UC Irvine
UC Irvine Previously Published Works
Title
Rapid submarine melting of the calving faces of West Greenland glaciers
Permalink
https://escholarship.org/uc/item/3hw228ts
Journal
Nature Geoscience, 3(3)
ISSN
1752-0894
Authors
Rignot, E
Koppes, M
Velicogna, I
Publication Date
2010-03-01
DOI
10.1038/ngeo765
Copyright Information
This work is made available under the terms of a Creative Commons Attribution License,
availalbe at https://creativecommons.org/licenses/by/4.0/
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
LETTERS
PUBLISHED ONLINE: 14 FEBRUARY 2010 | DOI: 10.1038/NGEO765
Rapid submarine melting of the calving faces of
West Greenland glaciers
Eric Rignot
1,2
*
, Michele Koppes
3
and Isabella Velicogna
1,2
Widespread glacier acceleration has been observed in
Greenland in the past few years
1–4
associated with the thinning
of the lower reaches of the glaciers as they terminate in
the ocean
5–7
. These glaciers thin both at the surface, from
warm air temperatures, and along their submerged faces
in contact with warm ocean waters
8
. Little is known about
the rates of submarine melting
9–11
and how they may affect
glacier dynamics. Here we present measurements of ocean
currents, temperature and salinity near the calving fronts of
the Eqip Sermia, Kangilerngata Sermia, Sermeq Kujatdleq and
Sermeq Avangnardleq glaciers in central West Greenland, as
well as ice-front bathymetry and geographical positions. We
calculate water-mass and heat budgets that reveal summer
submarine melt rates ranging from 0.7±0.2to3.9±0.8md
−1
.
These rates of submarine melting are two orders of magnitude
larger than surface melt rates, but comparable to rates of
iceberg discharge. We conclude that ocean waters melt a
considerable, but highly variable, fraction of the calving fronts
of glaciers before they disintegrate into icebergs, and suggest
that submarine melting must have a profound influence on
grounding-line stability and ice-flow dynamics.
In the past decade, surface melt around Greenland has markedly
increased in magnitude and spatial extent, whereas snowfall has
increased only slightly
12–14
. As a result, the ice-sheet mass deficit
tripled in the period 1996–2007 (ref. 15). Critically, 50–60% of
that loss was caused by an acceleration of the outlet glaciers;
the remainder resulting from an increase in surface melt. Hence,
glacier acceleration is a significant, if not dominant, response to
climate warming.
The widespread two- to threefold acceleration of the glaciers
cannot be explained solely by enhanced lubrication of the bed from
surface meltwater
16
, for seasonal variations in glacier velocity do
not exceed 8–10%, independent of latitude
1
. Glacier acceleration
is instead probably caused by the ungrounding of ice fronts from
the bed, which reduces buttressing of inland ice and entrains faster
rates of ice flow to the ocean
5
. To unground glaciers from the bed,
they must melt and thin. Warmer air temperatures thin the glaciers
from the surface and allow the ice flotation margin to migrate
inland. Such surface melting is well documented in Greenland
12–14
.
However, melting can also occur along the submarine termini of
the glaciers. A warmer ocean will erode submerged grounded ice
and cause the grounding line to retreat. In contrast, we know very
little about rates of submarine melting along calving fronts. The
only measurements of submarine glacier melting so far have been
conducted in Alaska
9,10
.
In August 2008, we deployed two InterOcean S4 conductivity,
temperature and depth (CTD)/current meters, a Seabird SBE-19
conductivity and temperature density profiler, and a Lowrance 18C
sonar depth sounder with a global positioning system in the fjords of
1
University of California, Earth System Science, Irvine, California 92617, USA,
2
Jet Propulsion Laboratory, Pasadena, California 91109, USA,
3
University of
British Columbia, Department of Geography, Vancouver, British Columbia, V6T 1Z2, Canada. *e-mail: erignot@uci.edu.
