Banda Arc Experiment—Transitions in the

Banda Arc-Australian Continental Collision

by Meghan S. Miller, Leland J. O’Driscoll, Nova
Roosmawati, Cooper W. Harris, Robert W. Porritt,
Sri Widiyantoro, Luis Teofilo da Costa, Eugenio
Soares, Thorsten W. Becker, and A. Joshua West

ABSTRACT
Eastern Indonesia is one of the least well-understood geological the subduction-related collision based upon seismic images
domains on Earth, yet the region is one of the few places which created by our broadband experiment. This highly seismogenic
is currently undergoing arc-continent collision. The Banda arc subduction zone setting and continental collision has produced
seismic experiment was designed to unravel some of the com- a very complex geologic history but is an ideal locale to study the
plex dynamics of convergent tectonics. This passive source transition from subduction to arc-continent collision.
experiment is composed of 30 broadband instruments that One observation of the transition from oceanic subduction
extend across the Nusa Tenggara Timur region of Indonesia to continental collision is through lithospheric deformation, as
and across Timor-Leste. This along-strike deployment allows reflected in seismicity (Fig. 1a). Wadati–Benioff zones can be
for seismic imaging and assessment of the spatiotemporal evo- traced down to below 600 km beneath much of the arc (Card-
lution of the collision of oceanic to continental lithosphere of well and Isacks, 1978; Das, 2004), but there are interesting
the Indo-Australian plate with the active volcanic arc. The ex- spatial gaps in shallow and intermediate depth seismicity such as
periment has been continuously recording broadband seismic beneath the island of Wetar between ∼70 and 350 km and be-
data since early 2014 and will continue through the end of neath the island of Timor (Fig. 1b). This lack of seismicity
2016, and the data will be archived at the Incorporated Re- beneath the island of Wetar correlates with the extinct volcanic
search Institutions for Seismology Data Management Center section of the arc and was termed the Wetar seismic gap (WSG;
under network code YS and will be available at the end of 2018. Sandiford, 2008; Ely and Sandiford, 2010). Up-dip from the
WSG, there are few shallow earthquakes beneath Timor in com-
INTRODUCTION parison with the rest of the region. These patterns have been
interpreted as being related to the collision and subsequent
In eastern Indonesia, the northern edge of the Australian conti- changes in lithospheric structure (Sandiford, 2008), and specifi-
nent is colliding with the active volcanic arc at ∼7:5 cm=yr (Bock cally to tearing of the oceanic slab away from continental litho-
et al., 2003), effectively capturing the initiation and continuation sphere. Further evidence for such tearing comes from the
of convergent orogenesis. To the west of the Banda arc seismic distribution of moment release within the slab (McCaffrey,
array, the oceanic lithosphere of the Indo-Australian plate is sub- 1988; Das, 2004; Sandiford, 2008; Ely and Sandiford, 2010),
ducting beneath the Eurasian plate, creating a long volcanic arc which indicates that the slab is deforming heterogeneously.
that extends from Sumatra to Flores Island (Fig. 1a). There, the Seismic tomography has been used extensively as a tool to
Indo-Australian plate composition changes to oceanic plateau investigate the dynamics of the study area (Widiyantoro and
(Scott plateau) and then to continental lithosphere of Australia van der Hilst, 1997; Fichtner et al., 2010; Spakman and Hall,
as it collides with the Banda arc, creating an uplifted forearc, and 2010; Widiyantoro et al., 2011), but due to the limited station
potentially, subducting continental lithosphere (Hamilton, 1979; coverage in the past, the resolution of the mantle and litho-
McCaffrey et al., 1991; Audley-Charles, 2004; Harris, 2011). The sphere is variable and in some places quite poor, especially
amount and extent of continental subduction as well as the tim- at the transition between oceanic and continental subduction
ing and rate of the continental collision are variable along the arc. in the uppermost mantle and lithosphere, which is fundamen-
The along-strike variation in the structure and composition of tal to our understanding of the collisional processes. Nonethe-
lithosphere allows for assessing the spatiotemporal evolution of less, a few of the prior published models suggest that slab

doi: 10.1785/0220160124 Seismological Research Letters Volume 87, Number 6 November/December 2016 1
(a) (b)

