Bull Volcanol (2013) 75:709
DOI 10.1007/s00445-013-0709-2
RESEARCH ARTICLE
Geodetic data shed light on ongoing caldera subsidence
at Askja, Iceland
Elske de Zeeuw-van Dalfsen & Hazel Rymer &
Erik Sturkell & Rikke Pedersen & Andy Hooper &
Freysteinn Sigmundsson & Benedikt Ófeigsson
Received: 30 May 2012 / Accepted: 4 March 2013
# Springer-Verlag Berlin Heidelberg 2013
Abstract Subsidence within the main caldera of Askja volcano in the North of Iceland has been in progress since
1983. Here, we present new ground- and satellite-based
deformation data, which we interpret together with new
and existing micro-gravity data, to help understand which
processes may be responsible for the unrest. From 2003 to
2007, we observe a net micro-gravity decrease combined
with subsidence and from 2007 to 2009 we observe a net
micro-gravity increase while the subsidence continues. We
infer subsidence is caused by a combination of a cooling and
contracting magma chamber at a divergent plate boundary.
Mass movements at active volcanoes can be caused by
several processes, including water table/lake level movements, hydrothermal activity and magma movements. We
suggest that, here, magma movement and/or a steam cap in
the geothermal system of Askja at depth are responsible for
the observed micro-gravity variations. In this respect, we
rule out the possibility of a shallow intrusion as an explanation for the observed micro-gravity increase but suggest
magma may have flowed into the residing shallow magma
chamber at Askja despite continued subsidence. In particular, variable compressibility of magma residing in the magma chamber as well as compressibility of the surrounding
rock may be the reason why this additional magma did not
create any detectable surface deformation.
Keywords Volcano deformation . Caldera unrest .
Micro-gravity . InSAR . Precise levelling . Iceland
Introduction
Editorial responsibility: T. Ohminato
E. de Zeeuw-van Dalfsen
Institut de Physique du Globe de Paris,
Paris, France
H. Rymer
Faculty of Science, The Open University, Milton Keynes, UK
E. Sturkell
Institute for Geosciences, Göteborg, Sweden
E. Sturkell : R. Pedersen : F. Sigmundsson : B. Ófeigsson
Nordic Volcanological Centre, Institute of Earth Sciences,
University of Iceland, Reykjavik, Iceland
A. Hooper
Delft University of Technology, Delft, the Netherlands
Present Address:
E. de Zeeuw-van Dalfsen (*)
Department 2 - Physics of the Earth,
GFZ German Research Centre for Geosciences,
Potsdam, Germany
e-mail: elske@gfz-potsdam.de
Long-term monitoring of caldera deformation is essential
for hazard evaluation and mitigation (Gottsmann et al.
2003). Deformation measurements are most often used to
quantify the caldera unrest. However, those measurements
alone can not differentiate between the different processes
which may be responsible for the unrest, such as magma
movements, hydrothermal activity, cooling and contraction
of a shallow source or lake level variations. It is only by the
combination of deformation and micro-gravity monitoring
that insights into which process may be responsible for the
caldera unrest, as mass movements within the system, can
be quantified.
The integration of micro-gravity and deformation data
has been successfully applied at several calderas worldwide.
At Campi Flegrei in Italy, Battaglia et al. (2006) found the
migration of fluids into and out of the hydrothermal system
within the caldera to be responsible for the caldera unrest
between 1980 and 1995. During a period of unrest (1982–
1999) at Long Valley caldera, similar measurements have
709, Page 2 of 13
been interpreted in terms of an intrusion of silicic magma with
a significant amount of volatiles beneath the caldera's resurgent dome (Battaglia and Hill 2009). Here, we present new
deformation and micro-gravity data collected between 2007
and 2010 at Askja, Iceland, and we provide insights into the
processes responsible for long-term caldera unrest there.
The Askja volcanic system in northern Iceland (Fig. 1a, b)
consists of a central volcano and at least three calderas. The
oldest caldera is difficult to observe as it is almost completely
filled by lavas. The main (early Holocene) Askja caldera itself
has a diameter of ∼10 km and is 200–300-m deep. The
youngest caldera which is 4.5 km across was formed in
association with a Plinian eruption in 1875 (Sturkell et al.
2006 and references therein) and continued to form afterwards
for decades (Hartley and Thordarson 2011). It is now filled by
lake Öskjuvatn, one of the deepest lakes in Iceland (Rist
1975). The last eruptive activity in the area occurred in 1961
when lava erupted from a fissure at the northern boundary of
the Askja caldera. Most of the lava flowed towards the
east out of the caldera onto the plains beyond (Sturkell and
Sigmundsson 2000).
Previous work
The first deformation measurements at Askja took place in
1966 when a precise levelling line, with 12 benchmarks
(extended to 30 in 1968), was installed in the northern part of
the caldera (see Fig. 1c) (Tryggvason 1989a). Measurements
show (Fig. 2) that the caldera subsided towards the centre,
compared with points outside the caldera, from 1968 to 1970.
This trend reversed to an observed inflation from 1970 to 1972
showing the largest deformation rate measured so far. From
1973 to 1983, no measurements were conducted, but after
1983 the measurements show a steady subsidence of the
caldera centre relative to the surroundings, and the rate of
subsidence is declining with time. In 1993, the first Global
Positioning System (GPS) measurements were conducted at 24
benchmarks in and around the Askja caldera. The network was
completely re-measured in 1998 after which selected benchmarks (see Fig. 1c) have been measured on an annual basis
(Sturkell et al. 2006) during measurement campaigns lasting
up to 24 h. The GPS data confirm the ongoing subsidence.
Interferometric Synthetic Aperture Radar (InSAR) images
formed using ESR-1 (Pagli et al. 2006) and RADARSAT-2
(de Zeeuw-van Dalfsen et al. 2012) show the ongoing subsidence within the caldera very clearly. Apart from the subsidence within the caldera, InSAR also shows broader scale
subsidence 20–25-km wide, along ∼50 km of the rift (Pagli
et al. 2006; Pedersen et al. 2009).
