Pure Appl. Geophys.

Ó 2021 The Author(s), under exclusive licence to Springer Nature Switzerland AG

https://doi.org/10.1007/s00024-021-02781-4 Pure and Applied Geophysics

The Current State of Earthquake Potential on Java Island, Indonesia

SUMANTA PASARI,1 ANDREAN V. H. SIMANJUNTAK,2 ANAND MEHTA,3 NEHA,1 and YOGENDRA SHARMA1

Abstract—Between 2006 and 2020, earthquakes and other 1. Introduction

geohazards on volcano-dotted Java Island have caused about 7000
deaths, and another 1.8 million people were injured, displaced, or
left homeless. In this study, we quantify the current state of The island of Java, the fourth largest island with
earthquake hazard for 29 cities of Java, using seismicity statistics the highest population density in Indonesia, is part of
of a cumulative number of small events (natural times) between
the complex convergence zone between the Eurasian
pairs of large earthquakes. This approach, known as earthquake
nowcasting (Rundle et al., 2016), rests on the key concepts of plate and the Indo-Australian plate (Schluter et al.,
elastic rebound and ergodic dynamics in earthquake fault networks. 2002; Metcalfe, 2011). The seismotectonic setting of
Our analysis of statistical inference shows that the estimated Java Island is illustrated in Fig. 1. This subduction
earthquake potential score (EPS) as on February 18, 2021 corre-
sponding to M C 6.5 events in a 300 km circular area ranges from zone regularly faces the hardship of many natural
43 to 94%, with the scores of Jakarta (43), Surabaya (89), Bandung disasters, including earthquakes, tsunamis, and vol-
(69), Semarang (48), Serang (47), and Yogyakarta (59). This canic eruptions as a result of the N-NE dipping Indo-
means, for example, that Surabaya has progressed significantly in
the regional cycle of large earthquakes, whereas Yogyakarta is
Australian plate beneath the Eurasian plate (Katili,
about midway in its seismic cycle. We observe that a change in 1975; Whittaker et al., 2007; Metcalfe, 2011; Harris
magnitude threshold or geographic area has a consistent impact on & Major, 2017). The plate convergence velocity from
the nowcast scores. These findings not only enable a rapid yet
geodetic studies is * 60 mm year-1 in an almost
meaningful way to rank several cities based on their current
exposure to earthquake hazards, but also empower earthquake perpendicular direction to the Java Trench that attains
scientists and policymakers towards better policymaking, land-use depths of 100–200 km in the southern part and sub-
planning, earthquake insurance, disaster risk mitigation, and social ducts beneath the Java Island until depths of 600 km
awareness with respect to the seismically active island of Java.
in the northern part (Abidin et al., 2009; Hanifa et al.,
Keywords: Earthquake potential, natural times, nowcast 2014; Koulali et al., 2017). Tectonic stresses on the
scores, Java Island, probability models. forearc region offshore and on the land of Java result
in two distinct sources of earthquakes—deep focus to
the north and shallow depth to the south (Tregoning
et al., 1994; Kopp et al., 2006; Abidin et al., 2009).
The extensive tectonic deformation on the island
of Java has led to several large-magnitude earth-
quakes in northwest and central Java along the
subduction zone of the Java Trench (Newcomb &
McCann, 1987; Harris & Major, 2017). A seismicity
1
Department of Mathematics, Birla Institute of Technology
plot of the study region during 1963–2021 is illus-
and Science Pilani, Pilani Campus, Jhunjhunu, Rajasthan 333031, trated in Fig. 2. Two major tsunamigenic earthquakes
India. E-mail: 86.sumanta@gmail.com; sumanta.pasari@pi- of recent times are the 1994 Java earthquake (Mw
lani.bits-pilani.ac.in; choudharyneha002@gmail.com; yogenmaths
7.8) and the 2006 Pangandaran earthquake (Mw 7.7)
2738@gmail.com
2
Meteorological, Climatological and Geophysical Agency (Polet & Kanamori, 2000, Abercrombie et al., 2001;
(BMKG) of Indonesia, Banda Aceh City, Aceh 23111, Indonesia. Fujii & Satake, 2006; Bilek & Engdahl, 2007; Kato
E-mail: andreansimanjuntak@gmail.com et al., 2007). Both earthquakes were generated by a
3
Department of Civil Engineering, Institute of Infrastructure
Technology, Research and Management, Ahmedabad, Gujarat thrusting mechanism at a shallow interface zone near
380026, India. E-mail: anandrajmehta@gmail.com the Java Trench (Ammon et al., 2006; Lay et al.,
 S. Pasari et al. Pure Appl. Geophys.

2011). While the June 3, 1994 Java earthquake and disrupted, but the fear of natural disasters loomed
consequent tsunami killed about 250 people, the July large over the millions of residents as well. To wor-
17, 2006 Pangandaran earthquake greatly destroyed sen the situation, on December 22, 2018, a tsunami
the south coast of Java, causing at least 802 human generated by the Anak Krakatau volcanic eruption
deaths and injuring another 10,000 people (Synolakis struck several coastal areas in Java, Banten, and
et al., 1995; Maramai & Tintin, 1997; Mori et al., adjacent regions. At least 426 people were killed and
2007; Sakamoto & Yamori, 2009). In 2006, several 14,059 were injured (Heidarzadeh, 2020; Ye et al.,
months prior to the Pangandaran event, another 2020; Paris et al., 2020). In total, between 2006 and
moderate yet disastrous earthquake of Mw 6.3 2020, earthquakes and other geohazards on the vol-
occurred on May 27 along a left-lateral strike-slip cano-dotted Java island caused about 7000 deaths,
fault zone located just 25 km away from Yogyakarta, and another 1.8 million people were injured, dis-
the largest city in central Java (Walter et al., 2008; placed, or left homeless (BNPB, 2019).
Tsuji et al., 2009; Diambama et al., 2019). The event Exciting lessons have been learnt from past
caused 5778 deaths, 36,299 injuries, and about 1.5 earthquakes in the region. For instance, the Yogya-
million people left homeless, marking an economic karta event in May 2006 puzzled the scientific
loss of * 3.1 billion USD (Muhari et al., 2007; community over the source fault. Seismic and
Lavigne et al., 2007; Gatignon et al., 2010). With the geodetic studies after the event revealed that neither
twin earthquakes in a year, not only was the socioe- surface deformations nor aftershock distributions
conomic activity of the entire island greatly aligned with the major Opak river fault thought to be

Figure 1
Tectonic settings of Java Island including some information of geological faults and major cities. The background color of the figure in the
lower panel represents the population density (Ciesin, 2005) of Java Island
 The Current State of Earthquake Potential on Java Island, Indonesia

the source of this earthquake (Tsuji et al., 2009). Cipamingkis, Kendeng, Jakarta, Wongsorejo, Mon-
Rather, several unmapped active faults with oblique tong and the Lasem faults that were previously
deformation lying parallel to the Opak river fault assumed to be dormant. This compressional fault
in * 10–20 km eastward hosted this strong earth- zone has a proximity to Jabodetabek (Jakarta-Bogor-
quake (Tsuji et al., 2009; Widjajanti et al., 2020). Depok-Tangerang-Bekasi), the most populous
Similarly, Gunawan and Widiyantoro, (2019), based metropolitan area in Indonesia. As a result, the pre-
on geodetic strain analysis, observed enormous con- sent understanding of active tectonism in the complex
temporary strain accumulation along the Cimandiri, regime of Java island and consequent fault-based

Figure 2
Summary of earthquake data (1963–2021) on which the NTA is developed. Subplots highlight several characteristics of the catalog, including
epicentral distributions on the map, cross-section views of hypocentral depth, magnitude of completeness, and occurrence time of large
earthquakes in the study region
 S. Pasari et al. Pure Appl. Geophys.

