SESolid EarthSESolid Earth1869-9529Copernicus PublicationsGöttingen, Germany10.5194/se-9-1035-2018Structure of the central Sumatran subduction zone revealed by local earthquake travel-time tomography using an amphibious networkStructure of the central sumatran subduction zoneLangeDietrichdlange@geomar.dehttps://orcid.org/0000-0003-2654-7963TilmannFrederikhttps://orcid.org/0000-0002-7439-8782HenstockTimRietbrockAndreasNatawidjajaDannyKoppHeidrunhttps://orcid.org/0000-0002-6898-1568Dynamics of the Ocean Floor, GEOMAR, Helmholtz Centre for Ocean Research Kiel, Kiel, GermanyHelmholtz-Zentrum Potsdam, Deutsches GeoForschungsZentrum GFZ, Potsdam, GermanyOcean and Earth Science, University of Southampton European Way, Southampton, SO14 3ZH, UKKarlsruhe Institute of Technology, Geophysical Institute, Karlsruhe, GermanyRC Geotechnology, Indonesian Institute of Sciences (LIPI), Bandung, IndonesiaInstitute of Geological Sciences, Freie Universität Berlin, Berlin, GermanyDepartment of Geosciences, Christian-Albrechts-Universität zu Kiel, Kiel, GermanyDietrich Lange (dlange@geomar.de)21August2018941035104923November20174January20187March201831July2018This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/4.0/This article is available from https://se.copernicus.org/articles/9/1035/2018/se-9-1035-2018.htmlThe full text article is available as a PDF file from https://se.copernicus.org/articles/9/1035/2018/se-9-1035-2018.pdf
The Sumatran subduction zone exhibits strong seismic and tsunamogenic
potential with the prominent examples of the 2004, 2005 and 2007 earthquakes.
Here, we invert travel-time data of local earthquakes for vp and
vp/vs velocity models of the central Sumatran forearc. Data were
acquired by an amphibious seismometer network consisting of 52 land stations
and 10 ocean-bottom seismometers located on a segment of the Sumatran
subduction zone that had not ruptured in a great earthquake since 1797 but
witnessed recent ruptures to the north in 2005 (Nias earthquake, Mw=8.7)
and to the south in 2007 (Bengkulu earthquake, Mw=8.5). The 2-D and 3-D vp
velocity anomalies reveal the downgoing slab and the sedimentary basins.
Although the seismicity pattern in the study area appears to be strongly
influenced by the obliquely subducting Investigator Fracture Zone to at least
200 km depth, the 3-D velocity model shows prevailing trench-parallel
structures at depths of the plate interface. The tomographic model suggests a
thinned crust below the basin east of the forearc islands (Nias, Pulau Batu,
Siberut) at ∼180 km distance to the trench. vp velocities
beneath the magmatic arc and the Sumatran fault zone (SFZ) are around 5 km s-1 at
10 km depth and the vp/vs ratios in the uppermost 10 km are low,
indicating the presence of felsic lithologies typical for continental crust.
We find moderately elevated vp/vs values of 1.85 at ∼150 km distance
to the trench in the region of the Mentawai Fault.
vp/vs ratios suggest an absence of large-scale alteration of the
mantle wedge and might explain why the seismogenic plate interface (observed
as a locked zone from geodetic data) extends below the continental forearc
Moho in Sumatra. Reduced vp velocities beneath the forearc basin
covering the region between the Mentawai Islands and the Sumatra mainland
possibly reflect a reduced thickness of the overriding crust.
Introduction
The largest earthquakes on Earth occur along subduction zones where the
oceanic plate is subducted beneath an upper continental plate and large
stress is accumulated during the interseismic phase of the seismic cycle.
Offshore Sumatra, the oceanic Indo-Australian plate subducts obliquely
beneath the Eurasian plate (Fig. 1). In the last decade, the margin hosted a
number of great earthquakes on the subduction thrust (Aceh-Andaman, 26 December 2004, Mw=9.2; Nias, 28 March 2005,
Mw=8.6; Bengkulu, 12 September 2007, Mw=8.5). Additionally, major events such as the intermediate depth
Mw=7.6 earthquake of 30 September 2009 (e.g., McCloskey et al.,
2010; Wiseman et al., 2012) and the shallow and slow rupture of the 25 October 2010
Mentawai tsunami earthquake (Mw=7.8; Bilek et al., 2011; Lay et al.,
2011;
Newman et al., 2011) were associated with significant loss of life. Yet, a
part of the margin near the northern Mentawai Islands (below Siberut)
remains unbroken since 1797 (Newcomb and McCann, 1987; Natawidjaja et al.,
2006; Konica et al., 2008; Chlieh et al., 2008; McCloskey et al., 2010). The
region is strongly coupled as inferred from GPS observations and coral data
(Chlieh et al., 2008). Further to the south, geodetic records suggest that
only half of the interseismic tectonic strain accumulated since the great
earthquake of 1833 (Fig. 1) might have been released by the 2007 Bengkulu
earthquake (Konca et al., 2008). Sieh et al. (2008) estimate the slip deficit
below Siberut Island since the large ruptures of 1797 and 1833 to be
∼8 m and a reduced slip deficit of ∼5 m for
the Batu Islands due to the lower degree of coupling in the region of the
Batu Islands (Fig. 2 and Chlieh et al., 2008). Therefore, the segment is in
an advanced stage of the seismic cycle, although east of Siberut Island
there has been significant intra-slab seismic activity, including the
Mw=7.6 Padang earthquake on 30 September 2009 (Fig. 2) at intermediate
depth (∼85 km), which caused significant damage in the city
of Padang. Based on Coulomb stress modeling, McCloskey et al. (2010)
suggest that the 2009 Padang earthquake did not significantly relax the
accumulated stress on the Mentawai segment leaving the threat of a great
tsunamogenic earthquake on the Mentawai patch below Siberut Island
unabated (e.g., Konca et al., 2008; Sieh et al., 2008).
The down-dip limit of subduction thrust earthquakes was suggested to be a
function of temperature at the plate interface and to be controlled by the
transition from unstable to stable sliding along the plate interface (e.g.,
Tichelaar and Ruff, 1993). Hyndman et al. (1997) estimate the maximum
temperature for seismic behavior to be 350 ∘C, while large
earthquakes may propagate with decreasing slip down to the 450 ∘C
isotherm. An additional limiting factor of the seismogenic zone results from
the presence of hydrated minerals (serpentinite) in the forearc mantle
wedge, suggesting that the down-dip limit of the seismogenic zone correlates
to the upper plate Moho (Oleskevich et al., 1999). However, for the Sumatran
margin the seismogenic zone is suggested to reach below the continental Moho
based on gravity surveys and wide-angle refraction and local earthquake
tomography (Siberut: Simoes et al., 2004; Kieckhefer et al., 1980,
∼30 km Moho depth; Aceh basin and Simeulue: Dessa et al.,
2009; Klingelhoefer et al., 2010; Tilmann et al., 2010, 21–25 km Moho depth;
southern Mentawai Islands: Collings et al., 2012, less than 30 km Moho
depth). For central Sumatra Chlieh et al. (2008) estimate the width of the
seismogenic zone based on geodetic data between 20 and 50 km, with the
largest width approximately alongside Siberut, and the smallest widths at
the Batu Islands and between Sipora and the Pagai islands.
Map showing the tectonic setting of the central Sumatran
subduction zone. The locations of earthquakes are indicated by red circles
(NEIC catalogue, M≥6, 1 January 1990 until 1 September 2017). The Mentawai Fault
(Diament et al., 1992; green line offshore) and the Sumatran Fault (Sieh and
Natawidjaja, 2000; black line onshore) are also shown. Bathymetry and
topography from the SRTM Plus database (Becker et al., 2009). Oceanic
fracture zones from Cande et al. (1989) and Tang et al. (2013). Rupture
zones of the great 1797 and 1833 earthquakes are based on uplift of coral
microatolls (Natawidjaja et al., 2006). Rupture areas of the 1861, 1935 and
1984 earthquakes are given by Rivera et al. (2002). Slip distribution of the
2004 earthquake from Chlieh et al. (2007). Slip distribution of the 2005 and
2007 earthquakes from Konca et al. (2007, 2008). Convergence between the
Australian plate and the forearc sliver from McNeill and Henstock (2014).
Volcanoes (Smithsonian Institute) shown with red triangles. Black boxes
indicate locations of Figs. 2, 6, 7 and 9. Abbreviations: Sim: Simeulue; BK:
Banyak Islands; Tb: Toba; N: Nias; B: Batu Islands; P: Pulau Pini; Sb:
Siberut Island; Sip: Sipora; NP: North Pagai; SP: South Pagai; Pdg: Padang.