▴ Figure 1. (a) Generalized tectonic setting of the Banda arc with active volcanoes plotted in red in the online version, seismicity (Das,
2004), and the Banda arc (YS) network shown by inverted white triangles. The dotted line box shows the location of the seismic array map
in (b) and the inset shows the location of the map on Earth. (b) Map of the YS network of broadband seismometers (2014–2016). Light
colored (yellow in online version) inverted triangles are instruments deployed in Timor-Leste and dark colored (red in online version)
inverted triangles are instruments deployed in Nusa Tenggara Timur region of Indonesia. The stations labeled with black font indicate
stations with STS-2/Q330 instrumentation and those with white font indicate stations with Trillium 120PA/Taurus instrumentation. The color
version of this figure is available only in the electronic edition.

tearing may occur (Widiyantoro and van der Hilst, 1997) and of the lithospheric structure and subducted slab morphology
that there may be a significant amount of continental based on broadband seismic data recorded at this array.
subduction beneath the active arc (Fichtner et al., 2010).
Even with these prior studies, many questions remain unan- INSTRUMENT DEPLOYMENT AND DETAILS
swered about the transition from subduction to arc-continental
collision, and our broadband passive source experiment thus We completed a temporary installation of 30 broadband
offers a rare glimpse into a fundamental process that has shaped seismometers during 2014, as part of the National Science Foun-
dation (NSF) funded Banda arc project; these instruments will
Earth’s evolution over geologic time. This deployment will pro-
remain in place until the end of 2016 (Fig. 1b). The first eight
vide new data to compare with the existing constraints on the instruments were deployed in March 2014 and have been con-
degree of continental subduction (e.g., location and chemistry of tinuously recording ground motion, spanning the geographic
active and recent volcanism, mapped thrust sheets, dating of range of Timor-Leste. The logistics and fieldwork was completed
uplifted coral terraces, Global Positioning System, and seismicity) in collaboration with scientists at the Instituto Petróleo e Geo-
and will make it possible to produce new high-resolution images logia of Timor-Leste. These eight instruments owned by the

▴ Figure 2. Photographs of example field sites and instrumentation. (a) Nanometrics instrumentation at TL05 in the village of Same,
Timor-Leste. (b) STS-2/Q330 instrumentation at station LEGO in Lelogama, West Timor, Indonesia. The color version of this figure is avail-
able only in the electronic edition.

2 Seismological Research Letters Volume 87, Number 6 November/December 2016

(a) (b)

db rel 1 m2/s4/Hz
db rel 1 m2/s4/Hz

ALRB - Alor Island LEGO - Lelogama, Timor Island, NTT

High / Low
Noise model High / Low Noise model
90th percentile 90th percentile bounds
bounds

(c)
db rel 1 m2/s4/Hz

TL05 - Same Village, Timor Leste

High / Low Noise model

90th percentile bounds

▴ Figure 3. Probability density function (PDF) of power spectral densities (PSDs) for stations (a) ALRB, (b) LEGO, and (c) TL05. LEGO is
equipped with a STS-2 sensor, and ALRB and TL05 are equipped with Trillium 120PA sensors. High- and low-noise models are those of
Peterson (1993). The color version of this figure is available only in the electronic edition.

