctic Peninsula
e.g. is due in part to these ice-dynamical
adjustments to the loss of ice shelves. It is believed that these ice-shelf
collapses along the Antarctic Peninsula have been attributed, at least in
part, to warming of the near-surface atmosphere . Warming has
been faster in this region than the global average since the 1950s
, in part caused by a very large regional decadal
variability . It has been hypothesized that enhanced meltwater
production at the ice-shelf surface can lead to hydrofracturing, whereby
meltwater-filled crevasses open up under the pressure exerted at the crevasse
tip by the column of standing meltwater , and/or where
drainage of meltwater lakes induces fracture by strong flexural stresses
. suggested that the conditions for ponding
and hydrofracturing depend on the local accumulation and melt fluxes, and
their effect on the vertical structure of the firn layer. Hence, there is a
need to describe the surface mass balance (SMB) of these ice shelves
accurately.
The Larsen C ice shelf (LCIS) is the largest ice shelf of the Antarctic
Peninsula, and is located to the east of the north–south-oriented Antarctic
Peninsula mountain range (Fig. ). As the dominant upper-air wind
direction is westerly, the LCIS is in the climatological leeside of the
mountains. This position gives rise to particular patterns in surface melt,
with more melt and the occurrence of meltwater ponding in the inlets directly
east of the mountains, and gradually less melt away from the mountains
towards the calving front in the Weddell Sea to the east
. The advection of warm, dry air masses over the ice
shelf during föhn winds is the likely cause of this surface melt
distribution . A notable expression of this surface melt gradient
is seen in the composition of the firn layer over the LCIS: the smallest amounts
of firn air are found in the inlets in the western part of the LCIS .
In Cainet Inlet (Fig. ), the firn contains a massive subsurface
ice layer , influencing ice-shelf temperatures, density, and
potentially, flow properties. The occurrence of föhn winds has increased in
recent decades , contributing to the destabilization of Larsen
B ice shelf through increased meltwater production.
However, although important, surface melt is not part of the SMB as we define
it in Eq. (). Spatially varying patterns of snowfall and other
SMB components can amplify or counteract the effect of surface melt on the
observed gradients of firn air. Moreover, along with the ocean-driven basal
mass balance, iceberg calving, and glacier inflow, SMB is an essential
component of the overall LCIS mass balance. In addition, SMB controls LCIS
firn properties such as density and temperature. Considering the importance
of SMB, the dearth of published in situ SMB over the LCIS is problematic. In the
most comprehensive compilation of Antarctic Peninsula SMB measurements to
date , none were available on the ice shelves themselves.
However, four observational records from sites on the grounded ice adjacent
to the LCIS are available. Two firn cores were collected on Dolleman Island
(70.58∘ S, 60.92∘ W, 398 m a.s.l.), yielding mean SMB values
of 0.390 m w.e. year-1 for 1962–1982 , and 0.404 m w.e. year-1. A firn core on Gipps Ice Rise (68.77∘ S,
60.93∘ W, 290 m a.s.l.) gives 0.349 m w.e. year-1.
Stake observations by revealed a mean SMB of 0.36 m w.e. year-1
on Jason Peninsula (66.25∘ S, 61.00∘ W), which forms
the northern boundary of the LCIS. Based on an interpolation between these
records, estimated annual SMB at < 0.50 m w.e. year-1 over
the LCIS. However, given the prominence of orographic gradients in this region,
one must be wary of extrapolating data from these elevated sites:
high-resolution climate modelling as well as radar observations show that the
SMB field around such features can deviate by up to ±50 % relative to
that of the flat surrounding terrain . Finally, a reanalysis
for the period 1979–1993 by indicates that solid
precipitation at a central point on the LCIS was 0.49 m w.e. year-1.
Accumulation is much lower on the eastern side of the Antarctic Peninsula
than on its western side. The latter is dominated by slow-moving low-pressure
systems over the Bellingshausen Sea, leading to a mostly north-westerly flow
of humid and relatively warm air . Subsequently, precipitation
is orographically driven by the Antarctic Peninsula mountains. The largest
precipitation events are associated with advection of moist air from
midlatitudes at times when a strong low-pressure system develops over the
Bellingshausen Sea. In contrast, a continental climate exists on the eastern
side. Barrier flow along the orography of the Antarctic Peninsula, resulting
from a climatological low-pressure area over the Weddell Sea, leads to
predominantly southerly flow over the eastern part of the peninsula
. Precipitation events over the LCIS are frequent and generally
small , and associated with (1) lee cyclogenesis over the LCIS and
the Weddell Sea immediately east of the LCIS, and (2) active fronts arriving from
the north-east or east .
Over the past decade, several field campaigns have been undertaken to collect
data on firn, SMB, meteorology, and climate over the LCIS. During various field
campaigns in 2008, 2009, 2010, 2011, 2014, and 2015, data have been collected
using a number of geophysical techniques, including ground-penetrating radar,
shallow firn coring, snow pits, and continuous snow height observations.
Moreover, airborne radar data have been collected over the LCIS in late 2009 and
2010 as part of the NASA Operation IceBridge campaign.
The aim of this paper is to bring together this suite of data sets in order
to provide a coherent picture of SMB over the LCIS, focusing on the intra-annual,
annual, and decadal timescales. These data can be used to evaluate the
performance of atmospheric models over the LCIS , such as RACMO2
, AMPS , or CESM . Models
assessing ice-shelf stability require an estimate of SMB as a boundary
condition e.g., and their performance is subject to a
correct representation of current melt and SMB over the ice shelf.
Data and methodsSurface mass balance
The specific surface mass balance (in metre water equivalent per year, m w.e. year-1) is defined as
SMB=∫yeardt(PR-SUs-RU-ERds-SUds),
where PR is precipitation, SUs is surface sublimation, RU is
meltwater run-off, and ERds and SUds represent erosion and
sublimation of drifting snow particles, respectively. Run-off depends on the
full liquid water balance, defined as
RU=RA+ME+CO-RF-RT,
i.e. the difference between surface melt (ME), rainfall (RA) and
condensation (CO) on one hand, and internal refreezing within the snow and
firn, RF, and retention by capillary forces, RT, on the other. Note that
SMB, as used in this study refers to the mass balance of the entire firn
column, not only of the surface.
Following , we add the definition of mass in an annual layer (MAL) of the snow cover:
MAL=SMB-m.
Here, m is the loss from an annual layer to lower or higher layers due to
vapour transport or meltwater percolation. If m is assumed to be small, MAL
can be used as an estimate for SMB.
Sonic height rangers
Since 2009, sonic height rangers at various automatic weather stations (AWS)
have been measuring snow surface height (see Table and Fig. ),
yielding records of 3 to 7 years in length. A correction to
account for the dependence of sound velocity on temperature is applied using
concurrent observations of air temperature. Sometimes, sonic pulses are
reflected from the AWS mast, from the datalogger box attached to the mast, or
from suspended snow particles during snowfall or snow drift, giving
erroneously low snow height readings. Such observations are filtered out, and
the resulting data gaps are filled by linear interpolation. Snow height
observations are corroborated by manual measurements of the distance between
the sonic ranger and the snow surface upon each annual maintenance visit. We
assume an uncertainty in the surface height ranger readings of 0.10 m, which
mainly represents noise due to small-scale surface roughness, as the accuracy
of individual measurements is of the order of 0.01 m.