Previous local earthquake tomography studies were conducted in northern
Sumatra focussing on the crustal structure of the region around Lake Toba
(Masturyono et al., 2001; Koulakov et al., 2009, 2016; Stankiewicz et al., 2010) or on the shallow crustal structure along the
Sumatran Fault (Muksin et al., 2013). Pesicek et al. (2010) imaged the deeper
slab geometry including the upper mantle and transition along the Sumatra,
Andaman and Burma subduction zones using a regional–global body wave
tomography. Offshore, the tomography study of Collings et al. (2012)
resolves the deeper structure beneath north and south Pagai where the
25 October 2010 tsunamogenic event occurred. Structural information is known
from active seismic reflection and refraction studies for a significant
number of profiles (e.g., Franke et al., 2008; Dean et al., 2010;
Klingelhoefer et al., 2010; Mukti et al., 2012; Shulgin et al., 2013). The
Mentawai Fault (Diament et al., 1992), located between the forearc islands
and the mainland, was recently imaged as a southwest dipping back-thrust
(e.g., Singh et al., 2010; Wiseman et al., 2011). However, there is only
limited information about the deep forearc structure and the seismogenic
zone (down to depths of ∼50 km) of the central Sumatran
margin.
Offshore central Sumatra, a ∼2500 km long NS trending
topographic feature, the Investigator Fracture Zone (IFZ), is situated on
the incoming Indo-Australian plate and is subducted at a rate of 57 mm yr-1
below the Sumatran mainland (Fig. 1). Seismicity occurring in the
prolongation of the IFZ down to depths of 200 km presumably reflects the
subducting trace of the IFZ (Fauzi et al., 1996; Lange et al., 2010). At
shallower depths, beneath the Batu Islands, both the forearc crust and the
plate interface are characterized by enhanced seismicity levels with a
number of persistent clusters. This region hosted several major events
during the last century (e.g., 1935 Mw=7.7 and 1984 Mw=7.2; Rivera et al., 2002) but was not affected by great
earthquakes in the last 220 years
at least (Konca et al., 2008). Together with the decreased locking this
justifies its identification as a persistent segment barrier (Natawidjaja et al., 2006).
The development of the forearc basin between the Sumatran mainland and the
island of Nias was described in Matson and Moore (1992). Overall, the
Sumatran margin is characterized by rapid accretion since the early
Oligocene with current trench fill ages from Quaternary to Eocene ages
(Moore et al., 1982). The uplift rates of Nias slope sediments is suggested
to be on the order of 100–300 m my and accreted material has been
uplifted by more than 800 m in the center of Nias island (Moore et al.,
1980).
In order to investigate the deep structure of the central Sumatran
subduction zone, a dense, temporary and amphibious (on–offshore) seismic network
was installed in central Sumatra in 2008. Besides local seismicity, the main
target of the seismometer network was to obtain velocity models of the
complete marine and continental forearc in order to decipher down-dip and
along-strike structural variations in the Sumatran subduction zone.
Station and event distribution, and ray coverage of the inversion.
Green circles indicate events used in the inversion and corresponding
ray paths are indicated by grey lines. Blue circles indicate the complete
local catalogue (e.g., with events outside the network, or excluded for other
reasons such as large RMS or a small number of picks per event). Yellow
triangles (grey triangles for 2-week OBS deployment) indicate stations
used in the study. The global Centroid-Moment-Tensor (gCMT) focal mechanism of the 30 September 2009 Padang
earthquake and its aftershocks (McCloskey et al., 2010) are indicated in red. The
plate coupling from Chlieh et al. (2008) is indicated with red contour lines
and labeled with the coupling degree. Light blue squares show the events
from the seismic crisis during 2008 occurring in a persistent seismicity
cluster SE of Siberut Island (Wiseman et al., 2011), including the Mw 7.2
main shock of 25 February (blue star). Other symbols as in Fig. 1.
Earthquake data
For the local earthquake tomography we use data from a dense amphibious
network of up to 62 stations covering the Sumatran forearc from the trench
to the volcanic arc (Lange et al., 2010). The 52 land stations from SEIS-UK were
installed in April 2008 between 1.8∘ S and 1.8∘ N on the
mainland and on the islands of Nias, Pulau Batu, Siberut and North Pagai.
Offshore, the network was complemented by 10 three-component ocean-bottom
seismometers (OBSs; Minshull et al., 2004) equipped with differential
pressure gauges from June 2008 to February 2009. During October 2008, 10
land stations were removed from the Sumatran mainland, leaving the remaining
42 land stations until February 2009. The land stations continuously
recorded three spatial components with sample rates of 50 and 100 Hz. We
also include data from eight permanent stations operated by BMKG
(Meteorological and Geophysical Agency of Indonesia, http://www.bmkg.go.id,
last access: 16 May 2010),
GEOFON (http://geofon.gfz-potsdam.de/, last access: 29 April 2010; FDSN code 1G) and
stations GSI and BKNI operated by the GEOFON network (FDSN network code GE;
GEOFON Data Centre, 1993) in the analysis. Furthermore, we include five
stations for strong events from a temporary deployment north of our project
area (Stankiewicz et al., 2010; GEOFON network code 7A-2008; Ryberg and
Haberland, 2008) and three stations from an adjacent temporary network to
the south (Collings et al., 2012). Additionally, data from 46 ocean-bottom
stations (OBS/H) from an active-source experiment offshore (25 May and
10 June 2008) were considered (Vermeesch et al., 2009). A summary of the
stations can be found in the supplementary material of Lange et al. (2010),
Table 1. The main sources of noise in the records were tree movement, rain
due to the tropical environment and anthropogenic noise (e.g., traffic),
affecting in particular the horizontal components. At the ocean-bottom
stations, S waves were very difficult to pick because, in addition to high
noise levels, the onset of S wave arrivals was usually poorly defined due to
basement conversions.
From the original dataset (Lange et al., 2010) with 1271 events and 32 4781 manually picked arrival times (20 251 P and 12 220 S-onsets), we selected
events with more than nine P and four S phase picks and RMS values smaller
than 1.5 s. Then, we removed all phase arrivals with residuals larger than 2 s.
Because of the large number of stations and events on or near the
Sumatran fault zone (SFZ) we applied a stricter selection criteria for these
crustal events (depths less than ∼20 km and distances of less
than 35 km from the fault trace) by excluding events with less than 11
recording stations and RMS values greater than 1 s. These selection criteria
were chosen to improve the numerical balance of events from different parts
of the study region (slab events: 9165 onsets, SFZ events: 7686 onsets).
Finally, we ignored stations with less than 15 high-quality observations.
OBS/H stations with high station residuals or dubious time corrections were
not included in the inversion in order to be sure that all the observed
travel times are accurate. After having checked the stability of the 2-D
inversions exclusively with events within the network (largest azimuthal gap
between azimuthally adjacent stations, gap ≤180∘), events with
gap < 200∘ were included in the inversion. We carefully
checked that the relaxation of the gap criterion to 200∘ did not
produce substantially different velocity models. Figure 2 shows the ray
coverage with many paths criss-crossing in the central part of the model.
The final dataset consists of 655 events with 9939 P- (therefrom 2626 with
the highest quality, using the quality assignment of Lange et al., 2010)
and 4859 S-arrivals (626 with highest quality).
Local earthquake tomography
We invert 2-D and 3-D velocity models of the Sumatran subduction zone
using local earthquake tomography (LET) techniques (Aki and Lee, 1976;
Kissling, 1988) by applying the well-established inversion code SIMUL2000
(Thurber, 1983; Evans et al., 1994) for the simultaneous inversion of
hypocentral parameters and velocity structure (vp, vp/vs). The
original algorithm by Thurber (1983) was subsequently modified and enhanced
with new features (e.g., Eberhart-Phillips, 1986, 1993; Um and Thurber, 1987;
Thurber and Eberhart-Phillips, 1999) and has been widely used for various
LET studies (e.g., Graeber and Asch, 1999; DeShon and Schwartz, 2004;
Haberland et al., 2009). For the inversion of the Sumatra data (located on
both the Southern Hemisphere and the Northern Hemisphere) SIMUL2000 needed to be modified to
operate across the Equator.