University of Southern California (USC) are Nanometrics Tril- plastic drums with a water-tight ring that purchased in the
lium 120PA with Taurus digitizers. The additional 22 broadband United States and shipped with instruments (Fig. 2). This style
seismometers in the Nusa Tenggara Timur region of Indonesia of vault also has a separate water-tight action packer or trans-
were deployed in September–November 2014 in collaboration port case that stores the electronics (data acquisition systems
with colleagues at the Institut Teknologi Bandung and Badan [DASs]) and battery that remains on the surface. The seismic
Meteorologi, Klimatologi, dan Geofisika (Fig. 1b). These 22 in- stations are powered by lead-acid car batteries purchased in
struments are composed of 7 additional Trillium 120PAs with country and are charged with two 36 W solar panels for
Taurus digitizers (owned by USC) and an additional 15 Program the Nanometrics stations (Fig. 2a) and a 65 W solar panel
for Array Seismic Studies of the Continental Lithosphere (PASS- for the STS-2 stations (Fig. 2b). Because of the low-latitude
CAL) instruments that consist of Streckeisen STS-2 sensors and location of the experiment, the 65 W panels were installed such
Kinemetrics Q330 dataloggers. All the stations are recording con- that they also provided shade for the DAS box.
tinuous time series data for three components at 50 samples=s. The stations were sited and installed with the help from our
The station design was based upon prior experience by the local collaborators and were often placed near municipal build-
team and consists of a buried sensor vault made out of large ings or buildings owned by friends and family of our local col-

Seismological Research Letters Volume 87, Number 6 November/December 2016 3

 (a)

(b)
Rayleigh
TL08 P PP, PcP sP ScP S
TL06
TL04
MTNG
TL07
TL03
ALRB
TL05

4400
TL02
HADA
SINA BAOP
TL1B

Distance (km)
BKOR
MALK
NOMU

LEGO
ENDE
ONIX
SARA

4600
PAGL

RENG

NAPU PPLA
MELO
GREG
SAVU
BULE KPAR

400 600 800 1000

Time (seconds since origin)

▴ Figure 4. (a) Plot of M w 5.5 and greater earthquakes recorded by the Banda arc array through 15 June 2016. (b) Record section of
band-pass-filtered (10–100 s) waveforms for the 30 May 2015 M w 7.8 Chichi-Jima ∼664-km-deep earthquake recorded on the Banda arc
network. Figure (B) made with SAC. The color version of this figure is available only in the electronic edition.

laborators to more effectively protect the instruments from theft for Seismology (IRIS) MUSTANG tool. The PSDs are com-
or damage (Fig. 2). Because of these limitations in site locations, puted from 17 Hanning tapered windows of the medium sam-
the time series are often noisy, as discussed in the next section, ple rate vertical channel (BHZ) per day, each with 50% overlap
but security has not been an issue during this experiment. with the neighboring windows, and then averaged via Welch’s
method. The PSDs are interpolated via a log 10 interpolation
algorithm and then stacked into probability space for the PDFs.
OVERALL DATA QUALITY AND AVAILABILITY
Station ALRB shows a relatively compact PDF, with signal just
above the low-noise model at the primary and secondary
The stations are serviced approximately every six months at the
end (March) and beginning (September) of the wet season. Be- microseismic peaks (∼8 and 16 s), but a long-period signal well
cause of the remote locations and poor road conditions, many of above the low-noise model at greater than ∼25 s period.
the stations would not be accessible at other times of the year. Stations LEGO and TL05 on the island of Timor show less
Examples of the background noise signal from three differ- compact PDFs with the 90th percentile values spanning a large
ent stations are shown in Figure 3: station ALRB on the rel- range. However, the largest probability peaks are near the low-
atively small island of Alor; and two stations on the island of noise model at the microseismic peaks and at longer periods,
Timor, LEGO and TL05. These plots show probability density suggesting relatively quiet long-period recording, which is rare
functions (PDFs) of power spectral densities (PSDs) of the sta- for island stations. Signals below the low-noise model are
tions from installation until 17 June 2016. The computations unlikely real, but may be due to recording errors, and signals
are slightly modified from Peterson (1993) and McNamara above the high-noise model may either be recording errors or
and Buland (2004) and Incorporated Research Institutions large signals from teleseismic earthquakes. Although some of

