Available via license: CC BY 4.0
Content may be subject to copyright.
The Cryosphere, 6, 113–123, 2012
www.the-cryosphere.net/6/113/2012/
doi:10.5194/tc-6-113-2012
© Author(s) 2012. CC Attribution 3.0 License.
The Cryosphere
Basal crevasses in Larsen C Ice Shelf and implications for their
global abundance
A. Luckman1, D. Jansen1, B. Kulessa1, E. C. King2, P. Sammonds3, and D. I. Benn4,5
1College of Science, Swansea University, UK
2British Antarctic Survey, High Cross, Cambridge, UK
3Department of Earth Sciences, University College London, UK
4University Centre in Svalbard, P.O. Box 156, 9171, Longyearbyen, Norway
5Department of Geography, University of St. Andrews, UK
Correspondence to: A. Luckman (a.luckman@swansea.ac.uk)
Received: 6 July 2011 – Published in The Cryosphere Discuss.: 28 July 2011
Revised: 15 December 2011 – Accepted: 4 January 2012 – Published: 24 January 2012
Abstract. 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 lit-
tle attention. We investigate basal crevasses in Larsen C Ice
Shelf using field radar survey, remote sensing and numeri-
cal modelling. We demonstrate that a group of features vis-
ible 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 down-
stream 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 al-
ways reflect this height, probably because of snow accumu-
lation 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 pre-
viously labelled simply as “crevasses”, where they are not
full thickness rifts, must be basal crevasse troughs, highlight-
ing a fundamental structural property of many ice shelves
that may have been previously overlooked.
1 Introduction
Basal (or bottom) crevasses are fractures that extend upwards
from the bottom of ice bodies (Jezek, 1984; van der Veen,
1998a). They differ from surface crevasses in their direction
of opening and in that they have the assistance of basal water
at pressure to promote their initiation and propagation. The
radar signatures of basal crevasses up to hundreds of meters
in height have been identified in (among others) the Ross Ice
Shelf (Jezek et al., 1979; Jezek and Bentley, 1983), a large
tabular iceberg calved from the same (Peters et al., 2007), and
the Fimbul Ice Shelf (Humbert and Steinhage, 2011). Basal
crevasses have also been observed in glacier ice where sub-
glacial water pressure is high enough (Christoffersen et al.,
2005, Harper et al., 2010). In early radar surveys the signa-
tures of basal crevasses were detected often enough to lead
Shabtaie and Bentle (1982) to suggest that such crevasses are
likely to be “abundant” in Antarctic ice shelves.
Basal crevasses have, however, received little attention in
the recent literature, although we are aware of a similar study
to this one (McGrath et al., 2012). Despite the current fo-
cus on ice shelves and their disintegration, few studies in the
last decade make direct reference to such features, which is
at odds with their previously inferred abundance and their
potential impact on both the ice and the ocean cavity be-
neath. By increasing the area of interface between ice and
water, basal crevasses may enable heat exchange with the
ocean deep inside the ice column where the ice would oth-
erwise be well insulated from external heat sources (Hellmer
Published by Copernicus Publications on behalf of the European Geosciences Union.
114 A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf
and Jacobs, 1992). This increased area of interface may also
moderate the exchange of mass with the ocean through melt-
ing and refreezing. Basal crevasses will additionally modify
local stresses in the ice, thereby potentially playing a key part
in ice shelf stability (Jezek, 1984; Shepherd et al., 2003). It is
therefore important to understand where and how frequently
they occur.
In this paper we use ground penetrating radar (GPR),
global positioning system (GPS) and remote sensing to in-
vestigate features in the Larsen C Ice Shelf that are visi-
ble in the MODIS (Moderate resolution imaging spectro-
radiometer) mosaic of Antarctica (MOA) (Fig. 1, Haran
et al., 2005). We differentiate between surface and basal
crevasses in terms of their GPR and satellite image signa-
tures. In the light of these observations we consider im-
plications for the abundance and potential impact of basal
crevasses elsewhere in the ice shelf and beyond. We also test
the ability of the linear elastic fracture mechanics (LEFM)
approach to predict the penetration height of basal crevasses
for which we have field data.
2 GPR observations from Larsen C Ice Shelf
We focus on two groups of features near the Joerg Peninsula
(Fig. 2a and b) which are visible in commonly-used satel-
lite images (Fig. 1), and which have previously been dis-
cussed simply as “crevasses” (Glasser et al., 2009). These
are aligned approximately perpendicular to the ice flow di-
rection and form part of two quasi-periodic series of features
which start near the grounding line and end at the calving
front. These features are defined by their planimetric dimen-
sions of width (the smaller dimension) and their length (the
longer dimension). The first series (Fig. 2a, referred to here
as Series 1) originates close to the grounding line, and is
characterised by features that are ∼4km in length, are per-
pendicular to the flow direction, occur on ice that is ∼350m
thick and have a regular spacing along flow of ∼1 km. Taking
into account flow speeds at this location (Jansen et al., 2010),
this implies that they are generated approximately every 5yr.
We highlight two features in this series for later attention and
refer to them as “S1C1” (series one, crevasse one, etc.) and
“S1C2” (Fig. 2a). The second series of features (Fig. 2b, Se-
ries 2) are ∼10 km long and are ∼30km from the grounding
line at their closest and are found on ic ∼320m thick. The
feature in this series nearest the grounding line (and we as-
sume therefore to be the most recent in origin) we refer to
as “S2C1” and the next one downstream (and therefore as-
sumed older) as “S2C2”. Although their periodicity is less
regular compared to that of Series 1, the most recently de-
veloped three features in Series 2 have a spacing of ∼5km
equivalent to a creation every 14 yr at a flow speed of approx.
360m a−1.
