Journal of Asian Earth Sciences 96 (2014) 123–131

Contents lists available at ScienceDirect

Journal of Asian Earth Sciences

journal homepage: www.elsevier.com/locate/jseaes

Detailed fault structure of the Tarutung Pull-Apart Basin in Sumatra,

Indonesia, derived from local earthquake data
Umar Muksin a,b,⇑, Christian Haberland a, Mochamad Nukman a,c, Klaus Bauer a, Michael Weber a,d
a
GFZ German Research Centre for Geosciences, Potsdam, Germany
b
Department of Physics, Syiah Kuala University, Banda Aceh, Indonesia
c
Program Studi Geofisika, FMIPA, Universitas Gadjah Mada, Yogyakarta, Indonesia
d
Institute of Earth and Environmental Science, University of Potsdam, Potsdam, Germany

a r t i c l e i n f o a b s t r a c t

Article history: The Tarutung Basin is located at a right step-over in the northern central segment of the dextral strike-
Received 11 March 2014 slip Sumatran Fault System (SFS). Details of the fault structure along the Tarutung Basin are derived from
Received in revised form 8 August 2014 the relocations of seismicity as well as from focal mechanism and structural geology. The seismicity dis-
Accepted 3 September 2014
tribution derived by a 3D inversion for hypocenter relocation is clustered according to a fault-like seis-
Available online 16 September 2014
micity distribution. The seismicity is relocated with a double-difference technique (HYPODD) involving
the waveform cross-correlations. We used 46,904 and 3191 arrival differences obtained from catalogue
Keywords:
data and cross-correlation analysis, respectively. Focal mechanisms of events were analyzed by applying
Focal mechanism
Seismicity
a grid search method (HASH code). Although there is no significant shift of the hypocenters (10.8 m in
Structural geology average) and centroids (167 m in average), the application of the double difference relocation sharpens
Extensional duplex the earthquake distribution. The earthquake lineation reflects the fault system, the extensional duplex
Flower structure fault system, and the negative flower structure within the Tarutung Basin. The focal mechanisms of
Sumatran fault events at the edge of the basin are dominantly of strike-slip type representing the dextral strike-slip
Pull-Apart Basin Sumatran Fault System. The almost north–south striking normal fault events along extensional zones
beneath the basin correlate with the maximum principal stress direction which is the direction of the
Indo-Australian plate motion. The extensional zones form an en-echelon pattern indicated by the pres-
ence of strike-slip faults striking NE–SW to NW–SE events. The detailed characteristics of the fault system
derived from the seismological study are also corroborated by structural geology at the surface.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction the examples of this case. The 1650 km long strike-slip Sumatran
Fault accommodates the trench-parallel shear component of this
The convergence between the Indo-Australian and Eurasian convergence along the magmatic arc (Bellier and Sébrier, 1994;
plates has produced the subduction zone beneath Indonesia. The McCaffrey, 2009; Sieh and Natawidjaja, 2000). Therefore, the shape
shape of the subduction zone beneath Sumatra collocates with the of the SFS corresponds to that of the Sumatran subduction zone and
volcanic arc (Fig. 1) as also often found in other areas (e.g. Tatsumi, the Sumatran volcanic arc (Sieh and Natawidjaja, 2000).
1989). In case of oblique subduction, a trench-parallel strike-slip A geomorphological study of Sieh and Natawidjaja (2000)
fault system can occur along the volcanic arc. The Liquiñe-Ofqui revealed that the SFS is partitioned into 19 major sections, mostly
Fault Zone accommodating the northward motion of a continental caused by the changes in the rate and direction of the Indo-Australian
forearc sliver relative to the South-American continent (Cembrano plate motion along Sumatra. This segmentation is characterized by
et al., 1996; Lange et al., 2008) and the San Andreas Fault System tak- several Pull-Apart Basins caused by dilatational step-overs
ing up most of the shear component (Teyssier et al., 1995) are among (Muraoka et al., 2010; Sieh and Natawidjaja, 2000). Though the
SFS is segmented, the magnitudes of earthquakes were large
enough to devastate the region, such as the 1994 Mw 6.8 Liwa
⇑ Corresponding author at: GFZ German Research Centre for Geosciences, and the 2013 Mw 6.2 Aceh earthquakes.
Geophysical Deep Sounding, Telegrafenberg, 14473 Potsdam, Germany. Tel.: +49 The segments of the SFS show different characteristics (Sieh and
17630447674.
Natawidjaja, 2000). In the southern part of the Sumatra island, the
E-mail addresses: [email protected], [email protected] (U.
Muksin). Sunda segment (6.75S–5.9S) is characterized by normal and

http://dx.doi.org/10.1016/j.jseaes.2014.09.009
1367-9120/Ó 2014 Elsevier Ltd. All rights reserved.
124 U. Muksin et al. / Journal of Asian Earth Sciences 96 (2014) 123–131