Micro-gravity data have been collected regularly since
1988 at Askja (Rymer and Tryggvason 1993; de Zeeuw-van
Dalfsen et al. 2005). The raw data for each location are
corrected for instrumental drift, tares, earth tides and station
Bull Volcanol (2013) 75:709
height changes (e.g., Rymer 1989) to calculate the net microgravity change with respect to the base station and referred to
values observed in 1988. De Zeeuw-van Dalfsen et al. (2005)
used a point source to model a net micro-gravity decrease of
115 μGal (1 μGal=10−8 m/s2) between 1988 and 2003 in the
centre of the caldera, in terms of a subsurface mass decrease of
1.6×1011 kg. Rymer et al. (2010) observed from 1988 to 2007
a net micro-gravity decrease of 140 μGal in the caldera centre
of Askja. They also reported a dramatic reversal in the net
micro-gravity change in 2007–2008 to an increase of
∼60 μGal (Rymer et al. 2010).
To be able to integrate deformation and micro-gravity
data sets, two factors are very important: (1) the times of
data acquisition should be as close as possible (since gravity
and elevation changes are clearly on-going) and (2) the
reference point used for comparison. At Askja, data are most
often collected from late June to early September, which is
the only time the caldera is not covered in snow. The first
condition is, therefore, normally met without problem.
However, with respect to the base station, the analysis
becomes more complicated. For the precise levelling data,
the cumulative height change between two chosen levelling
benchmarks (406 and 429) is reported with 1968 as the start
year (Sturkell et al. 2006). GPS data are referred to base
station DYNG located outside the caldera and referred to
1993. Micro-gravity data are usually reported with respect
to a station on the 1961 lava flow close to the caldera rim
(83001, at the same location as the new GPS station VIKR;
see Fig. 1c) (de Zeeuw-van Dalfsen et al. 2005). To overcome these differences, the micro-gravity benchmark at the
caldera rim (83001) was measured as part of the ‘extended’
levelling line in 2003, 2007, 2008 and 2009, and GPS has
been measured yearly at VIKR since 2009.
Previous modelling and interpretation
All previous inverse modelling of geodetic data at Askja
assumes one or more contracting point sources embedded in
a homogeneous elastic half space, thereby implying that the
entire vertical deformation signal should be attributed to processes within the magmatic plumbing system. Outcomes of
such modelling of geodetic data at Askja suggests a shallow
magma chamber directly below the centre of the caldera
(65.05°N 16.78°W) at 3–3.5 km depth (Pagli et al. 2006;
Rymer and Tryggvason 1993; Sturkell and Sigmundsson
2000). RADARSAT-2 InSAR analyses suggest that the concentric fringes centred in the Askja caldera can be modelled by
a point source at a depth of 3.2–3.8 km with a volume
decrease of 0.0012–0.0017 km3/year from 2000 to 2009.
The horizontal contraction detected by the GPS data
10–15 km outside the caldera was interpreted in terms
of a second deep source at ∼16 km depth (Sturkell et
al. 2006).
Bull Volcanol (2013) 75:709
a
North−American Plate
km
0
50
100
66˚N
K
A
65˚N
T
B
19.5 mm/yr
G
Vatnajökull
64˚N
20˚W
22˚W
14˚W
16˚W
24˚W
Eurasian Plate
18˚W
65˚10'
b
Askja
65˚00'
Mt. Upptyppingar
Öskjuvatn
64˚50'
Fig. 1 a Volcanic systems in
Iceland depicting the fissure
swarms in grey, with their
associated central volcanoes
and calderas (after Einarsson
and Saemundsson 1987). The
black box shows the coverage
of RADARSAT images in the
F5 frame, with the dotted line
displaying the area shown in
Fig. 4. The following volcanoes
are annotated: Askja (A),
Krafla (K), Bárdarbunga
(B), Grimsvötn (G) and
Tungnafjellsjökull (T).
Modified after de Zeeuw-van
Dalfsen et al. (2012). b Digital
elevation model of the study
area depicting the fissure swarm
in yellow. Lake Öskjuvatn is a
nested caldera, located within
the main Askja caldera (both
indicated by depression
contours). Mt. Upptyppingar is
located 20–25 km to the east of
Askja. Two purple stars show
the locations of the geothermal
areas, Viti in the NE and
Myvetningahraun in the SW.
The geodetic network is
depicted but is better visible and
described in Fig. 1c. Modified
after de Zeeuw-van Dalfsen et
al. (2012). c Close-up of survey
area showing the micro-gravity
benchmarks: the base station
with an open black square, the
'Northern' stations with open
red circles, the 'South-eastern'
stations with yellow squares
and the 'Centre' stations with
open blue squares. The precise
levelling line is shown with
small black squares and levelling
stations 406 and 429 are pointed
out. GPS stations are marked
with open black circles and the
geothermal areas, Viti and
Myvetningahraun, are depicted
Page 3 of 13, 709
Vatnajökull
−17˚00'
−16˚30'
709, Page 4 of 13
Bull Volcanol (2013) 75:709
Fig. 1 (continued)
c
De Zeeuw-van Dalfsen et al. (2012) show that a homogeneous elastic half-space may not be the best assumption
for modelling the deformation at Askja caldera. Their twodimensional Finite Element Models (FEM) include structural complexities in the crustal layers, such as a visco-elastic
‘ridge’ at the location of the Askja fissure system and a
mechanically weak caldera filling. The ridge represents an
area of visco-elastic material reaching to shallow (1–2.5 km)
depth (immediately beneath the plate boundary) with the
same visco-elastic properties as the layer representing the
lower crust. This configuration is supported by the results
from the deformation based study of the structure of the
plate spreading segment (Pedersen et al. 2009). The weak
layer is envisioned as the result of a combination of geothermal alteration and the presence of explosive eruption
products. The results indicate that the tectonic setting of
Askja plays a major role in the continuous, long-term high
subsidence rates observed.
Models using three-dimensional FEM have been developed (Dickinson 2010) that combine magma extraction
from a shallow, fluid-filled cavity with a plate spreading
model that depicts rheologic partitioning to simulate a rift
segment. They predict both the radially symmetric pattern of
subsidence observed within the Askja caldera and the elongated pattern of subsidence tracking the rift segment rather
well, suggesting the contraction of a deeper source (∼16 km)
is not needed to explain the InSAR data.