In the damped least-squares inversion, the velocity structure vp and
vp/vs are inverted from the observed travel times. The velocity model
is represented by velocity values specified on a rectangular grid of
irregularly spaced nodes. The velocity for a given point within the grid is
calculated by linearly interpolating the eight neighboring grid nodes. For
a fast calculation of the path integral, Thurber (1983) implemented the ray
tracer based on the ”Approximate Ray Tracing” technique (ART). Receiver
and source are connected with different circular arcs with varying radii and
inclinations. Then, the 2-D circular arcs are perturbed in three dimensions
to further minimize the travel time in an iterative process (Um and Thurber,
1987). Following common practice we applied a staggered inversion scheme
starting with inversions for a one-dimensional model, followed by an
inversion for a two-dimensional velocity model, and finally a 3-D inversion
using the 2-D model as a starting model. For each inversion, the arrival
times were weighted by their assigned pick uncertainties and all events were
relocated prior to each iteration.
The importance of careful selection of the minimum 1-D model was described
by various authors (e.g., Kissling, 1988; Eberhart-Phillips, 1990; Kissling
et al., 1994). As 1-D vp starting model, we used the ”minimum
one-dimensional model” from Lange et al. (2010) (Fig. 3, green line), which
was obtained from a brute force search of different one-dimensional input
models using the program VELEST (Kissling et al., 1994) and active source
studies (Vermeesch et al., 2009). For the inversion of the 2-D velocity model
we tested different vp starting models from an active source refraction
study (Vermeesch et al., 2009), from the seismicity study of Lange et al. (2010) and the LET of Collings et al. (2012) (Fig. 3). Based on these
different velocity models the inversion of the 2-D vp velocity model
leads to very similar results. For the inversion of the 2-D
vp/vs model we fixed (i.e., highly damped) the vp model and used a
constant vp/vs ratio of 1.77 derived from Wadati diagrams as starting
model.
Horizontal distances between nodes were 30 km in the trench-perpendicular
direction (x axis) and, for the 3-D inversion, 50 km in the trench-parallel
direction (y axis). In the vertical direction (z axis) node spacing is 10 km
down to 50 km depth with one additional node at 5 km depth. Below 50 km
depth, coarser node spacing is used with nodes at 70, 90, and 120 km depth
to account for the decreasing ray coverage with depth. The grid is rotated
relative to the trend of the north direction by 28∘ and centered at
0∘ N, 99∘ E. After carefully testing different spacing
parameters for 2-D and 3-D inversions in all three directions, we selected the
node spacing as a compromise between resolution and stability of the
inversion.
Following Evans et al. (1994), one additional node is introduced at all
edges of the model with a much larger distance for computational reasons.
The damping value of the damped least-squares inversion was carefully
determined by ”trade-off” curves between model variance and data variance
(Eberhart-Phillips, 1986) and is chosen such as to simultaneously minimize
the model variance and data variance. This is achieved by plotting model
variance versus data variance of one-step inversions with different damping
values for a given model geometry. SIMUL2000 uses one damping value for all
inversion steps and the model and data variance for the trade-off curve is
taken from the first inversion step. We made various inversions with
different damping values and found that the spatial distribution of
anomalies stays similar, but with varying amplitudes of the anomalies. The
final 3-D inversion yields a significant reduction of the data variance. The
P-wave data variance reduction is 76 % compared to the minimum 1-D velocity
model. The S-wave data variance reduction is only 18 % compared to a
homogeneous model with vp/vs values of 1.77. The small degree of
improvement for the 3-D velocity model relates mostly to the high noise
levels on the horizontal components resulting in S onsets of low quality. We
inverted 3-D velocity models for vp/vs ratios and conducted extensive 3-D
vp/vs checkerboard tests, synthetic modeling and parameter tests. However,
due to the low quality of S onsets, the 3-D vp/vs inversion was not robust
and the data variance reduction was always small. Therefore we only discuss
vp/vs ratios of the 2-D inversion.
Velocity models used as starting model for the two-dimensional
tomographic inversion of vp (red line and red circles).
Because of the large number of stations and events on or near the SFZ, the
minimum 1-D velocity model (green line) is dominated by the crustal structure
of the Sumatran mainland. This velocity model is not appropriate for the
events in the outer forearc, so we constructed a layered 1-D vp velocity
function based on an active source refraction study (Vermeesch et al.,
2009). Above 30 km depth, we adopted a one-dimensional, staircase-like
velocity–depth function based on an active source refraction study
(Vermeesch et al., 2009, orange line); for depths greater than 30 km we
adopted the one-dimensional velocity function from a minimum 1-D velocity
model (Lange et al., 2010). The blue line shows the minimum 1-D vp
velocity model from Collings et al. (2012) for the region adjacent to the
southeast of our study region but covering a similar part of the subduction
system.
Resolution and uniqueness
The method of LET tries to find a set of hypocenters and a velocity model,
which jointly fit the arrival times best. Therefore, any LET code has some
limitations, which include a finite number of synthetic recovery test and a
partially subjective choice of parameterization (e.g., grid spacing) of the
velocity model or the choice of the damping value. As discussed in the
previous chapter, SIMUL2000 uses a fixed velocity grid definition and a
constant damping value set according to finding a compromise between
obtaining a good data fit with low model variance, as judged by a trade-off
curve.
Dependency of 2-D inversion on 1-D input model
We tested the dependency of the 2-D inversion (constant values along the
y axis) on the 1-D input model in order to estimate the stability of the
inversion and its ability to converge. This was done by constructing
(realistic) randomized vp velocity models with increasing velocity for
increasing depths. These models were used as alternative starting models and
the inversion was otherwise carried out identically. We also tested
alternative vp starting models from the active source refraction study
of Vermeesch et al. (2009) and the minimum 1-D vp model of Collings et al. (2012) (Fig. 3). We then carefully checked the dependency of the 2-D
inversion on the velocity models and only found a minor dependency of the 1-D
input model, indicating a very stable result of the 2-D inversion, which
suggests a well-defined global minimum in the solution space for the 2-D
inversion. The independence of the inverted 2-D velocity model on the 1-D
input models alone does not necessarily point to a better imaging capacity
of the model and might also be related to oversimplification of reality.
We find this stability of the 2-D inversion for different velocity model
parameterizations (lateral and depth spacing) and a wide range of 1-D vp
velocity input models. Furthermore, the following 3-D inversion only results
in a modest further improvement of the fit. The trench-perpendicular
velocity heterogeneity (2-D structure) is thus more important than
trench-parallel heterogeneity (3-D structure).
Spread value
The spread function of the resolution matrix poses a possibility to assess
the resolution of the model nodes. The spread function (e.g., Toomey and
Foulger, 1989) summarizes the information contained in a single averaging
vector or row of the full-resolution matrix. For a peaked resolution, i.e.,
low smearing, the diagonal element is much larger than the off-diagonal
elements and the spread is low. The spread values (Fig. 4) show low values
in the central part of the model between the SFZ and the islands with a
reduced resolution in the region offshore Siberut and Nias. At depths larger
than 50 km, resolution is moderate as indicated by reduced spread values to
80 km depth. Below the Wadati–Benioff zone there is basically no penetration
and thus no resolution at all.
Checkerboard tests and synthetic recovery tests
Synthetic tests and checkerboard tests were carried out to evaluate the
resolution of the inversion. The procedure includes forward calculation of
the travel times for a synthetic velocity model and the actual source and
receiver distribution. In a second step, the calculated travel times are
then perturbed with Gaussian noise, with a standard deviation dependent on
the pick quality, from 0.05 s for the highest-quality observations to 0.2 s for
the lowest-quality observations. Finally, the perturbed travel times are
introduced into the inversion.
Spread values of 2-D vp(a) and
vp/vs(b)
inversion. Regions with low spread values and thus good resolution
(selection of a cut-off spread value based on analysis of the model
resolution matrix and synthetic tests) are encircled with a red line.
Cut-off spread values are 2.1 and 1.9 for vp and vp/vs, respectively. Circles
indicate hypocenters and grid nodes are shown with crosses. Stations are
indicated with triangles.
2-D checkerboard tests
The 2-D checkerboard tests were conducted for vp and vp/vs models
(Fig. 5). We used varying block sizes in which the input
models were perturbed by ±5 %. At the highest resolution (blocks
with one grid point dimension, equivalent to 30 km horizontal space and 10 km vertical space in the shallow part of the model), the pattern of
perturbations is restored in the central part but the maximum amplitude of
the recovered anomalies was 3.7 %, i.e., only about 75 % of the input
anomalies. The checkerboard tests with 2×2 blocks (60×20 km) and
lower resolution restore both the pattern and the amplitudes in the central
part of the model and beneath the SFZ.
2-D synthetic checkerboard models with 5 % velocity perturbation
input anomalies (a, d, g), the inversion restoration for vp(b, e, h) and
vp/vs(c, f, i) models. Crosses represent
nodes used in the inversion and the reconstructions are plotted with the
resulting hypocenter locations (black points). We calculated different
checkerboard inversions using 1×1 and 2×2 (shown in the rows from top to
bottom) grid node model perturbations. Noise was added to the synthetic data
depending on the quality of the arrivals.