94˚ 95˚ 96˚ 97˚ 98˚ 99˚ 100˚ 101˚ 102˚ 103˚ 104˚ 105˚ 106˚
6˚

5˚ Su
m
at
ra
n
Fa Indonesia
4˚ ul
t

3˚

R
Sumatra

en
Australia

Ta
un

ru
tu
2˚

ng
3.9 cm/yr

1˚

Bi
fu
rc
at
io
0˚

Si
n

ng
ka
ra
k
−1˚
4.8 cm/yr
In
do

−2˚
−A
us
tra
lia

−3˚
pl
at
e

−4˚
Ke
m
ur
in
−5˚ g

Su
5.2 cm/yr oh
Sund

−6˚
a

Trench Fault Volcano

Fig. 1. Regional tectonic setting of Sumatra and Tarutung district (box) is located between the Toba caldera and the bifurcation. The location of the Singkarak and the Suoh
Basins are represented by the white circles.

dextral faulting on the surface, as also indicated by normal fault The Tarutung Basin hosts several geothermal manifestations
earthquakes on the western side of the graben, while reverse slip (Fig. 2). At least 18 springs (i.e. cold and warm) discharge along
was found along the Kemuring segment (5.3–4.35 S). In central the Tarutung Basin margins (Nukman and Moeck, 2013) and about
Sumatra, the SFS is characterized by the presence of a bifurcation 13 hot springs discharging in the Sarulla area (Gunderson et al.,
around 0–1.7 N composed of segments (Sieh and Natawidjaja, 1995). It has been suggested that the geothermal system in the
2000). Weller et al. (2012) observed a dextral strike-slip Tarutung region is controlled by the fault system (Muksin et al.,
duplex system of the bifurcation linked by smaller sinistral 2013a). Along the SFS, normal faults along Pull-Apart Basins play
strike-slip faults. The mechanism of the extensional duplex system an important role as major discharge zones for geothermal fluids
along a bent strike-slip fault is responsible for the presence of a (Muraoka et al., 2010). Therefore, it is important to study the fault
flower structure as described in detail by Woodcock and Fischer system in the Tarutung region to understand the structural
(1986). controls on the geothermal system in the region. This paper
More to the north of the bifurcation, the SFS is characterized by presents the fault structure along the Tarutung Basin derived from
the Renun segment which is the longest segment and one of the local earthquake seismicity, as well as the focal mechanism
largest graben along the SFS traversing the western flank of the analysis enriched by geological field investigation.
Toba caldera (Sieh and Natawidjaja, 2000). The Tarutung district,
characterized by the Tarutung Pull-Apart Basin in the north and 2. Data and methods
the Sarulla graben in the south, is located between these two major
features, the Toba caldera and the bifurcation segments (Fig. 1). 2.1. Previous work
Bellier and Sébrier (1994) suggested that the Tarutung Pull-Apart
Basin was developed due to a larger step-over of the previous fault The research started with the deployment of 42 short period
system and as the latest evolution of the great Toba caldera within seismic instruments for 10 months starting in May 2011, in the
the older step-over region. district of North Tapanuli Sumatra, Indonesia (Fig. 2). The network
 U. Muksin et al. / Journal of Asian Earth Sciences 96 (2014) 123–131 125

98.7˚ 98.8˚ 98.9˚ 99˚ 99.1˚ 99.2˚ 99.3˚

2.3˚
Sumatra

Siborongborong

2.2˚

Imun V.

To
ba
2.1˚
Si
Tarutung
Pg
Ht
Tarutung
2˚

Martimbang V.

1.9˚
Na
mo
ra
L.