4 Seismological Research Letters Volume 87, Number 6 November/December 2016

the stations are noisy, the stations were all deployed in secure raphy has been utilized to prepare a shear-velocity model for the
locations allowing for overall high return on the data and clear upper 50 km (Porritt et al., 2016). From this model, we report low
teleseismic and local earthquake signals. However, one of the shear-velocity anomalies associated with sedimentary units from
first Timor-Leste stations installed (TL01) in Oecussi was dis- the incoming Australian plate vertically offset by the high shear-
covered to be particularly noisy after the first service run in velocity backstop of the Banda terrane (Fig. 5). We also observe
October 2015. The station was moved to another nearby that high shear velocities of Sumba Island are contiguous through
location TL1B in March 2015 a few kilometers away, which the Savu Sea, but are underlain by a low velocity associated with
reduced the amount of cultural noise. the top of the subducting Scott plateau. Finally, we suggest that
A few issues affected data availability for three stations dur- velocity offsets in the active inner arc are associated with the
ing the deployment. One of the Trillium 120PA sensors in lateral offset of oceanic subduction and continental collision.
Timor-Leste (station TL07) was found to have an audible
2–3 Hz pinging sound during recentering as well as observing
an inability of the sensor to stabilize (i.e., to stay centered) during (a)

Relative velocity (%)

the first service run in September 2014. It was shipped back to °

Latitude (°S)
be repaired by Nanometrics and was then reinstalled in the same
location in March 2015, which resulted in a ∼7-month data gap.
A wildfire occurred on the island of Sumba on 3 October 2015
°
that burned down the site of one of the PASSCAL instruments
installed in Tambolaka, Sumba Barat Daya (station BULE). Only
the STS-2 sensor survived the fire. The last data from this station ° ° ° °
was collected on 6 September 2015, which provided about a year (b)

Relative velocity (%)

°
worth of data in total. One of the STS-2 sensors, at station

Latitude (°S)
LMBR , had continuous issues with all channels beginning
mid-April 2015 just after the first service run. Problems incurred
include monsoonal flooding and accidental severing of cables °
after this service. After determining that the masses in the sensor
were not responding to any tests and replacing the cables during ° ° ° °
the next service run in September 2015, the sensor was replaced (c)

Relative velocity (%)

with the STS-2 from BULE in March 2016, but was still unable °
Latitude (°S)

to center and is still only recording digital noise.

°
INITIAL OBSERVATIONS

The deployment (as of June 2016) has already recorded over ° ° ° °

600 local earthquakes with magnitude greater than M w 4.5 in (d)

Relative velocity (%)

the Banda arc region and has also recorded ∼1000 teleseismic °
Latitude (°S)

earthquakes greater than 5.5 in the ∼2 years the experiment has

been active (Fig. 4a). An example record section of one of these Savu Sea
teleseisms, the 30 May 2015 M w 7.8 Chichi-Jima, ∼664-km- °
deep earthquake, is shown in Figure 4b. The large number of
local events will provide us with a wealth of data to be used for ° ° ° °
many types of analyses such as waveform analysis (Miller et al., (e)
2016), local shear-wave splitting (Harris et al., 2015), and im- ° Relative velocity (%)
Latitude (°S)

portantly a new, regional earthquake catalog. The teleseismic

data will be used for many types of structural seismology analy-
ses that will provide new high-resolution images and inference °
into structure and evolution of the crust and mantle. The ar- Scott Plateau
chived data will become openly available in 2 years (late 2018)
° ° ° °
via the IRIS Data Management Center (DMC) after the deploy- Longitude (°E)
ment has finished recording data as outlined by NSF–Earth Sci-
ences Division guidelines. ▴ Figure 5. Maps of relative shear velocity (reference velocity la-
beled in upper left) at depth (labeled in lower right corner) from an
INITIAL RESULTS inversion of ambient noise-derived phase velocity maps (repro-
duced from Porritt et al., 2016). Identification of the Scott plateau
As of this article, three publications are being prepared or are in is inferred from bathymetry and seismic velocity at depth. The color
press with the data collected so far. Ambient seismic noise tomog- version of this figure is available only in the electronic edition.