Map showing fieldwork
locations, Operation IceBridge flight lines, GPR tracks, and relevant
geographical names. The MODIS mosaic of Antarctica is shown in the background
.
At all locations, sonic height rangers were attached to an AWS mast. Between
2009 and 2011, additional height rangers were placed at AWS 14 and 15 (see
map in Fig. ), mounted at ∼ 2 m above the surface on a
separate triaxial construction consisting of three lightweight aluminium
poles drilled into the snow at an angle of about 30∘ to a depth of
2.5–3.0 m below the surface. After 2011, we make use of the sonic height
rangers attached to the AWS itself. We present a continuous snow surface time
series that accounts for the occasional raising of the sonic height ranger to
avoid burial. The sonic height rangers themselves were always located between
1 and 4 m above the surface.
All snow height data series exhibit data gaps, but unambiguous values for
summer, winter, and annual surface height change could be derived for all
years and all stations, except for one: the LAR1 time series was interrupted
from 16 October to 9 November 2013. We have filled this gap with the mean
elevation change of the other four sonic rangers, allowing us to compute a
summer and winter surface height change at LAR1 for 2013.
Overview of sonic height rangers operated over the LCIS and used in this
study. Locations are shown on the map in Fig. .
During various field campaigns between 2008 and 2015, snow pits were dug to a
depth of ∼ 2 m below the surface (Fig. ). In these
snow pits, detailed snow stratigraphy was logged, and vertical profiles of
density were recorded using a variety of tools and methods. In total, 22
shallow snow pits – all collected before melt onset – were used for the
analysis in this paper. From these 22 snow pits, we collected vertical
profiles of density, resampled into 5 cm depth bins. We assign an uncertainty
to the density observations of 20 kg m-3, which is based on observations
presented below.
In addition, three vertical profiles of density up to 11 m depth were
collected using a neutron-scattering probe in 2009, at
locations J1, J2, and J4, indicated on the map in Fig. . In
2015, a 90 m-long borehole was drilled with hot water and surveyed with an
optical televiewer at the site of AWS14 referred to as
CI-120 in. Using the vertical profile of luminosity as a proxy
for density , an estimate of firn density is available for the
entire length of the borehole.
All of these snow pits and firn cores provide an estimate of the density of
the uppermost layers of firn to convert from radar two-way travel time to
thickness and from actual layer thickness to SMB in water-equivalent
thickness.
Ground-penetrating radar
Between 2008 and 2015, five ground-based radar surveys were carried out
in different locations, covering northern and central portions of the
LCIS (Fig. ) . Surveys were
carried out using antennas of different frequencies (25, 100, or 200 MHz)
depending on the field campaign. Despite the different frequencies, a
distinct, spatially continuous reflector was identified in a large portion of
the ground surveys, at a median depth of 41 m below the surface. Several
assumptions are necessary in order to convert the measured two-way travel
time to this reflector (τ) to snow accumulation rates.
First, we use an empirical relation between firn density and its dielectric
constant (ϵ; ), which determines the effective wave
velocity through the firn. Thus, depth (d) is calculated as follows:
d=0.5cτϵ-0.5,
where c is the speed of light in a vacuum. As ϵ depends on the
firn density profile, and the accumulation rate (proportional to d) affects
the density profile, an iterative technique is required to solve for d in a
consistent manner. The firn density profile can exhibit strong spatial
variations due to differing surface melt rates and internal refreezing, snow
accumulation rates, and air temperatures; however, field observations on
Larsen C are sparse. Indirect measurements do exist, however, such as the map
of firn air content (FAC) over the LCIS derived by , which
provides insight into the spatial variations in firn density. This gridded
product represents the firn air content as a column thickness, which requires
a firn densification model in order to derive a density profile. Thus, we
have modified a scheme developed by to solve iteratively for
a density profile that is consistent with the radar-derived accumulation
rates and the FAC derived by . The method relies on the
semi-empirical densification model to produce steady-state
depth–density profiles. Starting with an initial accumulation rate estimate,
a constant initial density, and long-term average temperature from the
atmospheric model RACMO2 (Sect. below), we model the
depth–density profile using the relationship from .
Accumulation rates are then derived from the radar horizons using that
initial density profile, yielding a new long-term accumulation value. This
new accumulation rate then replaces our initial estimate and the process
repeats until convergence. This results in radar-derived accumulation rates
that are consistent with the prescribed density data (i.e. the density
profile is dependent on the accumulation rate, so it is necessary to ensure
they are mutually consistent).
Because we have additional independent information in the form of the FAC, we
use an additional iterative scheme to ensure the modelled FAC matches the
gridded FAC from . To accomplish this fitting, we iterate the
assumed initial density to reach the desired FAC. We are effectively lumping
all changes in the FAC into the surface density, which do not necessarily
represent real surface processes. In reality, surface melt is seasonal, and
meltwater infiltration and refreezing do not occur right at the surface,
resulting in complex ice structures that can be interspersed with firn
pockets throughout the vertical firn column . Therefore, in
areas of high melt, the scheme will yield unrealistically high surface
density estimates in order to accommodate for refreezing within the firn
column that is not accounted for in our dry snow densification model. A
modelling effort to include the complexity of meltwater infiltration,
retention, and refreezing is beyond the scope of this work and, while our
scheme is necessarily simplified, it provides reasonable bounds on the snow
accumulation over the LCIS sufficient to meet the needs of this work. Using our
method, we generate radar-derived accumulation rates that are physically
related to the modelled density profile, which, in turn, is consistent with
the FAC data of .
To convert the estimated reflector depth to mass, we assume that the
reflector is below the pore close-off depth (estimated at 3–9 m within
120 km from the grounding line and likely around 10–15 m
further downstream), and then subtract the firn air thickness of
from the reflector depth, yielding an ice-equivalent
thickness, which is converted to mass using an ice density of 910 kg m-3.
Column mass is corrected for differences in acquisition dates of
the various radar lines, assuming a mean accumulation of 0.45 m w.e. year-1
in recent years. Finally, we assume that the change in reflector depth due to
dynamical stretching is negligible.
Airborne snow radar and radar picking
Beginning in 2009, the NASA Operation IceBridge (OIB) campaign has annually
surveyed both the Greenland and Antarctic ice sheets using a variety of
instruments designed to map the geometry and internal structure of the ice.
Two frequency-modulated continuous-wave (FMCW) radar systems, developed by
the Center for Remote Sensing of Ice Sheets (CReSIS) at the University of
Kansas, are capable of imaging the near-surface stratigraphy of the firn
column, hereafter termed the snow and Ku-band radars. For the 2009 campaign,
the snow and Ku-band radars operated over the 4–6 and 14–16 GHz
frequency ranges respectively; thus, the systems have the same bandwidth (2 GHz)
and the same theoretical range resolution (∼ 6 cm in snow). While
accumulation studies have exclusively used the snow radar up to this point,
we use the Ku-band radar data because the snow radar was not operational
during one of the two flights covering the LCIS in 2009.