Ho
1.8˚
Su

po
Sarulla
ma

ng
Ca
tra

lde
nF

ra
au

1.7˚
lt

km

0 10
1.6˚
Stations Hot springs Volcanoes Elev. (m)
0 500 1000 1500 2000 2500

Fig. 2. Station distribution covering the Tarutung and the Sarulla Basin deployed for 10 months starting May 2011.

covered the area of the Tarutung Basin in the north and the Sarulla recordings were used in this inversion and subsequent steps.
graben in the south. The details of the seismic experiment and the More details of the inversion using VELEST can be found in
data preprocessing are explained in Muksin et al. (2013a, b). An Muksin et al. (2013a).
automatic picking algorithm (Nippress et al., 2010) was used to The final 1D layered velocity model obtained by VELEST was
detect the arrivals of P and S waves of earthquakes and then the used to construct an initial 3D velocity structure. The 3D model
arrivals were revised manually. consists of regularly spaced grid nodes with individual velocity val-
For simplification, the earth structure initially was assumed to ues. A tomographic inversion was then carried out to determine
consist of layers forming a 1D velocity model. We applied the optimum velocity structure and seismicity distribution within
HYPO71 (Lee and Valdes, 1985) based on the global 1D velocity this 3D model using the SIMUL2000 software (Eberhart-Phillips
model of IASP91 (Kennett and Engdahl, 1991) to produce the first and Michael, 1998; Evans et al., 1994; Thurber, 1983).
estimate location of 2586 earthquakes. The earthquake localiza- The iterative inversion procedure includes Approximate Ray
tion was further improved by using the simultaneous inversion Tracing (Thurber, 1983) to predict travel times, comparison with
for hypocenter relocation and 1D velocity structure (VELEST soft- observed travel time data, and updates of velocity values and
ware, Kissling et al., 1994). In extension to the previous step hypocenter locations using a damped least-squares algorithm.
(application of HYPO71 based on a fixed velocity-depth function), The iteration process is repeated until the RMS misfit between
the 1D velocity model was now modified during the inversion observed and predicted travel times is minimized. The details of
together with the hypocenter locations to better explain the the 3D tomographic inversion are presented in Muksin et al.
travel time data. Only events with a minimum of 10 station (2013a).
126 U. Muksin et al. / Journal of Asian Earth Sciences 96 (2014) 123–131

2.2. New relocation and focal mechanism cross-correlation less than 0.05 s and a correlation factor greater
than 70%.
The seismicity resulting from the 3D tomographic inversion We calculated the focal mechanism of the events based on the
was then relocated using a double difference relocation scheme visually inspected and manually picked first polarity of the P waves
(hypoDD) (Waldhauser and Ellsworth, 2000). In general, earth- using the HASH software (Hardebeck and Shearer, 2002). In the
quakes which occur along the same fault segment can be analysis of the focal mechanism, we only include seismic events
assumed to show similar source mechanisms, and, hence, produce having more than 8 polarity records. The take-off angles for each
similar waveforms at a receiver. The main idea of this technique observation are calculated based on the ray tracing in the 3D veloc-
is to minimize the residuals between the observed and the calcu- ity model obtained from SIMUL2000. The seismicity distribution
lated travel times of neighboring events observed at one particu- and the focal mechanisms are then compared with the surface geo-
lar station. Two closely located events are considered to be logical field survey to better analyze the fault structure of the
neighboring, if the distance between the events is much smaller region.
than the event-station distance and the size of the velocity heter-
ogeneity (Waldhauser and Ellsworth, 2000). High-resolution rela-
2.3. Structural geology observation
tive hypocenters will be obtained when the residuals are
minimized. The conjugate least squares inversion method is used
Fault kinematic results from Tarutung rock outcrops by
in this technique.
Nukman and Moeck (2013) are used to constrain the interpretation
We used the cross-correlation of waveforms in the relocation
of the fault system in the area. The fault kinematics are derived
procedure. The delineation of the resulting earthquake distribution
from a striations and subsidiary fractures analysis (Nukman and
seems to form a micro-fault system around the basin. We then
Moeck, 2013). The measured trend of fractures, veins, and
clustered the earthquakes into 10 groups based on this lineation
lineaments of thermal manifestations are also used in the fracture
(Fig. 3) indicated by different colors. The clustering was carried
analysis by comparing their fracture attitude to the main fault
out in space.
orientation (Nukman and Moeck, 2013). Most of the structural
The maximum distance between an earthquake pair in each
measurements are taken from rock unit of Miocene to Recent
cluster was set 1 km. This configuration is plausible since input
age.
for the HYPODD was from the result of a 3  3 km grid size
SIMUL2000 routine.
Using the double difference relocation, the residual between the 3. Results
observed and the predicted arrival times for pairs of earthquakes
recorded by a similar receiver is evaluated. The velocity model From the HYPODD relocation we found, that after 16 iterations
used in HYPODD relocation is the one obtained from VELEST which the average RMS of waveform cross-correlation and earthquake
was also used for the initial model for the 3D tomographic catalogue are 0.33 ms and 0.47 ms, respectively. The average
inversion (Muksin et al., 2013a). We used 46,904 and 3191 arrival change in the hypocenter is 10.8 m and the average shift of cen-
differences obtained from catalogue data and cross-correlation troids is 167 m. The largest change is contributed to the relocation
analysis, respectively. The neighbor threshold used was 10 of the first cluster which contains the largest number of earth-
links, with stations no more than 80 km from the event pairs. quakes. The use of the double difference relocation reduced the
We constrained the arrival time differences obtained from number of events from 809 to 735 because the outlier events were
excluded. The summary of the relative relocation statistics is pro-
vided in Table 1.
98.9˚ 98.95˚ 99˚ 99.05˚ Fig. 4 shows the hypocenter location resulting from the 3-D
2.1˚
SIMUL2000 inversion (SIMUL2000) and the relative relocation (HYPODD). In
general, the use of the relative relocation does not significantly
change the hypocenter location. However, it significantly sharpens
the lineation of the earthquakes along the main SFS and along the
1 9 secondary fault system around the Tarutung Basin.
2.05˚ 6
Given the ray azimuth and takeoff angles derived from a 3D
4
inversion, each focal mechanism is calculated for different trials
indicated by black thin lines on each fault plane solution in
Fig. 5a. A grid search procedure is performed to find fault plane
5 3 solutions which explain the data. Following Hardebeck and
2˚ Shearer (2002) we perform 50 trials with a 5° interval. The
preferred solution represents the most probable mechanism of
10 7 all acceptable mechanisms. Fig. 5a shows some examples of the
8
focal mechanisms with different quality defined in Table 2. After
removing outliers indicated by black dots on the white areas and
1.95˚ white dots on the black areas in Fig. 5a the average of acceptable
solutions is chosen as the preferred mechanism. In the procedure
of the focal mechanism calculation, the solution with the largest
2
difference from the average is excluded and the new average is
calculated until all remaining mechanisms are within 30° of the
1.9˚ average.
The mechanism quality criteria shown in Table 2 are similar
Fig. 3. The earthquakes were clustered based on a fault-like seismicity distribution to the criteria used by Hardebeck and Shearer (2002). We
obtained from the 3D inversion. The earthquakes are then relocated by using double
difference technique. Different colors represent different group of events. (For
assume that the focal mechanisms with quality B (31 events),
interpretation of the references to color in this figure legend, the reader is referred C (23 events) and D (376 events) are acceptable. From 725
to the web version of this article.) events, only 430 events have acceptable focal mechanisms which
 U. Muksin et al. / Journal of Asian Earth Sciences 96 (2014) 123–131 127