Seismological Research Letters Volume 87, Number 6 November/December 2016 5

 Longitude (°E)
115° 120° 125° 130°

Inferred Slab Su
ur fa
fac
ce
Conntour Lines
−5°

100 km
200 km
300
500

400

00 km
60
70

00 km

0 km

0 km
00 km
Latitude (°S)

−10° 1 sec. ∂t
0° 2 sec.
Station
90°
angle ϕ

▴ Figure 6. Splitting measurements for local earthquakes that occurred during the first half of the deployment (2014–2015) plotted as bars
at the station locations. Splitting time delay (δt ) scales to bar length and bar orientation is equal to the inferred fast axis (φ). Stars
represent earthquake epicenters and are color-coded to match relevant splitting measured. The inferred slab surface contour lines
are based upon seismicity (Das, 2004). The color version of this figure is available only in the electronic edition.

Although dispersion curves from ambient noise tomogra- structure of the downgoing plate, corner flow patterns imparted
phy have been shown to stabilize with 1–2 years worth of data by the collision, or a slab tear as suggested by previous studies (Ely
(Bensen et al., 2007; Porritt et al., 2011), studies requiring a and Sandiford, 2010). However, preliminary results indicate a
distribution of local or teleseismic earthquakes benefit from hav- general trend of arc-normal fast directions that are observed near
ing more data collection time. Therefore, the studies mentioned the middle of the array, and where the collision is most advanced,
in the Initial Observations section above are not complete; how- near Timor, we observe a discontinuous prevalence of roughly
ever, preliminary results analyzing the high-frequency character trench-perpendicular fast axes. This suggests that complex local
of local deep focus direct P waves (Miller et al., 2016) and seis- splitting measurements and discrepancies between local and tele-
mic anisotropy (Harris et al., 2015) are summarized here. seismic phases could be related to the ongoing collision of the
Through waveform analysis of the many deep earthquakes oc- Australian continental lithosphere with the Banda arc.
curring in the region, initial results indicate that the energy as-
sociated with the P-wave first arrival indicates a large amount of SUMMARY
variation between waveforms at different stations along the arc,
both in terms of frequency content and maximum amplitudes. Challenging logistics have been an obstacle for this experiment
These results do not appear to be a result of site/local geology in Nusa Tenggara Timur Indonesia and Timor-Leste, but ini-
effects, but suggest that we may be able to image a change in tial observations and results indicate that the broadband data
subducted lithospheric structure or composition along strike. collected will allow for addressing fundamental science ques-
Additional preliminary results on local and teleseismic tions about arc-continent collision. In addition, many other
shear-wave splitting have been presented at the American Geo- related and unrelated questions about deep Earth structure,
physical Union Fall 2015 meeting (Harris et al., 2015). These seismic hazard, and tectonics will be addressed with these data.
very initial observations from just a few shear-wave splitting
analyses of teleseismic core (SK
K S) phases reveal an approx- DATA AND RESOURCES
imately trench-parallel fast axis along strike, which is consistent
with other subduction settings and can be interpreted as mantle The seismograms and preliminary results shown here were col-
flow being deflected around a descending slab (e.g., Long and lected as part of the National Science Foundation (NSF) funded
Becker, 2010). Conversely, our splitting analysis of local (S) Banda arc experiment using 15 Program for Array Seismic
phases illustrates a more complicated distribution of fast axes that Studies of the Continental Lithosphere (PASSCAL) broadband in-
varies significantly along strike (Fig. 6). To first order, these mea- struments and 15 broadband instruments owned by University of
surements appear to vary more along the strike of the arc than Southern California (USC). Data will be available from the Incor-
the core phase splits. This could be related to heterogeneous porated Research Institutions for Seismology Data Management