The net micro-gravity decrease from 1988 to 2003 was
explained by de Zeeuw-van Dalfsen et al. (2005) as a
combination of cooling and contraction of magma in a
shallow chamber (30 %) and magma drainage from this
shallow reservoir (70 %). Rymer et al. (2010) suggested that
the net micro-gravity increase in 2008–2009 might be
caused by accumulation of magma beneath the caldera.
New deformation data
We present precise levelling data up to 2011, adding 7 more
years to the previously published data. The yearly difference
(Fig. 2a), between levelling benchmarks 404 and 429,
shows only slight variations through time. The 2009 measurement is slightly anomalous and it is unclear if its deviation is significant, but the overall cumulative vertical
displacement (Fig. 2b) reflects a smooth subsidence trend
since 1983. New GPS data (Figs. 1c and 3) covering the
2003–2010 period, for stations 404 and BATS, and covering
the 2003–2009 period, for stations OLAF and MASK, confirm this trend. The GPS data were processed with the
Bernese software, version 5.0, using double-differencebased analysis with quasi-ionosphere-free resolution strategy (Dach et al. 2007). The final network solution is a
minimum constraint solution, realised by three no-nettranslation conditions imposed on a set of reference coordinates of GPS stations derived by the International GNSS
Service (IGS). The set of IGS stations used includes stations
in North America, Iceland and Europe. The reference coordinates that are used are termed IGS05 and are a realisation of
the ITRF2005 reference frame (Altamimi et al. 2007). The
scatter in the data, as for example for the vertical movement of
station BATS, is due to atmospheric disturbances.
Bull Volcanol (2013) 75:709
Page 5 of 13, 709
The 2009 and 2010 InSAR data show less LOS subsidence
than observed with ground-based methods. This deviation is
comparable to that observed in the 2009 measurement of the
precise levelling data.
New micro-gravity data
Fig. 2 Precise levelling data from 1968 to 2010. Upper figure shows
the absolute yearly height difference in metres between two of the
precise levelling benchmarks (429 and 406, respectively; see Fig. 1c).
The lower figure shows the cumulative vertical displacement in m
between those benchmarks. Data were fitted with an exponential
function (Sturkell et al. 2006). Note the systematic decay of the
subsidence rate which has not changed considerably in the last decades
We created interferometric images using StaMPS
(Hooper 2008) following the method as described by de
Zeeuw-van Dalfsen et al. (2012). Three RADARSAT satellite images of three different frames were obtained during
the summer of 2010. Combined with the previously acquired images (de Zeeuw-van Dalfsen et al. 2012), we
formed 12 new interferograms with good coherence, all
covering the Askja caldera. Here, we present the descending
F5 frame as this is the most complete time series (Fig. 4).
The 2000–2010 unwrapped interferogram shows ∼22.3 cm
line-of-sight (LOS) deformation or about ∼2.2 cm LOS
deformation per year. There are two specific signatures
visible in the unwrapped interferograms: (1) a concentric
feature depicting subsidence in the main Askja caldera and
(2) an oval feature elongated along the rift indicating subsidence. Profiles tracing the green EW line in the unwrapped
interferograms (Fig. 4) show the decreasing LOS subsidence
with time (Fig. 5). Up to 2008, this decrease seems to be
similar to the decrease suggested from precise levelling data.
The micro-gravity benchmarks at Askja are grouped by
location in the ‘Centre’, ‘South-eastern’ and ‘Northern’ regions of the caldera. Each group consists of two or more
benchmarks which are then measured repeatedly during
each survey season. During the summer of 2007, a microgravity survey was carried out, but due to the poor weather
conditions and problems with the instrumentation, the data
for the Northern and South-eastern stations are not reliable.
To evaluate the micro-gravity data in terms of mass
changes, we correct the data for height changes (Fig. 6).
Due to the lack of real-time geodetic measurements, we are
forced to use a model (Sturkell et al. 2006) to calculate the
height change at each station relative to base station 83001.
It is evident that the micro-gravity data are influenced by
noise. To test for the statistical significance of the net microgravity variations, we apply a Student's t test using 95 %
confidence levels (Table 1) following the suggestions of Saibi
et al. (2010) at Unzen volcano. The error on net micro-gravity
data for location groups is less than ±20 μGal (Rymer et al.
2010). In Fig. 6, we show the best-fit location of the 2007
‘Centre’ point, which fits the trend between 1988 and 2005,
but has a larger error bar than the data points of other years.
The average net micro-gravity is displayed for each station
group, but only when the data are significant.
The net micro-gravity variations at the ‘Northern’ and
‘South-eastern’ stations lie within the uncertainty for the
data of ±20 μGal, and thus do not show any significant
change over the 1988–2010 period. The net micro-gravity at
the ‘Centre’ was measured at three different benchmarks
(OLAF, D-19 and MASK), all located close (856 m,
926 m and 742 m) to the centre of deformation. The
‘Centre’ stations show the biggest net micro-gravity variations: from 1988 to 2007, a decrease of ∼140 μGal,
followed by an increase of ∼40 μGal between 2007 and
2008. No changes occurred from 2008 to 2009, and finally a
decrease of ∼50 μGal occurred from 2009 to 2010. Overall,
from 1988 to 2007 the net micro-gravity change at Askja
shows a decrease, which reversed to an increase from 2007
to 2009. Our recent measurements indicate the reversal was
temporary and the net micro-gravity changes since then
(2009–2010) follow the previously reported decreasing
trend.
In all these analyses, it should be noted that there is the
potential for data aliasing. Data are obtained during the
summer months for logistical reasons and so variations
occurring during the winter months are masked.