Eqip Sermia (EQIP), Kangilerngata Sermia (KANGIL) and Sermeq
Kujatdleq and Avangnardleq (TOR), 100 km north of Ilulissat
(Fig. 1). Seven to ten casts were taken at 500–1,000 m intervals in
each fjord, between 200 and 1,000 m (EQIP and KANGIL) and
4 km (TOR) from the ice fronts. We sampled the water column
from 1 m below the surface to a depth of 80 m at EQIP, 150 m
at KANGIL and 200 m at TOR, which is not the entire water
column. Water velocity, temperature and salinity were measured
continuously, as well as averaged, at 1.5-m depth intervals in the top
50 m and at 15-m intervals below 50 m, because most of the velocity
structure was observed in the upper layers
9,11
. Fjord bathymetry
was also mapped, except in TOR where the fjord depth exceeded
our instrument capability. We complemented our bathymetric data
with a global bathymetric data set
17
, which agreed to within 10–15 m
at the fjord centres with our measurements, but differed by up
to 80 m along the fjord edges because of poor definition of the
land–ocean boundary in the global data set. During the survey, we
experienced mildly warm weather (5–15
◦
C), calm wind conditions
and no remnant seasonal sea-ice.
To model the forced convective flow into the fjord and derive
its mass and heat budgets, we modified the method of Motyka and
colleagues
9
. In this two-layer approach, subglacial water discharge
drives ocean convection, drawing deep, warm saline waters towards
the near-vertical terminal ice face, where the two components
turbulently mix and rise along the face to reach the surface and
flow away from the terminus in an overflow plume (Fig. 2). The
ascending waters melt ice along the calving face, and contribute to
the overflow plume. Subglacial meltwater typically rises to the water
surface within a kilometre of the ice front
11
. The overflow plume is
not necessarily homogeneous, because subglacial water is ejected at
discrete locations across the glacier front. Greenland’s fjords are also
fronted by shallow sills that limit the exchange of warm, saline water
between the ocean and the fjord. EQIP and KANGIL have sill depths
of 200 m. TOR is 450 m deep near the ice front, with a sill depth of
300 m at the fjord entrance
17
.
The casts revealed the presence of small jets in the upper 40–60 m
of the water column (Fig. 3). Salinity and temperature increase
with depth, except at EQIP where the unstratified temperature
data suggest an inflow of warm water to the glacier centre and
an outflow of colder water along the fjord sides. In EQIP, jets
at 10 and 30 m depth flow at 30–35 cm s
−1
away from the ice
to the northwest, along the deepest trough (Figs 1 and 3). With
inflow coming from the south to the glacier centre, this circulation
pattern is contrary to that observed in the other fjords and
to our simplified model. Below 50 m depth, however, current
speeds drop to a few centimetres per second, agreeing with a
two-layer structure. In KANGIL, high outflow is observed above
40 m depth along the western and eastern edges of the fjord. In
TOR, similar jets are found in the top 60 m to the north and
NATURE GEOSCIENCE | VOL 3 | MARCH 2010 | www.nature.com/naturegeoscience 187
© 2010 Macmillan Publishers Limited. All rights reserved.
LETTERS
NATURE GEOSCIENCE DOI: 10.1038/NGEO765
0
0
0
Sermeq
Avangnardleq
Sermeq Kujatdleq
Kangilern
gata Sermia
Torssukatak
Fjord
Eqip Sermia
TOR-8
TOR-3
TOR-4
TOR-5
TOR-7
TOR-2
KANGIL-9
KANGIL-7
KANGIL-3
KANGIL-10
EQIP-2
EQIP-6
300
300
200
200
500
400
100
EQIP-7
KANGIL-8
KANGIL-2
EQIP-3
EQIP-4
0
4
21
N
50° 0’ W50° 15’ W50° 30’ W
Longitude
Latitude
70° 0’ N
69° 45’ N
(km)
0
0
Figure 1 | Bathymetry and location of study area. Bathymetric contours (50-m interval, blue lines) and cast locations (yellow dots with cast names) for
Eqip Sermia, Kangilerngata Sermia, Sermeq Kujatdleq and Sermeq Avangnardleq, West Greenland, overlaid on an ASTER satellite image. Bathymetric
survey lines are thin red lines. Ocean–ice, land–ice and land–ocean boundaries are, respectively, delineated as thick red, thin black and thin white lines. The
black square in the inset map shows the location of the glaciers in central West Greenland.
middle of the fjord, all drawing water away from the glacier
front. In all three fjords, we had no visual evidence of upwelling
of turbid subglacial waters in front of the glaciers, although
these areas were covered in brash ice periodically stirred by the
calving of icebergs.