6 Seismological Research Letters Volume 87, Number 6 November/December 2016

Center (IRIS-DMC) for network YS at the end of 2018 (http:// from shear-wave splitting analysis of teleseismic and local slab events,
ds.iris.edu/gmap/YS?timewindow=2014‑2016, last accessed June AGU Annual Fall Meeting, San Francisco, California, DI21A–2596.
Long, M. D., and T. W. Becker (2010). Mantle dynamics and seismic
2016). Data on active volcanoes are obtained from the Smithsonian anisotropy, Earth Planet. Sci. Lett. 297, nos. 3/4, 341–354.
Volcano Database (Global Volcanism Program, 2013). McCaffrey, R. (1988). Active tectonics of the eastern Sunda and Banda
arcs, J. Geophys. Res. 93, no. B12, 15,163–15,182.
McCaffrey, R., M. Robert, and G. A. Abers (1991). Orogeny in arc-con-
ACKNOWLEDGMENTS tinent collision: The Banda arc and western New Guinea, Geology
19, no. 6, 563–566.
This work is supported by National Science Foundation (NSF) McNamara, D. E., and R. P. Buland (2004). Ambient noise levels in the
—Tectonics/Geophysics/Office of International Science and continental United States, Bull. Seismol. Soc. Am. 94, no. 4, 1517–1527.
Miller, M. S., A. F. Holt, and D. Sun (2016). Detecting slab structure
Engineering (OISE)—Global Venture Fund Grant EAR- beneath the Banda arc from waveform analysis of deep focus earth-
1250214. Many thanks are due to our colleagues at Instituto quakes, Seismol. Res. Lett. 87, no. 2B, Abstracts of the SSA Annual
Petróleo e Geologia (IPG), Badan Meteorologi, Klimatologi, Meeting, Reno, Nevada.
dan Geofisika (BMKG), and Institut Teknologi Bandung (ITB) Peterson, J. (1993). Observations and modeling of seismic background
as well as personnel at the U.S. Embassy in Jakarta, and Incor- noise, U.S. Geol. Surv. Open-File Rept. 93–322.
Porritt, R. W., R. M. Allen, D. C. Boyarko, and M. R. Brudzinski (2011).
porated Research Institutions for Seismology–Program for Investigation of Cascadia segmentation with ambient noise tomog-
Array Seismic Studies of the Continental Lithosphere (IRIS- raphy, Earth Planet. Sci. Lett. 309, nos. 1/2, 67–76.
PASSCAL) who made the work possible. R. Harris and P. Cum- Porritt, R. W., M. S. Miller, L. J. O’Driscoll, C. W. Harris, N. Roosma-
mins also provided valuable insight into working in the region. wati, and L. Teofilo da Costa (2016). Continent-arc collision in the
Figures were made with Generic Mapping Tools (GMT; Wessel Banda arc imaged by ambient noise tomography, Earth Planet. Sci.
Lett. 449, 246–258, doi: 10.1016.j.epsl.20116.06.011.
and Smith, 1991) and Seismic Analysis Code (SAC). Power Sandiford, M. (2008). Seismic moment release during slab rupture be-
spectral densities (PSDs) and probability density functions neath the Banda Sea, Geophys. J. Int. 174, no. 2, 659–671.
(PDFs) computed with station analysis tools software available Spakman, W., and R. Hall (2010). Surface deformation and slab–mantle
through IRIS Seiscode (https://seiscode.iris.washington. interaction during Banda arc subduction rollback, Nat. Geosci. 3,
edu/projects/station‑analysis‑tools, last accessed June 2016). no. 8, 562–566.
Wessel, P., and W. H. F. Smith (1991). Free software helps map and