earthquake hazard evaluation are inadequate, as (Luginbuhl et al., 2018; Rundle & Donnellan, 2020).
several large earthquakes have occurred unexpectedly It may be emphasized that the idea of earthquake
in regions considered to be of low risk, where no nowcasting which defines the current state of pro-
geological faults have been mapped previously, and gress of a region is conceptually distinct from
no evidence of historical earthquakes is available earthquake forecasting which characterizes future
(BNPB, 2019; Widjajanti et al., 2020). probabilities of earthquake occurrences (Holliday
In this study, we develop a statistical, data-driven, et al., 2016; Rundle et al., 2016, 2018).
natural time analysis (NTA) to examine the contem- Since its inception, the natural time-based earth-
porary state of earthquake hazard at several cities on quake nowcasting method has been implemented for
Java Island, Indonesia. The NTA (Varotsos et al., several global megacities, including San Francisco,
2011) uses seismicity statistics of interspersed small Los Angeles, Chile, Ankara, Tokyo, Taipei, Manila,
event counts (natural times) between pairs of large Jakarta, Dhaka, New Delhi, Kathmandu and Islam-
earthquakes to delineate the current progression of a abad, to determine the current state of their hazard
city in its seismic cycle of large events. The NTA cycle (Rundle et al., 2016, 2018; Pasari & Mehta,
approach in seismology, known as earthquake now- 2018; Pasari 2019b, 2020; Pasari & Sharma, 2020;
casting (Rundle et al., 2016), rests on two Pasari et al., 2019, 2021a, b). Sensitivity analysis of
fundamental concepts: elastic rebound principle and the parameters was also performed. At a few sites of
ergodic dynamics in earthquake fault networks. induced earthquakes, such as the Groningen gas field,
While the notion of earthquake cycle in a fault zone Geysers geothermal field and the fluid-injection areas
stems from the elastic rebound paradigm, the ergod- in Oklahoma, the NTA was employed to study the
icity in earthquake fault systems exhibits spatial time-dependent state of induced seismicity (Lugin-
invariance of ensemble natural time statistics over buhl et al., 2018). While the earthquake nowcasting
longer times (Tiampo et al., 2007; Rundle et al., method, in combination with Shannon information
2012, 2016; Luginbuhl et al., 2018). As a conse- entropy, was extended to compute hazards from
quence, the NTA for the current earthquake hazard global great-earthquake and mega-tsunami sources
evaluation gives rise to several contrasting properties (e.g., Rundle et al., 2019), the applicability of the
to the conventional seismic hazard assessment NTA has also been demonstrated to characterize
methods, such as: (1) the NTA tactfully substitutes earthquake processes by cycles of recharge and dis-
the clock/calendar time by a time domain of discrete charge through some noticeable patterns in the
event counts, (2) the NTA in a given area is unaf- seismicity clustering of aftershock sequences and
fected to the changes of background seismicity rate (a seismic swarms in Southern California (Rundle &
value in the G–R relation), influence of triggered or Donnellan, 2020).
induced seismicity, presence of any dependent events Influenced by the thoughtfulness and robustness
(e.g., foreshocks, aftershock sequences) or any such of the nowcasting approach, the present study pro-
type of heterogeneity in the database, as long as the vides a comprehensive analysis of natural times for
b-value in the region remains largely homogeneous in Java Island to statistically compute the earthquake
time and space (Rundle et al., 2016, 2018), (3) the potential score (EPS) that describes the current state
NTA and consequent nowcasting method consider of progress of a city through its nominal earthquake
earthquake ‘‘cycles’’ arising from a ‘‘fault system’’ to cycle. In the following section, Sect. 2, we define the
characterize the hazard (earthquake) cycle in a study area and discuss some general characteristics of
region, rather than focusing on the large earthquakes the earthquake catalog. Section 3 illustrates the
that repeatedly rupture a given part of a fault, and (4) methodology and results, together with the sensitivity
the NTA through small earthquake proxy data can testing of parameters. In Sect. 4, we provide a
establish an inherent linkage to the earthquake detailed discussion about the critical assumptions,
dynamics characterized by the renewal process of model formulations, and contemporary hazard
stress accumulation and release to determine the assessment for Java Island. The key findings of the
time-dependent state of stress in a defined region present analysis are summarized in Sect. 5.
 The Current State of Earthquake Potential on Java Island, Indonesia

2. Study Area and Earthquake Data province (Hamilton, 1979; Metcalfe, 2011). With
several tectonic features, the Island provides resi-
2.1. Geology and Tectonics dence to around 150 million people of Indonesia
(Fig. 1).
With a typical hot and humid equatorial climate,
Due to the persistent tectonic activity between the
the study area (102°–120° E, 2°–12° S) encompasses
Eurasian and the subducted Indo-Australian plates,
the calc-alkaline volcanic Java Island and its neigh-
Java Island is an assemblage of several active faults
boring regions of the Java Sea to the north, southern
(Katili, 1973; Hamilton, 1979). Figure 1 highlights
part of Sumatra to the west, Bali and Lombak Islands
some of the major and regional faults that are
to the east, and the Indian Ocean to the south (Fig. 1).
adjacent to the large cities. For example, the strike-
The subduction-induced Java Island forms the south-
slip Cimandiri fault, with rake angle * 15° and
ernmost part of the continental Sunda Plate,
dip * 70°, is a major active fault in the western part
overriding the Indo-Australian oceanic plate (Katili,
of Java (Dardji et al., 1994; Abidin et al., 2009). The
1973; Curray et al., 1982; Zahirovic et al., 2014;
SW–NE-stretched Opak river fault, near the special
Simons et al., 2015). Four major provinces, namely
region of Yogyakarta in Central Java, is another
West Java, Central Java, East Java, and Banten, along
widely studied active fault that connects several
with two special regions of Jakarta and Yogyakarta
regional active faults in Banten, Bogor and Tasik-
make up Java Island. While the West Java region is
malaya, Cimandiri active faults in Sukabumi,
located along the transition of frontal subduction
Lembang fault at * 20 km north of Bandung,
beneath Sumatra since Eocene rifting, the East Java
Baribis faults in Subang, and Majalengka faults in
province is situated on the southeastern edge of the
Garut (Meilano, 2012; Supendi et al., 2018). Several
Sundaland craton with Cretaceous to basal Tertiary
E–W-extending faults are situated in Central Java,
melange basement (Hamilton, 1979; Schluter et al.,
including the Cerebon, Tegal-Pekalongan, Semarang-
2002; Metcalfe, 2011). Tectonically, the western part
Demak-Cepu-Kendeng, Pati-Lasem, and Waru-Sur-
of Java is different from the eastern part due to the
ababy faults (Metcalfe, 2011). Even the small
variability in orientation pattern controlled by an
regional faults are observed to have the potential to
oblique dipping of the Indo-Australian plate beneath
generate a strong earthquake. For instance, the
the Sumatra subduction zone and a perpendicular
Lembang fault, with a length of * 24 km, slip
dipping of the same plate beneath Java (Malod et al.,
rate * 6 mm/year, and locking depth of * 3–15
1995). Such characteristics of fault kinematics and
km, is capable of producing a 6.5 magnitude event
displacement patterns are also evident in the crustal
every 400–600 years (Horspool et al., 2014; Afnimar
velocity field that depicts a homogeneous counter-
et al., 2015).
clockwise rotation with a NE–SW convergence
increasing from * 58.3 ± 0.5 mm year-1 in the
western part to * 65 ± 0.4 mm year-1 in the east- 2.2. Earthquake Data
ern part (Malod et al., 1995; Koulali et al., 2017). The
The compressive tectonics of Java Island lead to
Banten region, comprising several structural highs
high seismic activity because of the Oligocene along
and lows, lies in the westernmost part of Java. The
the forearc basin and Miocene on the regional faults
active Seribu fault zone separates Banten Block from
of Java that constitute a tertiary structure with
the neighboring Sunda Strait to the west. In the
complicated deformation patterns (Matthews et al.,
middle of Java Island, the Central Java province
1995; Pal et al., 2003. Susilohadi et al., 2005; Kopp
contains a bathymetric basin between the Java
et al., 2006; Hall, 2009). While some disastrous
volcanic arc and the non-volcanic outer ridge of the
events occur at a shallow depth, several earthquakes
Java Trench on its northern flank (Whittaker et al.,
originate at the deeper portion up to 150 km of the
2007). Filled with clastics of deep marine facies, two
asthenosphere (Fig. 3) (Tregoning et al., 1994; Kopp
Neogene sedimentary basins of Late Oligocene phase
et al., 2006; Abidin et al., 2009; Widiyantoro et al.,
of folding, faulting and volcanism are evident in this
2020).
 S. Pasari et al. Pure Appl. Geophys.