During November and December 2009 we carried out a
field campaign using GPR and dGPS (differential GPS) to
24
Figures 513
514
Figure 1. (a) Location of Larsen C Ice Shelf featuring an extract from MODIS MOA mosaic 515
(2003; Haran et al., 2005). Boxes indicate areas covered by later figures ; (b) Rotated part of 516
MOA image with dashed lines highlighting the two crevasse series investigated – Series 1 in 517
blue and Series 2 in green. 518
519
Fig. 1. (a) Location of Larsen C Ice Shelf featuring an extract from
MODIS MOA mosaic (2003; Haran et al., 2005). Boxes indicate
areas covered by later figures; (b) rotated part of MOA image with
dashed lines highlighting the two crevasse series investigated – Se-
ries 1 in blue and Series 2 in green.
investigate the structure of the ice shelf in this region includ-
ing ice heterogeneity, shelf thickness and the signature of
marine ice. Several hundreds of line kilometres of common-
offset 50MHz radar data were acquired using a Pulse-Ekko
PE100 GPR system towed behind a snow-scooter. Precise
planimetric and height location of the antennas was recorded
with a differential Leica System 1200 GPS. The GPS po-
sition wass recorded every second and the maximum base-
line was 15km. The GPR surveys were carried out with
a sampling interval of 0.8ns using 8 stacks. GPR traces
were acquired every 3s at a towing speed of approximately
5km h−1, yielding a mean trace spacing of ∼4.3 m. Here we
present ∼20 km of GPR data that cover features of interest in
the ice shelf.
The raw GPR data were processed using standard tech-
niques including automatic gain control, band pass filter-
ing, and correction for surface topography as recorded by
dGPS. Travel time was converted to depth assuming a radar-
wave velocity in ice of 0.175mns−1, a value based on fit-
ting diffraction hyperbolae to data from a common mid-point
(CMP) survey carried out on one of the crests between the
crevasses on the profile (Fig. 2a). We present the data as
The Cryosphere, 6, 113–123, 2012 www.the-cryosphere.net/6/113/2012/
A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf 115
25
520
(a) (b)
Figure 2. Landsat 7 sub-images from the southern end of Larsen C Ice Shelf showing 521
features investigated in 2009 by RES along transects shown in various colours, location of 522
CMP survey (yellow star), and location of GPS base-station (cyan star): (a) Series 1 of 523
closely-space features; (b) Series 2 of widely-spaced features. See figure 1 for location within 524
the ice shelf. Black diagonal stripes are due to a well-known Scan Line Correction fault. 525
526
Fig. 2. Landsat 7 sub-images from the southern end of Larsen C Ice Shelf showing features investigated in 2009 by RES along transects
shown in various colours, location of CMP survey (yellow star), and location of GPS base-station (cyan star): (a) Series 1 of closely-space
features; (b) Series 2 of widely-spaced features. See Fig. 1 for location within the ice shelf. Black diagonal stripes are due to a well-known
Scan Line Correction fault.
un-migrated profiles because features deep in the ice column
are more easily identified in printed figures by their charac-
teristic hyperbolic returns.
Figures 3 and 4 show the processed radargrams from pro-
files over the two series of features. A single profile of GPR
data was acquired from Series 1, while from Series 2, four
profiles were acquired from S2C1 and two from S2C2. All
radargrams collected from Series 2 are presented to demon-
strate the repeatability of the observations.
Following previous authors (e.g. Jezek and Bentley, 1983),
we interpret these data as exhibiting the signatures of basal
crevasses. The key characteristics of basal crevasses in GPR
data are the highly uneven basal return (in which the basal
openings are usually evident), and the reflection from the
crevasse tips (emphasized here by presenting un-migrated
data). The crevasse tips are clearer for Series 2 (Figs. 2b,
4) than for Series 1 features (Figs. 2a, 3). Our GPR profiles
were collected perpendicular to the crevasses so the typical
basal width of Series 1 crevasses appears to be ∼400m. For
S2C1 and S2C2 of Series 2, the complex basal echoes do not
allow such an estimate to be made. The penetration height
of Series 1 crevasses is not generally as clear as Series 2, but
S1C1 and S1C2 tips are visible in the GPR data and these
crevasses are chosen for further analysis (Fig. 3). Assum-
ing that these echoes originate from the highest penetration
point, the Series 1 crevasse heights are around 100m above
the base of the ice shelf, which is close to one third of the ice
shelf thickness at this location. The penetration height of Se-
ries 2 basal crevasses is between 180m and 230 m above the
base of the ice shelf, or around two-thirds of the ice thick-
ness. Table 1 gives more precise values for basal crevasse
dimensions. In the absence of better error estimates from
our own data, we assume that errors in basal crevasses pen-
etration height are proportional to velocity errors from CMP
surveys and are therefore ∼5% (Barrett et al., 2007).
In addition to the specific characteristics of the basal
crevasses, two further aspects of the radargrams (Figs. 3 and
4) are of note. Firstly, hyperbolic radar responses originating
near the surface, can be seen on the edges of some of the sur-
face troughs and these are highlighted in Fig. 4. Secondly,
the ice shelf surface (as measured by dGPS) has troughs di-
rectly over each basal crevasse, and shallow internal layer-
ing within the ice follows approximately the profile of these
surface troughs. Deeper layers appear to be more deformed
downwards above the basal crevasses than is the ice surface,
although we should note that this may partly be due to the
fact that we have assumed a constant radar velocity despite
the varying ice density. For Series 1 basal crevasses, the
surface troughs are 5–10m deep, and for Series 2 they are
∼10 m deep. The widths of these surface troughs are not well
defined in either the GPS data or the satellite data as there is
no clear break in slope. However, for both crevasse series
the apparent surface deformation extends more than 500m.
More precise characteristic measurements of ice thickness,
penetration height and trough depth for the four highlighted
crevasses are presented in Table 1.
3 Discussion of GPR observations
3.1 Basal crevasses
We have established that the features visible in MODIS
data (Fig. 1), illustrated using Landsat data (Fig. 2) and
www.the-cryosphere.net/6/113/2012/ The Cryosphere, 6, 113–123, 2012
116 A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf
Table 1. Measured and modelled properties of four basal crevasses on Larsen C Ice Shelf with estimates of error.