3-D checkerboard test
For the 3-D case we performed numerous checkerboard inversions using
different checkerboard sizes. The checkerboard anomaly with 8 nodes (2×2×2
checkerboard, equivalent to 60×100×20 km) is reconstructed in the central
part at depths between 5 and 50 km (Fig. 6). Below 50 km only the region
beneath the volcanic arc shows sufficient ray coverage, but the profile view
suggests vertical smearing below 50 km depth. In general, the resolution is
good between the forearc islands and the SFZ between 5 and 50 km for the
region above the Wadati–Benioff zone, so we will restrict our interpretation
to this depth range. The shallow (< 30 km) region beneath the
eastern part of Siberut is characterized by aseismic behavior during the
deployment and the limited ray coverage results in insufficient recovery of
the checkerboard in this region. A threshold for the spread values has been
chosen to discriminate regions with high and low resolution and is
superimposed on the resulting tomographic velocity models. The choice of
threshold was carefully determined based on checkerboard tests, the ray
coverage, and on the relative amplitudes of the spread values.
3-D synthetic checkerboard models with 5 % velocity perturbation
input anomalies and the inversion restoration for the 3-D vp model. Other
symbols as in Fig. 5.
3-D synthetic restoration test
Restoring resolution tests were conducted to estimate the capacity of the
data to resolve the geometry and amplitudes of potential velocity
structures. We constructed synthetic vp velocity models with similar
characteristics in amplitude and dimensions as the inversion results and
further models with velocity anomalies representing the subducted IFZ. A
possibly modified crust along the IFZ was incorporated as obliquely oriented
positive velocity anomalies at the expected position of the subducting crust
(but with a larger thickness varying between 15 and 30 km). Further tests
were conducted with shallow velocity anomalies. Figure 7 shows the 3-D
restoration of a synthetic model where we integrated different anomalies.
The figure shows the restoration of an oblique velocity anomaly oriented in
the direction of the subducted IFZ at 5 km depth and at depths from the
plate interface and trench-parallel velocity anomalies at 30 km depth. The
absolute values of vp for the synthetic features are adequately
reproduced (Fig. 7) in regions with good resolution as indicated by the
spread value.
3-D synthetic models with 5 % velocity vp perturbation input
anomalies and the inversion restoration. The model consists of north–south
trending anomalies (map view 30 km depth) and a NE–SW trending low velocity
anomaly for both the shallow part of the model in 5 km and for the
trace of the subducted IFZ. Red and green lines indicate the 5 % contour
lines of the input anomalies. The blue lines encircle regions with good
resolution defined by the spread value. Other symbols as in Fig. 5.
Results and discussion
The 2-D vp and vp/vs velocity model is shown in Fig. 8, and the
final 3-D vp velocity model is shown in Figs. 9 and 10.
In the following, we discuss the main features for the different tectonic
units, making use of the lower-case labels in Figs. 9 and 10.
2-D tomographic velocity models for vp(a) and vp/vs(b) models
(profile direction is trench perpendicular). Regions with good resolution
are encircled with a red line. Circles indicate hypocenters and grid nodes
are shown with crosses. Stations are indicated with triangles. The dashed
line in panel (a) indicates the vp 7.8 km s-1 contour line and is used as a
proxy for the Moho.
Accretionary prism, forearc islands and forearc basin
In the shallow part of the vp velocity model we observe regions of
reduced vp velocities alternating with higher vp values at shallow
depths (Fig. 8, ∼10 km depth and Fig. 10a, b and c).
In the following, we discuss these regimes starting at the trench and moving
towards the mainland of Sumatra. The accretionary wedge composes the frontal
prism adjacent to the deep-sea trench as well as the lower to middle
continental slope seaward of the forearc islands. The accretionary domain
(labeled a in Figs. 9 and 10) is characterized by moderate velocities of
∼5 km s-1 down to a depth of ∼15 km, increasing
to ∼6 km s-1 above the landward-dipping high-velocity zone
(labeled f, Fig. 10). Velocities in the upper 15 km increase underneath the
forearc islands (labeled b) with values of ∼6 km s-1, which
are also observed beneath the coast. The forearc basin between the islands
and the mainland (labeled c) shows moderately low velocities of
∼5 km s-1 down to 10 km depth. When considering the shallow
forearc structure (< 15 km), the trench-perpendicular shallow
structure variations are similar to the results of Collings et al. (2012)
for the southern Mentawai Islands, in a way that the slow and fast domains
alternate in the landward direction. The most obvious difference is that
Collings et al. (2012) found low velocity values of approximately 5 km s-1
beneath the Mentawai forearc islands (Sipora, North Pagai and South Pagai),
adjacent to faster material beneath the forearc basin. Our results image the
region beneath the forearc islands as a trench-parallel (labeled b, Figs. 10 and 11), elongated zone of increased velocities, sandwiched between the
relatively lower velocities of the trenchward accretionary prism (labeled
a, Figs. 10 and 11) and the landward forearc basin (labeled c) the fast
velocity anomalies below and between the islands might be interpreted as
occurrence of faster accreted IFZ material beneath the Batu Islands. On
geological timescales the intersection of the IFZ with the marine forearc
migrates southeast as the subducted plate descends, and thus might have
created margin-parallel accreted features north of the current intersection of
the IFZ with the trench (e.g., north of Siberut island). However, we cannot
find significant along-strike variations in vp between the Mentawai Islands
and the trench (e.g., labeled a in Figs. 10 and 11), which might equally be
explained by accretion of seamounts (Fig. 1; 4.5∘ S, 99.5∘ E).
The very shallow marine forearc at depths of 5 km is characterized by three
regions of relatively reduced vp velocities of between 5 and 6 km s-1.
Faster regions (∼6 km s-1) are spatially related to the
forearc islands Nias, Pulau Batu, Siberut, and Pulau Pini (Fig. 9b).
In-between the forearc islands the marine forearc is mostly characterized by
vp velocities of 5 km s-1.
At depths of 20–30 km and 25 km east of the Mentawai Fault, a trench-parallel
velocity anomaly of higher vp velocities (labeled d in
Figs. 9 and 10, indicative by the upwelling of contour lines) suggests a shallower
location of the Moho beneath the forearc basin and hence a reduced thickness
of the overriding crust. Alternatively, this velocity anomaly might reflect
a deep subducted seamount. Based on reflection data Singh et al. (2011)
image an undulation of the top of the subducting slab in the Sumatran
forearc to the south at 5∘ S and interpreted this as a subducted
seamount. We exclude the possibility of a subducted seamount due to the size
of the anomaly (200 km×80 km) and the absence of a similar feature in the
seismicity (Fig. 1). Alternatively, this trench-parallel velocity anomaly of
higher vp velocities (labeled d) might be explained by an accreted
mafic block.
Sumatran fault zone (SFZ) and volcanic arc
While the offshore forearc is made up of young sediments from the Eocene to
Holocene, the mainland shows a ∼130 km wide belt of different
rock units along the SFZ. The SFZ is characterized by high seismicity rates
(e.g., Weller et al., 2012) due to stress and strain partitioning from the
oblique subduction (McCaffrey et al., 2000). This belt is mostly composed of
Permian to Jurassic sedimentary rocks, Eocene volcanic rocks and Jurassic to
Eocene intrusive units (Crow and Barber, 2005). The 3-D
velocity model along the SFZ is characterized by only minor changes in
vp along strike. Seismic velocities of 7.8 km s-1 (indicative of
continental Moho) are reached at depths larger than 30 km and outside the
region of good resolution. A Moho depth between 28 and 40 km is inline with
Moho depths from receiver functions in the region of the caldera of Lake
Toba (Fig. 1; Sakaguchi et al., 2006; Kieling et al., 2011) and similar to
the Moho depths inferred from receiver functions (Gunawan et al., 2011).
vp/vs values beneath the SFZ (depths ≤20 km) are between 1.65 and
1.72 (Fig. 8) and similar to the minimum 1-D velocity model of Weller et al. (2012), which used the same stations to derive an optimum 1-D model for the
SFZ region only. These low vp/vs ratios seem to be characteristic
for the shallow crust in the region of the SFZ. Muksin et al. (2013)
conducted a LET for the shallow crust (< 15 km) at 2∘ N and found similar lower
vp/vs values away from
the SFZ. Equally, Koulakov et al. (2009) imaged predominantly lower
vp/vs ratios below 1.8 for the region 100 km northwest of our study
area (labeled Tb in Fig. 1). Our findings differ from the velocity model of Koulakov
et al. (2009, 2016), in that we find only weak indications of a patchy low-velocity
zone beneath the magmatic arc at 30 km depth only.