709, Page 6 of 13
Bull Volcanol (2013) 75:709
40
OLAF
MASK
A404
BATS
East [mm]
20
0
−20
−40
−60
Jan May Sep
Jan
May Sep
2004
Jan May Sep
Jan May Sep
2006
Jan May Sep
Jan May Sep
2008
Jan May Sep
Jan May Sep
2010
Jan May Sep
Jan
May Sep
2004
Jan May Sep
Jan May Sep
2006
Jan May Sep
Jan May Sep
2008
Jan May Sep
Jan May Sep
2010
Jan May Sep
Jan
May Sep
2004
Jan May Sep
Jan May Sep
2006
Jan May Sep
Jan May Sep
2008
Jan May Sep
Jan May Sep
2010
180
160
140
North [mm]
120
100
80
60
40
20
0
40
20
0
Up [mm]
−20
−40
−60
−80
−100
−120
−140
Fig. 3 Time series of horizontal and vertical displacement (in
millimetres) at Ólafsgígar (OLAF), Miðaskja (MASK), A404 and
Bátshraun (BATS) GPS stations, relative to the ITRF2005 reference
frame. Error bars show the uncertainties of each measurement. The
vertical measurements (bottom panel) show increased subsidence rates
toward the centre of the caldera. OLAF and MASK are located close to
the centre while A404 and BATS are at the caldera rim (see Fig. 1c)
Interpretation
observe a net micro-gravity decrease combined with subsidence and from 2007 to 2009, we observe a net microgravity increase while the subsidence continues albeit at a
decreased rate. The simplest model for these observations
Focusing on the 2003–2010 period, we can divide the data
into two groups. For 2003–2007 and 2009–2010, we
Bull Volcanol (2013) 75:709
Page 7 of 13, 709
Fig. 4 Unwrapped small
baseline interferograms of the
F5 frame, covering the 2000
to 2010 period. See text for
description of visible features.
Thin lines show the outlines
of the Askja fissure swarm
and central volcano, as well as
the Askja caldera and lake
Öskjuvatn. Interferograms
have been corrected for orbital
fringes by removal of a ramp
and unwrapping. Incoherent
areas, such as lake Öskjuvatn,
do not have any pixels and are
hence white. Time indication
refers to the master and slave
date, respectively. The green
line projects the profile of Fig. 5
would be that the same source and the same process are
responsible for all the observations. This might be appropriate if we had observed inflation in the 2007–2009 period.
However, all deformation data—precise levelling, GPS and
InSAR—confirm the ongoing, albeit decreasing, subsidence
at the caldera centre to the present time.
Mass movements at active volcanoes can be caused by
several processes such as water table/lake level movements,
hydrothermal activity and magma movements. It is important to evaluate the possible effects of each of these
Fig. 5 EW profile though the unwrapped interferograms displayed in
Fig. 4 showing the decreasing rate of subsidence with time
processes and to monitor parameters that may assist in their
quantification. From this perspective, it is unfortunate that
no continuous record exists of the lake level height of lake
Öskjuvatn. The controlling factors for the Öskjuvatn water
level are the amount of snow, its melting rate and permeability of the surrounding rock. The Mt Dyngjufjöll massif,
Fig. 6 Net micro-gravity data, corrected for height changes within the
caldera, using benchmark 83001 (same location as GPS station VIKR)
as a base and 1988 as a reference year, for each area within the Askja
caldera. As uncertainty for the 2007 point is high, dashed lines are used
for the 'Centre' stations around this point. See text for discussion
709, Page 8 of 13
Bull Volcanol (2013) 75:709
Table 1 Results of the Student’s t test applied to the net micro-gravity data using a set standard deviation of 25 μGal for each measurement and a
confidence level of 95 %
BM
Centre
Southeastern
Northern
005
D-19
D-18
IV16
Von Knebel
405
412
'88
to 89
'89
to '90
'90
to '91
'91
to '92
'92
to '94
'94
to '95
'95
to '97
'97
to '02
'02 to '03
'03
to '07
'07
to '08
'08
to '09
'09
to '10
1
1
1
1
0
1
1
1
1
1
1
0
0
0
1
0
1
0
0
0
1
1
0
1
1
1
0
1
1
1
1
–
0
1
0
1
1
1
–
1
1
0
1
1
1
1
0
0
1
1
0
1
1
1
1
1
1
1
0
1
1
1
1
1
1
1
–
1
–
–
1
1
1
–
1
–
–
1
1
0
–
1
1
1
1
1
0
–
1
0
1
A value of '1' indicates significant change, '0' indicates insignificant change and '–' indicates no data, at the station
hosting the Öskjuvatn lake, is build up by huge amounts of
hyloclastite which is permeable. It has been observed that
the lake level is highest towards the end of summer when
most of the snow has melted (Tryggvason 1989b). Seasonal
variation of the lake level is on the order of 1–2 m, while
during summer, the variation can be ∼0.5 m depending on
the presence of snow. The lake is drained through ground
water motion. From 1950 to 1968, sporadic relative measurements were noted by Sigurjón Rist (Tryggvason 1989b).
In 1968, 20 benchmarks were installed around the lake to
monitor its level. Several of these became submerged or lost
over the years (Tryggvason 1989a, b). Overall, up to 1961,
the lake level was more or less stable. After the 1961
eruption, the lake level decreased a total of 6 m due to the
reactivation of old cracks and possibly the opening of new
cracks, increasing the drainage. In 1968, this trend was
reversed to a gradual increase in lake level (Tryggvason
1989b) as drainage passages presumably became clogged.
It is believed that this slow increase continues up to the
present. The lake level data do not correlate with the observed deformation nor does it follow the opposite trend,
suggesting no direct influence. Tryggvason (1989b) also
observed ‘tilting’ in 1968 of the ‘highest’ lake shore data,
with a difference of 5 m between the North and South shore.
He interpreted this ‘tilting’ as caused by inflation to the
north of the caldera related to the 1961 eruption and that
this inflation must have been rather large compared to the
extruded material.
To estimate the expected gravity change (Δgwt), resulting
from water table and lake level movements (δz), we approximate an unconfined aquifer with density (ρw) and effective
porosity (φe) as an infinite slab such that:
Δgwt ¼ 2p G φe ρw dz
ð1Þ
This is an oversimplification as it assumes the lake level
reflects the water table level within the caldera. It therefore
under-estimates the necessary water table rise/fall required
to produce the observed gravity changes. To explain the
totality of the net gravity increase of 50 μGal, assuming
an effective porosity of 17 % (Franzson et al. 2001), a water
table rise of ∼7 m needs to take place. Although only limited
observations of the lake level have been carried out, such a
large change is highly implausible. Furthermore, if a rise in
water table/lake level was responsible for the net microgravity increase, the same increase should have been observed at the South-eastern stations, which are just as close
to the level of the lake, and this is not the case.
