Auxiliary Material for
Tomographic image of melt storage beneath Askja volcano, Iceland using local
microseismicity.
Michael A. Mitchell1, 2, Robert S. White1, Steve Roecker3, and Tim Greenfield1
(1 Department of Earth Sciences, Bullard Laboratories, University of Cambridge,
Cambridge, UK.)
(2 Department of Earth, Ocean, and Atmospheric Sciences, Geophysical Inversion
Facility (GIF), University of British Columbia, Vancouver, BC, Canada.)
(3 Department of Earth and Environmental Sciences, Rensselaer Polytechnic Institute,
Troy, New York, USA.)
Geophysical Research Letters, 2013
Introduction
Since it is important to have a reasonable 1-D reference velocity model for the 3-D
tomographic inversion, velocity models derived from regional seismic refraction surveys
were used to help guide our choice of an optimal 1-D reference model (See “ts01.pdf”).
For comparison a regional velocity model derived from the work of Martens et al., 2010
and Key et al., 2011 is provided in “ts02.pdf.” This model is based on data from two
regional seismic refraction surveys in the vicinity: ICEMELT [Darbyshire et al., 1998]
and RRISP [Gebrande et al., 1980 and Menke et al., 1996]. Figure 2 in the main article
provides a visual comparison of these 1-D models.
The S velocity model results shown in “fs01.jpg” provides x, y, and z sections which
correspond to the P velocity sections shown in Figure 2 of the main article. A direct
comparison of the Vp and Vs velocity models shows that there is a high degree of
correlation between the two models. This results in very small amplitude Vp/Vs variations
(see “fs02.jpg”). The structure of the Vp/Vs model, as shown in the 3 cross sections of
“fs02.jpg”, is dominated by a gradient, which increases with depth in keeping with the 1D reference model. The lack of a strongly elevated Vp/Vs anomaly beneath the caldera
suggests that much of the velocity anomaly is caused by high, but sub-solidus
temperature rock with melt distributed in many small sills rather than in a single large
magma chamber.
“ms01.mp4” is a animation which shows a full 360° rotation of the final tomographic
model. This animation helps us better visualize the spatial relationships between the
earthquake hypocenters, seismic stations, velocity model anomalies, and the location of
the caldera rim. The animation was produced in MATLAB but compressed using the free
HandBrake video transcoder.
The synthetic checkerboard model consisted of a series of uniformly distributed 4 km by
4 km prisms of alternating positive and negative velocity contrast with respect to the
starting 1-D model. Panels a), b), and c) of “fs03.jpg” illustrate this pattern. In this model
both the negative and positive velocity anomalies differ from the 1-D velocity model by
5%. Since the checkerboard model has an even distribution of anomalies throughout, it is
a useful tool for evaluating spatial variations in model resolution. By stepping through the
various depth slices, for instance, it is possible to see how the model resolution changes
with depth, while the x and y slices can be used to assess lateral variations in model
resolution. The results of this test (shown in panels d), e), and f) of “fs03.jpg”) indicate
that the tomographic inversion is capable of recovering the overall shape and distribution
of anomalies within the core of the survey region down to a depth of approximately 15
km. As was expected, based on the distribution of earthquakes and the ray coverage, the
resolution of the eastern and central portions of the model are much better than that of the
model periphery.
1. ts01.pdf (Table S1) The optimal 1-D velocity model that was developed for use as the
reference model in the tomographic inversion. See the Regional 1-D Velocity Model
section of the main article for the details of how this model was created and tested.
1.1. Column “Depth”, km, depth of the layer interface, positive downward from sea
level.
1.2. Column “Vp”, km/s, P wave velocity of the layer below the corresponding
interface.
1.3. Column “Vs”, km/s, S wave velocity of the layer below the corresponding
interface.
2. ts02.pdf (Table S2) A 1-D velocity model for the Askja region that was developed by
Martens et al., 2010 and Key et al., 2011. This model is based on data from two
regional seismic refraction surveys in the vicinity: ICEMELT [Darbyshire et al.,
1998] and RRISP [Gebrande et al., 1980 and Menke et al., 1996].
2.1. Column “Depth”, km, depth of the layer interface, positive downward from sea
level.
2.2. Column “Vp”, km/s, P wave velocity of the layer below the corresponding
interface.
2.3. Column “Vs”, km/s, S wave velocity of the layer below the corresponding
interface.
3. fs01.jpg (Figure S1) Section views of the tomographic model (VS velocities) that slice
through the imaged magma chamber anomaly. Panel a) is a north-south section along
x = -5 km; panel b) is an east-west section along y = -0.5 km; and panel c) shows a
depth slice from z = 6 km. Panel c) also shows an outline of the Askja and Öskjuvatn
calderas (solid black lines) and the location of the x and y sections (broken black
lines). Only cells with a ray hit count above 10 are colored.
4. fs02.jpg (Figure S2) A series of x, y, and z sections through the tomographic model
which show the mapped variation in Vp/Vs. As in the Figure S1 panel a) is a northsouth section along x = -5 km; panel b) is an east-west section along y = -0.5 km; and
panel c) shows a depth slice from z = 6 km. The outline of the Askja and Öskjuvatn
calderas (solid black lines) and the location of the x and y sections (broken black
lines) are superimposed on the z-section in panel c). Only cells with a ray hit count
above 10 are colored.
5. “ms01.mp4”(Figure S3) This animation shows a full 360° rotation of the final
tomographic model. Here the sections x = -5 km, y = -0.5 km, z = 6 km and the
isosurface enclosing all velocity anomalies more negative than -5.75% are plotted
along with the hypocentral locations of the earthquakes.
6. fs03.jpg (Figure S4) This figure shows 3 cross-sections from the checkerboard
recovery test, that transect the zone in which the sub-caldera magma chamber was
imaged. Panels a), b), and c) show sections through the original model, while panels
d), e), and f) show the corresponding recovered sections. The results of this test
indicate that the resolution of the model appears to be sufficient to image anomalies
as small as 2 km square with amplitudes of only ±5%. Since the shallow magma
storage region imaged in this study is larger in size and amplitude than the recovered
checkers we have confidence that this structure is well resolved and represents an
actual geologic structure which is required to fit the data.