Table 1
Statistics of hypoDD relocation of 10 main clusters after 16 iterations. CC refers to cross-correlation and CT means catalogue.

Cluster Number of events Events mean shift CC–RMS (ms) CT–RMS (ms)
x (m) y (m) z (m) t (ms)
1 188 33.5 25.5 9.9 0.1 0.596 0.486
2 151 2.5 3.2 16.7 0.6 0.201 0.471
3 76 7.7 7.2 7.9 0.1 0.455 0.533
4 58 1.8 0.9 14.7 0.4 0.347 0.440
5 58 4.6 5.7 1.3 0.0 0.239 0.472
6 52 4.3 3.6 1.2 0.1 0.405 0.498
7 39 1.1 1.5 11.5 0.3 0.310 0.432
8 39 0.2 1.8 4.1 0.1 0.315 0.470
9 34 1.9 1.6 4.7 0.4 0.322 0.550
10 21 0.9 1.0 1.8 0.2 0.152 0.424

98.9˚ 98.95˚ 99˚ 99.05˚ 98.9˚ 98.95˚ 99˚ 99.05˚

2.1˚
SIMUL2000 HYPODD

2.05˚

2˚

1.95˚

1.9˚
km km
2 4 6 8 10 12 2 4 6 8 10 12

Fig. 4. The comparison between the hypocenters obtained from SIMUL2000 (LEFT) and HYPODD relocation (RIGHT). Color indicates earthquake depths. (For interpretation of
the references to color in this figure legend, the reader is referred to the web version of this article.)