Figure 3
a Three vertical cross-sections (AA’, BB’, CC’) highlight the depth distribution of the considered earthquakes comprising upper-plate shallow
crust, interface, and intraslab events in the study region. Several sources of earthquakes are not a cause for concern in nowcasting analysis that
considers earthquake occurrence as a unified threshold system. b In the left panel, graph of A–A’ profile that perpendicularly crosses the slab
beneath Java shows the ‘‘locking’’ zone in the interface area that is assumed to be a seismic gap and typical consistent seismogenic zone for
large subduction zone earthquakes (Widiyantoro et al., 2020). The red line represents the slab model which marks the subduction system,
whereas the dashed lines show the oceanic and continental moho depths (Kopp et al., 2002, Shulgin et al., 2011, Wölbern & Rümpker, 2016,
Suhardja et al., 2020). In the right panel, graph of B-B’ profile provides similar information with A–A’ profile and shows that the interface
events have dominantly occurred above moho depth in the upper plate crust while the intraslab events have occurred below 50 km depth. In
the bottom panel, graph of C–C’ profile also bears similar information with A–A’ and B–B’ profiles; the interface events are generally
observed to occur above moho depth, whereas the intraslab events are located below 50 km depth. The three profiles clearly show different
seismogenic sources between interface and intraslab events, whereas the crustal events from active faults are quite less in number
 The Current State of Earthquake Potential on Java Island, Indonesia

To develop the natural time statistics in the study ensemble natural time statistics requires the descrip-
region, we compile an instrumental earthquake tion of a large geographic area, small local area,
catalog combining global and regional seismic net- large-magnitude event and small-magnitude event to
works, as available in the public ANSS (Advanced determine the EPS for the region. While the ‘‘large’’
National Seismic System), ISC (International Seis- geographic area usually comprises the entire study
mological Centre), and the national BMKG region, a smaller circular city area embedded in the
(Meteorological, Climatological and Geophysical larger region demarcates the ‘‘local’’ area. Similarly,
Agency of Indonesia) catalog. While the data for while the magnitude of completeness in an instru-
the past decade are taken from BMKG, the earlier mental catalog defines the threshold of ‘‘small’’
data are obtained from the global network coverage. earthquakes ðMr Þ, the ‘‘large’’ earthquake threshold
During February 21, 1963 and February 18, 2021, ðMk Þ is determined based on the extent of possible
there are a total of 11,530 events in the region having damage or destruction to the study area (Rundle et al.,
magnitude 4.0 or higher within the considered 2016, 2018; Pasari 2019b, 2020; Pasari & Sharma
maximum focal depth of 150 km. The data set 2019, 2020). In the present analysis, we consider
contains 31 cycles of large earthquakes ðM  6:5Þ small- and large-magnitude thresholds Mr ¼ 4,
with an average of 355 small ð4:0  M\6:5Þ event Mk ¼ 6:5, and a city radius R ¼ 300 km to calculate
counts. The observed natural times are the key input nowcast scores for 29 cities (population larger than
in the NTA of the present work. A summary of the 1,00,000) on Java island.
data set in terms of epicentral locations on a map, The basic methodology for earthquake nowcast-
cross-section views of hypocentral depth, and mag- ing comprises two key steps. First, the natural time
nitude of completeness graph is provided in Fig. 2. statistics in the larger area are developed using
While the cross-section map describes the depth several reference probability distributions. Then, the
distribution and locations of subduction zone in the same seismicity statistics are applied to a number of
study region, the least-squares regression on the smaller circular city regions to compute their EPS,
Gutenberg-Richter frequency–magnitude linear rela- also known as nowcast score, through the cumulative
tionship suggests a completeness magnitude distribution function evaluated for the current number
threshold of M 4.2 ± 0.1. We also provide a time (since the last strong event) of small event counts in
versus cumulative number of events ðM  3:5Þ graph, the local area (Rundle et al., 2016). To illustrate this
highlighting the occurrence time of large-magnitude idea, let Ai ; i ¼ 1; 2; . . .; k be the k number of selected
earthquakes. It should be emphasized that in a city regions contained in the broader seismic region
nowcasting process, the earthquakes within a depth A. Further, let fX1 ; X2 ; . . .; Xn g represent a random
of 0–150 km comprising upper-plate shallow crust, sample of natural times ð X Þ corresponding to n cycles
interface, and intraslab events (Fig. 3) are collec- of large earthquakes in the region A. On this random
tively treated as a unified threshold system (e.g., sample, a number of potential probability distribu-
Rundle et al., 2016), rather than considering a specific tions are developed to calculate the underlying
earthquake source corresponding to a specific fault in natural time statistics. This step also necessitates
the study region. carrying out parameter estimations and model per-
formance analysis based on the goodness-of-fit
measures. With the known natural time statistics of
3. Methodology and Results the larger area, it is now desirable to compute EPS for
a local area Ai ði ¼ 1; 2; . . .; kÞ as EPSi 
3.1. Formulation of the Earthquake Nowcasting FX ðmi ðtÞÞ ¼ PfX  mi ðtÞg; where F ð:Þ is the cumu-
Method lative distribution corresponding to the best-fit
probability model and mi ðtÞ; i ¼ 1; 2; . . .; k is the
Like traditional fault-based hazard assessment
current count of small events at clock/calendar time t
methods (e.g., Scholz, 2009; Field et al., 2015),
since the time of the last large event in a circular city
evaluation of the current hazard state through
area Ai ði ¼ 1; 2; . . .; kÞ.
 S. Pasari et al. Pure Appl. Geophys.

Table 1
Probability distribution, parameter estimation and model selection results corresponding to the observed natural times of the study region

Distribution Density function ðt [ 0Þ Statistical inference

MLE K–S A–D

1 at
Exponential a e b1 a^ ¼ 340.2903 0.2079 1.9215 (* 0)
1 t at
Gamma CðbÞ ab e a^ ¼ 617.4373 0.1290 0.4281
  2  b^ ¼ 0.5511 (0.3112)
1 ffiffiffiffi
Lognormal p
tb 2p
exp 12 ln ta
b a^ ¼ 4.9226 0.0985 0.3626 (0.4424)
b^ ¼ 1.5445
b b1 ð t b
Weibull t e Þ
a a^ ¼ 287.5976 0.1020 0.3673 (0.4317)
ab
qffiffiffiffiffiffiffi h i b^ ¼ 0.7599
b bðtaÞ2
Inverse Gaussian 2pt3 exp  2a2 t a^ ¼ 340.2903 0.2154 2.7686 (* 0)
b^ ¼ 40.8992
t b
Inverse Weibull bab tb1 eð Þ a a^ ¼ 61.5610 0.1457 0.9306 (0.002)
b^ ¼ 0.6084
Exponentiated exponential abð1  eat Þb1 eat a^ ¼ 452.1387 0.1191 0.4771 (0.2374)
 b

c1 b^ ¼ 0.6620
bc t b1 ðat Þ t b
Exponentiated Weibull a a e 1  eðaÞ a^ ¼ 116.5869 0.0934 0.3264 (0.9858)
(best-fit model) b^ ¼ 0.5364
c^ ¼ 1.9285

In the present study, we consider six continuous (Johnson et al., 1995; Pasari, 2015). Both tests
probability distributions, namely exponential, expo- perform well when a wide range of reference
nentiated exponential, gamma, lognormal, Weibull, distributions (e.g., time-dependent, time-independent,
and exponentiated Weibull, to develop the underlying heavy-tailed, and exponentiated family of distribu-
seismicity statistics of the study region based on the tions) are applied to fit a data set (Johnson et al.,
observed random sample of size 31. These distribu- 1995; Pasari, 2015). The probability density func-
tions have demonstrated their usefulness in several tions, estimated parameter values and model selection
analyses of statistical seismology, including earth- results are summarized in Table 1. Based on the
quake inter-event time distributions, probabilistic observed goodness-of-fit statistics, we find that the
seismic hazard analysis, and earthquake nowcasting exponentiated Weibull distribution provides the best
(Utsu, 1984; Vere-Jones et al., 2005; Pasari and fit to the natural time counts in the study region.
Dikshit 2014a, b, 2015a, b, 2018; Field et al., 2015;
Rundle et al., 2016; Pasari 2015, 2018,
3.2. Earthquake Potential Score at Several Cities
2019a, b, 2020; Pasari et al., 2021a, b). While we
apply the method of maximum likelihood estimation Considering a circular region of 300 km radius
(MLE) for parameter estimation, the best-fit proba- around the city center, we now derive the EPS for 29
bility distribution is identified from a pair of cities on Java Island and adjacent regions. The EPS
nonparametric model selection criteria, namely the for a city region at a given time t is simply the
Kolmogorov–Smirnov (K–S) test and the Anderson– probability PfX  mðtÞg of the current count of small
Darling (A–D) test (Johnson et al., 1995; Pasari & events m since the last large event in the city region.
Dikshit 2015a, b, 2019). As earthquake data and In this way, the nowcasting analysis provides a
consequent natural times are subjected to large, unique numerical score between 0 and 100% for a
seemingly outlier values, this prevents any a priori region which enables hazard ranking of cities in
symmetric distributional assumption (e.g., Gaussian) terms of their current ‘‘progression’’ in the regional
of the underlying statistical population (Scholz, 2009; cycle of large earthquakes. The nowcast scores for 29
Vere-Jones et al., 2005). As a consequence, we cities as on February 18, 2021, are summarized in
employ nonparametric K–S and A–D tests to prior- Table 2 and are graphically presented in Fig. 4.
itize the model suitability in the present analysis
 The Current State of Earthquake Potential on Java Island, Indonesia

Table 2
Nowcast values for Mk  6:5 earthquakes (as on February 18, 2021) in major cities on Java Island with Mr ¼ 4 and R ¼ 300 km