The Ku-band radar data set consists of two comprehensive surveys of the
entire extent of the LCIS on 4 and 16 November 2009 and one smaller survey along
the western inlets on 31 October 2009. An analysis of active scatterometer
observations reveals that the OIB data were collected prior to the onset of
surface melt. Prior work with the snow radar e.g.
showed that strong, continuous radar reflections are observed in West
Antarctica and Greenland over hundreds of kilometres and to ∼ 30 m depth.
However, over the LCIS we find a single, strong reflection horizon about 1 m
below the surface (see sample radar echogram in Fig. ). This
horizon is the only apparent feature in both the snow and Ku-band radar data
sets.
Sample radar echogram obtained from CReSIS Ku-band radar
(14–16 GHz) on board the 16 November 2009 Operation IceBridge flight.
Automated surface and subsurface picks are overlaid. Darkest colours indicate
air above the snow surface. On the vertical axis, we show the vertical
direction relative to an arbitrary level. Nearly all echograms look similar,
with very strong surface and a strong to very strong subsurface reflection
horizon. The subsurface horizon is interpreted as the end of the last melt
season, and it physically represents the transition from dry snow (above) and
snow that has strong melt features (ice lenses).
The radar data were not stacked because the strength of the reflection was
well above the noise, making it easier to develop a simple, automated picking
scheme. In an initial step, the surface reflection is picked automatically,
and the radar two-way travel time for each trace is zeroed to the two-way
travel time of the surface pick. Thus, here τ is the two-way travel time
relative to the surface. The picking scheme finds the next strongest
reflection below the surface, which in this case is likely generated by the
presence of ice lenses and metamorphosed firn created during the prior melt
season. The subsurface picks are then filtered to exclude any extreme
outliers using a running median filter. Finally, we inspect the result
visually to ensure the automated picks are consistent with the visible
stratigraphy.
Regional atmospheric climate model
RACMO2 is a regional climate model, optimized for simulation of climate and
SMB in polar regions. For example, RACMO2 contains a multi-layer snow model
with dedicated parameterizations for drifting snow and snow albedo. In this
study, we use data from a RACMO2 simulation in a domain covering the
Antarctic Peninsula and surrounding polar ocean at a horizontal resolution of
5.5 × 5.5 km, covering the period 1979–2014
.
Complete, filtered time series of cumulative surface height change
(m) recorded with sonic height rangers at five locations. LAR1, 2, and 3 were
equipped with two sonic rangers, plotted in the same colour. Grey bars
indicate the summer season used in this paper
(1 November–31 March).
ResultsSMB estimates from sonic height rangers
We present the complete time series of snow height for the five sonic rangers
on the LCIS (Fig. ).
At all locations, the surface height increased quite gradually, suggesting
that precipitation occurs in frequent small-magnitude events, rather than in
a small number of large accumulation events per year. In the summer months,
surface lowering is observed at all stations, consistent with expected
melting of the surface snow, and refreezing in the firn below, and with
enhanced firn compaction at elevated snow temperatures.
The multi-year records from sonic height rangers can be used to establish
in situ estimates of annual and seasonal SMB, provided that a reliable
estimate of snow and firn density is available to convert surface elevation
increase to mass accumulation. First, we estimate the density of the winter
accumulation (i.e. before the start of the melt season) from the 22 snow
pits investigated (Sect. above). We excluded the part of the
density profile below melt layers that likely originated from the previous
melt season, thereby obtaining the density of the winter accumulation. The 22
pits show no discernable temporal or spatial variability in the density of
the winter accumulation. We therefore computed one mean vertical profile,
shown along with the individual density profiles in Fig. a.
The vertically averaged (± 1 standard deviation) density of the winter
accumulation is 360 ± 20 kg m-3. Combining this value with
wintertime (April–October) sonic height ranger observations, we can construct
annual time series of wintertime SMB. Figure shows that
interannual variability of wintertime SMB is large (standard deviation at
19–44 % of the mean for stations with time series longer than 3 years), but
all stations show similar year-to-year variations. For example, the winters
of 2009 and 2013 stand out as high-SMB winters at all stations. In this
time span, we found no link between SMB and the southern annular mode (SAM,
shown as grey bars in Fig. ), with values of R2 lower
than 0.3. Föhn is enhanced during negative SAM, which also increases
temperature and summer melt . However, our data show that SAM
is not a good indicator for the occurrence of precipitation on the LCIS.
Vertical profiles of snow and firn density used to convert the
observed surface height changes to mass fluxes. Thin grey lines represent all
individual profiles, and thick lines indicate mean density profiles. (a) Snow
pit observations between 0 and 1.50 m. Grey shading indicates a vertical
profile of standard deviation. Vertical bars in the top of the panel are
vertically integrated densities, including the mean vertically integrated
density as a thick bar; (b) firn core observations (gravity coring and OPTV
logging) up to 7 m below the surface. Note different depth and density scales
between the panels.
The multi-year mean winter SMB at each site is summarized in Table ,
with confidence intervals reflecting measurement accuracy, not
interannual variability shown in Fig. . The differences
between the stations are small and mostly within error bounds. The highest
mean winter SMB is seen at AWS14 at 0.23 ± 0.02 m w.e. year-1 for
2009–2015, and the lowest at LAR3 at 0.17 ± 0.02 m w.e. year-1 for
2009–2011. The winter SMB is somewhat higher near the ice-shelf margin than
farther inland.
Time series of winter (April–October) SMB (m w.e.) derived from
sonic height rangers, with left vertical axis. Magnitude of the southern
annular mode (SAM) is shown as grey bars, with the right vertical axis.
A lack of detailed density profiles at the end of the melt season precludes
the construction of an annually resolved record of summer SMB (November–March) for
each site. However, a multi-year mean annual SMB can be computed by using the
cumulative height signal over multiple years and combining that with
firn-core-derived density profiles. The mean summer SMB at each site can then
be calculated by subtracting the mean winter SMB from the mean annual SMB.
This assumes there is no summer run-off and that all surface melt water
refreezes within the annual layer of snowfall (which is typically
0.70–1.40 m thick). In Eqs. () and (), this
corresponds to assuming that RU =0 and m=0. As an estimate of the density
profile required to convert height change to SMB, we take the mean density
profile of four available deep density profiles: three firn cores at LAR1, 2,
and 3, and an OPTV density log at AWS14. This mean density profile is shown
in Fig. b.
Estimates of mean winter (April–October), summer (November–March) and annual
SMB (in m w.e. year-1) for five sites with continuous sonic height ranger
observations. The rightmost column shows the period on which the estimated
SMB is based. The bottom five rows show results for the longest common period
of all records (April 2009–November 2011). The confidence interval reflects the
measurement accuracy, not the interannual variability. Winter and annual SMB
are derived from sonic height ranger and density observations; summer SMB is
derived as the difference between them.
Annual SMB (see Table ) is highest for LAR1
(0.50 ± 0.05 m w.e. year-1) and lowest for LAR3
(0.40 ± 0.06 m w.e. year-1). Only the annual SMB value from
LAR1 is substantially higher than at the four other locations. If we consider
the common period only (2009–2011), LAR1 stands out even more at
0.56 ± 0.07 m w.e. year-1. For summer SMB (1 November–31 March), we
see that LAR1 shows the largest values by far at
0.37 ± 0.06 m w.e. year-1, which is ∼ 40 % more summer
mass gain than the other locations. At the other locations, the summer SMB
values are almost identical at 0.19–0.22 m w.e. year-1.