are located at latitude larger than 1.85°. Because of poor station with the strike-slip fault of the main SFS. The seismicity distribu-
coverage, earthquakes located in Sarulla (southern part of Taru- tion also confirms that the Tarutung Basin is a Pull-Apart Basin.
tung) have low quality focal mechanisms (E and F criteria). The The two active flanks (the eastern and the western) of the SFS
majority of the earthquakes along the main fault is right lateral are delineated by the seismicity. The events along these two flanks
strike-slip associated with the NW–SE striking main fault (Figs. 4 are mostly strike-slip similar to the SFS. The western flank of the
and 5b). Along the secondary fault system within the Tarutung fault seems to be more active than the eastern flank.
Basin the earthquakes are more complex containing NW–SE The earthquake distribution in and around the Tarutung Basin
strike-slip, NS strike-slip and (oblique) normal mechanisms reflects the geometry of the basin. The boundary line of the basin
(Fig. 6). in Fig. 6 is obtained from Nukman and Moeck (2013). A large num-
ber of earthquakes occurred at the northwestern and southeastern
corners of the basin. The normal fault system at the southeastern
4. Discussion corner of the basin (Fig. 6 in Nukman and Moeck (2013)) is also
reflected by the seismicity pattern. The high seismicity in the cor-
4.1. Fault system ner of the basin is associated with the fault bend. A Pull-Apart
Basin along a strike-slip fault normally starts to develop at the fault
4.1.1. The seismicity along Sumatran Fault and the basin margin bend (e.g. Woodcock and Fischer, 1986). The large number of the
We focus the analysis of the fault structure on the area sur- earthquakes at the fault bends (at both corners) indicates that
rounding the Tarutung Basin. Since the stations along the Sarulla the basin is very active and the basin is still developing.
Basin (the southern part of the network) were only located along
the fault, we could not derive good quality focal mechanisms for
the events in the south. As shown in Fig. 6, in the south of Tarutung 4.1.2. Extensional zones
(outside of the basin), the earthquakes are distributed along the In addition to the boundary faults, we notice large number of
main SFS. This earthquake cluster is mostly strike-slip associated earthquakes related to a set of secondary faults. The secondary
128 U. Muksin et al. / Journal of Asian Earth Sciences 96 (2014) 123–131

extensional faults correlates with the maximum principal stress

direction. The almost N–S compressional stress would be induced
11 521 444 11 811 21 5 11 925 1530 11 929 1243
by the nearly N–S Indo-Australian plate motion. The extensional
faults form an en-echelon pattern indicated by the NW–SE dextral
B strike-slip earthquake along the northern extensional zones.

11 519 1235 11 521 543 111112 2 5 12 219 444 4.1.3. The conceptual model
In order to study the fault system at depth we rotate the seis-
C micity, the faults, and the basin geometry so that the secondary
fault system (the extensional zones) is perpendicular to the X-axis
as shown in Fig. 7a. All earthquakes are then projected onto the Y
11 518 1658 11 519 1252 11 614 1231 11 7 4 946 axis showing the distribution of the earthquakes at depth (Fig. 7b).
We deduce the vertical fault system from the depth to the surface
D based on the seismicity pattern. The derived fault system is indi-
(a) cated by the thick transparent light blue lines overlaying the seis-
micity. In Fig. 7b we exclude the seismicity along the main SFS and
along the basin margins in order to analyze the vertical pattern of
the seismicity within the Tarutung Basin. The earthquakes did not
NORTH occur at shallow depth, most likely because the basin is filled with
unconsolidated material until a depth of 2.5 km, as indicated in the
Vp images (Muksin et al., 2013a).
The secondary fault system appears to be very steep as also
found by the geological studies (Nukman and Moeck, 2013). The
vertical seismicity pattern also correlates with the location of
the extensional zones at the surface as observed by the geological
mapping. The black squares at 0 km in Fig. 7b indicate the loca-
tion of the fault zones observed at the surface. We determine
WEST EAST
the location of the other extensional zones at the surface (blue
squares at 0 km in Fig. 7b) according to the pattern of the
seismicity.
From the seismicity pattern and the focal mechanisms we pro-
pose a simplified conceptual model of the fault system in the Taru-
tung Basin as shown in Fig. 8. The layers in Fig. 8 are taken from
seismic velocity derived from the Vp and Vp/Vs tomography study
(Muksin et al., 2013a – Fig. 15). Different layers represented by the
SOUTH 0.2
variation of seismic velocities are caused mostly by the changes in
rock types. The map view of the seismicity shows imbricate fault
(b) Strike direction arrays in strike-slip systems splaying on the Tarutung Pull-Apart
Basin. This indicates the presence of almost symmetrically exten-
Fig. 5. The results of focal mechanism analysis. (a) Examples of preferred solutions
sional duplexes occurring at a releasing bend between 1.96° and
of focal mechanisms characterized as B–D defined in Table 2. White and black dots 2.07° N (Fig. 6). Normal or oblique normal faults should accompany
represent down and up polarities, respectively. The thin lines show 10 different the extensional duplexes to accommodate the extension caused by
possible solutions with higher RMS values. (b) The total strike direction of all focal the NW–SE dextral strike-slip main SFS. The conceptual model
mechanisms.
(Fig. 8) shows dip-slip faults resulting in the extensional duplexes
in the cross-section view. In a fault-perpendicular cross-section,
fault system associated with the extensional zones from Nukman this structure would be recognized as a negative flower structure.
and Moeck (2013) is also indicated by the almost north–south Nukman and Moeck (2013) suggested that several parallel WNW–
direction strike-slip and (oblique) normal fault earthquakes ESE striking normal faults in the east of the Tarutung Basin indicate
(Fig. 6). Some extensional zones within the basin can be observed the presence of negative flower structure (see Fig. 3 in Nukman and
in the topography. The extensional zones derived from the linea- Moeck, 2013). Following Nukman and Moeck (2013), the strike
tion of seismicity are consistent with those obtained from the geo- direction of these parallel normal faults has rotated clockwise to
logical observation. The black and blue nearly N–S striking WNW–ESE.
extensional zones (Fig. 6) represent the fault zones derived from Although the extensional duplex system is not observed clearly
geological observation and seismicity lineation, respectively. The on the surface it is revealed by the accurate hypocenter locations
N–S and NNE–SSW strike of fault plane solutions along the in the small Tarutung Basin. Extensional duplex faults are often