Crevasse Measured deformation Ice thickness Mean ice Density Measured crevasse Modelled crevasse Measured surface Modelled surface Measured surface
rate (RS) (GPR) (seismics) penetration height (GPR) penetration height trough depth (GPS) trough depth trough width (GPS)
±0.0005 [a−1]±5 % [m] ±10 [kg m−3]±5 % [m] ±10 [m] ±0.02 [m] ±1 [m] ±10 [m]
S1C1 0.0055 358 870 108 100 6.7 10.7 580
S1C2 0.0055 350 870 119 102 8.8 12.2 580
S2C1 0.0086 328 804 217 140 12.3 23.2 375
S2C2 0.0086 315 824 180 145 10.5 18.6 435
26
527
528
529
Figure 3. GPR data from Series 1 crevasses along the transect shown in Fig. 2a: Centre panel 530
is the full 20 km profile, and right and left-hand panels show enlargement of S1C1 and S1C2 531
for clarity. Ice flow direction is left to right. X and y axes on different scales: x-dimension 532
assumes one radar pulse every 5 m of travel and y-dimension assumes a constant velocity in 533
ice of 0.175 m/ns. 534
535
Fig. 3. GPR data from Series 1 crevasses along the transect shown in Fig. 2a: centre panel is the full 20 km profile, and right and left-hand
panels show enlargement of S1C1 and S1C2 for clarity. Ice flow direction is left to right. X and y-axes on different scales: x-dimension
assumes one radar pulse every 5m of travel and y-dimension assumes a constant velocity in ice of 0.175m ns−1.
investigated using GPR (Figs. 3 and 4) are basal crevasses.
In this section we discuss their origin, geometrical character-
istics and associated GPR signatures.
The potential for the formation of basal crevasses in ice
shelves is well established from a theoretical perspective
(e.g. van der Veen, 1998a) as well as from previous radio-
echo-sounding (RES) observations (e.g. Jezek and Bentley,
1983). The formation of basal crevasses requires tensile de-
viatoric stresses and basal water pressures at or close to over-
burden pressure, conditions that occur readily below float-
ing ice shelves and transiently during glacier surges (Weert-
man, 1980; van der Veen, 1998a; Christoffersen et al., 2005).
Basal crevasses may form where the stress intensity exceeds
a critical value and, once a critical flaw size is exceeded, are
able to grow quickly to their maximum extent. The size of
these initial flaws is on the order of a meter and it is likely that
such cracks will be found readily in the base of ice shelves,
especially near the grounding line (van der Veen, 1998a).
Marine ice might be expected to concentrate in basal
crevasses because of the buoyant nature of super-cooled, low
salinity source water (Khazendar and Jenkins, 2003), and
certain regions of Larsen C Ice Shelf are subject to marine
ice accretion (Holland et al., 2009). Outside these regions
basal crevasses will continue to creep open through the same
stresses that initiated them, and melting of the ice faces in
contact with the ocean may enhance this process where the
water is sufficiently warm (Jenkins and Doake, 1991), as is
believed to be the case for Larsen C Ice Shelf (Shepherd et
al., 2003). For crevasse heights of 100–200 m in Series 1 and
2, we observe maximum crevasse widths at the ice shelf base
of hundreds of meters (Figs. 3 and 4).
Where basal crevasses do not freeze shut again quickly
after opening (Weertman, 1980) or fill significantly with
marine ice, hydrostatic forces can no longer completely sup-
port the ice above them. As a result, the surface of the ice
shelf can deform viscously downwards to provide a surface
expression of the basal crevasse below in the form of a trough
(Shabtaie and Bentley, 1982). This process explains the sur-
face troughs we measured, and also the deformation of inter-
nal layers: deeper (older) layers may show more deformation
than the surface layers because more recent layers are likely
to be subject to uneven accumulation which would tend to
favour the trough. The trough widths we observe (Figs. 3
and 4) are consistent with the maximum widths of the basal
crevasses below them (hundreds of meters).
The Cryosphere, 6, 113–123, 2012 www.the-cryosphere.net/6/113/2012/
A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf 117
27
536 (a) Four passes over S2C1
(b) Two passes over S2C2
Figure 4. Multiple GPR profiles over crevasses S2C1 and S2C2 along transects shown in 537 Fig. 2b. Axes are in meters and not to scale: x-dimension assumes one radar trace every 4.3 m 538 of travel and y-dimension assumes a constant velocity in ice of 0.175 m/ns. Multiple passes 539 are illustrated to demonstrate measurement repeatability and correspond to coloured transects 540 in Fig. 2b as follows; 1 and 4:blue, 2 and 3:red, 5:yellow and 6:purple. Flow direction in each 541 case is left to right. Surface crevasse reflections highlighted in red. 542
27
536 (a) Four passes over S2C1
(b) Two passes over S2C2
Figure 4. Multiple GPR profiles over crevasses S2C1 and S2C2 along transects shown in 537 Fig. 2b. Axes are in meters and not to scale: x-dimension assumes one radar trace every 4.3 m 538 of travel and y-dimension assumes a constant velocity in ice of 0.175 m/ns. Multiple passes 539 are illustrated to demonstrate measurement repeatability and correspond to coloured transects 540 in Fig. 2b as follows; 1 and 4:blue, 2 and 3:red, 5:yellow and 6:purple. Flow direction in each 541 case is left to right. Surface crevasse reflections highlighted in red. 542
Fig. 4. Multiple GPR profiles over crevasses S2C1 and S2C2 along transects shown in Fig. 2b. Axes are in meters and not to scale: x-
dimension assumes one radar trace every 4.3m of travel and y-dimension assumes a constant velocity in ice of 0.175m ns−1. Multiple
passes are illustrated to demonstrate measurement repeatability and correspond to coloured transects in Fig. 2b as follows; 1 and 4: blue; 2
and 3: red; 5: yellow and 6: purple. Flow direction in each case is left to right. Surface crevasse reflections highlighted in red.
www.the-cryosphere.net/6/113/2012/ The Cryosphere, 6, 113–123, 2012
118 A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf
The depths of surface troughs above basal crevasses are in-
fluenced by a number of factors, some of which are not well
known for our study. These include the penetration height of
the crevasse, the density profile through the ice column, the
differential surface accumulation patterns since the trough
was formed, how much marine ice has collected within the
crevasse, and finally whether there is stress-bridging (other-
wise known as flexural stresses) across the crevasse. If we
assume no stress bridging, surface accumulation to be homo-
geneous, ice density to be consistent with that measured else-
where by CMP survey, and marine ice not to be present, then
hydrostatic equilibrium would predict surface trough depths
approximately 1.5 times as deep as we measure (Table 1).