To evaluate the errors in the simplified two-layer model, we
considered both the uncertainty in the positioning of the transition
boundary and uncertainties in the average temperature and salinity
of the inflow layer (see the Methods section). All calculations
assume steady-state conditions and negligible influence of tidal
currents. As the flow pattern in EQIP calls our simplified model into
question, we also calculated the mass and heat budget of the entire
water column in EQIP and KANGIL fjords. The results from the
two approaches agreed quite well (within 2–20%), which provides
a verification of our model.
In EQIP, the calculated freshwater outflux of 116 ± 11 m
3
s
−1
is 3% of the incoming flux of 3,362 ± 278 m
3
s
−1
. The heat
budget indicates a submarine meltwater production of 5± 2m
3
s
−1
for a net submarine melt rate of 0.7 ± 0.2md
−1
across the
0.6-km
2
submerged ice face. In KANGIL, the freshwater outflux
is 286 ± 61 m
3
s
−1
with a submarine melt flux of 18 ± 4m
3
s
−1
.
Q
p
, H
p
Q
s
, H
s
Deep water
Sill
W
Overflow plume
Glacier fjord
Glacier
Q
m
, H
m
Q
sg
, H
sg
Figure 2 | A simplified two-layer model of forced convective flow in a
glacier fjord. Deep-water access is guarded by a sill and terminated by a
calving front
9
. The incoming mass flux from the deep ocean, Q
s
, and from
subglacial water, Q
sg
, is balanced by the mass flux from the overflow plume,
Q
p
, and the submarine meltwater, Q
m
. The incoming deep-ocean heat flux,
H
s
, and subglacial water heat flux, H
sg
, melt submarine ice with a heat flux,
H
m
, to yield an overflow plume with a heat flux, H
p
. The overflow plume is
not homogeneous in velocity structure.
188 NATURE GEOSCIENCE | VOL 3 | MARCH 2010 | www.nature.com/naturegeoscience
© 2010 Macmillan Publishers Limited. All rights reserved.
NATURE GEOSCIENCE DOI: 10.1038/NGEO765
LETTERS
0
EQIP-6
KANGIL-9
TOR-7
TOR-6
TOR-5
TOR-4
TOR-8
TOR-2
TOR-7
TOR-6
TOR-5
TOR-4
TOR-8
TOR-2
TOR-7
TOR-6
TOR-5
TOR-4
TOR-8
TOR-2
KANGIL-7
KANGIL-8
KANGIL-2
KANGIL-1
KANGIL-9
KANGIL-7
KANGIL-8
KANGIL-2
KANGIL-1
KANGIL-9
KANGIL-7
KANGIL-8
KANGIL-2
KANGIL-1
EQIP-4
EQIP-3
EQIP-2
EQIP-7
EQIP-6
EQIP-4
EQIP-3
EQIP-2
EQIP-7
EQIP-6
EQIP-4
EQIP-3
EQIP-2
EQIP-7
20
40
60
80
100
1,000 2,000
North¬south distance (m)
North–south distance (m) North–south distance (m)
North–south distance (m)
North–south distance (m)
Depth (m)
Depth (m)
Depth (m)
Depth (m)
Depth (m)
Depth (m)
Depth (m)
Depth (m)
Depth (m)
North–south distance (m) North–south distance (m)
North¬south distance (m) North–south distance (m)
3,000 4,000 5,0000
0
20
40
60
80
100
1,000 2,000 3,000 4,000 5,0000
1,000 2,000 3,000 4,000 5,0000
0
200
150
100
50
0
200
150
100
50
0
20
40
60
80
100
1,000 2,000 3,000 4,0000
1,000 2,000 3,000 4,0000
400
300
200
100
0
0 2,000 4,000 6,000 8,000
0 2,000 4,000 6,000 8,000
0 2,000 4,000 6,000 8,000
400
300
200
100
0
400
300
200
100
0
a
d
g
e
h
f
i
bc
0.2 0.4 0.6 0.8
Temperature (°C)
1.0 1.2
30
¬10 ¬3 4 11
Velocity (cm s
¬1
)
18 25
¬10 ¬3 4 11 18 25
31 32
Salinity
33 34 35
Temperature (°C)
Salinity
35
0.2 0.4 0.6 0.8 1.0 1.2
Temperature (°C)
0.2 0.4 0.6 0.8 1.0 1.2