Table 2
The mechanism quality criteria used in the focal mechanism calculation.

Quality Average misfit RMS fault plane uncertainty Station distribution ratio Mechanism probability Number of events
A 60.15 625° P0.5 P0.8 0
B 60.20 635° P0.4 P0.7 31
C 60.30 645° P0.3 P0.6 23
D Maximum azimuthal gap 690°, maximum takeoff angle gap 660° 376
E Maximum azimuthal gap >90°, maximum takeoff angle >60° 26
F Fewer than 8 polarities 269
 U. Muksin et al. / Journal of Asian Earth Sciences 96 (2014) 123–131 129

98.91˚ 98.94˚ 98.97˚ 99˚ 99.03˚ 99.06˚

2.1˚

Pg Eastern flank
Si
2.07˚

Extens
ional z
2.04˚

Hb

ones
2.01˚
Weste
rn flan

1.98˚
k

Ma
rtim
ban
gV

1.95˚
.
South

1.92˚
of Taru

Namora
tung

Si=Siriaria
−I−L.

1.89˚ Hb=Hutabarat
Pg=Panabungan

Elev. (m)
600 800 1000 1200 1400 1600

Fig. 6. The seismicity pattern and selected focal mechanisms plotted onto the topographic map. Topography is ASTER G-DEM (30 m resolution). The basin geometry and the
extensional zones are represented by the black solid lines (after Nukman and Moeck (2013)). The blue lines indicate the extensional zones derived from the seismicity. The
red stars, the white circles, and the blue ellipses respectively represent the hot springs, the seismicity, and the travertines. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)

observed at a bend or a step-over offset of a strike-slip fault by Halldórsson et al. (2013) shows a low value (i.e. R/Ra 1.6–1.7).
(Woodcock and Fischer, 1986). A slightly different type of an exten- This low R/Ra indicates that a heat source in the crust is more dom-
sional duplex system is observed in the vicinity of the bifurcation inant than one from the mantle. Niasari et al. (2012) suggested that
just south of the Tarutung district (Weller et al., 2012). Other duplex the geothermal resource in Tarutung is a result of deep fluid circu-
systems might also be found along the SFS since several fault bends lation according to the interpretation of the magnetotelluric data.
and step-over offsets are visually observed (e.g. Singkarak and Suoh At the eastern junctions of the extensional zones and the east-
Pull-Apart Basins). Similar extensional duplex systems are found in ern margin of the basin we found travertine (indicated by blue
other areas for example along the Liquiñe-Ofqui Fault Zone, southern ellipses in Fig. 6). Nukman and Moeck (2013), Hochstein and
Chile (Cembrano et al., 1996) and York Cliffs strike-slip fault system, Sudarman (1993) suggested that these travertine dykes explain
southern coastal Maine, USA (Swanson, 1990). the characteristics of the geothermal reservoir. The presence of
the travertine dykes in the eastern margin indicates that high
4.2. Fault controlled geothermal system hydrogen-carbonate rich hot fluid is fed from depth (6–10 km) to
the east (Nukman and Moeck, 2013). At depth, the fluid uses the
It has been proposed that the geothermal resource in the Taru- faults as pathways while at shallow depth the fluid is transported
tung area is controlled by the fault system (Muksin et al., 2013a, b; along a more permeable zone in the east (Hochstein and
Nukman and Moeck 2013). This interpretation is based on the high Sudarman, 1993). At the surface, Nukman and Moeck (2013)
Vp/Vs structure and the high seismic attenuation anomalies observed large number of west-dipping normal faults (e.g. around
(Muksin et al., 2013a, b). The ratio of helium isotope as measured Panabungan) and fractures in the east that ease the fluid transport.
130 U. Muksin et al. / Journal of Asian Earth Sciences 96 (2014) 123–131