Therefore not all of these assumptions can be correct and
one or more effect must be in operation to limit the crevasse
trough depth.
In summary, both series of surface features that we inves-
tigated on Larsen C Ice Shelf, and which represent just a
few of an extensive series of such features stretching from
near the grounding line all the way to the calving margin,
can be entirely explained by the presence of basal crevasses.
The signatures in satellite imagery of these features are the
surface troughs that form above the basal crevasses in re-
sponse to hydrostatic forces, and which are limited in depth
by differential surface or basal accumulation, or by stress
bridging.
3.2 Surface crevasses
Surface crevasses may be initiated by the same stresses as
basal crevasses and may therefore be expected to form in
similar parts of an ice shelf (van der Veen, 1998b). With-
out the influence of water at pressure, however, surface
crevasses cannot penetrate nearly as far through the ice col-
umn as basal crevasses unless a plentiful source of surface
melt-water is invoked (van den Broeke, 2005). One esti-
mate of the typical ratio of surface crevasse penetration depth
to basal crevasse penetration height under equivalent stress
conditions is around 1:10 (Weertman, 1980). If we make
the assumption that basal and surface crevasses have similar
width-to-depth aspect ratios, the maximum width of a surface
crevasse will therefore be at most ∼10% of that of a basal
crevasse in the same part of an ice shelf. This ignores the
likelihood of melt widening within a basal crevasse, which
would make the ratio of basal to surface crevasse widths
still greater. Sustained melt widening in surface crevasses
is highly unlikely given the mean air temperatures where ice
shelves form. Thus, where the surface expression of a basal
crevasse is expected to have a visible width of hundreds of
meters, a neighbouring surface crevasse will at most be me-
ters to tens of meters wide. The implications of this differ-
ence for their visibility in satellite data is discussed later.
The near-surface hyperbolic radar responses highlighted in
Fig. 4 are characteristic of surface crevasses (Woodward and
Burke, 2007). Despite their prevalence in our common-offset
GPR data, surface crevasses were only seen occasionally in
the field on the down-stream side of basal crevasse troughs.
Where they were not obscured by accumulated snow, their
width was a few meters and thus consistent with the expected
ratio of surface to basal crevasse widths. It is not known
whether these surface crevasses were generated at the same
time as the basal crevasses as a result of the same stresses that
initiated the latter, or during a later period as a result of the
local downwards deformation responsible for the formation
of the trough.
4 Satellite observations from Larsen C Ice Shelf
The widths of the basal crevasse troughs observed in MODIS
(Fig. 1) and Landsat (Fig. 2) imagery is entirely consistent
with our field data. At the low sun-angles experienced at this
latitude, the pattern of reflectance in the satellite images is
also consistent with the known depths and shape of the sur-
face troughs. There are no other features in Fig. 2 that can-
not be explained by the presence of basal crevasses in this
region. Elsewhere on the ice shelf, however, recent satellite
images exhibit features that are unlike those we investigated.
Figure 5 shows Landsat 7 panchromatic images (15×15 m
pixels) of an area in the North of Larsen C Ice Shelf con-
taining features that closely match Series 1 crevasses (S1C1
and S1C2) in their size, shape and periodicity (∼18 yr) and
we therefore interpret these also as the surface expressions
of basal crevasses. Alongside these troughs are much nar-
rower surface features that are only a pixel or two wide, are
longer than the surface troughs, but have the same orienta-
tion (Fig. 5a). Similar diverse sets of features are also found
near the grounding line in the South of Larsen C Ice Shelf
(Fig. 5b) and have been noted on the Fimbul Ice Shelf (Hum-
bert and Steinhage, 2011).
Figure 5 also shows that between 1986 and 2010, a new
basal crevasse trough formed in this series. Despite a care-
ful inspection of all available images, including US Depart-
ment of Defence declassified intelligence satellite photogra-
phy (DISP) from 1963, no such development of new features
in crevasse Series 2 was detected.
5 Discussion of satellite observations
5.1 Background
Surface features on ice shelves have been observed in
remotely-sensed images since the earliest satellite missions
(e.g. Crabtree and Doake, 1980). Their shape, pattern
and development through time have been used to infer ice
source, flow direction and speed. A variety of features have
been described including “rifts”, “flow-lines”, “crevasses”,
“crevasse fields” and “crevasse plumes”, and most studies of
such features have used data from the Landsat or MODIS
The Cryosphere, 6, 113–123, 2012 www.the-cryosphere.net/6/113/2012/
A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf 119
28
(a) (b)
(c) (d)
Figure 5. Landsat sub-images from parts of Larsen C Ice Shelf indicated in Figure 1 543 showing: Basal crevasse troughs and adjacent narrow features towards the North (a) and 544 South (b) of the ice shelf; and (c, d) advection of the surface expression of basal crevasses 545 between 1986 and 2010 and development of a further basal crevasse mid-shelf (flow is left to 546 right and the two 1986 features have been advected ~5km in this period). 547
548
28
(a) (b)
(c) (d)
Figure 5. Landsat sub-images from parts of Larsen C Ice Shelf indicated in Figure 1 543 showing: Basal crevasse troughs and adjacent narrow features towards the North (a) and 544 South (b) of the ice shelf; and (c, d) advection of the surface expression of basal crevasses 545 between 1986 and 2010 and development of a further basal crevasse mid-shelf (flow is left to 546 right and the two 1986 features have been advected ~5km in this period). 547
548
Fig. 5. Landsat sub-images from parts of Larsen C Ice Shelf indicated in Fig. 1 showing: basal crevasse troughs and adjacent narrow features
towards the North (a) and South (b) of the ice shelf; and (c, d) advection of the surface expression of basal crevasses between 1986 and
2010 and development of a further basal crevasse mid-shelf (flow is left to right and the two 1986 features have been advected ∼5km in this
period).
satellite missions (e.g. Skvarca, 1994; Fahnestock et al.,
2000; Glasser et al., 2009).