City City center Date of last large Magnitude of last large Current small event No of past large EPS
event event count events (%)
Lat Long
(°S) (°E)

Bandung 6.9175 107.6191 12/15/2017 6.6 05 327 69

Banjar 7.3707 108.5342 12/15/2017 6.6 05 314 68
Batu 7.8831 112.5334 09/28/1998 6.6 03 1114 93
Bekasi 6.2383 106.9756 08/02/2019 6.9 05 124 43
Blitar 8.0955 112.1609 09/28/1998 6.6 02 1155 94
Bogor 6.5971 106.8060 08/02/2019 6.9 06 140 46
Cilegon 6.0187 106.0558 08/02/2019 6.9 07 134 45
Cimahi 6.8841 107.5413 12/15/2017 6.6 05 332 69
Cirebon 6.7320 108.5523 12/15/2017 6.6 02 251 62
Denpasar 8.6705 115.2126 08/19/2018 6.8 05 430 76
Depok 6.4025 106.7942 08/02/2019 6.9 06 136 45
Jakarta 6.2088 106.8456 08/02/2019 6.9 05 127 43
Kediri 7.8480 112.0178 09/28/1998 6.6 01 1056 93
Madiun 7.6311 111.5300 09/28/1998 6.6 01 1022 92
Malang 7.9666 112.6326 09/28/1998 6.6 03 1129 94
Mojokerto 7.4705 112.4401 09/28/1998 6.6 02 957 91
Pasuruan 7.6469 112.8999 09/28/1998 6.6 02 968 92
Pekalongan 6.8898 109.6746 12/15/2017 6.6 02 204 56
Probolinggo 7.7764 113.2037 09/28/1998 6.6 02 974 92
Salatiga 7.3305 110.5084 12/15/2017 6.6 02 177 52
Semarang 7.0051 110.4381 12/15/2017 6.6 02 152 48
Serang 6.1169 106.1539 08/02/2019 6.9 07 143 47
Sukabumi 6.9277 106.9300 08/02/2019 6.9 06 155 49
Surabaya 7.2575 112.7521 09/28/1998 6.6 02 829 89
Surakarta 7.5755 110.8243 09/28/1998 6.6 01 975 92
Tangerang 6.1702 106.6403 08/02/2019 6.9 07 134 45
Tasikmalaya 7.3506 108.2172 12/15/2017 6.6 05 314 68
Tegal 6.8797 109.1256 12/15/2017 6.6 02 224 59
Yogyakarta 7.7956 110.3695 12/15/2017 6.6 02 230 59

Table 2 shows that the nowcast values corre- through its cycle. The EPS values can also be
sponding to M  6:5 events in a 300 km circular area interpreted as a measurement of stress accumulation
range from 43 to 94%, with the scores of the seven since the last major event (Rundle et al., 2016, 2020;
largest cities (one million population) of Bandung Rundle & Donnellan, 2020). However, it is important
(69), Bekasi (43), Depok (45), Jakarta (43), Semarang to note that the nowcast values which describe the
(48), Surabaya (89), and Tangerang (45). Other cities, contemporary earthquake hazards of a region do not
namely Batu, Blitar, Kediri, Madiun, Malang, Mojok- necessarily provide any information on future seis-
erto, Pasuruan, Probolinggo, and Surakarta, have an micity and earthquake forecasting. More discussion
EPS of more than 90%; Banjar, Cimahi, Cirebon, and in this regard is provided in a later section.
Tasikmalaya have a nowcast value between 60 and
75%; and Bogor, Cilegon, Magelang, Pekalongan,
Salitaga, Serang, Sukabumi, Tegal, and Yogyakarta 4. Discussion
have an EPS between 45 and 60%. Physically, the
EPS scores provide a yardstick of a city’s seismic Earthquake hazard assessment can be regarded as
progression. For example, Surabaya is about 89% of a set of methods for drawing hazard-related impli-
the way through its cycle for magnitude 6.5 or greater cations from earthquake data in a spatially defined
earthquakes, whereas Jakarta is about 43% of the way region (Utsu 1984; Scholz, 2009; Field et al., 2015).
 S. Pasari et al. Pure Appl. Geophys.

Figure 4
Nowcast scores (as on February 18, 2021) for different cities in the study region for a variety of combinations of magnitude thresholds and
local geographic area. Note that the EPS value corresponding to a radius of 250 km in Semarang is not reported, as the region did not
experience any large event during the time span of the considered catalog. The stacked bar-chart in the lower-panel provides an easy
interpretation of the current number of small event counts at several local regions

In the present work, the use of cumulative small event state of earthquake hazards at several cities on Java
counts between successive large earthquakes has Island, Indonesia. The earthquake nowcasting tech-
been demonstrated to determine the contemporary nique adopted herein relaxes at least two critical
 The Current State of Earthquake Potential on Java Island, Indonesia

assumptions of traditional fault-based hazard assess- 4.1. Validity of the Key Assumption in Earthquake
ment methods, namely (i) removal of aftershocks, Nowcasting
foreshocks, or any dependent events in hazard eval-
As noted earlier, the earthquake nowcasting
uation, and (ii) the notion of ‘‘segmentation’’ that
approach assumes that the natural time statistics of
considers repeated earthquakes on a specified fault
a large area are no different from the associated
section rather than focusing on recurrent earthquakes
statistics of a local region encompassed within the
in a defined fault system comprising a miscellany of
larger region (e.g., Rundle et al., 2016; 2018). This
faults within the geographic area. These a priori
characteristic of ensemble natural time statistics
assumptions in turn prevent the consideration of
stems from a key consequence of the classical
several emerging problems in geophysics, including
ergodic theory in earthquake fault networks (Tiampo
the interaction of earthquake fault networks in a
et al., 2003, 2007; Rundle et al., 2003, 2012). To
time–space domain (e.g., Chen et al., 2013), hierar-
verify this basic assumption in the specific space and
chical geometry of the faulting mechanism (e.g.,
time regions that are investigated in this study, we
Ouillion et al., 1996), information encoded in after-
generate frequency–magnitude statistics for a variety
shock sequences (e.g., Boyd, 2012; Yukutake & Lio,
of space–time combinations, as illustrated in Fig. 5
2017), the interlink between an ordered pair (e.g.,
below.
large and small earthquakes) of time series in a
In temporal variation, primarily to examine the
nonlinear dynamic system (e.g., Varotsos et al.,
possible inclusion of potential bias in earthquake
2011), or the unpredictable fault switching behavior
catalog, such as whether small events in aftershock
of some earthquakes in a complex tectonic regime
sequences are properly recorded or not due to the
(e.g., Hamling et al., 2017). Subsequently, although
instrumental defects or whether there is a bit of roll-
the concept of spatially and temporally dependent
off at a higher magnitude in the early instrumental
events was incorporated in a series of methods such
data, we have computed several b value estimates
as earthquake clustering in hazard assessment (e.g.,
corresponding to seven different time periods of
Toro & Silva, 2001, Rundle et al., 2003; Bhatia et al.,
earthquake catalog. It is observed that the b value
2018), aftershock probabilistic seismic hazard anal-
estimates are generally homogeneous in the time
ysis (e.g., Wiemer, 2000; Yeo & Cornell, 2009),
domain. Similarly, in spatial variation, to determine
time-dependent probabilistic seismic hazard assess-
whether the b value estimates change significantly
ment (e.g., Boyd 2012), and sequence-based
over different sub-regions of the study area, we have
probabilistic earthquake hazard evaluation involving
estimated b values corresponding to 29 local circular
several possible combinations of fault ruptures (Ier-
city regions. We observe that the b value estimates
volino et al., 2014; Field et al., 2015), quantification
are generally homogeneous in space. As a conse-
of the ‘‘current’’ progression of a city in the network
quence, the estimated b values as shown in Fig. 5
of regional faults was still unclear until the inception
largely validate the theoretical assumption of earth-
of the earthquake nowcasting approach (e.g., Rundle
quake nowcasting. Such assumption of b value in a
et al., 2016). In this paper, an attempt was made to
region essentially brings flexibility to the NTA and
assess the current state of hazard cycles at 29 popu-
consequent seismic nowcasting scheme that is inde-
lous cities on the seismically active island of Java.
pendent of the background seismic activity rate
We found that the numerical EPS values vary from a
(Rundle et al., 2016, 2018).
minimum of 43% to as high as 94%. In the following
sections, we discuss the basic assumption of the
nowcasting application, consistency of nowcast 4.2. Sensitivity Analysis
scores with respect to the changes in parameter val-
The ability to determine the current state of
ues, societal implications of nowcasting results in the
earthquake hazard through EPS scores is a crucial
study region, and recent developments in earthquake
component in city planning and disaster preparation.
nowcasting in light of earthquake forecasting in a
In fact, the EPS scores that are developed for a
defined region.
 S. Pasari et al. Pure Appl. Geophys.