Geothermal activity at Askja is focused along the boundaries of the newest caldera collapse and along the shore of
lake Öskjuvatn. It includes the explosion crater Víti, which
is filled with lukewarm water. One geothermal vent is
known on the lake bottom (Fig. 1c), offshore, at the West
side of the lake, next to Myvetningahraun (Ólafsson 1980).
In winter, some years, thermal activity is sufficient to maintain
an opening in the ice cover of the lake (Sigvaldason 1964 and
later observations). This has also been observed in recent
(April 2012, June 2011, June 2010 and May 2009 for example) ASTER satellite images (http://igg01.gsj.jp/vsidb/image/
proto_header.html). The distance from Myvetningahraun to
the ‘Centre’ stations is ∼1.5 km, short enough to influence
micro-gravity observations there. In 2012, the whole
Öskjuvatn lake became ice free unusually early in the year,
after fast progressive enlarging in the lake opening above the
geothermal vent. This may have been a response to increased
heat flow from below, although other possibilities cannot be
ruled out, such as convection of the lake bringing heat from
the water body, contributing to the ice melt.
Unrest at other calderas has been (partly) attributed to
hydrothermal activity [e.g., Campi Flegrei (Gottsmann et al.
2006) and Yellowstone (Battaglia et al. 2003)]. A steam cap
in the geothermal system of Askja at depth, of variable size
and with a link to the surface of variable efficiency may
induce micro-gravity variations. Such a steam cap could
develop above the centre of the magmatic source—literally
under the ‘Centre’ group of micro-gravity stations, causing
Bull Volcanol (2013) 75:709
variations in micro-gravity measurements there. Reduction
of the steam cap (replacing steam with water) in 2007–2009
could possibly explain the micro-gravity increase. Increase
in the steam cap would then cause a reduced density after
2009 resulting in a micro-gravity decrease. If part of the
geothermal system is close to critical conditions then steam
can change to liquid water (or vice versa) without much
pressure changes. This process would cause little to no
detectable deformation at the surface but large microgravity changes and is considered more plausible than the
vertical migration of the water table as the cause of the
observed gravity changes. At the resolution of this survey,
the gravitational effect observed at the surface of an infinite
slab and a vertical cylinder with a radius of >2.5 km is the
same. Therefore, although we cannot pinpoint the lateral
extent of the steam cap exactly, its radius certainly cannot
be larger than 5 km, as it would then also have influenced
the Northern and South-eastern stations. A steam cap with a
much smaller radius would require an unrealistically large
vertical movement and is therefore not considered. In our
case, the steam cap can, as a first approximation, be represented by the same infinite slab model. Thus, of the order of
7 m of increased/decreased depth of the steam layer would
be required to account for the magnitude of the gravity
changes.
Another possible explanation for the net micro-gravity
increase is magma movement. We can envisage such magma movement as a shallow magma intrusion into the crust
or as an intrusion of magma into the residing magma chamber. A shallow intrusion would be expected to produce
additional signatures at the surface, such as increased heat
flow and localised deformation. However, a deeper intrusion, focused within the existing magma chamber, is a
plausible explanation of the data presented. The geodetic
data are consistent with contributions from both processes: a
magma intrusion into the magma chamber and fluctuations
in a steam cap underneath the centre of the caldera. It is not
possible to determine the relative contributions from either
process using the current data. In the following section, we
will discuss in more detail how magma intrusion can explain
the micro-gravity increase without large surface deformation.
Discussion and implications
Askja caldera has been continuously subsiding at least since
1983 but quite likely since 1973. The rate of subsidence has
decayed exponentially from 1983 to 2003 (Sturkell et al.
2006), and this continues up to 2010 (see Fig. 2). At Krafla
caldera, 70 km North of Askja (Fig. 1a), subsidence decayed
much more rapidly after its last rifting episode (1975–1984)
and the decay was attributed to relaxation of the crust
(Sturkell et al. 2008). The most recent activity at Askja
Page 9 of 13, 709
consisted of relatively gentle magma extrusion compared
with Krafla's rifting episode. Furthermore, simple 2D FEM
models show that rheological complexities, arising from the
tectonic setting of the Askja caldera at a divergent plate
boundary, may have large effects on the surface deformation
(de Zeeuw-van Dalfsen et al. 2012; Pedersen et al. 2009).
These rheological complexities can be envisaged as a viscoelastic ‘ridge’ at the location of the Askja fissure system and
a ‘weak’ layer filling the caldera. The ‘ridge’ represents an
area of visco-elastic material reaching to shallow depth
(immediately beneath the plate boundary) with the same
visco-elastic properties as the layer representing the lower
crust. As Askja is located closer to the inferred centre of upwelling in Iceland than Krafla, a ‘ridge’ underneath Askja
may be able to intrude to relatively shallow levels hence
enhancing the effect of such a rheological complexity.
Inversion of our data with 2D models, as suggested by de
Zeeuw-van Dalfsen et al. (2012), are beyond the scope of
this work because the resolution of the gravity data do not
support this level of detail. However, based on the results
from that paper, we suggest the subsidence at Askja is
caused by a combination of the plate spreading and
cooling/contraction of the shallow magma chamber. The
plate spreading at Askja, during the current inter-rifting
phase, is a steady process (e.g., Camitz et al. 1995); hence,
the decay may be attributed to the cooling/contraction of the
shallow magma chamber.
During the 2000–2010 period, several episodes of
unrest occurred within a 100-km radius of Askja. The
Grimsvötn volcano, located 70 km to the SW of Askja
beneath the Vatnajökull icecap, erupted in November 2004
(Sigmundsson and Gudmundsson 2005). An ∼0.05 km3 magma dike intruded in 2007–2008 at ∼15 km depth in the
Upptyppingar area, just 20–25 km to the East of Askja
(Fig. 1b) (Hooper et al. 2011; Jakobsdóttir et al. 2008).
In 2006, for the first time, very deep (>20 km) microearthquakes were detected in the Askja area using a
temporary dense network of seismometers. These earthquakes are located at the NE edge of the Askja caldera,
in a NE–SW-trending belt ∼10-km long, as well as in two
zones >40 km away (Soosalu et al. 2010). Each of these three
zones is interpreted as foci of melt supply from where magma
feeds the volcanic rift zone (Key et al. 2011).