Estimating long-term SMB using ground-penetrating radar
Example radargram from ground-penetrating radar, with the reflection
horizon at 35–45 m below the surface clearly visible, partly indicated with
a red line. The lowermost reflections are from the ice–ocean interface at
about 350 m depth.
In all ground-based radar surveys of the LCIS, we find a particularly strong
reflection horizon at 35–45 m (median 41 m) depth below the surface (Fig. ).
Radar reflection horizons in firn and ice are related to
strong contrasts in dielectric properties of the firn, originating, for example, from
changes in the firn properties, such as density, fabric, grain size, and
chemical constituents. Based on the distinctiveness and spatial continuity of
this layer, and lack of east–west depth gradient that would suggest
progressive burial with seaward ice advection, we interpret it as having
formed synchronously over the ice shelf, within a single melt season. The
strong reflection is at the same depth and of similar signature in radargrams
at cross-over points. Unfortunately, the reflection horizon is undated, which
precludes the conversion from accumulated mass to SMB.
Map of the LCIS showing estimated total mass (m w.e.) above the
strong, undated reflection horizon along the GPR tracks.
Total accumulated mass is shown in Fig. for all available
radar lines. The lowest values (26–28 m w.e.) are found in the northern part
of the LCIS along the MIDAS 2014 radar lines. The highest values (40–45 m w.e.)
are concentrated near the southern end of the McGrath radar survey, and to
the south-west in the SOLIS 2009 survey. The multi-decadal SMB estimates from
the radar surveys confirm the observations made by the sonic height rangers,
in that the SMB is lower in the north than in the south.
Airborne radar and spatial wintertime snowfall
Airborne radar is another, independent method capable of mapping spatial
variability in winter SMB across the ice shelf. The OIB Ku-band shows a
spatially persistent, single reflection horizon approximately 0.70 to 1.40 m
below the surface, across the entire ice shelf (Fig. ). The
presence of one single strong reflector is anomalous compared to data
collected over other parts of Antarctica, where multiple reflectors provide
information on the vertical layering in the top few metres of the firn
e.g.. Unlike low-frequency GPR, the OIB radar data over
the LCIS only show a single strong subsurface horizon. This is likely related to
differing scattering characteristics of the firn including melt features at
the much higher OIB radar frequency.
Map of the LCIS showing Operation IceBridge reflector depths (in
metres) for the 2009 flight lines. Stars indicate the locations of automatic
weather stations, with solid stars at locations where snow pit observations
were made concurrent with the OIB overpasses.
To test our earlier assumption that this horizon represents the top of the
melt layer formed during the previous melt season (Sect. 2.5), we compared
the OIB reflector depth with the four sonic height rangers for which data are
available in 2009. Specifically, we extracted the thickness of the snow layer
that had accumulated since the last significant melt event of the melt season
preceding the OIB flights. The melt-season termination dates were established
using QuikSCAT and ASCAT microwave data . Uncertainty in the
sonic height ranger data is computed based on instrument measurement error
and uncertainty in the melt termination date derived from the microwave
satellite data. In most cases, this amounts to a few centimetres, depending on the
snow height variability around the melt season termination date. At AWS15 in
2009, however, the approximated end-of-melt date coincides with a major
snowfall event, raising the uncertainty in the sonic height ranger data to
0.17 m. Since we assume that the single OIB reflector represents the melt
horizon of the last melt season, a snow density of 360 kg m-3,
representative of the winter accumulation and derived from in situ
snow pit observations (Sect. ), is used to convert the two-way
travel time to the subsurface reflection to depth. Again, we use Eq. ()
after to relate firn density to its
dielectric constant ϵ, which determines the effective wave velocity
through the firn. We calculate the depth uncertainty by combining the
dominating uncertainty due to the picking scheme (±2 range bins) with
that of the density assumption (±20 kg m-3). Crossover analysis of
the OIB radar-derived depths showed close agreement with an RMSE just under 2
range bins, which provides the basis for our uncertainty estimate due to
layer picking. The comparison between sonic height ranger and OIB data,
compiled in Table , shows that, at three locations, there is
agreement within 0.02 m between the observed snow accumulation since the
previous melt season and the depth of the OIB-observed reflection horizon.
There is a discrepancy between OIB-estimated and sonic height ranger depth at
LAR1 that exceeds the uncertainty estimates. The LAR1 sonic height ranger
appears out of pattern, but we could not establish the cause for this
discrepancy.
Comparison of reflector depths from OIB radar data with snow height
changes for the corresponding period. End-of-melt date is estimated from
space-borne scatterometry ; distance denotes the minimum
distance between the OIB flight track and the sonic height ranger locations.
NameEnd of melt dateOIB flight dateOIB depth (m)SHR depth (m)Distance (km)AWS1412 Feb 20094 Nov 20091.05 ± 0.071.06 ± 0.031AWS153 Feb 20094 Nov 20091.03 ± 0.071.04 ± 0.174LAR129 Jan 200916 Nov 20091.16 ± 0.070.97 ± 0.051LAR23 Feb 20094 Nov 20091.11 ± 0.071.13 ± 0.052
In Fig. , we compare OIB-derived accumulation for the period
February–October 2009 with RACMO2-simulated SMB for the same period (assuming again a
density of 360 kg m-3). At a total of 415 equally spaced locations along
the OIB flight lines, the mean bias is 0.10 m (RACMO2: 0.97 m, and OIB: 1.07 m).
RACMO2 seems to underestimate at low-accumulation locations and
overestimate at high-accumulation locations, leading to a slope that is > 1
but not significantly so. Nonetheless, RACMO2 seems to capture the magnitude
and the spatial pattern of accumulation for this winter season.
Comparison between Operation IceBridge-derived and
RACMO2-simulated winter accumulation (in m) between February and November 2009, based on 415
equally spaced observations along the OIB flight lines. The 1:1-line is
dashed; the linear fit is represented by a solid line, with associated
uncertainty of the fit in dashed lines. Slope of the fit line is m and
intercept with the vertical axis is b.
Having established the OIB reflection horizon over the LCIS to approximate
snowfall since the last snowmelt event of the previous melt season, we are
able to map the spatial variability of snow accumulation for the austral
winter of 2009 at locations where OIB data are available (Fig. ).
We see diminished winter accumulation in the northern part of
the ice shelf, with smallest values in the north-western inlets of the LCIS.
Higher accumulation is found in the southern part, with particularly high
values around LAR1 in the south-western part of the ice shelf. In the far
south of the ice shelf (south of the Kenyon Peninsula), we again see much
lower amounts of accumulation.