Subducting oceanic lithosphere
Where the slab is still in contact with the overriding plate, the oceanic
Moho is imaged as the inclined 7.8 km s-1vp contour line
(Figs. 8c
and 10f). The plate interface, inferred from seismicity, is
located at approximately 25 km depth below the forearc islands (Fig. 8), a
little deeper than beneath the Pagai Islands at 3∘ S, where it was
found at 20 km depth (Collings et al., 2012), but significantly deeper than
the plate interface from seismicity and refraction seismicity found at 15 km
depth beneath Simeulue Island at 2.5∘ N (Tilmann et al., 2010;
Shulgin et al., 2013).
Seismicity 25 km west of Nias (Fig. 2) is part of a coast-parallel band of
seismicity. This band of high seismicity corresponds to the transition
between regions of significant coseismic (down-dip) and aseismic slip (up-dip)
of the 2005 earthquake (Hsu et al., 2006) and extends northwestwards until
Simeulue Island, roughly following the 500 m isobath contour lines (Tilmann
et al., 2010). The depth variations in seismicity along this seismicity band
suggest that the seismicity transition from aseismic to seismic behavior in
the down-dip direction (Lange et al., 2007, 2010; Tilmann et al., 2010) might not be controlled by depth and hence lithostatic pressure.
The inclination of the subducting plate is approximately 25∘
within the depth range between 40 and 80 km, also based on the seismicity,
as the resolution and grid spacing is insufficient for imaging subducting
oceanic crust. There are hints of the contrast between the subducting high-velocity slab
and the mantle wedge in the form of a dipping velocity contour
(e.g., Fig 10d), but it is only imaged in a patchy way at the limit of the
resolved area. At larger depths, seismicity can be traced down to 220 km
with an inclination of approximately 36∘ (Lange et al., 2010) but
the velocity structure is no longer resolved (Fig. 9e).
Figure 9f shows a section through the 3-D vp velocity model following the
plate interface (defined by the SLAB1.0 model; Hayes et al., 2012). The
dominant feature in this panel is the contrast between crust and mantle,
allowing us to identify the position of the toe of the mantle wedge just
landward of the forearc islands (except Pulau Pini, which is already well
above the mantle wedge). No obvious along-strike change can be identified in
the mantle wedge. In particular, the velocity model does not reveal
indications of velocity anomalies in the direction of the subducted IFZ,
although the trace of the subducted IFZ is reflected by seismicity down to
∼200 km depths (Fauzi et al., 1996; Lange et al., 2010); in
Fig. 9f it is visible as a band of seismicity striking north (i.e., to top
right in the figure) from Pulau Batu. The synthetic restoration tests (Fig. 7) document that the inversion is capable to resolve a ∼40 km
wide velocity anomaly, if there would be any. Considering such large-scale
structures, we conclude that the subducted IFZ did not disturb the velocity
structure at depths of the plate interface, e.g., by releasing fluids and
enhancing melt production. However, the IFZ clearly had a significant impact
on the rheological conditions within the slab since it enhances intermediate
depth seismicity down to large depths (Lange et al., 2010). Some of the
events, labeled with f in Fig. 10, panel (c) are located 10–15 km below the
plate interface defined by the global slab model (Hayes et al., 2012). Based
on their hypocentral depths we interpret them as being spatially related to
the oceanic crust to mantle transition (e.g., near the oceanic Moho) or even
possibly occurring in the uppermost oceanic mantle. For the North Chilean
subduction zone, Bloch et al. (2014) found a similar group of events
∼8 km below the plate interface for the North Chilean
subduction zone and at depths between 30 and 60 km and proposed them to be
spatially related to the oceanic Moho.
The 3-D vp model: depth sections (a–e) and curved section
along the plate interface as defined by the global SLAB1.0 model (Hayes et al., 2012) (f).
Red lines encircle regions of good resolution defined
by a cut-off spread value of 1.5. White circles indicate events within 10 km
of the section depth, except (f), where all events used for the
inversion are shown. Volcanoes (Smithsonian Institute) shown with red
triangles. The Mentawai Fault (blue line offshore) and the Sumatran Fault
(red line onshore) are also shown. See text for explanation of characters.
Other symbols as in Fig. 6.
Cross sections along trench-perpendicular trending profiles
through the 3-D vp model. See Fig. 9a for location of cross
sections. White circles indicate events within 10 km of the profile and
stations closer than 25 km to the profile are shown by white triangles, the
remaining ones by black triangles. The 46 OBS stations of the 2-week
deployment are shown with smaller triangles. Grid nodes are shown with
crosses. Red lines encircle regions of good resolution defined by a cut-off
spread value of 1.5. Green line in panel (c) indicates the plate interface as
defined by the global SLAB1.0 model (Hayes et al., 2012). The 7.8 km s-1vp
contour line is indicated by a black line. See text for explanation of
characters. Other symbols as in Fig. 6. Note that the geographic labels at
the top refer to all profiles, but that only profiles C (Batu Islands) and E
(Siberut) actually cross a forearc island.
vp/vs model of the forearc
As discussed in Sect. 2 S onsets are of lower quality due to tropical
conditions and anthropogenic noise. Therefore, we only present the 2-D vp/vs
inversion results (Fig. 8b). In our study region around Nias and Siberut,
we only find mostly moderately elevated vp/vs values (up to 1.85, Fig. 8b), whereas Collings et al., 2012 find strongly elevated
vp/vs values (up to 2.0) down to the plate interface below the Pagai
Islands. The largest values are found west of the forearc basins in the
region of the Mentawai Fault just landward of the forearc islands. Since
rays of the 2-D vp/vs velocity model mostly sample the region northeast
of Pulau Batu (Fig. 2) this likely reflects a local vp/vs anomaly
close to the Equator rather than being a feature present along the whole
along-strike length of the study region. The reason for this region of
elevated vp/vs remains enigmatic. Possible explanations include
fluids related to pathways created by the Mentawai Fault or structural
differences due to subducted material from the IFZ. Although
vp/vs ratios are moderately elevated (up to 1.85) we cannot identify
large-scale alteration of the mantle wedge due to surplus liquids from a
strongly hydrated IFZ because serpentinized material is characterized by
clearly elevated vp/vs and reduced vp values
(e.g., Carlson and Miller, 2003). Because mantle serpentinization favors
aseismic sliding and is related to the down-dip extent of the seismogenic
zone (e.g., Hyndman et al., 1997; Oleskevich et al., 1999), the lack of
large-scale serpentinization could explain why the seismogenic plate
interface extends into the forearc mantle off Sumatra (e.g., Simoes et al.,
2004; Collings et al., 2012). In particular, the stalling of the 2005
rupture was suggested to be limited by the subducted IFZ and reduced
coupling of the plate interface (Fig. 2 and Chlieh et al., 2008) and might
be related to rheological properties and heterogeneities along the plate
interface. Based on multi-channel seismic data, Henstock et al. (2016)
identified an isolated 3 km basement high close to the 2005 slip termination as well as along-strike
variations of basement relief. Such features are large enough to affect the
rheological behavior of the plate interface such as coupling but are below
the resolution of our LET.
Conclusions
We present 2-D and 3-D velocity models from a local earthquake tomography (LET)
using data from a dense network of seismic stations covering the onshore and
offshore domain of the northern Sumatra forearc close to the Equator. The
models resolve the structure of the forearc including the accretionary
prism, forearc islands, the forearc basin, the mantle wedge and the
volcanic arc down to a maximum depth of ∼60 km. The down-going
slab is traced by inclined velocity contour lines at depths < 40 km.
The oceanic crust has a velocity of ∼7 km s-1 and is located at
a depth of ∼25 km beneath the forearc islands (based on the
seismicity depth distribution). vp velocities beneath the magmatic arc,
which spatially coincides with the SFZ, are around 5 km s-1 at 10 km depth and
the vp/vs ratios in the uppermost 10 km are low, indicating the
presence of felsic lithologies typical for continental crust.
The forearc basins west and east of the Mentawai Islands are characterized
by velocities of ∼5 km s-1 down to 15 km depth. Although the
region is characterized by the subducted IFZ, which influences seismicity
down to depths of 200 km, the 3-D velocity model at depths of the plate
interface shows prevailing trench-parallel structures suggesting that the
subducted IFZ did not significantly modify the velocity structure at
seismogenic depths. At very shallow depths (∼5 km) and below
the forearc islands (Pulau Batu, Siberut, Nias) higher vp velocities of
∼6 km s-1 are found.