subjective choice of geographic areas and magnitude regions usually result in higher EPS values, though
thresholds should be consistent with the minor there is some contrasting evidence due to the
changes in parameter values (Pasari, inclusion of recent large events in the wider area;
2019b, 2020; Pasari and Sharma, 2020; Pasari and (c) lowering the large-magnitude threshold to
et al., 2021a, b; Rundle et al., 2016). For this, we Mk ¼ 6:0 generally provides higher EPS values for
perform a basic sensitivity analysis in the following the scores of Jakarta (78), Surabaya (81), Bekasi (77),
tripartite setting: (a) with static magnitude thresholds Bandung (95), Depok (79), Semarang (82), Serang
Mr ¼ 4, Mk ¼ 6:5, we determine the impact of local (81), Tangerang (79), and Yogyakarta (91). Overall,
area changes ðR ¼ 250; 300; 350 kmÞ on the EPS the numerical nowcast scores are broadly consistent
score; (b) with fixed local region ðR ¼ 300 kmÞ and with the changes in threshold magnitude and local
large-magnitude threshold Mk ¼ 6:5, we observe the geographic area in the seismic-prone Java Island.
impact of small earthquake threshold changes
ðMr ¼ 3:5; 4:0; 4:5Þ; and (c) with fixed local region
4.3. Contemporary Earthquake Hazards on Java
ðR ¼ 300 kmÞ and small-magnitude threshold
Island
Mr ¼ 4, we determine the impact of lowering the
large-magnitude threshold to Mk ¼ 6:0. For each of The analysis of natural times tabulated from an
the above combinations, calculations are repeated and instrumental earthquake data set of about 60 years for
the resultant EPS values are graphically illustrated in Java Island and adjacent areas provides nowcast
Fig. 4. Several noteworthy observations are evident scores of between 43 and 94%, with the values of
from Fig. 4. For instance, (a) a gradual increase in the Jakarta (43), Surabaya (89), Bandung (69), Semarang
small-magnitude threshold leads to a consistent (48), Serang (47), and Yogyakarta (59). In addition,
decrease in EPS scores; (b) enlarged local city we found that the current state of seismic

Figure 5
The b value estimates for a variety of space–time regions in the study region. In the upper panel, b value estimates corresponding to seven
different time periods, while in the lower panel, we have estimated b values corresponding to 29 local circular city regions
 The Current State of Earthquake Potential on Java Island, Indonesia

‘‘progression’’, analogous to the tectonic stress accu- to previous three cycles commencing on same dates
mulation in an area, is terribly high (* 90%) for are about 0.56, 0.86, and 0.76, respectively; for
several city regions, such as Batu, Blitar, Kediri, Bandung, three previous nowcast values correspond-
Madiun, Malang, Mojokerto, Pasuruan, Probolinggo ing to earthquake cycles of the 2009 West Java
and Surakarta. However, the formulation of earth- earthquake (M 7.0, depth 46 km), 2006 Pangandaran
quake nowcasting follows the concept of seismic earthquake on July 17 (M 7.7, depth 20 km) and the
cycles, rather than the tectonic stress build-up in 2000 Sunda Strait earthquake (M 6.8) are 0.87, 0.76,
‘‘supercycles’’ with a long-term fault memory and 0.58, respectively; for Surabaya, Semarang and
(Rundle et al., 2016; Salditch et al., 2020). As a Yogyakarta cities that have experienced only one
consequence, the nowcast score restarts from zero at previous large event on July 14, 1976 (M 6.5),
the onset of each seismic cycle, unlike the notion of September 28, 1998 (M 6.6) and September 28, 1998
‘‘leftover’’ stress amount in a major earthquake (M 6.6), respectively, prior to the latest strong event,
(Scholz, 2009; Pasari, 2015; Rundle et al., 2016; the EPS values read as 64%, 85%, and 93%,
Salditch et al., 2020; Sharma et al., 2020). respectively.
While the nowcast scores as ‘‘thermometer’’ At this juncture, we reiterate that earthquake
readings help identify the cities at greater risk, the forecasting, which is more concerned with the
inherent variation (randomness) in natural time probability of future seismic activity, is an altogether
statistics may observe occasional early failures, as different approach from earthquake nowcasting,
commonly evident in several stochastic processes which delineates the current state of the hazard cycle
(Johnson et al., 1995; Rundle et al., 2016). For in a defined area (Rundle et al., 2016). Nonetheless,
instance, since 1963, the city of Bogor has experi- can earthquake nowcasting lead to earthquake fore-
enced five large earthquakes ðMk  6:5Þ prior to the casting in a given area? To answer this question,
recent strong 2019 Sunda Strait earthquake on August Holliday et al., (2016), with the basic formulation of
02 (M 6.9, depth 11 km). These five earthquakes are Rundle et al. (2012) employed natural time Weibull
December 15, 2017 Java earthquake (M 6.6, depth (NTW) forecasting based on a simple conversion of
115 km), September 02, 2009 West Java earthquake the natural time domain to a clock/calendar time
(M 7.0, depth 46 km), June 27, 2002 Sunda Strait domain using a Poisson rate—a long-term average
earthquake (M 6.5, depth 11 km), October 25, 2000 rate of small event counts. Similar to the classical
Sunda Strait earthquake (M 6.5, depth 11 km), and time-dependent conditional probability of major
the December 21, 1999 Mainland Java earthquake earthquakes in a region (e.g., Utsu, 1984), the
(M 6.5, depth 56 km). The natural times, in a reverse natural-time-based conditional probability of a large
chronological order, prior to the onset of those cycles event with a time interval Dt in the future, given that
were 203, 705, 409, 58 and 30, which result in mðtÞ number of small events have already occurred
nowcast values of 0.56, 0.87, 0.75, 0.26, and 0.16, since the last large earthquake, is defined as:
respectively. This shows that occasional occurrence (    )
of large earthquakes with a smaller EPS score (e.g., mðtÞ b mðtÞ þ xDt b
PðDt=tÞ ¼ 1  exp 
26% or even 16%) cannot be ruled out due to the M M
stochastic nature of the process (Pasari and Sharma,
Here, x represents the time-averaged rate of small
2020). Yet, within the statistical variations, the usual
event counts with magnitude lying in ½Mr ; Mk Þ and M
interpretation of EPS values is meaningful. For
is the mean natural time counts. Similarly, extended
example, the nowcast scores for Jakarta and Bekasi
prediction schemes to generalize nowcasting into
corresponding to previous three seismic cycles com-
forecasting (e.g., Perez-Oregon et al., 2020) are
mencing on December 15, 2017 (M 6.6, depth
proposed based on the knowledge of the current state
115 km), September 02, 2009 (M 7.0, depth 46 km)
of the system and the assumptions of the probability
and October 25, 2000 (M 6.5, depth 11 km) turnout to
distributions. Using Gutenberg–Richter law for con-
be 0.54, 0.85 and 0.73, respectively; for Serang,
stants a and b in a geographic region, it is trivial to
Depok and Tangerang, the EPS values corresponding
 S. Pasari et al. Pure Appl. Geophys.

show that M ¼ 10bðMk Mr Þ  1 (Rundle et al., 2012). provides the best fit to the observed natural
Thus, the ensemble natural time ðM Þ statistics is times.
independent of the background seismicity rate, (v) The seismic nowcasting method has no direct im-
though the NTW forecast depends on the ‘‘produc- plications for the future seismicity of a region,
tivity’’ of a region and the current count of small although, through a natural time Weibull (NTW)
events. In addition, in an NTW forecast model, projection, the nowcasting method may lead to
similar to a ‘‘renewal’’ or ‘‘characteristic earthquake’’ earthquake forecasting in terms of projected
model, the conditional probability steadily increases conditional probability curves.
with time until an earthquakes occurs, and then it
As a whole, the above analysis offers a rapid yet
decreases abruptly, but not necessarily dropping to
meaningful way of ranking several cities based on
zero (Rundle et al., 2012; Holliday et al., 2016). As a
their current progression within the regional earth-
consequence, the NTW forecast appears to be char-
quake cycles. Various practical applications and
acterized by earthquake ‘‘supercycles’’ (e.g., Salditch
considerations in decision-making, earthquake insur-
et al., 2020), rather than the adherence of seismic
ance, disaster preparation, and social awareness with
cycles in earthquake nowcasting. Our future work
regard to Java Island may benefit from the estimated
will consider the practical implementation of the
nowcast scores.
NTW forecast for Java Island.