A map of SMB and its origin
At all timescales examined in this study – from seasonal to multidecadal –
we find greater SMB values in the middle and southern sectors of the LCIS, and
lower SMB values in the north. In order to expand our coverage to unsurveyed
areas of the LCIS, we combined the regional climate model RACMO2 (Sect. )
and observations from GPR and the sonic height rangers. The
average SMB from RACMO2 over the period 1979–2014 guided the extrapolation
of the GPR data. Subsequently, RACMO2 SMB was adjusted to match the observed
annual SMB from the sonic height rangers. This was done as follows: the
1979–2014 average SMB from RACMO2 was normalized with respect to its spatial
mean and so were the GPR data. Next, we determined a linear regression of
the normalized RACMO2 SMB values to the normalized GPR data. We used this
regression to adjust the RACMO2 SMB to maximize its match to the GPR data
while conserving the spatial mean SMB. The result, shown in Fig. c,
is a RACMO2-guided extrapolation of the GPR over the
unsurveyed portions of the LCIS, in which the spatial pattern of RACMO2 SMB is
adjusted to the spatial pattern of the GPR observations.
The next step was to adjust the absolute values of RACMO2 SMB to available
sonic height ranger observations. We converted RACMO2 SMB back from
normalized to absolute values, again using the spatial mean SMB. We
determined a weighted mean bias between RACMO2 SMB and all available sonic
height ranger observations, selecting the periods for which both were
available. We used the length of the height ranger observation period as a
weight for the averaging, reflecting how short-term variability plays a
smaller role in longer time series. Compared to the sonic height rangers,
RACMO2 underestimated SMB by 14 ± 10 %. Applying a bias adjustment leads to
the gridded SMB shown in Fig. a. An estimate of SMB
uncertainty was based on (1) the fit between normalized GPR and RACMO2 SMB,
and (2) the 10 % uncertainty of the RACMO2 bias. The resulting uncertainty
is typically 15 % of the SMB value, shown in Fig. b.
The underestimation of RACMO2 snowfall over the LCIS was noted by
and may be the result of the representation of snow formation in clouds or
with underestimated evaporation in the Weddell Sea, the most important source
region for moisture precipitated over the LCIS. The underestimation of RACMO2
snowfall is also apparent in the comparison with Operation IceBridge radar
data and amounts to -13± 10 % (Fig. ), reinforcing the
robustness of our bias estimate.
The SMB pattern in Fig. a provides a broader context to the
various data sets presented above. In the area of GPR observations, the
RACMO2-guided interpolation suggests that the SMB gradient is not strictly
north–south but tilted in the north-east to south-west direction, with the
lowest values near Bawden Ice Rise in the north-east and highest values of
SMB in the inlets in the west to south-west part of the ice shelf.
(a) Map of the Larsen C ice shelf showing a reconstruction
of annual SMB (in mm w.e. year-1), obtained by adjusting 1979–2014
RACMO2 SMB to the spatial pattern of ground-penetrating radar and
observations of SMB from sonic height rangers. (b) Estimated
uncertainty of the SMB values in panel (a) (in
mm w.e. year-1). (c) Coloured dots indicate depth of the
reflection horizon, detected by GPR, normalized with respect to the mean
depth (unitless). The background shows a map of relative SMB (unitless) from
RACMO2.
We used RACMO2 to study the origin of this spatial distribution of SMB. In
the absence of notable run-off, SMB is dictated by snowfall, and by
sublimation. Figure a illustrates the spatial
coherence of snowfall across the LCIS, by showing the fraction of snowfall
occurring simultaneously with snowfall events exceeding 5 mm w.e. day-1 in
the southern area of the LCIS (results are relatively insensitive to the exact
location on the LCIS). The pattern shows that snowfall is coherent across the
shelf and that only a small fraction of snowfall on the western side of the
Peninsula occurs when there is snowfall on its eastern side. Figure b
shows the mean circulation pattern during these
snowfall events (wind speed and direction at the 850 hPa pressure level,
along with temperature at 850 hPa in the background). Snowfall on the LCIS is
thus strongly associated with low pressure centred near its northern end
over the Weddell Sea. These low-pressure systems source water vapour from the
Weddell Sea and from more northerly regions to produce snowfall on the LCIS. This
circulation pattern can explain the relatively high snowfall rates in the
south and south-west, where snowfall is orographically enhanced. As a peculiar
smaller-scale expression of this orographic effect, the south-eastern side of
the Kenyon Peninsula receives more snowfall than its north-western side,
situated in the lee of the flow associated with snowfall. The same pattern
can be seen around the promontories clockwise around the ice shelf (Cole and
Churchill peninsulas, Veier Head, and Argo Point), where reduction of
snowfall is seen at the obstacle's lee side. The snowfall minimum in the
north can be explained by the fact that the low-pressure centre will
fluctuate around the location illustrated in Fig. b.
If the low-pressure system is located farther to the south, the north-eastern
part of the LCIS will experience an offshore wind, in the lee of the Jason
Peninsula and its promontories, extending to the south (Churchill Peninsula,
Veier Head, and Argo Point). Such a southerly position of the low-pressure
system would only exert a notable influence on snowfall in the north-east
sector of the ice shelf. Thus, snowfall in the north-east of the LCIS is more
restricted during local offshore wind than in other places.
Maps showing (a) the fraction of annual snowfall on days
when snowfall on southern LCIS (location given by black dot in the left
panel) exceeds 5 mm w.e. day-1 and (b) the mean wind speed
and direction, and air temperature on those days.
Sublimation exerts a secondary control over SMB. Over the LCIS, föhn winds are
frequent, and the combination of high wind speed and dry air increases the
sublimation rate. During föhn, a pattern of alternating higher and lower
wind speed emerges in the western part of the LCIS, where low-elevation inlets
are separated by higher-elevation promontories that protrude from the
Antarctic Peninsula mountains . An estimate of annual
mean sublimation rate from RACMO2 is shown in Fig. . While
sublimation rates from RACMO2 are poorly evaluated over the LCIS, an estimate of
sublimation from in situ AWS observations reveals that it amounts to
∼ 25–30 mm w.e. year-1 at AWS 14 (see map in Fig. 1) and between 17
and 64 mm w.e. year-1 at the site of a newly installed AWS in Cabinet
Inlet. Comparing the sublimation flux (Fig. ) with the total
SMB (Fig. ) shows that sublimation spatially modulates
SMB in the western part of the LCIS, likely by föhn. According to RACMO2,
annual mean sublimation typically removes 5–15 % of the annual snowfall over
the LCIS. This fraction could be larger if RACMO2 underestimates the sublimation
flux. It is conceivable that the SMB in certain inlets (Cabinet, Mill,
Whirlwind, Mobiloil) has decreased in recent decades following
intensification of föhn due to enhanced sublimation.
Annual mean sublimation (mm w.e. year-1) simulated by RACMO2 over the LCIS, averaged over 1979–2014.
Conclusions
We have combined several geophysical techniques along with a regional climate
model to constrain spatial and temporal patterns of SMB over the Larsen C ice
shelf. Results have been integrated to show that SMB is larger towards the
south of the ice shelf, overprinted by an increase in SMB toward the west.
Assuming that run-off is negligibly small over the LCIS , the
spatial pattern of SMB is dominated by spatial differences in snowfall and
sublimation. Thus, our results indicate that snowfall is larger in the south
than in the north of the LCIS.
Previous studies have indicated a strong gradient in firn air content from
west to east across the LCIS , with the lowest values in the west.