significantly. The seismicity distribution images well the geometry

of the Tarutung Pull-Apart Basin at depth. Along the western and
eastern margin as well as along the fault south of the Tarutung
Basin the earthquakes are mostly strike-slip associated with the
dextral strike-slip SFS.
The delineation of the events shows the presence of the strike-
slip extensional duplex at the releasing bend within the Tarutung
Pull-Apart Basin which is not clearly observed at the surface. The
extensional duplex system forms a negative flower structure of
the Tarutung Basin. Two duplex faults derived from the seismicity
distribution coincide with extensional zones found by geological
observation. The array of parallel extensional zones accommodates
the almost N–S maximum stress associated with the direction of
the Indo-Australian and Sunda plates convergence. The geothermal
system in the Tarutung area seems to be a fault-controlled system
which uses the fault zones (at depth) as the fluid pathways. We
suggest that the hot fluid is transported to the more dilatational,
more fractured and more permeable region in the east.

Acknowledgments

This work is part of the Project ‘‘Sustainability concepts for

exploitation of geothermal reservoirs in Indonesia-capacity build-
Fig. 7. (a) The rotated map view of the seismicity. (b) The seismicity at all depth ing and methodologies for site deployment’’ funded by the German
projected to X coordinate. The seismicity along the main fault is excluded in Fig. b. Federal Ministry of Education and Research (BMBF) under Grant
The thick transparent light-blue lines indicate the vertical fault system deduced 03G0753A. On the Indonesian side, we wish to thank the Center
from the seismicity pattern. (For interpretation of the references to color in this
for Geological Resources (the Indonesian governmental agency
figure legend, the reader is referred to the web version of this article.)
responsible for geothermal exploration) and the local government
of North Tapanuli (Tarutung), North Sumatra for their support in
the deployment of the instruments. The scientific equipment for
Su the experiment was provided by the Geophysical Instrument Pool
ma
tra Potsdam (GIPP). Particular thanks go also to all colleagues who par-
nF N
au ticipated in the field work. We wish to thank the reviewers A.J. Bar-
lt
E ber and Chris Morley for their comments and suggestions to
improve the paper.
W
S
References

Bellier, O., Sébrier, M., 1994. Relationship between tectonism and volcanism along
Sum
atra the Great Sumatran fault zone deduced by SPOT image analyses.
n Fau Tectonophysics 233, 215–231.
l t Cembrano, J., Hervé, F., Lavenu, A., 1996. The Liquiñe Ofqui fault zone: a long-lived
intra-arc fault system in southern Chile. Tectonophysics 259, 55–66.
Eberhart-Phillips, D., Michael, A.J., 1998. Seismotectonics of the Loma Prieta,
Sumatran Fault Sumatran Fault California, region determined from three-dimensional Vp, Vp/Vs, and seismicity.
J. Geophys. Res. 103, 21099–21120.
Evans, J.R., Eberhart-Phillips, D., Thurber, C., 1994. User’s manual for simulps12 for
imaging VP and Vp/Vs: a derivative of the ‘‘Thurber’’ tomographic inversion
simul3 for local earthquakes and explosions. US Department of the Interior, US
Geological Survey.
Gunderson, R., Dobson, P., Sharp, W., Pudjianto, R., Hasibuan, A., 1995. Geology and
thermal features of the Sarulla contract area, North Sumatra, Indonesia. Proc.
World Geothermal Congr. 2, 687–692.
Halldórsson, S.A., Hilton, D.R., Troll, V.R., Fischer, T.P., 2013. Resolving volatile
sources along the western Sunda arc, Indonesia. Chem. Geol. 339, 263–282.
Hardebeck, J.L., Shearer, P.M., 2002. A new method for determining first-motion
focal mechanisms. Bull. Seismol. Soc. Am. 92, 2264–2276.
Fig. 8. Simplified conceptual model of the fault system in the Tarutung Basin. The Hochstein, M.P., Sudarman, S., 1993. Geothermal resources of Sumatra. Geothermics
layers are derived from seismic velocity layers indicating different rock types 22, 181–200.
(Muksin et al. (2013a)). Kennett, B., Engdahl, E., 1991. Traveltimes for global earthquake location and phase
identification. Geophys. J. Int. 105, 429–465.
Kissling, E., Ellsworth, W.L., Eberhart-Phillips, D., Kradolfer, U., 1994. Initial
reference models in local earthquake tomography. J. Geophys. Res. 99,
5. Conclusion 19635–19646.
Lange, D., Cembrano, J., Rietbrock, A., Haberland, C., Dahm, T., Bataille, K., 2008. First
seismic record for intra-arc strike-slip tectonics along the Liquiñe-Ofqui fault
We derived the fault structure at depth from the seismicity zone at the obliquely convergent plate margin of the southern Andes.
relocation and focal mechanism analysis and supported by geolog- Tectonophysics 455, 14–24.
ical evidence. The application of the relative relocation scheme Lee, W.H.K., Valdes, C., 1985. HYPO71PC: a personal computer version of the
HYPO71 earthquake location program. US Geological Survey.
sharpens the seismicity distribution reflecting the fault lineation McCaffrey, R., 2009. The tectonic framework of the Sumatran subduction zone.
more clearly, although the cluster centroids do not change Annu. Rev. Earth Planet. Sci. 37, 345–366.
 U. Muksin et al. / Journal of Asian Earth Sciences 96 (2014) 123–131 131