AT depths of 20–30 km and ∼25 km east of the Mentawai Fault,
a trench-parallel velocity anomaly of higher vp velocities might
suggest a shallower location of the Moho beneath the forearc basin and
hence a reduced thickness of the overriding crust.
Elevated vp/vs ratios of 1.85 are found in the overriding crust in the
region of the Mentawai Fault, which might be related to fluids. However,
vp/vs ratios are still too small to support a large-scale
serpentinization of the continental mantle and could explain why the
seismogenic plate interface (observed as a locked zone from geodetic data)
extends below the continental forearc Moho in Sumatra.
Seis-UK data are available from IRIS (https://www.iris.edu) using the network
code ZB (2007–2009) (Lange et al., 2010). The GEOFON data with network codes GE and 7A (2008) (GEOFON Data Centre, 1993) are stored at
the GEOFON data centre (https://geofon.gfz-potsdam.de/). GFZ instruments were provided by the Geophysical Instrument Pool Potsdam
(GIPP). The data from the permanent Indonesian network (network code IA) are stored at BMKG (http://www.bmkg.go.id, last access: 16 May 2010).
DL, FT, TH, AR and DH were involved in the installation of the seismological
stations. FT, TH, AR and DH designed the experiment. AR and DH supervised
the project. DL processed the data, conducted the inversions and prepared
the artwork. DL led the development of the manuscript supported by
significant contributions from FT, TH and HK who contributed to the ideas,
concepts and interpretation presented in this manuscript.
The authors declare that they have no conflict of
interest.
Acknowledgements
We thank the SeisUK facility in Leicester for the loan of the instruments
and the logistic support during this project, loan 828 (Brisbourne, 2012).
We acknowledge the support of the colleagues at Geoteknologi LIPI for this
project. LIPI-EOS let us share some of the sites of the SuGaR GPS network.
We thank the captain and crew of the vessel Andalas for excellent work in the
field. Furthermore, we thank the master and crew of R/V SONNE cruises SO-198
and SO-200 for the deployment and recovery of the OBS. OBS instruments were
provided by OBIF. We thank Lisa C. McNeill and Penny Barton who participated
in the acquisition of the OBS data. The project was funded by NERC
(NE/D00359/1). EOS (Earth Observatory of Singapore) is thanked for
supporting logistical costs of deployment on Mentawai and Batu Islands. We
thank the Indonesian BMKG and German GEOFON for the station data from their
permanent networks. Furthermore, we are indebted to all field crews for
their excellent work under tropical conditions. We thank Imam Suprihanto,
Bambang Suwargadi and Rachel Collings for support during the fieldwork.
Finally, we gratefully acknowledge the cooperation of many Sumatran
landowners, communities, and institutions for support and for allowing us to
install the seismic stations on their property. Our special thanks go to
Sylvain Barbot, Ivan Koulakov and an anonymous reviewer for their
constructive comments and suggestions.
Edited by: Tarje Nissen-Meyer
Reviewed by: Ivan Koulakov and one anonymous referee
ReferencesAki, K. and Lee, W. H. K.: Determination of three-dimensional velocity
anomalies under a seismic array using first P-arrival times from local
earthquakes, 1. A homogeneous initial model, J. Geophys. Res., 81,
4.381–4.399, 10.1029/JB081i023p04381, 1976.Becker, J. J., Sandwell, D. T., Smith, W. H. F., Braud, J., Binder, B.,
Depner, J., Fabre, D., Factor, J., Ingalls, S., Kim, S-H., Ladner, R.,
Marks, K., Nelson, S., Pharaoh, A., Sharman, G., Trimmer, R., VonRosenburg,
J., Wallace, G., and Weatherall, P.: Global Bathymetry and Elevation Data at
30 Arc Seconds Resolution: SRTM30_PLUS, Mar. Geod., 32, 355–371,
10.1080/01490410903297766, 2009.Bilek, S. L., Engdahl, E. R., DeShon, H. R., and El Hariri, M.: The 25
October 2010 Sumatra tsunami earthquake: Slip in a slow patch, Geophys. Res.
Lett., 38, L14306, 10.1029/2011GL047864, 2011.Bloch, W., Kummerow, J., Salazar, P., Wigger, P., and Shapiro, S. A.:
High-resolution image of the North Chilean subduction zone: seismicity,
reflectivity and fluids, Geophys. J. Int., 197, 1744–1749, 10.1093/gji/ggu084, 2014.
Brisbourne, A.: How to store and share geophysical data, Astron. Geophys.,
53, 4.19–4.20, 2012.
Cande, S. C., LaBrecque, J. L., Larson, R. L., Pitman, W. C., Golovchenko,
X., and Haxby, W. F.: Magnetic lineations of World's Ocean Basins (one
chart), Amer. Ass. Petrol. Geol., Tulsa, 1989.Carlson, R. L. and Miller, D. J.: Mantle wedge water contents estimated
from seismic velocities in partially serpentinized periodites, Geophys. Res.
Lett., 30, 1250, 10.1029/2002GL016600, 2003.Chlieh, M., Avouac, J.-P., Hjorleifsdottir, V., Song, T.-R. A., Ji, C.,
Sieh, K., Sladen, A., Hebert, H., Prawirodirdjo, L., Bock, Y., and Galetzka,
J.: Coseismic Slip and Afterslip of the Great Mw 9.15 Sumatra-Andaman
Earthquake of 2004, B. Seismol. Soc. Am., 97, S152–S173,
10.1785/0120050631, 2007.Chlieh, M., Avouac, J. P., Sieh, K., Natawidjaja, D. H., and Galetzka, J.:
Heterogeneous coupling of the Sumatran megathrust constrained by geodetic
and paleogeodetic measurements, J. Geophys. Res., 113, B05305,
10.1029/2007JB004981, 2008.Collings, R. E., Lange, D., Rietbrock, A., Tilmann, F., Natawidjaja, D. H.,
Suwargadi, B., Miller, M., and Saul, J.: Structure and seismogenic properties
of the Mentawai segment of the Sumatra subduction zone revealed by local
earthquake travel time tomography, J. Geophys. Res., 117, B01312, 10.1029/2011JB008469, 2012.Crow, M. J. and Barber, A. J: Map: Simplified geological map of Sumatra
Geological Society, London, Memoirs, 2005, 31:NP,
10.1144/GSL.MEM.2005.031.01.17, 2005.Dean, S. M., McNeill, L. C., Henstock, T. J., Bull, J. M., Gulick, S. P. S.,
Austin, J. A., Bangs, N. L. B., Djajadihardja, Y. S., and Permana, H.:
Contrasting Décollement and Prism Properties over the Sumatra 2004–2005
Earthquake Rupture Boundary, Science, 329, 207–210,
10.1126/science.1189373, 2010.DeShon, H. R. and Schwartz, S. Y.: Evidence for serpentinization of the
forearc mantle wedge along the Nicoya Peninsula, Costa Rica, Geophys. Res.
Lett, 31, L21611, 10.1029/2004GL021179, 2004.Dessa, J. X., Klingelhoefer, F., Graindorge, D., Andre, C., Permana, H.,
Gutscher, M. A., Chauhan, A., and Singh, S. S.: Megathrust earthquakes can
nucleate in the forearc mantle: Evidence from the 2004 Sumatra events,
Geology, 37, 659–662, 10.1130/G25653A.1, 2009.
Diament, M., Harjono, H., Karta, K., Deplus, C., Dahrin, D., Zen, M. T.,
Gérard, M., Lassal, O., Martin, A., and Malod, J.: Mentawai fault zone
off Sumatra: A new key to the geodynamics of western Indonesia, Geology, 20, 259–262, 1992.
Eberhart-Phillips, D.: Three-dimensional structure in northern California
coast ranges from inversion of local earthquake arrival time, B. Seismol.
Soc. Am., 76, 1.025–1.052, 1986.Eberhart-Phillips, D.: Three-Dimensional P and S Velocity Structure in the
Coalinga Region, California, J. Geophys. Res., 95, 15.343–15.363,
10.1029/JB095iB10p15343, 1990.
Eberhart-Phillips, D.: Local earthquake tomography: earthquake source
regions, in: Seismic Tomography: Theory
and practice, edited by: Iyer, H. M. and Hirahara, K., Chapman and Hall, London, 630–642, 1993.
Evans, J. R., Eberhart-Phillips, D., and Thurber, C. H.: User's Manual for
SIMULPS12 for Imaging Vp and Vp/Vs: A derivative of the “Thurber”
tomographic inversion SIMUL3 for local earthquakes and Explosions. U.S.