It has been suggested that this reflects enhanced melt and subsequent
refreezing, directly at the foot of the Antarctic Peninsula mountains in the
western part of the LCIS caused by föhn winds descending from
the mountains . Using observations of SMB, this study shows an
east-to-west SMB gradient that, in the absence of melt, would lead to the
highest, rather than the lowest, values of firn air in the west. We therefore
conclude that the gradient in firn air content is caused by melt and
refreezing, rather than by spatial patterns of snowfall or SMB.
We interpret a strong, shallow reflection horizon in the OIB radar data as
the top of the melt layer formed during the last melt season. The presence of
sufficiently thick melt layers, like on the LCIS, precludes airborne observations
of multi-year firn stratigraphy like in the dry snow areas of Antarctica and
Greenland. Still, we demonstrate that OIB radar data can be used to track the
shallowest melt layer and derive winter SMB immediately prior to the radar
survey. This opens up the possibility of acquiring winter SMB estimates over
other Antarctic ice shelves and over parts of the Greenland Ice Sheet that
experience small to moderate amounts of melt, as is typical for percolation
zones.
Much recent work has focused on the stability of the LCIS in a warming climate,
with hydrofracturing suggested as a potential mechanism for ice-shelf
collapse . Using models to test hypotheses
that link atmospheric change to ice-shelf stability is challenging, given the
complexity of terrain and climate in this region. A substantial portion of
total melt is governed by the occurrence of small-scale föhn winds, which
are captured reliably only in models with kilometre-scale horizontal resolution
. Summertime melt is also observed frequently on other days
, and its representation in models depends on subtle changes
in albedo and on the correct simulation of all components of
the radiation balance at the surface .
Precipitation depends on the ability of existing weather fronts to cross the
Antarctic Peninsula mountain range but also on local lee cyclogenesis and
low-pressure systems crossing the Weddell Sea. The moisture content of the
latter depends on both sea ice extent and the presence of polynyas. Thus, the
LCIS is an important and challenging test bed for regional, global, and Earth
system models. The performance of these models in predicting ice-shelf collapse is
subject to a correct representation of observed climate. Estimates of SMB
presented in this study can guide model evaluation and development with the
aim of improving our capacity to predict the stability of the LCIS.
The gridded SMB data are available from Kuipers Munneke et
al. (2017) (10.15784/601056). Underlying GPR and part of the sonic
height ranger data are available from January 2018 at the NERC Polar Data
Centre hosted at https://www.bas.ac.uk/data/uk-pdc. Operation IceBridge
data sets are available at
ftp://data.cresis.ku.edu/data/kuband/2009_Antarctica_DC8/CSARP_qlook/.
PKM, DM and BM conceived this study,
and performed the analysis and synthesis of the data sets. PKM led
the writing of the manuscript. DM, BM, SB, BK, DJ, AB, PS, DA, AS,
and NG processed and provided observational data sets. AL, SB, BK, AB, BH, DA,
and HS collected data in two MIDAS fieldwork campaigns. All authors
contributed to discussions on writing this manuscript.
The authors declare that they have no conflict of interest.
Acknowledgements
This work is funded by the Netherlands Polar Programme, Netherlands Earth
System Science Centre (NESSC), NSF OPP research grant 0732946, NERC/GEF
grants NE/L006707/1, NE/L005409/1, NE/E012914/1, GEF loans 863, 890, 1028. We
thank logistical support from the British Antarctic Survey during the various
field campaigns. We also acknowledge the generous contribution of faculty,
staff, and students at CReSIS in collecting and processing the Ku-band data
as well as NASA's Operation IceBridge team in collecting and disseminating
data to the public. We acknowledge the efforts from two
anonymous reviewers and the editor to improve this manuscript. Use of trade,
product, or firm names is for descriptive purposes only and does not imply
endorsement by the U.S. Government. Edited by:
Kenny Matsuoka Reviewed by: two anonymous referees
ReferencesAshmore, D. W., Hubbard, B., Luckman, A., Kulessa, B., Bevan, S., Booth, A.,
Kuipers Munneke, P., O'Leary, M., Sevestre, H., and Holland, P. R.: Ice and
firn heterogeneity within Larsen C Ice Shelf from borehole optical
televiewing, J. Geophys. Res.-Earth, 122, 1139–1153,
10.1002/2016JF004047, 2017.Barrand, N. E., Vaughan, D. G., Steiner, N., Tedesco, M., Kuipers Munneke,
P., van den Broeke, M. R., and Hosking, J. S.: Trends in Antarctic
Peninsula surface melting conditions from observations and regional climate
modeling, J. Geophys. Res.-Earth, 118, 315–330, 10.1029/2012JF002559,
2013.Berthier, E., Scambos, T. A., and Shuman, C. A.: Mass loss of Larsen B
tributary glaciers (Antarctic Peninsula) unabated since 2012, Geophys.
Res. Lett., 39, L13501, 10.1029/2012GL051755, 2012.Bindschadler, R., Choi, H., Wichlacz, A., Bingham, R., Bohlander, J., Brunt,
K., Corr, H., Drews, R., Fricker, H., Hall, M., Hindmarsh, R., Kohler, J.,
Padman, L., Rack, W., Rotschky, G., Urbini, S., Vornberger, P., and Young,
N.: Getting around Antarctica: new high-resolution mappings of the grounded
and freely-floating boundaries of the Antarctic ice sheet created for the
International Polar Year, The Cryosphere, 5, 569–588,
10.5194/tc-5-569-2011, 2011.Cape, M. R., Vernet, M., Skvarca, P., Marinsek, S., Scambos, T., and Domack,
E.: Foehn winds link climate-driven warming to ice shelf evolution in
Antarctica, J. Geophys. Res.-Atmos., 120, 11037–11057,
10.1002/2015JD023465, 2015.Cook, A. J. and Vaughan, D. G.: Overview of areal changes of the ice shelves
on the Antarctic Peninsula over the past 50 years, The Cryosphere, 4, 77–98,
10.5194/tc-4-77-2010, 2010.
De Angelis, H. and Skvarça, P.: Glacier surge after ice shelf collapse,
Science, 299, 1560–1562, 2003.DeConto, R. M. and Pollard, D.: Contribution of Antarctic to past and
future sea-level rise, Nature, 531, 591–597, 10.1038/nature17145,
2016.Dupont, T. K. and Alley, R. B.: Assessment of the importance of ice-shelf
buttressing to ice-sheet flow, Geophys. Res. Lett., 32, L04503,
10.1029/2004GL022024, 2005.Elvidge, A. D., Renfrew, I. A., King, J. C., Orr, A., Lachlan-Cope, T. A.,
Weeks, M., and Gray, S. L.: Föhn jets over the Larsen C Ice Shelf,
Antarctica, Q. J. Roy. Meteor. Soc., 141, 698–713,
10.1002/qj.2382, 2014.