¬10 ¬3 4 11 18 25
30 31 32 33 34
Salinity
30 31 32 33 34 35
Velocity (cm s
¬1
)
Velocity (cm s
¬1
)
0
200
150
100
50
1,000 2,000 3,000 4,0000
Figure 3 | Water characteristics in three West Greenland glacial fjords. a–i, Water temperature (
◦
C), salinity and velocity normal to the flux gate (cm s
−1
,
positive outward, negative towards the glacier front) for EQIP (a–c), KANGIL (d–f) and TOR (g–i) as a function of water depth (m) and distance from north
to south (m). The position of individual casts is noted as a white vertical line. The purple horizontal lines at 47 m (EQIP, KANGIL) and 57 m (TOR) depth
denote the transition boundary between the overflow plume and the deeper ocean inflow used in our calculations. The horizontal and vertical scales vary
from one glacier plot to the next. Water-column areas with no data are coloured black. Areas with no ocean water (that is, at the edges of the fjord) are
coloured grey.
The average melt rate across the 0.6-km
2
submerged area is
2.6±0.5md
−1
. In TOR, the freshwater outflux of 874±46 m
3
s
−1
is
3% of the mass inflow. The submarine melt flux of 255 ± 33 m
3
s
−1
is 30% of the freshwater outflux and yields a submarine melt rate
of 3.9 ± 0.8md
−1
across the 5.7-km
2
submerged area of the two
ice fronts. Ocean thermal forcing, that is, the difference between
the water temperature and the freezing point of sea water at the
base of the calving face, is 2.8, 3.0 and 3.8
◦
C, respectively, for
the three fjords.
Ice–ocean interaction models suggest that submarine melting
increases with the square of thermal forcing from the ocean
18
.At
Leconte Glacier, summer submarine melting was 12 m d
−1
(ref. 9)
NATURE GEOSCIENCE | VOL 3 | MARCH 2010 | www.nature.com/naturegeoscience 189
© 2010 Macmillan Publishers Limited. All rights reserved.
LETTERS
NATURE GEOSCIENCE DOI: 10.1038/NGEO765
Table 1 | Mass and heat budgets of three West Greenland
glacial fjords from a two-layer model.
Fjord EQIP KANGIL TOR
Q
s
3,362 ±278 5,521± 394 26,967 ± 1,000
Q
p
3,478± 280 5,807±409 27,841± 1,018
Q
sg
111± 10 268± 60 619± 33
Q
m
5± 218± 4 255± 33
H
s
7.6± 0.515.7±0.2 160.6± 14
H
p
5.9± 0.29.4 ± 0.874.8± 3
H
m
1.7± 0.36.3± 0.985.8 ±14
V
m
0.7± 0.22.6± 0.53.9 ± 0.8
Q
r
132± 65 160± 76 516 ±236
Q
f
25± 524± 4 338± 15
Water influx from the deep ocean, Q
s
, outfluxes from the overflow plume, Q
p
, the subglacial
water, Q
sg
, and the submarine meltwater, Q
m
, in cubic metres per second. Heat influx from the
ocean, H
s
, and outfluxes from the overflow plume, H
p
, and to melt submarine ice, H
m
in 10
9
watts. Submarine melt rate, V
m
, in metres per day. July–August 2008 average runoff production,
Q
r
, and ice-front discharge, Q
f
, in cubic metres per second of water equivalent. The submerged
areas of the three glaciers are 0.612 km
2
, 0.626 km
2
and 5.68 km
2
(see Supplementary
Information). The full water-column budget calculation at EQIP yields Q
sg
= 118 ± 10 m
3
s
−1
,
Q
m
= 4 ± 2m
3
s
−1
, H
m
= 1.3 ± 0.3 × 10
9
W and V
m
= 0.6 ± 0.2md
−1
. At KANGIL, we find
Q
sg
= 297±60 m
3
s
−1
, Q
m
= 18± 4m
3
s
−1
, H
m
= 6.2± 0.8×10
9
W and V
m
= 2.6± 0.5md
−1
.