Dept. of the Interior, U.S. Geological Survey; Books and Open-File Reports
Section, distributor, open File Report 94–431, 1994.Fauzi, McCaffrey, R., Wark, D., Sunaryo, and Haryadi, P. Y. P.: Lateral
variation in slab orientation beneath Toba Caldera, northern Sumatra,
Geophys. Res. Lett., 23, 443–446, 10.1029/96GL00381, 1996.Franke, D., Schnabel, M., Ladage, S., Tappin, D. R., Neben, S.,
Djajadihardja, Y. S., Mueller, C., Kopp, H., and Gaedicke, C.: The great
Sumatra–Andaman earthquakes – Imaging the boundary between the ruptures of
the great 2004 and 2005 earthquakes, Earth Planet. Sc. Lett., 269,
118–130, 10.1016/j.epsl.2008.01.047, 2008.GEOFON Data Centre: GEOFON Seismic Network, Deutsches GeoForschungsZentrum
GFZ, Other/Seismic Network, 10.14470/TR560404, 1993.Graeber, F. and Asch, G.: Three-dimensional models of P wave velocity and
P-to-S velocity ratio in the southern central Andes by simultaneous
inversion of local earthquake data, J. Geophys. Res., 104,
20.237–20.256, 1999.
Gunawan, A., Tilmann, F., Lange, D., Collings, R., Rietbrock, A.,
Natawidjaja, D., and Widiyantoro, S.: Moho Depth Estimation beneath Sumatera
and Mentawai Islands Using Receiver Functions Recorded with a Temporary
Array, EGU General Assembly 2011, Geophysical Research Abstracts, vol. 13,
EGU2011-8072, 2011.Haberland, C., Rietbrock, A., Lange, D., Bataille, K., and Dahm, T.:
Structure of the seismogenic zone of the southcentral Chilean margin
revealed by local earthquake traveltime tomography, J. Geophys. Res., 114,
B01317, 10.1029/2008JB005802, 2009.Hayes, G. P.,Wald, D. J., and Johnson, R. L.: Slab1.0: A three-dimensional
model of global subduction zone geometries, J. Geophys. Res., 117, B01302,
10.1029/2011JB008524, 2012.Henstock, T. J., McNeill, L. C., Bull, J. M., Cook, B. J., Gulick, S. P. S.,
Austin, J. A., Permana, H., and Djajadihardja, Y. S.: Downgoing plate
topography stopped rupture in the A.D. 2005 Sumatra earthquake, Geology, 44,
71–74, 10.1130/G37258.1, 2016.Hsu, Y.-J., Simons, M., Avouac, J.-P., Galetzka, J., Sieh, K., Chlieh, M., Natawidjaja, D., Prawirodirdjo, L., and Bock, Y.: Frictional Afterslip
Following the 2005 Nias-Simeulue Earthquake, Sumatra, Science, 312, 1921–1926, 10.1126/science.1126960, 2006.Hyndman, R. D., Yamano, M., and Oleskevich, D. A.: The seismogenic zone of
subduction thrust faults, Isl. Arc, 6, 244–260,
10.1111/j.1440-1738.1997.tb00175.x, 1997.
Kieckhefer, R. M., Sho, G. G., and Curray J. R.: Seismic refraction studies
of the Sunda trench and forearc basin, J. Geophys. Res., 85, 863–889, 1980.Klingelhoefer, F., Gutscher, M. A., Ladage, S., Dessa, J. X., Graindorge,
D., Franke, D., Andre, C., Permana, H., Yudistira, T., and Chauhan, A.:
Limits of the seismogenic zone in the epicentral region of the 26 December
2004 great Sumatra-Andaman earthquake: Results from seismic refraction and
wide-angle reflection surveys and thermal modeling, J. Geophys. Res., 115,
B01304, 10.1029/2009JB006569, 2010.
Kissling, E.: Geotomography with local earthquake data, Rev. Geophys., 26,
659–698, 1988.
Kissling, E., Ellsworth, W. L., Eberhart-Phillips, D., and Kradolfer, U.:
Initial reference models in local earthquake tomography, J. Geophys. Res.,
99, 19.635–19.646, 1994.Kieling, K., Roessler, D., and Krueger, F. J.: Receiver function study in
northern Sumatra and the Malaysian peninsula, J. Seismol., 15, 235–259, 10.1007/s10950-010-9222-7, 2011.Konca, A. O., Hjorleifsdottir, V., Song, T. R. A., Avouac, J. P.,
Helmberger, D. V., Ji, C., Sieh, K., Briggs, R., and Meltzner, A.: Rupture
Kinematics of the 2005 Mw 8.6 Nias-Simeulue Earthquake from the Joint
Inversion of Seismic and Geodetic Data, B. Seismol. Soc. Am., 97,
S307–322, 2007.
Konca, A. O., Avouac, J. P., Sladen, A., Meltzner, A. J., Sieh, K., Fang,
P., Li, Z., Galetzka, J., Genrich, J., Chlieh, M., Natawidjaja, D. H., Bock,
Y., Fielding, E. J., Ji, C., and Helmberger, D. V.: Partial rupture of a
locked patch of the Sumatra megathrust during the 2007 earthquake sequence,
Nature, 456, 631–635, 2008.Koulakov, I., Yudistira, T., Luehr, B.-G., and Wandono: Velocity and
vp/vs ratio beneath the Toba caldera complex (Northern Sumatra) from
local earthquake tomography, Geophys. J. Int., 177, 1121–1139,
10.1111/j.1365-246X.2009.04114.x, 2009.Koulakov, I., Kasatkina, E., Shapiro, N. M., Jaupart, C., Vasilevsky, A.,
Khrepy, S. E., Al-Arifi, N., and Smirnov, S.: The feeder system of the Toba
supervolcano from the slab to the shallow reservoir, Nat. Commun., 7, 12228,
10.1038/ncomms12228, 2016.Lange, D., Rietbrock, A., Haberland, C., Bataille, K., Dahm, T., Tilmann, F.,
and Flüh, E. R.: Seismicity and geometry of the south Chilean subduction
zone (41.5∘ S–43.5∘ S): Implications for controlling
parameters, Geophys. Res. Lett., 34, L06311, 10.1029/2006GL029190,
2007.Lange, D., Tilmann, F., Rietbrock, A., Collings, R., Natawidjaja, D. H.,
Suwargadi, B. W., Barton, P., Henstock, T., and Ryberg, T.: The Fine
Structure of the Subducted Investigator Fracture Zone in Western Sumatra as
Seen by Local Seismicity, Earth Planet. Sc. Lett., 298,
47–56, 10.1016/j.epsl.2010.07.020, 2010.Lay, T., Ammon, C. J., Kanamori, H., Yamazaki, Y., Cheung, K. F., and Hutko,
A. R.: The 25 October 2010 Mentawai tsunami earthquake (Mw 7.8) and the
tsunami hazard presented by shallow megathrust ruptures, Geophys. Res. Lett,
38, L06302, 10.1029/2010GL046552, 2011.Masturyono, McCaffrey, R., Wark, D. A., Roecker, S. W., Fauzi, Ibrahim, and
G., Sukhyar: Distribution of magma beneath the Toba caldera complex, north
Sumatra, Indonesia, constrained by three-dimensional P wave velocities,
seismicity, and gravity data, Geochem. Geophy. Geosy., 2, 10.1029/2000GC000096, 2001.
Matson, R. and Moore, G. F.: Structural controls on forearc basin subsidence
in the central Sumatra forearc basin. In Geology and Geophysics of
Continental Margins, Am. Assoc. Petrol. Geol. Memoir, 53, 157–181, 1992.
McCaffrey, R., Zwick, P., Bock, Y., Prawirodirdjo, L., Genrich, J.,
Stevens, C. W., Puntodewo, S. S. O., and Subarya, C.: Strain partitioning during
oblique plate convergence in northern Sumatra: Geodetic and seismologic
constraints and numerical modeling, J. Geophys. Res., 105, 28363–28376,
2000.McCloskey, J., Lange, D., Tilmann, F., Nalbant, S. S., Bell, A. F.,
Natawidjaja, D. H., and Rietbrock, A.: The September 2009 Padang earthquake,
Nat. Geosci., 3, 70–71, 10.1038/ngeo753, 2010.McNeill, L. C. and Henstock, T. J.: Forearc structure and morphology along
the Sumatra-Andaman subduction zone, Tectonics, 33, 112–134, 10.1002/2012TC003264, 2014.
Minshull, T. A., Sinha, M. C., and Peirce, C.: Multi-disciplinary, sub-seabed
geophysical imaging – A new pool of 28 seafloor instruments in use by the
United Kingdom Ocean Bottom Instrument Consortium, Sea Technol., 46, 27–31,
2004.Moore, G. F., Billman, H. G., Hehanussa, P. E., and Karig, D. E.:
Sedimentology and paleo- bathymetry of Neogene trench-slope deposits, Nias
Island, Indonesia, J. Geol., 88, 161–180,
10.1086/628489, 1980.
