You are on page 1of 8

The application of ambient noise tomography method at Opak River Fault region,

Yogyakarta
Dafi’ Imamul Khair Fatharani Sadat, Tedi Yudistira, and Andri Dian Nugraha

Citation: AIP Conference Proceedings 1987, 020028 (2018); doi: 10.1063/1.5047313


View online: https://doi.org/10.1063/1.5047313
View Table of Contents: http://aip.scitation.org/toc/apc/1987/1
Published by the American Institute of Physics

Articles you may be interested in


Seismic microzonation of Bandung basin from microtremor horizontal-to-vertical spectral ratios (HVSR)
AIP Conference Proceedings 1987, 020004 (2018); 10.1063/1.5047289

Introduction to stress triggers, stress shadows and their implications for seismic hazard in Indonesia
AIP Conference Proceedings 1987, 020033 (2018); 10.1063/1.5047318

A simple approach of enhancing ambient noise array analysis as bedrock depth proxy
AIP Conference Proceedings 1987, 020001 (2018); 10.1063/1.5047286

Preface: International Symposium on Earth Hazard and Disaster Mitigation 2017


AIP Conference Proceedings 1987, 010001 (2018); 10.1063/1.5047285

Aftershock location determination of the 27 May 2006, M 6.4 Yogyakarta earthquake using a non-linear
algorithm: A preliminary results
AIP Conference Proceedings 1987, 020049 (2018); 10.1063/1.5047334

Preliminary result: Earthquake rates analysis from seismic and geodetic strain at Nusa Tenggara and Banda
region, Indonesia
AIP Conference Proceedings 1987, 020013 (2018); 10.1063/1.5047298
The Application of Ambient Noise Tomography Method at
Opak River Fault Region, Yogyakarta
Dafi’ Imamul Khair Fatharani Sadat1,a), Tedi Yudistira2,b), Andri Dian Nugraha 2,c)
1
Geophysical Engineering, Faculty of Mining and Petroleum Engineering, Institute of Technology Bandung
2
Global Geophysics Research Group, Faculty of Mining and Petroleum Engineering, Institute of Technology
Bandung
a)
dafiikfs16@gmail.com
b)
tedi@geoph.itb.ac.id
c)
nugraha@gf.itb.ac.id

Abstract. In seismology, earth’s subsurface imaging is usually use earthquake or explosive as its main sources. Ambient
noise tomography is a relatively new technique to image velocity variations from natural vibrations of Earth. In this
study, ambient noise tomography method is applied to Opak River Fault region, Yogyakarta, Indonesia. The fault
existence was previously assumed as the source of Yogyakarta earthquake on 27 May 2006 which make it interesting to
be further investigated. This research uses temporary seismometer array managed by Geo Forschungs Zentrum (GFZ)
and consists of 12 stations for three months recordings (June, 1st to August, 31st 2006). Interstation Green’s function is
calculated using crosscorrelation process of two records data. Then, group velocity curve is determined by Multiple
Filtering Technique (MFT). The set of interstation group velocity curve are then used to create group velocity maps as a
function of periods. The obtained maps show velocity contrast variations in the research area ranging from 0.4 – 2.3
km/s. The Opak River Fault is clearly indicated by velocity contrast in the western part of the research area. There is also
a velocity contrast that indicated the existence of another fault located ±11 km distance eastward of Opak River Fault.
This fault is possibly associated with the source of Yogyakarta earthquake on 27 May 2006.

Keywords: ambient noise, tomography, Opak River Fault

INTRODUCTION

Yogyakarta Earthquake that happened on 27th May 2006 had caused death for almost six thousand people.
Consultative Group on Indonesia stated that the economic lost by this tragedy reached 3 billion US dollars. People
previously believed that Opak River Fault, which is the major fault in that area, was the cause of the earthquake
because areas that suffered the most severe damage were found alongside the fault. However, many studies over the
year discovered that the source of this deadly earthquake was not generated by Opak River Fault movement. It was
in fact generated by another fault located several kilometers eastward from Opak River Fault referred as blindfault
[1].
Ambient noise tomography is a method used to image velocity variations of the subsurface using ambient
vibration of Earth. We use three months of seismometer recording data around the earthquake area. The main aim of
this research is to produce group velocity variation map of the research area and hopefully locate the existence of the
blindfault.

International Symposium on Earth Hazard and Disaster Mitigation (ISEDM) 2017


AIP Conf. Proc. 1987, 020028-1–020028-7; https://doi.org/10.1063/1.5047313
Published by AIP Publishing. 978-0-7354-1703-8/$30.00

020028-1
DATA AND METHOD

The data used in this study is a daily seismic recordings data that had been recorded for three months starting
from 1st June to 31st August 2006. Figure 1 below is a map showing seismometer distribution around Opak River
Fault.

FIGURE 1. Seismometer distribution. Red triangle is active seismometer and black triangle is inactive seismometer.