with a thermal forcing of 7.3
◦
C. This would imply rates of 1.8, 2.0
and 3.2md
−1
for EQIP, KANGIL and TOR fjords based on similar
thermal forcing, which is reasonably close to our calculations.
Similarly, submarine melting depends on the sine of the basal slope
of the ice face
18
. Beneath floating ice shelves, melt rates range from
centimetres to tens of metres per year
19
, but the sine of their basal
slope is typically only 0.01–0.02, two orders of magnitude lower
than for the vertical faces of tidewater glaciers, where much larger
melt rates are expected.
The average July–August 2008 runoff at the three glaciers
12,14
is estimated at 132 ± 65, 160 ± 76 and 516 ± 236 m
3
s
−1
versus
calculated subglacial water fluxes of 111 ± 10, 267 ± 60 and
619 ± 33 m
3
s
−1
(Table 1). This agreement within error bars of the
two estimates provides an independent validation of our water
budget calculations. We also calculated ice-front discharge in water
equivalent using a standard approach
15
to obtain 25± 5, 24± 4 and
338 ± 15 m
3
s
−1
, respectively, versus submarine meltwater fluxes
of 5±2, 18 ± 4 and 255 ± 33 m
3
s
−1
(Table 1). We conclude from
this comparison that 20–80% of the summer ice-front fluxes are
directly melted by the ocean. Submarine melting is therefore a
major form of ice ablation, which also varies significantly from
one fjord to the next.
In winter, subglacial water discharge will decrease because runoff
production ceases, forced convection in front of the glacier will
slow down and submarine melting will decrease. This has been
observed at Columbia Glacier
10
. Our summer submarine melt rates
are therefore unlikely to be representative of year-round melt rates,
and the importance of submarine melting in the mass budget may
be reduced in winter.
Surface melt rates in the lower 5 km of the three glacier
systems averaged 4–18 cm d
−1
in July–August 2008 (highest value
in July; refs 12, 14). Our inferred submarine melting rates are
two orders of magnitude larger. Submarine melting is therefore
a major ablation process across tidewater glacier fronts. This
has several important consequences for glacier dynamics. First,
submarine melting is more likely to dislodge glaciers from their
beds than surface melting. Enhanced submarine melting, either
from warmer ocean waters or enhanced forced convection by
increased subglacial discharge, will melt grounded ice directly
and cause grounding-line retreat. Our data show that a thermal
forcing of 3
◦
C melts ice at a rate of several metres per day,
that is, hundreds of metres in one summer. In comparison, an
increase in surface melting will be effective at ungrounding a
glacier only if the glacier surface slope is low so that the line of
hydrostatic equilibrium retreats rapidly with a small change in ice
thickness. Furthermore, if the glacier retreats into deeper waters,
submarine melting will increase because the submerged area and
the pressure-dependent melting point of ice will both increase,
two positive feedbacks. Moreover, submarine melting must have
an enormous influence on ice calving mechanics. Pronounced
submarine melting will undercut the submerged ice faces and
promote calving from below the water surface, a mode of calving
frequently observed at tidewater termini
9,20
. In winter, reduced
submarine melting may explain the reduced iceberg production
noted at many outlet glaciers
21
.