Moore, G. F., Curray, J. R., and Emmel, F. J.: Sedimentation in the Sunda
Trench and forearc region, Geol. Soc. London Spec. Publ., 10, 245–258,
1982.Muksin, U., Bauer, K., and Haberland, C.: Seismic vp and vp/vs
structure of the geothermal area around Tarutung (North Sumatra, Indonesia)
derived from local earthquake tomography, J. Volcanol. Geoth. Res., 260,
27–42, 10.1016/j.jvolgeores.2013.04.012, 2013.Mukti, M. M., Singh, S. C., Deighton, I., Hananto, N. D., Moeremans, R., and
Permana, H.: Structural evolution of backthrusting in the Mentawai Fault
Zone, offshore Sumatran forearc, Geochem. Geophy. Geosy., 13,
10.1029/2012GC004199, 2012.Natawidjaja, D. H., Sieh, K., Chlieh, M., Galetzka, J., Suwargadi, B. W.,
Cheng, H., Edwards, R. L., Avouac, J.-P,. and Ward, S. N.: Source parameters
of the great Sumatran megathrust earthquakes of 1797 and 1833 inferred from
coral microatolls, J. Geophys. Res, 111, B06403, 10.1029/2005JB004025,
2006.
Newcomb, K. R. and McCann, W. R.: Seismic history and seismotectonics of the
Sunda arc, J. Geophys. Res, 92, 421–439, 1986.Newman, A. V., Hayes, G., Wei, Y., and Convers, J.: The 25 October 2010
Mentawai tsunami earthquake, from real-time discriminants, finite-fault
rupture, and tsunami excitation, Geophys. Res. Lett., 38,
10.1029/2010GL046498, 2011.
Oleskevich, D. A., Hyndman, R. D., and Wang, K.: The updip and downdip limits
to great subduction earthquakes: Thermal and structural models of Cascadia,
south Alaska, SW Japan, and Chile, J. Geophys. Res., 104,
14.965–14.991, 1999.Pesicek, J. D., Thurber, C. H., Zhang, H., DeShon, H. R., Engdahl, E. R.,
and Widiyantoro, S.: Teleseismic double-difference relocation of earthquakes
along the Sumatra-Andaman subduction zone using a 3-D model, J. Geophys.
Res., 115, B10303, 10.1029/2010JB007443, 2010.Rivera, L., Sieh, K., Helmberger, D., and Natawidjaja, D.: A comparative
study of the sumatran Subduction-Zone earthquakes of 1935 and 1984, B.
Seismol. Soc. Am, 92, 1721–1736, 10.1785/0120010106, 2002.Ryberg, T. and Haberland, C.: Lake Toba seismic network, Sumatra, Indonesia.
Deutsches GeoForschungsZentrum GFZ, Other/Seismic Network,
10.14470/2N934755, 2008.Sakaguchi, K., Gilbert, H., and Zandt, G.: Converted wave imaging of the Toba
Caldera, Indonesia, Geophys. Res. Lett., 33, L20305, 10.1029/2006GL027397,
2006.Shulgin, A., Kopp, H., Klaeschen, D., Papenberg, C., Tilmann, F., Flueh,
E. R., Franke, D., Barckhausen, U., Krabbenhoeft, A., and Djajadihardja,Y.:
Subduction system variability across the segment boundary of the 2004/2005
Sumatra megathrust earthquakes, Earth Planet. Sc. Lett., 365, 108–119,
10.1016/j.epsl.2012.12.032, 2013.Singh, S. C., Hananto, N. D., Chauhan, A. P. S., Permana, H., Denolle, M.,
Hendriyana, A., and Natawidjaja, D.: Evidence of active backthrusting at the
NE Margin of Mentawai Islands, SW Sumatra, Geophys. J. Int., 180, 703–714,
10.1111/j.1365-246X.2009.04458.x, 2010.Singh, S. C., Hananto, N., Mukti, M., Robinson, D.P., Das, S., Chauhan, A.,
Carton, H., Gratacos, B., Midnet, S., Djajadihardja, Y., and Harjono, H.:
Aseismic zone and earthquake segmentation associated with a deep subducted
seamount in Sumatra, Nat. Geosci., 4, 308–311, 10.1038/ngeo1119, 2011.
Sieh, K. and Natawidjaja, D.: Neotectonics of the Sumatran fault, Indonesia,
J. Geophys. Res., 105, 28295–28326, 2000.Sieh, K., Natawidjaja, D. H., Meltzner, A. J., Shen, C., Cheng, H., Li, K.,
Suwargadi, B. W., Galetzka, J., Philibosian, B., and Edwards, R. L.:
Earthquake Supercycles Inferred from Sea-Level Changes Recorded in the
Corals of West Sumatra, Science, 322, 1674–1678, 10.1126/science.1163589, 2008.Stankiewicz, J., Ryberg, T., Haberland, C., Fauzi, and Natawidjaja, D.: Lake
toba volcano magma chamber imaged by ambient seismic noise tomography,
Geophys. Res. Lett., 37, L17306, 10.1029/2010GL044211, 2010.Simoes, M., Avouac, J. P., Cattin, R., and Henry, P.: The Sumatra subduction
zone: A case for a locked fault zone extending into the mantle, J. Geophys.
Res., 109, B10402, 10.1029/2003JB002958, 2004.Tang, G., Barton, P. J., McNeill, L. C., Henstock, T. J., Tilmann, F., Dean, S. M., Jusuf,
M. D., Djajadihardja, Y. S., Permana, H., Klingelhoefer, F., and Kopp, H.: 3-D active
source tomography around Simeulue Island offshore Sumatra: Thick crustal
zone responsible for earthquake segment boundary, Geophys. Res. Lett., 40,
48–53, 10.1029/2012GL054148, 2013.
Tilmann, F. J., Craig, T. J., Grevemeyer, I., Suwargadi, B., Kopp, H., and
Flueh, E.: The updip seismic/aseismic transition of the Sumatra megathrust
illuminated by aftershocks of the 2004 Aceh-Andaman and 2005 Nias events,
Geophys. J. Int., 181, 1261–1274, 10.1111/j.1365-246X.2010.04597.x,
2010.Thurber, C. and Eberhart-Phillips, D.: Local earthquake tomography with
flexible gridding, Comput. Geosci., 25, 809–818,
10.1016/S0098-3004(99)00007-2, 1999.Thurber, C. H.: Earthquake locations and three-dimensional crustal structure
in the Coyote Lake area, Central California, J. Geophys. Res., 88,
8.226–8.236, 10.1029/JB088iB10p08226, 1983.
Tichelaar, B. W. and Ruff, L. J.: Depth of seismic coupling along subduction
zones, J. Geophys. Res., 98, 2.017–2.037, 1993.
Toomey, D. R. and Foulger, G. R.: Tomographic inversion of local earthquake
data from the Hengill-Grensdalur central volcano complex, Iceland, J.
Geophys. Res., 94, 17.497–17.510.2, 1989.
Um, J. and Thurber, C. H.: A fast algorithm for two-point ray tracing, B.
Seismol. Soc., Am, 77, 972–986, 1987.
Vermeesch, P. M., Henstock, T. J., Lange, D., McNeill, L. C., Barton, P. J.,
Tang, G., Bull, J. M., Tilmann, F., Dean, S. M., Djajadihardja, Y., and
Permana, H.: 3-D tomographic seismic imaging of the southern rupture barrier
of the great Sumatra-Andaman 2005 earthquake, Geophysical Research
Abstracts, Vol. 11, EGU2009-11509, EGU General Assembly Vienna, 2009.Weller, O., Lange, D., Tilmann, F., Natawidjaja, D., Rietbrock, A.,
Collings, R., and Gregory, L.: The structure of the Sumatran Fault revealed
by local seismicity, Geophys. Res. Lett., 39, L01306, 10.1029/2011GL050440, 2012.Wiseman, K., Banerjee, P., Sieh, K., Bürgmann, R., and Natawidjaja, D.
H.: Another potential source of destructive earthquakes and tsunami offshore
of Sumatra, Geophys. Res. Lett., 38, L10311, 10.1029/2011GL047226, 2011.Wiseman, K., Banerjee, P., Bürgmann, R., Sieh, K., Dreger, D. S., and
Hermawan, I.: Source model of the 2009 Mw 7.6 Padang intraslab earthquake
and its effect on the Sunda megathrust, Geophys. J. Int., 190, 1710–1722,
10.1111/j.1365-246X.2012.05600.x, 2012.