It is not implausible that a small pocket of magma intruded into shallower levels rather than flowing into the diverging rift zone at depth. Such an intrusion, outside the
previously existing magma chamber, would likely cause an
earthquake swarm and ground deformation. To explain the
net micro-gravity increase of ∼50 μGal, just ∼0.01 km3 of
magma would need to pond 2 km below the surface, assuming a point source, or ∼0.02 km3, assuming a penny-shaped
crack. This is only 17–34 % of the volume extruded during
the most recent eruption in 1961. However, forward
709, Page 10 of 13
calculations of the surface deformation potentially caused
by such an intrusion predict inflation on the order of 10–
50 cm which we have not observed in any of the deformation data. Furthermore, the time lag between the earthquake
activity at depth and the possible magma intrusion into
shallower levels seems too large. Hence, an intrusion of this
kind is an unlikely explanation for our observations.
There are slight deviations from the average subsidence
rate observed in the precise levelling (2009 data) and InSAR
images and profiles (2009 and 2010 data). They may show
the beginning of what could have been a reversal of deformation data had the magma intrusion persisted. We postulate that the mass increase was too small and the overprint of
the larger subsidence signal related to the divergent plate
boundary was too large to result in detectable net inflation at
the surface. To test this hypothesis and to observe if there is
any hidden ‘signal’ in the InSAR data, we calculate the
residual surface deformation (Fig. 7). The average LOS−
velocity field estimated from all interferograms before 2009,
multiplied by the time span, gives the ‘secular’ deformation
for the 2000–2009 period. We subtract this deformation
from the original 2000–2009 interferogram to get the residual. For the 2000–2009 period, the residual shows a variation of ∼2 cm, within the atmospheric noise expected.
Similarly, the residual for 2008 and for 2010 were calculated
Fig. 7 Residual surface
deformation for 2009 (c, d)
calculated by subtracting the
secular deformation (b), found
by multiplying the velocity for
the correct number of years,
from the original interferogram
(a). Small black dots
approximate the outline of lake
Öskjuvatn. Open squares
represent the location of the
micro-gravity stations: 'Centre'
in blue, 'South-eastern' in
yellow and 'Northern' in red.
Open black square shows the
location of the micro-gravity
reference station on the 1961
lava flow
Bull Volcanol (2013) 75:709
(Fig. 8). For 2010, the residual shows a variation of ∼1.5–
2 cm, within the expected atmospheric noise; however, the
residual of 2008 shows a variation of 5 cm, suggesting there
was extra uplift at this time. We invert this uplift signal
using a spherical chamber in a homogeneous elastic halfspace, represented by a point pressure source (Mogi 1958),
and a coin-shaped crack in a homogeneous, elastic halfspace (Fialko et al. 2001). We find reasonable fits for both,
but neither of the best-fit models can explain more than
20 % of the observed micro-gravity increase. This implies
that the observed micro-gravity change was not associated
with detectable surface deformation and that an intrusion into shallower levels cannot explain the observed
micro-gravity data.
The tectonic setting of Askja plays an important role in
the continuous, long-term high subsidence rates observed
there (de Zeeuw-van Dalfsen et al. 2012). If indeed part of
the observed subsidence at the Askja caldera is caused by
the ongoing divergence of the plate boundary, then the
amount of subsidence related to the shallow source may be
smaller than previously suggested. Continuing this reasoning, we postulate that the use of a more realistic rheological
model of the area would suggest a shallower source depth.
This has direct consequences for the calculated mass
changes needed to explain the observed net gravity changes.
Bull Volcanol (2013) 75:709
Page 11 of 13, 709
magma already in the chamber and elastic compressibility
of the surrounding rock, hence not inducing any detectable
surface deformation. This would represent the other side of
the spectrum of the model proposed to explain the bigger
volume of dike intrusions compared to deflating source
volumes at Kilauea (Rivalta and Segall 2008). The mass
input, ∆M, associated with an increase of magma chamber
volume, ∆V, is not simply ∆M = ρ0∆V (where ρ0 is the
density of magma before the intrusion). In fact, mass input
into an existing magma chamber compresses both the surrounding rock (creating more space for the chamber) and the
resident magma (causing a density change, ∆ρ, of magma
residing in the chamber). Differentiation and substitution of
equations (see Johnson et al. 2000; Segall et al. 2001;
Rivalta 2010; Anderson and Segall 2011) suggests that for
a spherical magma chamber:
ΔM ¼ rV ρ0 ΔV
ð2Þ
with rv =1 + βm/βc
and βm is magma compressibility and βc is the elastic
compressibility of a magma chamber.
This equation states that a mass of magma bigger than
ρ0∆V can be squeezed into a magma chamber and still result
in an apparent volume change ∆V, while its ‘true’ volume
(before injection or once erupted) is actually rv times larger.
For gas-poor magmas and spherical or cigar-shaped chambers, rv may be in the range of 1–5; for gas-rich magmas,
this could be even larger (Mastin et al. 2008).
Fig. 8 Residual surface deformation for 2008 (a) and 2010 (b) calculated in the same way as for Fig. 7. Annotations as for Fig. 7
All observations could be explained with smaller mass
changes when the assumed source depth is shallower. For
example, to explain a net micro-gravity increase of
∼50 μGal, using a source at 3 km depth, ∼0.7×1011 kg mass
inflow is needed. When the source is at 2 km depth, only ∼0.
3×1011 kg mass inflow is needed to explain the same net
micro-gravity change.