Acknowledgements
5. Summary and Conclusions
The authors thankfully acknowledge constructive
Using an earthquake nowcasting approach, the
suggestions and useful comments from two anony-
present study quantifies the current state of earth-
mous reviewers. Some of the figures were prepared
quake hazards at 29 cities on the seismic-prone Java
using Generic Mapping Tools (GMT). The fourth
Island. Our data and analyses lead to the following
author [Neha] thankfully acknowledges the financial
conclusions:
support from the CSIR-UGC-NET (Ref. No:
(i) The estimated EPS as on February 18, 2021 1197/CSIR-UGC NET JUNE 2017).
corresponding to M  6:5 events in a 300 km
circular area lies between 43 and 94%, with the
scores of Jakarta (43), Surabaya (89), Bekasi (43), Funding
Bandung (69), Depok (45), Semarang (48),
Serang (47), Tangerang (45), and Yogyakarta This research did not receive any specific grant from
(59). funding agencies in the public, commercial, or not-
(ii) The analysis corresponding to M  6 events in a for-profit sectors.
300 km circular region results in nowcast scores
ranging from 77 to 99%, with the scores of
Jakarta (78), Surabaya (81), Bekasi (77), Ban- Data Availability
dung (95), Depok (79), Semarang (82), Serang
(81), Tangerang (79), and Yogyakarta (91). Earthquake data were downloaded from public and
(iii) It is observed that a slight change in magnitude regional catalogs: Advanced National Seismic Sys-
threshold or geographic area has a consistent tem comprehensive catalog (http://www.ncedc.org/
impact on the resulting nowcast scores in the anss/catalog-search.html), Meteorological, Climato-
study region. logical and Geophysical Agency (BMKG) of
(iv) Among the six studied probability models, we Indonesia catalog (http://repogempa.bmkg.go.id/
found that exponentiated Weibull distributions repo_new/), and International Seismological Centre
catalog (http://www.isc.ac.uk/iscbulletin/search/
 The Current State of Earthquake Potential on Java Island, Indonesia

catalogue/). All websites were last accessed in Indonesia. Journal of Southeast Asian Earth Sciences, 9(1–2),
3–11.
February, 2021.
Diambama, A. D., Anggraini, A., Nukman, M., Lühr, B. G., &
Suryanto, W. (2019). Velocity structure of the earthquake zone
Declarations
of the M6.3 Yogyakarta earthquake 2006 from a seismic
tomography study. Geophysical Journal International, 216(1),
Conflict of interest The authors declare that they have no
439–452.
competing interests.
Field, E. H., Biasi, G. P., Bird, P., Dawson, T. E., Felzer, K. R.,
Jackson, D. D., Johnson, K. M., Jordan, T. H., Madden, C.,
Publisher’s Note Springer Nature remains neutral Michael, A. J., & Milner, K. R. (2015). Long-term time-depen-
with regard to jurisdictional claims in published maps dent probabilities for the third uniform California earthquake
rupture forecast (UCERF3). Bulletin of Seismological Society of
and institutional affiliations. America, 105, 511–543.
Fujii, Y., & Satake, K. (2006). Source of the July 2006 West Java
tsunami estimated from tide gauge records. Geophysical
Research Letters. https://doi.org/10.1029/2006GL028049
REFERENCES Gatignon, A., Van Wassenhove, L. N., & Charles, A. (2010). The
Yogyakarta earthquake: Humanitarian relief through IFRC’s
Abercrombie, R. E., Antolik, M., Felzer, K., & Ekström, G. (2001). decentralized supply chain. International Journal of Production
The 1994 Java tsunami earthquake: Slip over a subducting sea- Economics, 126(1), 102–110.
mount. Journal of Geophysical Research: Solid Earth, 106(B4), Gunawan, E., & Widiyantoro, S. (2019). Active tectonic defor-
6595–6607. mation in Java, Indonesia inferred from a GPS-derived strain
Abidin, H. Z., Andreas, H., Kato, T., Ito, T., Meilano, I., Kimata, rate. Journal of Geodynamics, 123, 49–54.
F., Natawidjaya, D. H., & Harjono, H. (2009). Crustal defor- Hall, R., B. Clements, & Smyth, H. R. 2009. Sundaland: basement
mation studies in Java (Indonesia) using GPS. Journal of character, structure and plate tectonic development.. In Indone-
Earthquake and Tsunami, 3(02), 77–88. sian Petroleum Association Thirty-Third Annual Convention and
Afnimar, E. Y., & Rasmid. (2015). Geological and tectonic Exhibition, IPA09- G-134: 1–27.
implications obtained from first seismic activity investigation Hamilton, W. B. (1979). Tectonics of the Indonesian region. Ge-
around Lembang fault. Geoscience Letters, 2(1), 4. ological Survey, 1078, 352.
Ammon, C. J., Kanamori, H., La, T., & Velasco, A. A. (2006). The Hamling, I. J., Hreinsdóttir, S., Clark, K., Elliott, J., Liang, C.,
17 July 2006 Java tsunami earthquake. Geophysical Research Fielding, E., Litchfield, N., Villamor, P., Wallace, L., Wright, T.
Letters, 33(24), L24308. J., & D’Anastasio, E. (2017). Complex multifault rupture during
Bhatia, A., Pasari, S., Mehta, A. (2018). Earthquake forecasting the 2016 Mw7.8 Kaikoura earthquake. Science, 356, 154ss.
using artificial neural networks. ISPRS-International Archives of Hanifa, N. R., Sagiya, T., Kimata, F., Efendi, J., Abidin, H. Z., &
the Photogrammetry, Remote Sensing and Spatial Information Meilano, I. (2014). Interplate coupling model off the south-
Sciences, XLII-5, 823–827. https://doi.org/10.5194/isprs- western coast of Java, Indonesia, based on continuous GPS data
archives-XLII-5-823-2018. in 2008–2010. Earth and Planetary Science Letters, 401,
Bilek, S. L., & Engdahl, E. R. (2007). Rupture characterization and 159–171.
aftershock relocations for the 1994 and 2006 tsunami earth- Harris, R., & Major, J. (2017). Waves of destruction in the East
quakes in the Java subduction zone. Geophysical Research Indies: The Wichmann catalogue of earthquakes and tsunami in
Letters. https://doi.org/10.1029/2007GL031357 the Indonesian region from 1538 to 1877. Geological Society,
BNPB-National Disaster Management Agency. 2019. Tsunami London, Special Publications, 441(1), 9–46.
Selat Sunda. (https://bnpb.go.id/publikasi/infografis/tsunami- Heidarzadeh, M., Ishibe, T., Sandanbata, O., Muhari, A., &
selat-sunda.html). Accessed Feb 2021. Wijanarto, A. B. (2020). Numerical modeling of the subaerial
Boyd, O. S. (2012). Including foreshocks and aftershocks in time- landslide source of the 22 December 2018 Anak Krakatoa vol-
independent probabilistic seismic-hazard analyses. Bulletin of the canic tsunami, Indonesia. Ocean Engineering, 195, 106733.
Seismological Society of America, 102, 909–917. Holliday, J. R., Graves, W. R., Rundle, J. B., & Turcotte, D. (2016).
Chen, K. H., Burgmann, R., & Nadeau, R. M. (2013). Do earth- Computing earthquake probabilities on global scales. Pure and
quakes talk to each other? Triggering and interaction of repeating Applied Geophysics, 173, 739–748.
sequences at Parkfield. Journal of Geophysical Research, 118, Horspool, N., Pranantyo, I., Griffin, J., Latief, H., Natawidjaja, D.
165–182. H., Kongko, W., Cipta, A., Bustaman, B., Anugrah, S. D., &
Ciesin, C. (2005). Gridded population of the world version 3 Thio, H. K. (2014). A probabilistic tsunami hazard assessment
(gpwv3): Population density grids, socioeconomic data and for Indonesia. Natural Hazards and Earth System Sciences,
applications center (SEDAC). Columbia University. 14(11), 3105.
Curray, J. R., Emmel, F. J., Moore, D. G., & Raitt, R. W. (1982). Iervolino, I., Giorgio, M., & Polidoro, B. (2014). Sequence-based
Structure, tectonics, and geological history of the northeastern probabilistic seismic hazard analysis. Bulletin of the Seismolog-
Indian Ocean. The ocean basins and margins (pp. 399–450). ical Society of America, 104(2), 1006–1012.
Boston: Springer. Johnson, N. L., Kotz, S., & Balakrishnan, N. (1995). Continuous
Dardji, N., Villemin, T., & Rampnoux, J. P. (1994). Paleostresses univariate distributions. Wiley-Interscience.
and strike-slip movement: The Cimandiri fault zone, West Java, Katili, J. A. (1973). Geochronology of West Indonesia and its
implication on plate tectonics. Tectonophysics, 19(3), 195–212.
 S. Pasari et al. Pure Appl. Geophys.