Gagliardini, O., Durand, G., Zwinger, T., Hindmarsh, R. C. A., and Meur, E. L.:
Coupling of ice-shelf melting and buttressing is a key process in ice-sheet
dynamics, Geophys. Res. Lett., 37, L14501, 10.1029/2010GL043334, 2010.Haran, T., Bohlander, J., Scambos, T., Painter, T., and Fahnestock, M.:
MODIS Mosaic of Antarctica 2008–2009 (MOA2009) Image
Map, version 1, 10.7265/N5KP8037,
available at: http://nsidc.org/data/NSIDC-0593 (last access: 11 October 2017), 2014.Harig, C. and Simons, F. J.: Accelerated West Antarctic ice mass loss
continues to outpace East Antarctic gains, Earth Planet. Sci. Lett., 415,
134–141, 10.1016/j.epsl.2015.01.029, 2015.
Herron, M. M. and Langway Jr., C. C.: Firn densification: an empirical model,
J. Glaciol., 25, 373–385, 1980.Holland, P. R., Corr, H. F. J., Pritchard, H. D., Vaughan, D. G., Arthern,
R. J., Jenkins, A., and Tedesco, M.: The air content of Larsen Ice
Shelf, Geophys. Res. Lett., 38, L10503, 10.1029/2011GL047245, 2011.
Hubbard, B., Roberson, S., Samyn, D., and Merton-Lyn, D.: Digital optical
televiewing of ice boreholes, J. Glaciol., 54, 1–8, 2008.Hubbard, B., Tison, J.-L., Philippe, M., Heene, B., Pattyn, F., Malone, T., and
Freitag, J.: Ice shelf density reconstructed from optical televiewer borehole
logging, Geophys. Res. Lett., 40, 5882–5887, 10.1002/2013GL058023,
2013.Hubbard, B., Luckman, A., Ashmore, D., Bevan, S., Kulessa, B., Kuipers
Munneke, P., Philippe, M., Jansen, D., Booth, A., Sevestre, H., Tison,
J.-L., O'Leary, M., and Rutt, I.: Massive subsurface ice formed by refreezing
of ice-shelf melt ponds, Nat. Comm., 7, 11897, 10.1038/ncomms11897, 2016.Jansen, D., Luckman, A., Kulessa, B., Holland, P. R., and King, E. C.: Marine
ice formation in a suture zone on the Larsen C ice shelf and its
influence on ice shelf dynamics, J. Geophys. Res.-Earth, 118, 1628–1640,
10.1002/jgrf.20120, 2013.King, J. C., Gadian, A., Kirchgaessner, A., Kuipers Munneke, P.,
Lachlan-Cope, T. A., Orr, A., Reijmer, C., van den Broeke, M. R., van Wessem,
J. M., and Weeks, M.: Validation of the summertime surface energy budget of
Larsen C Ice Shelf (Antarctica) as represented in three
high-resolution atmospheric models, J. Geophys. Res.-Atmos., 120, 1335–1347,
10.1002/2014JD022604, 2015.Koenig, L. S., Ivanoff, A., Alexander, P. M., MacGregor, J. A., Fettweis, X.,
Panzer, B., Paden, J. D., Forster, R. R., Das, I., McConnell, J. R., Tedesco,
M., Leuschen, C., and Gogineni, P.: Annual Greenland accumulation rates
(2009–2012) from airborne snow radar, The Cryosphere, 10, 1739–1752,
10.5194/tc-10-1739-2016, 2016.
Kovacs, A., Gow, A. J., and Morey, R. M.: The in-situ dielectric constant of
polar firn revisited, Cold Reg. Sci. Technol., 23, 245–256, 1995.Kuipers Munneke, P., van den Broeke, M. R., Lenaerts, J. T. M., Flanner, M. G.,
Gardner, A. S., and van de Berg, W. J.: A new albedo parameterization for use
in climate models over the Antarctic ice sheet, J. Geophys. Res.-Atmos.,
116, D05114,
10.1029/2010JD015113, 2011.Kuipers Munneke, P., Picard, G., van den Broeke, M. R., Lenaerts, J. T. M., and
van Meijgaard, E.: Insignificant change in Antarctic snowmelt volume since
1979, Geophys. Res. Lett., 39, L01501, 10.1029/2011GL050207, 2012.Kuipers Munneke, P., Ligtenberg, S. R. M., van den Broeke, M. R., van Angelen,
J. H., and Forster, R. R.: Explaining the presence of perennial liquid water
bodies in the firn of the Greenland Ice Sheet, Geophys. Res. Lett.,
41, 476–483,
10.1002/2013GL058389, 2014a.Kuipers Munneke, P., Ligtenberg, S. R. M., van den Broeke, M. R., and Vaughan,
D. G.: Firn air depletion as a precursor of Antarctic ice-shelf collapse,
J. Glaciol., 60, 205–214, 10.3189/2014JoG13J183, 2014b.Kuipers Munneke, P., McGrath, D., Medley, B., Van den Broeke, M. R., and Van
Wessem, J. M.: Mean surface mass balance over Larsen C ice shelf, Antarctica
(1979–2014), assimilated to in situ GPR and snow height data [Data set],
U.S. Antarctic Program (USAP) Data Center, 10.15784/601056, 2017.Kulessa, B., Jansen, D., Luckman, A. J., King, E. C., and Sammonds, P. R.:
Marine ice regulates the future stability of a large Antarctic ice shelf,
Nat. Comm., 5, 4707, 10.1038/ncomms4707, 2014.Lenaerts, J. T. M., Brown, J., van den Broeke, M. R., Matsuoka, K., Drews, R.,
Callens, D., Philippe, M., Gorodetskaya, I. V., van Meijgaard, E., Reijmer,
C. H., Pattyn, F., and van Lipzig, N. P. M.: High variability of climate and
surface mass balance induced by Antarctic ice rises, J. Glaciol., 60,
1–10, 10.3189/2014JoG14J040, 2014.Lenaerts, J. T. M., Vizcaino, M., Fyke, J., van Kampenhout, L., and van den
Broeke, M. R.: Present-day and future Antarctic ice sheet climate and
surface mass balance in the Community Earth System Model, Clim.
Dynam., 47, 1367–1381,
10.1007/s00382-015-2907-4, 2016.Luckman, A., Jansen, D., Kulessa, B., King, E. C., Sammonds, P., and Benn, D.