Is it possible that magma flows into the shallow chamber
without any detectable effects in the deformation observations? If there is a well-developed conduit or pocket for
magma, then it is possible to generate measurable gravity
changes without any deformation signals. Such a change
was for example observed at Izu-Oshima, Japan, after the
1986 eruption (Watanabe et al. 1998). However, the processes at this open conduit volcano, displaying effusive
activity, are very different from unrest at a caldera. Maybe
magma flowing into the residing shallow magma chamber at
Askja is accommodated by the compressibility of the
Conclusions
Deformation data show that Askja caldera continues to
subside up to 2010. The subsidence is smooth but slowly
decaying and is strongest in the centre of the main Askja
caldera. This subsidence is caused by a combination of the
cooling and contracting magma chamber at a divergent plate
boundary. The net micro-gravity decrease observed with this
trend indicates a mass decrease, e.g., mass moving away
from the centre of subsidence for example into the diverging
rift. Statistical tests show the net micro-gravity increase
observed in 2007–2009 is significant. This mass increase
was not accompanied by detectable net inflation at the
surface nor can it be explained by an intrusion at shallow
depth. We suggest two plausible explanations: (1) the observed mass increase was caused by magma flow into the
existing magma chamber at Askja. In particular, compressibility of magma residing in the chamber as well as compressibility of the surrounding rock may explain why this
intrusion remained undetectable through surface deformation. (2) Alternatively, variations within a steam cap in the
geothermal system of Askja at depth may be responsible for
709, Page 12 of 13
the observed micro-gravity variations. We suggest that 2D
modelling or detailed 3D FEM modelling, taking into account the rheological complexities of the area, may help to
estimate the relative contribution that each process makes to
the ongoing unrest at Askja caldera.
Acknowledgements This project was financed by a Marie Curie
intra-European fellowship (from EDZ). EDZ thanks Claude Jaupart,
Eleonora Rivalta, Petar Marinkovic, Judicael Decriem, Florian
Lhuillier, Rósa Ólafsdóttir and Thom Warmerdam for discussion, help
with data processing, MATLAB programming, GMT plotting, DEM
preparation and other computer issues. EDZ thanks Rósa Ólafsdóttir
for preparation of Fig. 1c. The CSA provided RADARSAT images for
this project as part of a DRU proposal. MDA assisted with the data
selection and ordering procedures. Financial support to RP was received from Rannís. The DEM and xy data files were produced by the
Icelandic Geodetic Survey. GMT public domain software was used to
prepare Fig.1a and b. We thank Glyn Williams-Jones, Takao Ohminato
and an anonymous reviewer for their constructive comments.
References
Altamimi Z, Collilieux X, Legrand J, Garayt B, Boucher C (2007)
ITRF2005: a new release of the International Terrestrial Reference
Frame based on time series of station positions and Earth Orientation
Parameters. J Geophys Res 112:B09401. doi:10.1029/
2007JB004949
Anderson K, Segall P (2011) Physics-based models of ground deformation and extrusion rate at effusively erupting volcanoes. J
Geophys Res 116:B07204. doi:10.1029/2010JB007939
Battaglia M, Hill DP (2009) Analytical modeling of gravity changes
and crustal deformation at volcanoes: the Long Valley caldera,
California, case study. Tectonophysics 471:45–57. doi:10.1016/
j.tecto.2008.09.040
Battaglia M, Segall P, Roberts C (2003) The mechanics of unrest at
Long Valley caldera, California. 2. Constraining the nature of the
source using geodetic and micro-gravity data. J Volcanol Geotherm
Res 127:219–245. doi:10.1016/S0377-0273(03)00171-9
Battaglia M, Troise C, Obrizzo F, Pingue F, De Natale G (2006)
Evidence for fluid migration as the cause of unrest at
Campi Flegrei caldera (Italy). Geophys Res Lett 33:L01307.
doi:10.1029/2005GL024904
Camitz J, Sigmundsson F, Foulger G, Jahn CH, Volksen C, Einarsson P
(1995) Plate boundary deformation and continuing deflation of
the Askja volcano, North Iceland, determined with GPS, 1987–
1993. Bull Volcanol 57(2):136–145. doi:10.1007/BF00301404
Dach R, Hugentobler U, Fridez P, Meindl M (2007) Bernese GPS
software version 5.0. Astronomical Institute, University of Bern
de Zeeuw-van Dalfsen E, Pedersen R, Hooper A, Sigmundsson F
(2012) Subsidence of Askja caldera 2000–2009: modelling of
deformation processes at an extensional plate boundary,
constrained by time series InSAR analysis. J Volcanol Geotherm
Res 213–214:72–82. doi:10.1016/j.jvolgeores.2011.11.004
de Zeeuw-van Dalfsen E, Rymer H, Sigmundsson F, Sturkell E (2005)
Net gravity decreases at Askja volcano, Iceland: constraints on
processes responsible for continuous caldera deflation, 1988–
2003. J Volcanol Geotherm Res 139:227–239. doi:10.1016/
j.jvolgeores.2004.08.008
Dickinson H (2010) Finite element models for the deformation of the
Askja volcanic complex and rift segment, Iceland. M.Sc. thesis.
The University of Alabama, Tuscaloosa, AL, USA
Bull Volcanol (2013) 75:709
Einarsson P, Saemundsson K (1987) Earthquake epicenters 1982–1985
and volcanic systems in Iceland (map). In: Sigfusson T (ed) I
Hlutarins Edli. Menningarsjodur, Reykjavik
Fialko Y, Khazan Y, Simons M (2001) Horizontal circular crack in an
elastic half-space, with applications to volcano geodesy. Geophys
J Int 146:181–190. doi:10.1046/j.1365-246X.2001.00452.x
Franzson H, Gudlaugsson S, Fridleifsson G (2001) Petrophysical properties of Icelandic rocks. Proceedings of the 6th Nordic
Symposium on Petrophysics. Trondheim, Norway
Gottsmann J, Berrino G, Rymer H, Williams-Jones G (2003) Hazard
assessment during caldera unrest at the Campi Flegrei, Italy: a
contribution from gravity/height gradients. Earth Planet Sci Lett
211:295–309. doi:10.1016/S0012-821X(03)00225-5
Gottsmann J, Folch A, Rymer H (2006) Unrest at Campi Flegrei: a
contribution to the magmatic versus hydrothermal debate from
inverse and finite element modeling. J Geophys Res 111:B07203.
doi:10.1029/2005JB003745
Hartley M, Thordarson T (2011) Formation of Öskjuvatn caldera at
Askja, north Iceland: evolution of caldera collapse and implications for the lateral flow hypothesis. Geophys Res Abstr 13:
EGU2011–10053
Hooper A (2008) A multi-temporal InSAR method incorporating both
persistent scatterer and small baseline approaches. Geophys Res
Lett 35:L16302. doi:10.1029/2008GL034654
Hooper A, Ófeigsson B, Sigmundsson F, Lund B, Einarsson P,
Geirsson H, Sturkell E (2011) Increased crustal capture of magma
at volcanoes with retreating ice caps. Nat Geosci 4:783–786.