Ambient seismic noise interferometry is a method that can be used to image the subsurface’s velocity variation.
Generally, the source used in seismic tomography method is an explosive source. However, ambient noise also
propagates through the Earth, hence it supposed to contain the same information about the subsurface as much as
waves generated from explosive source. The basic principle of this method is that by cross correlating recorded data
from two different stations, one can produce a new signal called Green’s Function. This new signal can then be
processed using Multiple Filtering Technique to extract group velocity. The set of inter-station group velocity curves
then can be used as an initial data to produce group velocity maps as a function of periods.

DATA PROCESSING

We have applied ambient noise tomography to noise data recorded by 12 seismometers around Opak River Fault
area. The procedure on processing ambient noise tomography is as presented in Figure 2. The processing steps in
this research are done by following the general procedure from Bensen et al. [2].

FIGURE 2. Data Processing Procedure

Because we use the noise recording so that every event recorded must be removed or reduced. In Single Data
Preparation, we expect to obtain a signal where dominance of event was reduced significantly by doing remove
mean, remove trend, time domain normalization, and spectral whitening. After the free event signal is obtained,
cross-correlation is performed on all possible station pairs and the results are then stacked over the total available
days for each pair of stations. We manage to get 66 cross-correlations from 12 active stations. The main purpose of
this step is to produce empirical Green’s function which later is used to analyze the dispersive characteristics of the
Rayleigh wave. Cross-correlogram of all inter-station Green’s functions over interstation distance filtered in 2.5 – 12

020028-2
second is then plotted to calculate the group velocity move out by tracing the gradient of the coherent pattern. We
can also use the cross-correlations to analyze the source orientation in this research. Figure 3 shows a cross-
correlation between station KEM and WAN with the arriving energy dominant on the negative part of the cross-
correlation. It indicates that ambient noise travelled dominantly in a direction from station WAN to KEM or from
north to south in general. Nevertheless, detail study still need to perform to identify the dominant source location.

FIGURE 3. Dominantly acausal cross-correlation between three months of ambient noise recorded by station KEM and WAN
(left). Map of seismometer distribution and direction of travelled ambient noise showed by black arrow (right).

From cross-correlation step, we then apply Multiple Filtering Technique (MFT) to extract the group velocity of
the Rayleigh wave. This technique is basically done simply by applying Gaussian filter window with desired central
frequency to the Green’s function so that the travelled energy at each frequency can be calculated. The group
velocity can be determined by tracing the maximum amplitude of the envelope signal [3]. From 66 Green’s function,
we manage to produce 54 considerably clear dispersion curves. The example of dispersion curve extracted from
station KEM-WAN and all of the dispersion curve selection can be seen in Figure 4 below.

FIGURE 4. Dispersion curve extracted from seismometer KEM and WAN (left). Multi dispersion curve plot for 54 data (right).

To construct velocity variation maps, we use the average of group velocity obtained beforehand as the initial
velocity model at each period which can be seen at Table 1. The tomographic process is done using FMST package
by Rawlinson [4]. From our experiment using several different grid size, the most optimum grid size is 3.5 x 4.3 km
with total number of grid is 15 x 15. Another significant parameters are smoothing/damping factor, and number of

020028-3
iteration which are determined all by trial and error with still considering the geological condition of the study area.
Tomogram from tomographic result will have different resolution area coverage depending on the ray distribution.
Generally, the more ray paths that pass through an area the higher the resolution will be. The resolution test is
performed using checkerboard test by creating synthetic data model shaped as checkerboard. The resolution zone
can be seen from the area that forms the checkerboard pattern of the initial model created (Figure 5).

TABLE 1. Initial group velocity model for each periods were determined by averaging all available dispersion curves.

FIGURE 5. Checkerboard model (left) and resolved area inside black dot boundary (right) for period 3 s.

RESULT AND DISCUSSION

The tomographic results are Rayleigh wave velocity variation maps at period range 3 - 7 s shown in Figure 6.
The color represents the absolute velocity value. Red color on the scale bar correlates with the low velocity anomaly
while blue color correlates with high velocity anomaly. The velocity variation values in these periods range between
0.4 – 2.5 km/s. The velocity contrast in the study area can be seen quite clear at period 3 to 5 s. Velocity image for
period 6 and 7 almost homogeneous with velocity around 2.3 km/s.
Tomogram at period 3 and 4 s almost similar on velocity variation, where there is a velocity contrast in the study
area. It can be identified low velocity anomaly in the western and northeast part and high velocity anomaly in the
eastern part of the research area. These anomaly contrast form a southwest-northeast trending that is thought to be
associated with Opak River Fault. The velocity variation values in these periods range between 0.4 – 2.1 km/s. At 5
second period, it appears that the low velocity anomaly in the western part of the study area begin to fade while the
high velocity anomaly widen. The velocity range in this period is 1 – 2.16 km/s. Tomogram at period 6 and 7 second
show relatively high velocity anomaly about 2.3 km/s. This maybe indicates the limit of the fault depth or due to the
lack of data for longer period.