display data, Eos Trans. AGU 72, 441 and 445–446.
REFERENCES Widiyantoro, S., and R. van der Hilst (1997). Mantle structure beneath
Indonesia inferred from high-resolution tomographic imaging, Geo-
Audley-Charles, M. G. (2004). Ocean trench blocked and obliterated by phys. J. Int. 130, no. 1, 167–182.
Banda forearc collision with Australian proximal continental slope, Widiyantoro, S., J. D. Pesicek, and C. H. Thurber (2011). Complex struc-
Tectonophysics 389, nos. 1/2, 65–79. ture of the lithospheric slab beneath the Banda arc, eastern Indonesia
Bensen, G. D., M. H. Ritzwoller, M. P. Barmin, A. L. Levshin, F. Lin, M. depicted by a seismic tomographic model, Res. Geophys. 1, no. 1, 1.
P. Moschetti, N. M. Shapiro, and Y. Yang (2007). Processing seismic
ambient noise data to obtain reliable broad-band surface wave Meghan S. Miller
dispersion measurements, Geophys. J. Int. 169, no. 3, 1239–1260. Leland J. O’Driscoll
Bock, Y., L. Prawirodirdjo, J. F. Genrich, C. W. Stevens, R. McCaffrey, C. Nova Roosmawati
Subarya, S. S. O. Puntodewo, and E. Calais (2003). Crustal motion
in Indonesia from Global Positioning System measurements, J. Geo- Cooper W. Harris
phys. Res. 108, no. B8, doi: 10.1029/2001jb000324. Robert W. Porritt
Cardwell, R. K., and B. L. Isacks (1978). Geometry of the subducted litho- Thorsten W. Becker
sphere beneath the Banda Sea in eastern Indonesia from seismicity A. Joshua West
and fault plane solutions, J. Geophys. Res. 83, no. B6, 2825–2838. Department of Earth Sciences
Das, S. (2004). Seismicity gaps and the shape of the seismic zone in the
Banda Sea region from relocated hypocenters, J. Geophys. Res. 109, University of Southern California
no. B12, doi: 10.1029/2004jb003192. 3651 Trousdale Parkway
Ely, K. S., and M. Sandiford (2010). Seismic response to slab rupture and Los Angeles, California 90039 U.S.A.
variation in lithospheric structure beneath the Savu Sea, Indonesia, [email protected]
Tectonophysics 483, nos. 1/2, 112–124.
Fichtner, A., M. De Wit, and M. J. Van Bergen (2010). Subduction of
continental lithosphere in the Banda Sea region: Combining evi- Sri Widiyantoro
dence from full waveform tomography and isotope ratios, Earth Global Geophysics Research Group
Planet. Sci. Lett. 297, nos. 3/4, 405–412. Institut Teknologi Bandung
Global Volcanism Program (2013). Volcanoes of the World, v. 4.5.0, Ven- Jalan Ganesa 10
zke, E. (Editor), Smithsonian Institution, downloaded 14 June Bandung 40132, Indonesia
2016, doi: 10.5479/si.GVP.VOTW4-2013.
Hamilton, W. B. (1979). Tectonics of the Indonesian region, U.S. Geo-
logical Survey Professional Papers, 1078, 345 pp. Luis Teofilo da Costa
Harris, R. (2011). The nature of the Banda arc-continent collision in the Eugenio Soares
Timor region, in Arc-Continent Collision, D. Brown and P. D. Ryan Institute of Petroleum and Geology
(Editors), Vol. 2, Springer, Berlin, Germany, 163–211. 2nd floor, Elemoloi Building, Rua Delta 1 Aimutin
Harris, C. W., M. S. Miller, L. J. O’Driscoll, R. W. Porritt, and N. Roosma-
wati (2015). Mantle dynamics of Australia-Banda collision as inferred Dili, Timor-Leste

Seismological Research Letters Volume 87, Number 6 November/December 2016 7