MODIS images provides a pixel size of 250m. Early
Landsat Multi-Spectral Scanner (MSS) images provided a
pixel size of 60–80m in the visible bands, which was im-
proved to 30m through the development of the satellite mis-
sion and the launch of new Landsat Thematic Mapper instru-
ments, but it was not until the launch of Landsat 7 in 1999
that data with a 15m pixel size was readily available. Pixel
size is not equivalent to spatial resolution, but the spatial res-
olution of these sensors is not significantly different from the
spacing of the data grid on which they are distributed. Fea-
tures smaller than the spatial resolution may be resolved in
such images, but only as long as they exhibit adequate ra-
diometric contrast to their background and are sufficiently
widely spaced. The various surface features identified in pre-
vious ice shelf studies are, of course, formed from the same
basic material and therefore cannot be discriminated by any
inherent spectral properties. Their contrast is owed entirely
to their three-dimensional shape, and the way in which this
modifies the reflectance of sunlight that naturally has a low
solar azimuth at polar latitudes.
5.2 Observations
Considering the spatial resolution of early Landsat and re-
cent MODIS data, and the likely shape and widths of surface
crevasses compared to basal crevasses, it is highly unlikely
that ice shelf features identified from such satellite data and
described as “crevasses” are surface crevasses. For surface
www.the-cryosphere.net/6/113/2012/ The Cryosphere, 6, 113–123, 2012
120 A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf
crevasses only meters wide, detection in data sampled at 60
or 250m would require an unreasonable level of contrast
given the homogeneity of materials. However, it is conceiv-
able that panchromatic data from Landsat 7 (15m pixels)
could resolve surface crevasses as well as the much wider
troughs associated with basal crevasses. The narrow features
in Fig. 5a and b, therefore, we interpret as surface crevasses.
These were not visible in MODIS or earlier Landsat data
of the northern part of Larsen C Ice Shelf, nor were they
detected in any Landsat images near basal crevasses in Se-
ries 1 or Series 2 (Fig. 2). However, similar narrow features
can be seen in Landsat 7 (15m pixels) images closer to the
grounding line in the region of the ice shelf that we surveyed
(Fig. 5b).
Finally, if we accept, by similarity of morphology to S2C1
and S2C2, that the large features in Fig. 5c and 5d are the
surface expressions of basal crevasses, then the development
of a new feature in this series between 1986 and 2010 is of
note. It demonstrates that basal crevasses penetrating up to
two-thirds of the way through the ice column may be initiated
and propagated due to stresses within the floating part of an
ice shelf a considerable distance downstream of the ground-
ing line.
In summary, we have shown that surface crevasses may
be observed in Landsat 7 data, but that these are highly un-
likely to be the features referred to in many previous papers
as “crevasses” as they are too narrow to have been observed
in any but the highest spatial resolution satellite data which
have only been available since 1999. We have demonstrated
that all of the features observed in satellite images that we
investigated by GPR may be explained by the presence of
basal crevasses and their surface expressions. We have es-
tablished that no feature in MODIS or early Landsat data is
likely simply to be a surface crevasse. By elimination, such
features and groups of features must therefore be either the
surface expressions of basal crevasses, or rifts which propa-
gate laterally and penetrate the entire ice column (Bassis et
al., 2005; Hulbe et al., 2010).
5.3 Implications
Within satellite images of medium spatial resolution such as
MODIS, surface features are widely observed in most parts
of the Larsen C Ice Shelf (Glasser et al., 2009; Fig. 1), and in
many ice shelves elsewhere (e.g. Crabtree and Doake, 1980;
Fahnestock et al., 2000). If all of these features correspond
to basal crevasses (penetrating up to two thirds of the ice
column) or rifts (penetrating the entire ice column), rather
than surface crevasses (with limited penetration), then collec-
tively they may be significant. Such abundant faults within
ice shelves will increase the surface area available for the ex-
change of mass and heat with the ocean cavity and may have
a substantial influence on stresses within the shelf.
6 Crevasse penetration height modelling
6.1 Introduction
A better understanding of where basal crevasses may form,
and how far they are likely to penetrate is desirable for fu-
ture prediction of ice shelf dynamics and stability. A num-
ber of approaches have previously been used to model basal
crevasse penetration heights (e.g. Weertman, 1980; Jezek,
1984; van der Veen, 1998a; Rist et al., 1999, 2002; Nick
et al., 2010). The approach adopted by Nick et al. (2010) as-
sumes that the crevasses are closely-spaced and uses the zero
stress criterion in which the balance of tensile stress, water
pressure and lithostatic stress is considered. We found this
method to significantly under-predict the penetration heights
of both series of crevasses investigated in this study, possi-
bly because they are more than 1km apart (over three times
the ice thickness) and therefore cannot be considered to be
“close”. Here we choose to test the LEFM (linear elastic
fracture mechanics) approach (Rist et al., 2002) in its abilities
to predict penetration depths because it is well established
for crevasse propagation in ice. Since crevasses are known
to propagate very rapidly under high tensile strain rates (Pa-
terson, 1994; Scott et al., 2010), it is often appropriate to
model ice as an elastic solid, despite its longer-term viscous
properties (Nye, 1970). The LEFM approach has been used
to demonstrate that basal crevasses can form readily in ice
shelves, but rarely on previous occasions to compare mea-
sured to predicted penetration heights (e.g. Rist et al., 2002),
probably because of the difficulty in finding crevasse obser-
vations coincident with known stress fields (van der Veen,
1998a).