doi:10.1038/ngeo1269
Jakobsdóttir SS, Roberts MJ, Gudmundson GB, Geirsson H, Slunga R
(2008) Earthquake swarms at Upptyppingar, Northeast Iceland: a
sign of magma intrusion? Stud Geophys Geod 52:513–528.
doi:10.1007/s11200-008-0035-x
Johnson DJ, Sigmundsson F, Delaney PT (2000) Comment on Volume
of magma accumulation or withdrawal estimated from surface
uplift or subsidence, with application to the 1960 collapse of
Kilauea volcano by P.T. Delaney and D.F. McTigue. Bull
Volcanol 61:491–493. doi:10.1007/s004450050006
Key J, White RS, Soosalu H, Jakobsdóttir SS (2011) Multiple melt
injection along a spreading segment at Askja, Iceland. Geophys
Res Lett 38:L05301. doi:10.1029/2010GL046264
Mastin L, Roeloffs E, Beeler NM, Quick JE (2008) Constraints on the
size, overpressure, and volatile content of the Mount St. Helens
magma system from geodetic and dome-growth measurements
during the 2004–2006+ eruption. In: Sherrod DR, Scott WE,
Stauffer PH (eds) A volcano rekindled: the renewed eruption of
Mount St. Helens, 2004–2006, chapter 22. U.S. Government
Printing Office, Washington DC
Mogi K (1958) Relations between the eruptions of various volcanoes
and the deformation of the ground surface around them. B Earthq
Res I 36:99–134
Ólafsson J (1980) Temperature structure and water chemistry of the
caldera lake Öskjuvatn, Iceland. Limnol Oceanogr 25(5):779–
788
Pagli C, Sigmundsson F, Arnadottir T, Einarsson P, Sturkell E (2006)
Deflation of the Askja volcanic system: constraints on the deformation source from combined inversion of satellite radar interferograms and GPS measurements. J Volcanol Geotherm Res
152:97–108. doi:10.1016/j.jvolgeores.2005.09.014
Pedersen R, Sigmundsson F, Masterlark T (2009) Rheologic controls
on inter-rifting deformation of the Northern Volcanic Zone,
Iceland. Earth Planet Sci Lett 281:14–26. doi:10.1016/
j.epsl.2009.02.003
Rist S (1975) Stöduvötn OS-ROD7519. Orkustofnun, Reykjavik
Rivalta E, Segall P (2008) Magma compressibility and the missing
source for some dike intrusions. Geophys Res Lett 35:L04306.
doi:10.1029/2007GL032521
Bull Volcanol (2013) 75:709
Rivalta E (2010) Evidence that coupling to magma chambers controls
the volume history and velocity of laterally propagating intrusions. J Geophys Res 115:B07203. doi:10.1029/2009JB006922
Rymer H (1989) A contribution to precision microgravity analysis
using LaCoste and Romberg gravity meters. Geophysics J
97:311–322. doi:10.1111/j.1365-246X.1989.tb00503.x
Rymer H, Tryggvason E (1993) Gravity and elevation changes at
Askja, Iceland. Bull Volcanol 55:362–371. doi:10.1007/
BF00301147
Rymer H, Locke C, Ófeigsson BG, Einarsson P, Sturkell E (2010) New
mass increase beneath Askja volcano, Iceland—a precursor to
renewed activity? Terra Nova 22(4):309–313. doi:10.1111/
j.1365-3121.2010.00948.x
Saibi H, Gottsmann J, Ehara S (2010) Post-eruptive gravity changes
from 1999 to 2004 at Unzen volcano (Japan): a window into shallow
aquifer and hydrothermal dynamics. J Volcanol Geotherm Res
191:137–147. doi:10.1016/j.jvolgeores.2010.01.007
Segall P, Cervelli P, Owen S, Lisowski M, Miklius A (2001)
Constraints on dike propagation from continuous GPS measurements. J Geophys Res 106:19301–19317. doi:10.1029/
2001JB000229
Sigmundsson F, Gudmundsson MT (2005) Eruption of Grimsvötn
volcano, November 1–6, 2004. Jokull 54:135–138
Sigvaldason GE (1964) Some geochemical and hydrothermal aspects
of the 1961 Askja eruption. Beiträge zur Mineralogie und
Petrographie 10:263–274
Page 13 of 13, 709
Soosalu H, Key J, White RS, Knox C, Einarsson P, Jakobsdóttir SS
(2010) Lower-crustal earthquakes caused by magma movement
beneath Askja volcano on the north Icelandic rift. Bull Volcanol
72:55–62. doi:10.1007/s00445-009-0297-3
Sturkell E, Sigmundsson F (2000) Continuous deflation of the Askja
Caldera, Iceland, during the 1983–1998 non-eruptive period. J
Geophys Res 105(B11):25671–25684. doi:10.1029/2000JB900178
Sturkell E, Sigmundsson F, Slunga R (2006) 1983–2003 decaying rate
of deflation at Askja caldera: pressure decrease in an extensive
magma plumbing system at a spreading plate boundary. Bull
Volcanol 68:727–735. doi:10.1007/s00445-005-0046-1
Sturkell E, Sigmundsson F, Geirsson H, Ólafsson H, Theodórsson T
(2008) Multiple volcano deformation sources in a post-rifting
period: 1989–2005 behaviour of Krafla, Iceland constrained by
levelling, tilt and GPS observations. J Volcanol Geotherm Res
177:405–417. doi:10.1016/j.jvolgeores.2008.06.013
Tryggvason E (1989a) Ground deformation in Askja, Iceland: its source
and possible relation to flow of the mantle plume. J Volcanol
Geotherm Res 39:61–71. doi:10.1016/0377-0273(89)90021-8
Tryggvason T (1989b) Measurement of ground deformation in Askja
1966 to 1989. Nordic Volcanological Institute 8904, University of
Iceland
Watanabe H, Okubo S, Sakashita S, Maekawa T (1998) Drain-back
process of basaltic magma in the summit conduit detected by
microgravity observation at Izu-Oshima Volcano, Japan.
Geophys Res Lett 25(15):2865–2868. doi:10.1029/98GL02216