Muksin, U., Bauer, K., Haberland, C., 2013a. Seismic Vp and Vp/Vs structure of the Sieh, K., Natawidjaja, D., 2000. Neotectonics of the Sumatran Fault, Indonesia. J.
geothermal area around Tarutung (North Sumatra, Indonesia) derived from Geophys. Res.: Solid Earth (1978–2012) 105, 28295–28326.
local earthquake tomography. J. Volcanol. Geoth. Res. 260, 27–42. Swanson, M.T., 1990. Extensional duplexing in the York Cliffs strike-slip fault
Muksin, U., Haberland, C., Bauer, K., Weber, M., 2013b. Three-dimensional upper system, southern coastal Maine. J. Struct. Geol. 12, 499–512.
crustal structure of the geothermal system in Tarutung (North Sumatra, Tatsumi, Y., 1989. Migration of fluid phases and genesis of basalt magmas in
Indonesia) revealed by seismic attenuation tomography. Geophys. J. Int. 195, subduction zones. J. Geophys. Res. 94, 4697–4704.
2037–2049. Teyssier, C., Tikoff, B., Markley, M., 1995. Oblique plate motion and continental
Muraoka, H., Takahashi, M., Sundhoro, H., Dwipa, S., Soeda, Y., Momita, M., Shimada, tectonics. Geology 23, 447–450.
K., 2010. Geothermal Systems Constrained by the Sumatran Fault and its Thurber, C.H., 1983. Earthquake locations and three-dimensional crustal structure
Pull-Apart Basins in Sumatra, Western Indonesia. World Geothermal Congress, in the Coyote Lake Area, Central California. J. Geophys. Res. 88, 8226–8236.
Bali, Indonesia. Waldhauser, F., Ellsworth, W.L., 2000. A double-difference earthquake location
Niasari, S.W., Muñoz, G., Kholid, M., Suhanto, E., Ritter, O., 2012. Magnetotelluric algorithm: method and application to the northern Hayward Fault, California.
Exploration of the Sipoholon Geothermal Field, Indonesia. Geophys. Res. Abst. Bull. Seismol. Soc. Am. 90, 1353–1368.
14, EGU2012-9405-1. Weller, O., Lange, D., Tilmann, F., Natawidjaja, D., Rietbrock, A., Collings, R., Gregory,
Nippress, S., Rietbrock, A., Heath, A., 2010. Optimized automatic pickers: application L., 2012. The structure of the Sumatran Fault revealed by local seismicity.
to the ANCORP data set. Geophys. J. Int. 181, 911–925. Geophys. Res. Lett. 39, L01306.
Nukman, M., Moeck, I., 2013. Structural controls on a geothermal system in the Woodcock, N.H., Fischer, M., 1986. Strike-slip duplexes. J. Struct. Geol. 8,
Tarutung Basin, north central Sumatra. J. Asian Earth Sci. 74, 86–96. 725–735.