We model crevasse penetration height with the LEFM ap-
proach for all four crevasses for which the penetration height
of the crevasse tip could be estimated from the GPR data.
Here we have suitable in-situ measurements of crevasse pen-
etration heights and ice densities, and we calculate the depth-
dependent stress from observed surface strain rates using
manual feature tracking of Landsat satellite images. Al-
though the crevasses will have been advected downflow since
their initiation, we choose to compare measured to mod-
elled penetration heights using the tensile stress calculated
at their present locations. Here, since the ice continues to
accelerate towards the calving margin, the tensile stress is
at a maximum for the crevasse history. It is possible that
open basal crevasses maintain a crack tip sharp enough to
continue to propagate (aided by tidal flexure) as they are ad-
vected downflow and thereby would be able to adapt to the
current stress environment. Choosing to use the present loca-
tion strain rates will thus make the modelled crevasses pene-
tration heights an upper bound.
The model implementation follows Rist et al. (2002) and
is briefly described here. To calculate basal crevasse penetra-
tion depth we consider the balance of total stresses across a
crevasse (which include longitudinal and lithostatic stresses)
The Cryosphere, 6, 113–123, 2012 www.the-cryosphere.net/6/113/2012/
A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf 121
and the water pressure within it. Longitudinal stresses are de-
rived from the satellite-derived strain rates using Glen’s con-
stitutive flow relation (e.g. Paterson, 1994) and are dependent
on depth through the relationship to temperature. Following
Sandh¨
ager et al. (2005), the temperature is approximated by
the function:
T (a) =(TS−Tb)·a
H1
3+Tb(1)
where T=Temperature, ais the position in the ice column,
TSis surface temperature, Tbis basal temperature, and H is
ice thickness. The surface temperature we used is the annual
mean for Larsen C Ice Shelf at −16◦C and the basal tem-
perature is −2◦C, or approximately the freezing point of sea
water. This approach gives a temperature profile with a steep
gradient at the base typical of a shelf undergoing basal melt-
ing (Paterson, 1994; Shepherd et al., 2003). Ice thickness is
taken from the field measurements (Figs. 3, 4) and the calcu-
lation of lithostatic stress requires a density profile through
the ice column which is derived from seismic measurements
in the field (Jansen et al., 2010).
Following Fett et al. (1990), we calculate the stress inten-
sity factor for the geometry of an edge crack within a plate
of finite thickness, an approach which is valid until the ratio
between ice thickness and crevasse height is smaller than 0.9.
The calculation is complicated by the fact that temperature
and density vary within the ice column in such a way that
the stress balance is nonlinear with depth. The numerical
approach approximates the non-linear stress balance with a
polynomial from which a solution for the stress intensity fac-
tor can be calculated (Rist et al., 2002).
The process of elastic linear fracturing is illustrated in
Fig. 6 for the four basal crevasses discussed. Once the crit-
ical stress intensity of 1.5×105Pa m0.5(Rist et al., 2002) is
exceeded, a crevasse can start to propagate upwards. At first
the stress intensity factor at the crevasse tip begins to increase
with height, giving even more impetus to the fracture. Even-
tually it decreases until it returns once again to the critical
stress intensity, causing the crevasses tip to stop propagating.
The penetration height at which this happens is dependent
on the strain rate and ice thickness, and might therefore dif-
fer between the crevasses. Figure 6 highlights the fact that
longitudinal stress is a stronger controlling factor than ice
thickness in this part of the shelf (Series 2 crevasses are pre-
dicted to be deeper than Series 1 crevasses). Table 1 gives
values for crevasse penetration heights modelled using this
approach, along with their estimates from GPR data.
29
549
550
551
Figure 6. Relationships between stress intensity and crevasses penetration height for the four 552
crevasses studied in detail.. Once initiated from a flaw of ~ 1m, a crevasse grows until the 553
stress intensity returns to the critical value, and its height is determined by the ice thickness, 554
density and deformation rate. 555
556
Fig. 6. Relationships between stress intensity and crevasses pene-
tration height for the four crevasses studied in detail. Once initiated
from a flaw of ∼1m, a crevasse grows until the stress intensity re-
turns to the critical value, and its height is determined by the ice
thickness, density and deformation rate.
6.2 Discussion of LEFM model results
The LEFM approach underestimates the penetration height
of all four basal crevasses, even though these predictions are
an upper bound on the assumption that the crevasses may
have adapted to the present minimum backstress conditions
(Table 1). Nevertheless, the modelled height of S1C1 and
S1C2 in crevasse Series 1 (Fig. 2a) are within 10–20% of
their measured values. Given the combined errors in model
parameters and field measurements, we infer that LEFM may
be a good predictor of crevasse penetration height for these
crevasses.
The penetration heights of C2C1 and C2C2 in crevasse Se-
ries 2 (Fig. 2b) are underestimated using LEFM by between
20 and 35 %, a discrepancy which falls outside even our most
conservative estimates of error. A likely explanation for this
discrepancy lies in the nature of stresses in this part of the
shelf. Crevasses S2C1 and S2C2 appear to be en echelon
features which form under significant shear stresses that ro-
tate the central part to form the crevasse into a sigmoid shape
(Benn and Evans, 2010). Shear stresses at this location calcu-
lated from flow model results (Jansen et al., 2010) are of the
same order of magnitude as the longitudinal stresses. The
modelling approach we have taken considers only mode I
stresses and does not take into account such shear processes,
which may act to further propagate basal crevasses in regions
with a non-simple stress regime.
www.the-cryosphere.net/6/113/2012/ The Cryosphere, 6, 113–123, 2012
122 A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf
7 Conclusions
Using geophysical survey and remote sensing on the Larsen
C Ice Shelf, we have demonstrated that a group of features
visible in MODIS images, and previously described simply
as “crevasses”, are completely explained as the surface ex-
pressions of basal crevasses in the form of surface troughs.