It was recently suggested that a warming of Greenland’s fjord
waters in the late 1990s may have triggered glacier acceleration
8
.
For Jakobshavn Isbrae, previous studies of ice-shelf melting suggest
that a 3
◦
C increase in ocean temperature would increase submarine
melting by 30–70 m yr
−1
(ref. 19), which is large enough to
explain the observed ice-shelf disintegration. For tidewater glaciers
in southeast Greenland, a 3
◦
C thermal forcing could increase
submarine melting to rates comparable to those calculated herein,
that is, hundreds of meters in one summer, with considerable spatial
variability. Such retreat rates could destabilize the glaciers rapidly.
In Greenland, tidewater glaciers control 90% of the ice discharge
into the ocean. Submarine melting may therefore have a large
indirect impact on the entire ice-sheet mass budget. If we are to
determine the future of the Greenland ice sheet more reliably, it is
essential to document these ice–ocean interactions more completely
and in greater detail.
Methods
Current velocities were averaged over 5 min at each depth interval to obtain stable
solutions. Ocean current measurements were corrected for boat drift calculated
from the global positioning system data. We did not correct for tidal currents,
but we estimate them to be low in this area. Using the Arctic Ocean barotropic
model (AOTIM-5; ref. 22), we calculate tides of ±60 cm and tidal currents of
±3cms
−1
during our time period (8–11 August 2008) at 70
◦
N, 50.5
◦
W. Noise in
our velocity data is ±2cms
−1
.
Vertical profiles of temperature and salinity were collected continuously both
during descent and ascent of the instruments. Comparison of one S4 CTD with
the calibrated SBE-19 CTD on one cast (we subsequently lost the SBE-19 CTD to
iceberg impingement) showed an excellent agreement in temperature and salinity
for the data retrieved during ascent. The agreement was poorer during descent
because the S4 CTDs took longer to stabilize after entering the water. The S4 CTDs,
which were deployed in parallel, were subsequently cross-calibrated and absolutely
calibrated to match the SBE-19 measurements. Measurements disturbed by boat
motion from a calving event or human intervention were discarded. The S4 probes
experienced difficulties in the presence of extensive amounts of brash ice; in such
instances, casts were repeated until stable solutions were obtained. The precision in
temperature is 0.05
◦
C and salinity is 0.1.
Errors associated with temporal changes in ocean current, salinity and
temperature measurements are difficult to estimate because our survey was limited
to one week. We re-occupied KANGIL-1, EQIP-2 and EQIP-3 three days apart and
obtained temperature and salinity within error bars and mass and heat budgets
not significantly different from the previous calculations, hence providing indirect
support for our steady-state assumption.
The errors in mass and heat budget are quoted in Table 1. We varied the
position of the transition boundary between the overflow plume and ocean inflow
in 5-m increments by ±10 m, averaged the results and calculated the standard
deviation. The depth of the boundary was determined to be 47 ± 10 m in EQIP,
47 ± 10 m in KANGIL and 57 ± 10 m in TOR. Another source of error is in the
selection of a mean temperature and salinity for the deep ocean. We extrapolated
our data to the sea bed using a linear scheme to calculate the depth-averaged
temperature and salinity of the ocean below the transition boundary. We compared
these values with those measured midway in the lower section to quantify the
uncertainty in temperature and salinity. The uncertainty associated with salinity is
small. In TOR, only one cast was deep enough to cover the midway point, so we
assumed the same thermal forcing on all casts, with an uncertainty of ±0.1
◦
C. The
errors in boundary position and mean temperature were treated as independent
errors when calculating the total error.
Errors of omission may be significant because part of the outflow may have
exceeded the edges of our survey, yet the sea bed is also shallower along the fjord
sides. We did not include this error source in our estimates.
190 NATURE GEOSCIENCE | VOL 3 | MARCH 2010 | www.nature.com/naturegeoscience
© 2010 Macmillan Publishers Limited. All rights reserved.