Katili, J. A. (1975). Volcanism and plate tectonics in the Indone- Ouillon, G., Castaing, C., & Sornette, D. (1996). Hierarchical
sian island arcs. Tectonophysics, 26(3–4), 165–188. geometry of faulting. Journal of Geophysical Research, 101(B3),
Kato, T., Ito, T., & Abidin, H. Z. (2007). Preliminary report on 5477–5487.
crustal deformation surveys and tsunami measurements caused Pal, T., Chakraborty, P. P., Gupta, T. D., & Singh, C. D. (2003).
by the July 17, 2006 South off Java Island Earthquake and Geodynamic evolution of the outer-arc–forearc belt in the
Tsunami, Indonesia. Earth, Planets and Space, 59(9), Andaman Islands, the central part of the Burma-Java subduction
1055–1059. complex. Geological Magazine, 140(3), 289–307.
Kopp, H., Flueh, E. R., Petersen, C. J., Weinrebe, W., & Wittwer, Paris, A., Heinrich, P., Paris, R., & Abadie, S. (2020). The
A. (2006). The Java margin revisited: Evidence for subduction December 22, 2018 Anak Krakatau, Indonesia, landslide and
erosion off Java. Earth and Planetary Science Letters, 242(1–2), tsunami: Preliminary modeling results. Pure and Applied Geo-
130–142. physics, 177(2), 571–590.
Kopp, H., Klaeschen, D., Flueh, E. R., Bialas, J., & Reichert, C. Pasari, S. 2015. Understanding Himalayan tectonics from geodetic
(2002). Crustal structure of the Java margin from seismic wide- and stochastic modeling. PhD Thesis, Dept. Civil Engg., IIT
angle and multichannel reflection data. Journal of Geophysical Kanpur, India: 376.
Research. https://doi.org/10.1029/2000JB000095 Pasari, S. (2018). Stochastic modeling of earthquake interoccur-
Koulali, A., McClusky, S., Susilo, S., Leonard, Y., Cummins, P., rence times in northwest Himalaya and adjoining regions.
Tregoning, P., Meilano, I., Efendi, J., & Wijanarto, A. B. (2017). Geomatics, Natural Hazards and Risk, 9(1), 568–588. Taylor &
The kinematics of crustal deformation in Java from GPS obser- Francis.
vations: Implications for fault slip partitioning. Earth and Pasari, S. (2019a). Inverse Gaussian versus lognormal distribution
Planetary Science Letters, 458, 69–79. in earthquake forecasting: keys and clues. Journal of Seismology,
Lavigne, F., Gomez, C., Giffo, M., Wassmer, P., Hoebreck, C., 23(3), 537–559. https://doi.org/10.1007/s10950-019-09822-5.
Mardiatno, D., Prioyono, J., & Paris, R. (2007). Field observa- Pasari, S. (2019b). Nowcasting earthquakes in the Bay-of-Bengal
tions of the 17 July 2006 Tsunami in Java. Natural Hazards and region. Pure Applied Geophycis, 23, 537–559.
Earth System Sciences, 7, 177–183. Pasari, S. (2020) Stochastic modeling of earthquake interevent
Lay, T., Ammon, C. J., Kanamori, H., Yamazaki, Y., Cheung, K. counts (Natural Times) in Northwest Himalaya and adjoining
F., & Hutko, A. R. (2011). The 25 October 2010 Mentawai regions. In S. Bhattacharyya, J. Kumar, K. Ghoshal (Eds.)
tsunami earthquake (Mw 7.8) and the tsunami hazard presented Mathematical modeling and computational tools, ICACM 2018,
by shallow megathrust ruptures. Geophysical Research Letters, Springer proceedings in mathematics & statistics (Vol. 320,
38(6), L06302. pp. 495–501). Singapore: Springer. https://doi.org/10.1007/978-
Luginbuhl, M., Rundle, J. B., & Turcotte, D. L. (2018). Natural 981-15-3615-1_35
time and nowcasting induced seismicity at the Groningen gas Pasari, S., & Dikshit, O. (2014a). Impact of three-parameter Wei-
fields in the Netharlands. Geophysical Journal International, bull models in probabilistic assessment of earthquake hazards.
215, 753–759. Pure and Applied Geophysics, 171(7), 1251–1281. https://doi.
Malod, J. A., Karta, K., Beslier, M. O., & Zen, M. T., Jr. (1995). org/10.1007/s00024-013-0704-8
From normal to oblique subduction: Tectonic relationships Pasari, S., & Dikshit, O. (2014b). Three-parameter generalized
between Java and Sumatra. Journal of Southeast Asian Earth exponential distribution in earthquake recurrence interval esti-
Sciences, 12(1–2), 85–93. mation. Natural Hazards, 73, 639–656. https://doi.org/10.1007/
Maramai, A., & Tinti, S. (1997). The 3 June 1994 Java tsunami: A post- s11069-014-1092-9.
event survey of the coastal effects. Natural Hazards, 15(1), 31–49. Pasari, S., & Dikshit, O. (2015a). Distribution of earthquake
Matthews, S. J., & Bransden, P. J. (1995). Late cretaceous and interevent times in northeast India and adjoining regions. Pure
cenozoic tectono-stratigraphic development of the East Java Sea Applied Geophysics, 172, 2533–2544.
Basin, Indonesia. Marine and Petroleum Geology, 12(5), Pasari, S., & Dikshit, O. (2015b). Earthquake interevent time dis-
499–510. tribution in Kachchh, northwestern India. Earth Planets and
Meilano, I., Abidin, H. Z., Andreas, H., Gumilar, I., Sarsito, D., Space, 67, 129. https://doi.org/10.1186/s40623-015-0295-y.
Hanifa, R. R., Harjono, H., Kato, T., Kimata, F., & Fukuda, Y. Pasari, S., & Dikshit, O. (2018). Stochastic earthquake interevent
(2012). Slip rate estimation of the Lembang Fault West Java time modeling from exponentiated Weibull distributions. Natural
from geodetic observation. Journal of Disaster Research, 7(1), Hazards, 90(2), 823–842. https://doi.org/10.1007/s11069-017-
12–18. 3074-1.
Metcalfe, I. (2011). Tectonic framework and Phanerozoic evolution Pasari, S., & Mehta, A. (2018). Nowcasting earthquakes in the
of Sundaland. Gondwana Research, 19(1), 3–21. northwest Himalaya and surrounding regions. International
Mori, J., Mooney, W. D., Kurniawan, S., Anaya, A. I., & Archives of the Photogrammetry, Remote Sensing, and Spatial
Widiyantoro, S. (2007). The 17 July 2006 tsunami earthquake in Information Sciences, XLII–5, 855–859.
west Java, Indonesia. Seismological Research Letters, 78(2), Pasari, S., & Sharma, Y. (2020). Contemporary earthquake hazards
201–207. in the west-northwest Himalaya: A statistical perspective through
Muhari, A., Diposaptono, S., & Imamura, F. (2007). Toward an natural times. Seismological Research Letters, 91(6), 3358–3369.
integrated tsunami disaster mitigation: Lessons learned from https://doi.org/10.1785/0220200104.
previous tsunami events in Indonesia. Journal of Natural Pasari, S., Sharma, Y., & Neha (2021a). Quantifying the current
Disaster Science, 29(1), 13–19. state of earthquake hazards in Nepal. Applied Computing and
Newcomb, K. R., & McCann, W. R. (1987). Seismic history and Geosciences, 10, 100058. https://doi.org/10.1016/j.acags.2021.
seismotectonics of the Sunda Arc. Journal of Geophysical 100058.
Research: Solid Earth, 92(B1), 421–439.
 The Current State of Earthquake Potential on Java Island, Indonesia