We have shown that such features cannot be surface crevasses
because they would not be resolved at the spatial resolu-
tion of MODIS. We have shown that it is possible for basal
crevasses to be generated as a result of stresses well down-
stream of the grounding line and that they can sometimes
penetrate as much as two thirds of the ice column.
Having surveyed and modelled four basal crevasses in
Larsen C Ice Shelf, we find that the LEFM approach is an
appropriate predictor of penetration height where stresses are
predominantly tensile, but can significantly underestimate
penetration height in regions affected by shear. The surface
trough depth for the crevasses we measured is less than might
be expected through hydrostatic equilibrium, most likely be-
cause of preferential surface accumulation within them, ma-
rine ice accretion in the basal crevasses below or stress bridg-
ing from the surrounding ice.
Surface features are common in MODIS images of the
Larsen C and other ice shelves. By extrapolation, these fea-
tures must either be the surface expressions of basal crevasses
(which can penetrate a significant proportion of the ice col-
umn) or rifts that have propagated laterally (penetrating the
entire ice column) rather than surface crevasses (which pen-
etrate relatively small distances through the ice). Although
they have received little attention to date, such abundant and
sometimes deep faults within ice shelves have implications
for both their thermodynamic and structural properties. By
increasing the area of interface between ice and water, they
may enhance heat exchange with the ocean deep inside the
shelf where the ice would otherwise be well insulated from
external sources of heat.
Through their potential to be initiated mid-shelf, and by
modifying the bulk mechanical properties of the ice, basal
crevasses may play an important part in ice shelf stability.
Penetration heights may be enhanced by surface densifica-
tion as the climate warms, and it is possible that they have
contributed to the break-up of ice shelves on the Antarctic
Peninsula (Holland et al., 2011). To assess the current stabil-
ity and predict the future longevity of ice shelves, the abun-
dance and penetration height of basal crevasses should be
considered within ice shelf models.
Acknowledgements. This project was funded by the UK Natural
Environment Research Council through the Antarctic Funding
Initiative (AFI) (NE/E012914/1). GPR and GPS equipment was
kindly loaned by the NERC Geophysical Equipment Facility
(GEF-905). We would particularly like to thank Neil Glasser for
sharing DISP data with us, and Catrin Thomas for assistance during
the field campaign.
Edited by: A. Klein
References
Barrett, B., Murray, T., and Clark, R.: Errors in radar CMP velocity
estimates due to survey geometry, and their implication for ice
water content estimation, J. Environ. Eng. Geophys., 12, 101–
111, doi:10.2113/JEEG12.1.101, 2007.
Bassis, J. N., Coleman, R., Fricker, H. A., and Minster,
J. B.: Episodic propagation of a rift on the Amery Ice
Shelf, East Antarctica, Geophys. Res. Lett., 32, L06502,
doi:10.1029/2004GL022048, 2005.
Benn, D. I. and Evans D. J. A.: Glaciers and Glaciation, 2nd Ed.,
Hodder Education, London, 2010.
Christoffersen, P., Piotrowski, J. A., Nicolaj, K., and Larsen, N. K.:
Basal processes beneath an Arctic glacier and their geomorphic
imprint after a surge, Elisebreen, Svalbard, Quaternary Res., 64,
125–137, 2005.
Crabtree, R. D. and Doake, C. S. M.: Flow lines
on Antarctic ice shelves, Polar Rec., 20, 31–37,
doi:10.1017/S0032247400002898, 1980.
Fahnestock, M. A., Scambos, T. A., Bindschadler, R. A., and
Kvaran, G.: A millennium of variable ice flow recorded
by the Ross Ice Shelf, Antarctica, J. Glaciol., 46, 652–664,
doi:10.3189/172756500781832693, 2000.
Fett, T., Munz, D., and Neumann, J.: Local stress intensity factors
for surface cracks in plates under power-shaped stress distribu-
tions, Eng. Fract. Mech., 36, 647–651, 1990.
Glasser, N. F. and Scambos, T. A: A structural glaciological analysis
of the 2002 Larsen B ice-shelf collapse, J. Glaciol., 54, 3–16,
doi:10.3189/002214308784409017, 2008.
Glasser, N. F., Kulessa, B., Luckman, A., Jansen, D., King,
E. C., Sammonds, P. R., Scambos, T. A., and Jezek,
K. C.: Surface structure and stability of the Larsen C
Ice Shelf, Antarctic Peninsula, J. Glaciol., 55, 400–410,
doi:10.3189/002214309788816597, 2009.
Haran, T., Bohlander, J., Scambos, T., Painter, T., and Fahnestock,
M.: MODIS mosaic of Antarctica (MOA) image map, Boulder,
Colorado USA: National Snow and Ice Data Center, Digital me-
dia, 2006.
Harper, J. T., Bradford, J. H., Humphrey, N. F., and
Meierbachtol, T. W.: Vertical extension of the subglacial
drainage system into basal crevasses, Nature, 467, 579–582,
doi:10.1038/nature09398, 2010.
Hellmer, H. H. and Jacobs, S. S.: Ocean interactions with the base
of Amery Ice Shelf, Antarctica, J. Geophys. Res., 97, 1925–
1931, doi:10.1029/92JC01856, 1992.
Holland, P. R., Corr, H. F. J., Vaughan, D. G., Jenkins, A., and
Skvarca, P.: Marine ice in Larsen Ice Shelf, Geophys. Res. Lett.,
36, L11604, doi:10.1029/2009GL038162, 2009.
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,
doi:10.1029/2011GL047245, 2011.
Hulbe, C. L., LeDoux, C., and Cruikshank, K.: Propagation of long
fractures in the Ronne Ice Shelf, Antarctica, investigated using
a numerical model of fracture propagation, J. Glaciol., 56, 459–
472, doi:10.3189/002214310792447743, 2010.