020028-4
FIGURE 6. Rayleigh wave velocity map at period 3 – 7 second

Zulfakriza et al. [5] apply the same method with used data from regional network of Central Java (MERAMEX
network). Their results for this area is broader compare with our results that have much more detail image around
Opak River Fault region. It produced more or less similar result that is a low velocity anomaly in the western part of
the study area and a high velocity anomaly in the eastern part of the area. The research area is located in the
Southern Mountain Zone dominated by alluvial deposits, volcanic clastic, volcanic rocks, and carbonate rocks [6].
The low velocity anomaly is assumed to be associated with alluvial and volcanic deposit whilst the high velocity
anomaly associated with carbonate rocks of Wonosari Formation. Another conducted research in Opak River Fault
region is done by Walter et al. [7]. He stated that volcanic deposit hold a significant role in amplifying the ground
which caused deformation on the surface. The areas that suffered the most severe damage are located in high PGA
zone indicated by red color in Figure 7. We can see that it coincides with the low velocity anomaly shown in
tomogram.

FIGURE 7. Damage area that correlate with PGA amplification by Walter et al. [7] (left) velocity map at period 3 second
showing low velocity anomaly that correlate with amplification zone (right)

We also can see in the PGA map a constellation of aftershocks located several kilometers eastward Opak River
Fault that suspected to be associated with the earthquake-causing fault. There are many studies that actually
investigated this fault [1, 8, 9]. Figure 8 below shows another velocity contrast located east of the study area. This
velocity contrast is approximately 11 kilometers from Opak River Fault and we interpret it as the existence of the
blind-fault marked by blue dashed line.

020028-5
FIGURE 8. Tomogram at period 3 – 5 second show another velocity contrast located ± 11 km distance eastward of Opak River
Fault

CONCLUSION

We have obtained 66 empirical Green’s function from cross-correlating the vertical components of 66 station
pairs. In general, cross-correlogram shows a relatively clear dispersive wave packet. We have selected 54 clear
dispersion curves and used to create velocity maps of Rayleigh wave at period range 3 – 7 second. The obtained
velocity map shows a clear horizontally velocity contrast that indicated the existence of the major fault in the
research area, that is Opak River Fault. There is also another velocity contrast emerged in the resulted tomogram
that assumed to be associated with the Yogyakarta earthquake-causing fault located ±11 kilometer distance eastward
of Opak River Fault.

ACKNOWLEDGEMENTS

The authors would like to acknowledge Geo Forschungs Zentrum (GFZ) for the data waveform used in this
research and also Near Surface Geophysical Laboratory of Institut Teknologi Bandung which have fully supported
this research.

REFERENCES

[1] Tsuji, T., Yamamoto, K., Matsuoka, T., Yamada, Y., Onishi, K., Bahar, A., Meilano, I., and Abidin, H. Z.,
“Earthquake fault of the 26 May 2006 Yogyakarta earthquake observed by SAR interferometry,” Earth,
Planets and Space, 61(7), 29-32 (2009).
[2] Bensen, G. D., Ritzwoller, M. H., Barmin, M. P., Levshin, A. L., Lin, F., Moschetti, M. P., Shapiro, N. M., and
Yang, Y., “Processing seismic ambient noise data to obtain reliable broad-band surface wave dispersion
measurements,” Geophysical Journal International, 169(3), 1239-1260 (2007).
[3] Yudistira, T., “The crustal structure beneath The Netherland inferred from ambient seismic noise,” Ph.D.
thesis, Utrecht University, (2015).
[4] Rawlinson, N., “FMST: Fast Marching Surface Tomography package,” Research School of Earth Sciences,
Australian National University, (2005).
[5] Z Zulfakriza, Erdinc Saygin, PR Cummins, Sri Widiyantoro, Andri Dian Nugraha, B-G Lühr, T Bodin, Upper
crustal structure of central Java, Indonesia, from transdimensional seismic ambient noise tomography,
Geophysical Journal International, 197, pp. 630-635
[6] Smyth, H. R., Hall, R., and Nichols, G. J., “Cenozoic volcanic arc history of East Java, Indonesia: The
stratigraphic record of eruptions on an active continental margin,” Geological Society of America Special
Papers, 436(10), 199-222 (2008).
[7] Walter, T. R., Wang, R., Luehr, B. G., Wassermann, J., Behr, Y., Parolai, S., Anggraini, A., Gunther, E.,
Sobiesiak, M., Grosser, H., Wetzel, H. U., Milkereit, C., Sri Brotopuspito, P. J. K., Harjadi, P., and Zschau, J.,

020028-6
“The 26 May 2006 magnitude 6.4 Yogyakarta earthqauke south of Mt. Merapi volcano: Did lahar deposits
amplify ground shaking and thus lead to the disaster?,” Geochemistry, Geophysics, Geosystems, 9(5), (2008).
[8] Abidin, H. Z., Andreas, H., Meilano, I., Gamal, M., Gumilar, I., and Abdullah, C. I., “Deformasi koseismik
dan pasca seismik gempa Yogyakarta 2006 dari Hasil Survey GPS,” Jurnal Geologi Indonesia, 4, 275-284
(2009).
[9] Kristyawan, S., “Identifikasi sesar aktif dan mekanisme sumber gempabumi di Yogyakarta,” Master thesis,
Institut Teknologi Bandung, (2017).

020028-7

You might also like