I.: Basal crevasses in Larsen C Ice Shelf and implications for their global
abundance, The Cryosphere, 6, 113–123,
10.5194/tc-6-113-2012, 2012.Luckman, A., Elvidge, A., Jansen, D., Kulessa, B., Kuipers Munneke, P., King,
J., and Barrand, N. E.: Surface melt and ponding on Larsen C Ice
Shelf and the impact of föhn winds, Antarct. Sci., 26, 625–635,
10.1017/S0954102014000339, 2014.MacAyeal, D. R. and Sergienko, O. V.: The flexural dynamics of melting ice
shelves, Ann. Glaciol., 54, 1–10, 10.3189/2013AoG63A256, 2013.Marshall, G. J., Orr, A., van Lipzig, N., and King, J. C.: The impact of a
changing Southern Hemisphere AnnularMode on Antarctic Peninsula
summer temperatures, J. Climate, 19, 5399–5405, 10.1175/Jcli3844.1,
2006.McGrath, D., Steffen, K., Scambos, T. A., Rajaram, H., Casassa, G., and
Rodriguez Lagos, J. L.: Basal crevasses and associated surface crevassing
on the Larsen C ice shelf, Antarctica, and their role in ice-shelf
instability, Ann. Glaciol., 58, 10–18, 10.3189/2012AoG60A005, 2012.McGrath, D., Steffen, K., Holland, P. R., Scambos, T., Rajaram, H., Abdalati,
W., and Rignot, E.: The structure and effect of suture zones in the Larsen
C Ice Shelf,, J. Geophys. Res.-Earth, 119, 588–602,
10.1002/2013JF002935, 2014.Medley, B., Joughin, I., Das, S. B., Steig, E. J., Conway, H., Gogineni, S.,
Criscitiello, A. S., McConnell, J. R., Smith, B. E., van den Broeke, M. R.,
Lenaerts, J. T. M., Bromwich, D. H., and Nicolas, J. P.: Airborne-radar and
ice-core observations of annual snow accumulation over Thwaites Glacier,
West Antarctica confirm the spatiotemporal variability of global and
regional atmospheric models, Geophys. Res. Lett., 40, 3649–3654,
10.1002/grl.50706, 2013.Medley, B., Ligtenberg, S. R. M., Joughin, I., van den Broeke, M. R., Gogineni,
S., and Nowicki, S.: Antarctic firn compaction rates from repeat-track
airborne radar data: I. Methods, Ann. Glaciol., 56, 155–166,
10.3189/2015AoG70A203, 2015.Morris, E. M.: A theoretical analysis of the neutron-scattering method for
measuring snow and ice density, J. Geophys. Res.-Earth, 113, F03019,
10.1029/2007JF000962, 2008.Morris, E. M. and Vaughan, D. G.: Spatial and temporal variation of surface
temperature on the Antarctic Peninsula and the limit of viability of ice
shelves, vol. 79, 61–68, AGU, Washington, DC, 10.1029/AR079p0061,
2003.
Mulvaney, R. and Wolff, E. W.: Evidence for winter/spring denitrification of
the stratosphere in the nitrate record of Antarctic firn cores, J. Geophys.
Res., 98, 5213–5220, 1993.Parish, T. R.: The influence of the Antarctic Peninsula on the wind field
over the western Weddell Sea, J. Geophys. Res.-Oceans, 88, 2684–2692,
10.1029/JC088iC04p02684, 1983.
Peel, D. A. and Clausen, H. B.: Oxygen-isotope and total beta-radioactivity
measurements on 10 m ice cores from the Antarctic Peninsula, J. Glaciol.,
98, 43–55, 1982.Powers, J. G., Manning, K. W., Bromwich, D. H., Cassano, J. J., and Cayette,
A. M.: A decade of Antarctic science support through AMPS, B. Am. Meteorol. Soc., 93, 1699–1712, 10.1175/bams-d-11-00186.1, 2012.
Rott, H., Rack, W., Nagler, T., and Skvarça, P.: Climatically induced
retreat and collapse of the northern Larsen Ice Shelf, Antarctic
Peninsula, Ann. Glaciol., 27, 86–92, 1998.
Scambos, T. A., Hulbe, C., Fahnestock, M., and Bohlander, J.: The link between
climate warming and break-up of ice shelves in the Antarctic Peninsula,
J. Glaciol., 46, 516–530, 2000.
Scambos, T. A., Hulbe, C., and Fahnestock, M.: Climate-induced ice shelf
disintegration in the Antarctic Peninsula, in: Antarctic Peninsula
Climate Variability: Historical and Paleoenvironmental Perspectives, edited
by: Domack, E., vol. 79 of Antarct. Res. Ser., 79–92, AGU,
Washington, DC, 2003.Scambos, T. A., Bohlander, J. A., Shuman, C. A., and Skvarça, P.: Glacier
acceleration and thinning after ice shelf collapse in the Larsen B
embayment, Antarctica, Geophys. Res. Lett., 31, L18402, 10.1029/2004GL020670,
2004.
Scambos, T. A., Haran, T., Fahnestock, M., Painter, T., and Bohlander, J.:
MODIS-based Mosaic of Antarctica(MOA) data sets:
continent-wide surface morphology and snow grain size, Rem. Sens. Environ.,
111, 242–257, 10.1016/j.rse.2006.12.020, 2007.Shepherd, A., Ivins, E. R., Guero, A., and the IMBIE Project
Group: A reconciled estimate if ice-sheet mass balance, Science, 338,
1183–1189, 10.1126/science.1228102, 2012.Trusel, L. D., Frey, K. E., Das, S. B., Kuipers Munneke, P., and van den
Broeke, M. R.: Satellite-based estimates of Antarctic surface meltwater
fluxes, Geophys. Res. Lett., 40, 6148–6153, 10.1002/2013GL058138, 2013.
Turner, J., Leonard, S., Lachlan-Cope, T. A., and Marshall, G. J.:
Understanding Antarctic Peninsula precipitation distribution and
variability using a numerical weather prediction model, Ann. Glaciol., 27,
591–596, 1998a.
Turner, J., Marshall, G. J., and Lachlan-Cope, T. A.: Analysis of
synoptic-scale low pressure systems within the Antarctic Peninsula sector
of the circumpolar trough, Int. J. Climatol., 18, 253–280,
1998b.Turner, J., Lachlan-Cope, T. A., Marshall, G. J., Morris, E. M., Mulvaney, R.,
and Winter, W.: Spatial variability of Antarctic Peninsula net surface
mass balance, J. Geophys. Res.-Atmos., 107, 4173, 10.1029/2001JD000755, 2002.Turner, J., Barrand, N. E., Bracegirdle, T. J., Convey, P., Hodgson, D. A.,
Jarvis, M., Jenkins, A., Marshall, G., Meredith, M. P., Roscoe, H., Shanklin,
J., French, J., Goosse, H., Guglielmin, M., Gutt, J., Jacobs, S., Kennicutt,
M. C., Masson-Delmotte, V., Mayewsky, P., Navarro, F., Robinson, S., Scambos,
T., Sparrow, M., Summerhayes, C., Speer, K., and Klepikov, A.: Antarctic
climate change and the environment: an update, Polar Record, 50, 237–259,
10.1017/S0032247413000296, 2014.Turner, J., Lu, H., White, I., King, J. C., Phillips, T., Hosking, J. S.,
Bracegirdle, T. J., Marshall, G. J., Mulvaney, R., and Deb, P.: Absence of
21st century warming on Antarctic Peninsula consistent with natural
variability, Nature, 535, 411–415, 10.1038/nature18645, 2016.Van der Veen, C. J.: Fracture propagation as means of rapidly transferring
surface meltwater to the base of glaciers, Geophys. Res. Lett., 34, L01501,
10.1029/2006GL028385, 2007.van Wessem, J. M., Ligtenberg, S. R. M., Reijmer, C. H., van de Berg, W. J.,
van den Broeke, M. R., Barrand, N. E., Thomas, E. R., Turner, J., Wuite, J.,
Scambos, T. A., and van Meijgaard, E.: The modelled surface mass balance of
the Antarctic Peninsula at 5.5 km horizontal resolution, The Cryosphere, 10,
271–285, 10.5194/tc-10-271-2016, 2016.