Humbert, A. and Steinhage, D.: The evolution of the western rift
area of the Fimbul Ice Shelf, Antarctica, The Cryosphere, 5, 931–
944, doi:10.5194/tc-5-931-2011, 2011.
The Cryosphere, 6, 113–123, 2012 www.the-cryosphere.net/6/113/2012/
A. Luckman et al.: Basal crevasses in Larsen C Ice Shelf 123
Jansen, D., Kulessa, B., Sammonds, P. R, Luckman, A., King, E.
C., and Glasser, N. F.: Present stability of the Larsen C Ice Shelf,
Antarctic Peninsula, J. Glaciol., 56, 593–600, 2010.
Jezek, K. C.: A modified theory of bottom crevasses used as
a means for measuring the buttressing effect of ice shelves
on inland ice sheets, J. Geophys. Res., 89, 1925–1931,
doi:10.1029/JB089iB03p01925, 1994.
Jezek, K. C. and Bentley, C. R.: Field studies of bottom crevasses
in the Ross Ice Shelf, Antarctica, J. Glaciol., 29, 118–126, 1983.
Jenkins, A. and Doake, C. S. M.: Ice-ocean interaction on
Ronne Ice Shelf, Antarctica, J. Geophys. Res., 96, 791–813,
doi:10.1029/90JC01952, 1991.
Jezek, K. C., Bentley, C. R., and Clough, J. W.: Electromagnetic
sounding of bottom crevasses on the Ross Ice Shelf, Antarctica,
J. Glaciol., 24, 321–330, 1979.
Khazendar, A. and Jenkins, A.: A model of marine ice formation
within Antarctic ice shelf rifts, J. Geophys. Res., 108, 3235,
doi:10.1029/2002JC001673, 2003.
McGrath, D., Steffen, K., Scambos, T., Rajaram, H., Casassa,
G., and Rodriguez, J.: Basal crevasses and associated surface
crevassing on the Larsen C Ice Shelf, Antarctica and their role in
ice shelf instability, Ann. Glaciol., 53, in press, 2012.
Nick, F. M., van der Veen, V. A., and Benn, D. I.: A physically
based calving model applied to marine outlet glaciers and im-
plications for the glacier dynamics, J. Glaciol., 56, 781–794,
doi:10.3189/002214310794457344, 2010.
Nye, J. F.: Glacier sliding without cavitation in a linear vis-
cous approximation, P. R. Soc. Lond. A. Ma., 315, 381–403,
doi:10.1098/rspa.1970.0050, 1970.
Paterson, W. S. B.: The Physics of Glaciers3rd Edition, Reed Edu-
cational and Professional Publishing, Oxford, 1994.
Peters, M. E., Blankenship, D. D., Smith, D. E., Holt, J.
W., and Kempf, S. D.: The distribution and classifica-
tion of bottom crevasses from radar sounding of a large
tabular iceberg, IEEE Geosci. Remote Sens., 4, 142–146,
doi:10.1109/LGRS.2006.887057, 2007.
Rist, M. A., Sammonds, P. R., Murrell, S. A. F., Meredith, P. G.,
Doake, C. S. M., Oerter, H., and Matsuki, K.: Experimental
and theoretical fracture mechanics applied to Antarctic ice frac-
ture and surface crevassing, J. Geophys. Res., 104, 2973–2987,
doi:10.1029/1998JB900026, 1999.
Rist, M. A., Sammonds, P. R., Oerter, H., and Doake, C. S. M.:
Fracture of Antarctic shelf ice, J. Geophys. Res., 107, 2002,
doi:10.1029/2000JB000058, 2002.
Sandh¨
ager, H., Rack, W., and Jansen, D.: Model investigations of
Larsen B Ice Shelf dynamics prior to the breakup, Forum for
Research into Ice Shelf Processes (FRISP), Report, 16, 5–12,
Bjerknes Cent. For Clim. Res., Bergen, Norway, 2005.
Scott, J. B. T., Smith, A. M., Bingham, R. G., and Vaughan, D.
G.: Crevasses triggered on pine island glacier, west Antarctica,
by drilling through an exceptional melt layer, Ann. Glaciol., 51,
65–70, doi:10.3189/172756410791392763, 2010.
Shabtaie, S. and Bentley, C. R.: Tabular icebergs: implications
from geophysical studies of ice shelves, J. Glaciol., 28, 413–430,
1982.
Shepherd, A., Wingham, D., Payne, T., and Skvarca, P.: Larsen
Ice Shelf has progressively thinned, Science, 302, 856–859,
doi:10.1126/science.1089768, 2003.
Skvarca, P.: Changes and surface features of the Larsen Ice Shelf,
Antarctica, derived from Landsat and kosmos mosaics, Ann.
Glaciol., 20, 6–12, 1994.
Thomas, R. H.: The creep of ice shelves: Theory, J. Glaciol., 12,
45–53, 1973.
van den Broeke, M.: Strong surface melting preceded collapse of
Antarctic Peninsula Ice Shelf, Geophys. Res. Lett., 32, L12815,
doi:10.1029/2005GL023247, 2005.
van der Veen, C.: Fracture mechanics approach to penetration of
bottom crevasses on glaciers, Cold Reg. Sci. Technol., 27, 213–
223, doi:10.1016/S0165-232X(98)00006-8, 1998a.
van der Veen, C.: Fracture mechanics approach to penetration of
surface crevasses on glaciers, Cold Reg. Sci. Technol., 27, 31–
47, doi:10.1016/S0165-232X(97)00022-0, 1998b.
Weertman, J.: Bottom crevasses, J. Glaciol., 25, 185–188, 1980.
Woodward, J. and Burke, M. J.: Applications of ground-penetrating
radar to glacial and frozen materials, J. Environ. Eng. Geophys.,
12, 69–85, doi:10.2113/JEEG12.1.69, 2007.
www.the-cryosphere.net/6/113/2012/ The Cryosphere, 6, 113–123, 2012