Pasari, S., Simanjuntak, A. V. H., Mehta, A., Neha, & Sharma, Y. ancient tectonic plate beneath the southern Indian Ocean. Geo-
(2021b). A synoptic view of the natural time distribution and physical Research Letters, 42(21), 9270–9278.
contemporary earthquake hazards in Sumatra, Indonesia. Natural Suhardja, S. K., Widiyantoro, S., Métaxian, J. P., Rawlinson, N.,
Hazards. https://doi.org/10.1007/s11069-021-04682-0. (in print) Ramdhan, M., & Budi-Santoso, A. (2020). Crustal thickness
Perez-Oregon, J., Angulo-Brown, F., & Sarlis, N. V. (2020). beneath Mt. Merapi and Mt. Merbabu, Central Java, Indonesia,
Nowcasting avalanches as earthquakes and the predictability of inferred from receiver function analysis. Physics of the Earth and
strong avalanches in the olami-feder-christensen model. Entropy, Planetary Interiors, 302, 106455.
22, 1228. https://doi.org/10.3390/e22111228 Supendi, P., Nugraha, A. D., Puspito, N. T., Widiyantoro, S., &
Polet, J., & Kanamori, H. (2000). Shallow subduction zone earth- Daryono, D. (2018). Identification of active faults in West Java,
quakes and their tsunamigenic potential. Geophysical Journal Indonesia, based on earthquake hypocenter determination, relo-
International, 142(3), 684–702. cation, and focal mechanism analysis. Geoscience Letters, 5(1),
Rundle, J. B., & Donnellan, A. (2020). Nowcasting earthquakes in 1–10.
southern California with machine learning: Bursts, swarms and Susilohadi, S., Gaedicke, C., & Ehrhardt, A. (2005). Neogene
aftershocks may reveal the regional tectonic stress. Earth Space structures and sedimentation history along the Sunda forearc
Science, 58, 147. basins off southwest Sumatra and southwest Java. Marine
Rundle, J. B., Giguere, A., Turcotte, D. L., Crutchfield, J. P., & Geology, 219(2–3), 133–154.
Donnellan, A. (2019). Global seismic nowcasting with Shannon Synolakis, C., Imamura, F., Tsuji, Y., Matsutomi, H., Tinti, S.,
information entropy. Earth Space Science, 6, 456–472. Cook, B., Chandra, Y. P., & Usman, M. (1995). Damage, con-
Rundle, J. B., Holliday, J. R., Graves, W. R., Turcotte, D. L., ditions of East Java tsunami of 1994 analyzed. EOS,
Tiampo, K. F., & Klein, W. (2012). Probabilities for large events Transactions American Geophysical Union, 76(26), 257–257.
in driven threshold systems. Physical Review E., 86, 021106. Tiampo, K. F., Rundle, J. B., Klein, W., Martins, J. S., & Ferguson,
Rundle, J. B., Luginbuhl, M., Giguere, A., & Turcotte, D. L. C. D. (2003). Ergodic dynamics in a natural threshold system.
(2018). Natural time, nowcasting and the physics of earthquakes: Physical Review Letters, 91, 238501.
Estimation of seismic risk to global megacities. Pure Applied Tiampo, K. F., Rundle, J. B., Klein, W., Martins, J. S., & Ferguson,
Geophysics, 175, 647–660. C. D. (2007). Ergodicity in natural fault systems. Physical
Rundle, J. B., Luginbuhl, M., Khapikova, P., Turcotte, D. L., Review E, 75, 0666107.
Donnellan, A., & McKim, G. (2020). Nowcasting great global Toro, G. R., & Silva, W. J. (2001). Scenario earthquakes for Saint
earthquake and tsunami sources. Pure Applied Geophysics, 177, Louis, MO, and Memphis, TN, and seismic hazard maps for the
359–368. central United States region including the effect of site condi-
Rundle, J. B., Turcotte, D. L., Donnellan, A., Grant-Ludwig, L., tions. Boulder: Risk Engineering.
Luginbuhl, M., & Gong, G. (2016). Nowcasting earthquakes. Tregoning, P., Brunner, F. K., Bock, Y., Puntodewo, S. S. O.,
Earth Space Science, 3, 480–486. McCaffrey, R., Genrich, J. F., Calais, E., Rais, J., & Subarya, C.
Rundle, J. B., Turcotte, D. L., Shchebakov, R., Klein, W., & (1994). First geodetic measurement of convergence across the
Sammis, C. (2003). Statistical physics approach to understanding Java Trench. Geophysical Research Letters, 21(19), 2135–2138.
the multiscale dynamics of earthquake fault systems. Reviews of Tsuji, T., Yamamoto, K., Matsuoka, T., Yamada, Y., Onishi, K.,
Geophysics, 41, 1019. Bahar, A., Meilano, I., & Abidin, H. Z. (2009). Earthquake fault
Sakamoto, M., & Yamori, K. (2009). A study of life recovery and of the 26 May 2006 Yogyakarta earthquake observed by SAR
social capital regarding disaster victims—A case study of Indian interferometry. Earth, Planets and Space, 61(7), 29–32.
Ocean tsunami and Central Java earthquake recovery. Journal of Utsu, T. (1984). Estimation of parameters for recurrence models of
Natural Disaster Science, 31(2), 49–56. earthquakes. Bulletin of Earthquake Research Institute, Univer-
Salditch, L., Stein, S., Neely, J., Spencer, B. D., Brooks, E. M., sity of Tokyo, 59, 53–66.
Agnon, A., & Liu, M. (2020). Earthquake supercycles and long- Varostos, P. A., Sarlis, N. V., & Skordas, E. S. (2011). Natural time
term fault memory. Tectonophysics. https://doi.org/10.1016/j. analysis: The new view of time. Springer.
tecto.2019.228289 (in print). Vere-Jones, D., Ben-Zion, Y., & Zuniga, R. (2005). Statistical
Schlüter, H. U., Gaedicke, C. H., Roeser, H. A., Schreckenberger, seismology. Pure and Applied Geophysics, 162, 1023–1026.
B., Meyer, H., Reichert, C. H., Djajadihardja, Y., & Prexl, A. Walter, T. R., Wang, R., Luehr, B. G., Wassermann, J., Behr, Y.,
(2002). Tectonic features of the southern Sumatra-western Java Parolai, S., Anggraini, A., Günther, E., Sobiesiak, M., Grosser,
forearc of Indonesia. Tectonics, 21(5), 11–1. H., & Wetzel, H. U. (2008). The 26 May 2006 magnitude 6.4
Scholz, C. H. (2009). The mechanics of earthquakes and faulting. Yogyakarta earthquake south of Mt. Merapi volcano: Did lahar
Cambridge University Press. deposits amplify ground shaking and thus lead to the disaster?
Sharma, Y., Pasari, S., Ching, K. E., Dikshit, O., Kato, T., Malik, J. Geochemistry, Geophysics, Geosystems, 9(5), Q05006.
N., Chang, C. P., & Yen, J. Y. (2020). Spatial distribution of Whittaker, J. M., Müller, R. D., Sdrolias, M., & Heine, C. (2007).
earthquake potential along the Himalayan arc. Tectonophysics, Sunda-Java trench kinematics, slab window formation and
791, 228556. https://doi.org/10.1016/j.tecto.2020.228556. overriding plate deformation since the Cretaceous. Earth and
Shulgin, A., Kopp, H., Mueller, C., Planert, L., Lueschen, E., Planetary Science Letters, 255(3–4), 445–457.
Flueh, E. R., & Djajadihardja, Y. (2011). Structural architecture Widiyantoro, S., Gunawan, E., Muhari, A., et al. (2020). Implica-
of oceanic plateau subduction offshore Eastern Java and the tions for megathrust earthquakes and tsunamis from seismic gaps
potential implications for geohazards. Geophysical Journal south of Java Indonesia. Scientific Reports, 10, 15274. https://doi.
International, 184(1), 12–28. org/10.1038/s41598-020-72142-z
Simmons, N. A., Myers, S. C., Johannesson, G., Matzel, E., & Widjajanti, N., Pratama, C., Sunantyo, T. A., Heliani, L. S., Ma’ruf,
Grand, S. P. (2015). Evidence for long-lived subduction of an B., Atunggal, D., Lestari, D., Ulinnuha, H., Pinasti, A., & Ummi,
 S. Pasari et al. Pure Appl. Geophys.

R. F. (2020). Present-day crustal deformation revealed active Yeo, G. L., & Cornell, C. A. (2009). A probabilistic framework for
tectonics in Yogyakarta, Indonesia inferred from GPS observa- quantification of aftershock ground-motion hazard in California:
tions. Geodesy and Geodynamics, 11, 135–142. Methodology and parametric study. Earthquake Engineering &
Wiemer, S. (2000). Introducing probabilistic aftershock hazard Structural Dynamics., 38(1), 45–60.
mapping. Geophysical Research Letters, 27(20), 3405–3408. Yukutake, Y., & Lio, Y. (2017). Why do aftershocks occur?
Wölbern, I., & Rümpker, G. (2016). Crustal thickness beneath Relationship between mainshock rupture and aftershock
Central and East Java (Indonesia) inferred from P receiver sequence based on highly resolved hypocenter and focal mech-
functions. Journal of Asian Earth Sciences, 115, 69–79. anism distributions. Earth, Planets and Space, 69(1), 68.
Ye, L., Kanamori, H., Rivera, L., Lay, T., Zhou, Y., Sianipar, D., & Zahirovic, S., Seton, M., & Müller, R. D. (2014). The cretaceous
Satake, K. (2020). The 22 December 2018 tsunami from flank and cenozoic tectonic evolution of Southeast Asia. Solid Earth,
collapse of Anak Krakatau volcano during eruption. Science 5(1), 227.
Advances, 6(3), eaaz377.

(Received November 16, 2020, revised April 15, 2021, accepted May 31, 2021)