Available via license: CC BY 4.0
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=tgsi20
Geo-spatial Information Science
ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tgsi20
Preliminary identification of earthquake triggered
multi-hazard and risk in Pleret Sub-District
(Yogyakarta, Indonesia)
Aditya Saputra , Christopher Gomez , Ioannis Delikostidis , Peyman Zawar-
Reza , Danang Sri Hadmoko & Junun Sartohadi
To cite this article: Aditya Saputra , Christopher Gomez , Ioannis Delikostidis , Peyman Zawar-
Reza , Danang Sri Hadmoko & Junun Sartohadi (2020): Preliminary identification of earthquake
triggered multi-hazard and risk in Pleret Sub-District (Yogyakarta, Indonesia), Geo-spatial
Information Science, DOI: 10.1080/10095020.2020.1801335
To link to this article: https://doi.org/10.1080/10095020.2020.1801335
© 2020 Wuhan University. Published by
Informa UK Limited, trading as Taylor &
Francis Group.
Published online: 07 Aug 2020.
Submit your article to this journal
View related articles
View Crossmark data
Preliminary identication of earthquake triggered multi-hazard and risk in
Pleret Sub-District (Yogyakarta, Indonesia)
Aditya Saputra
a
, Christopher Gomez
b
, Ioannis Delikostidis
c
, Peyman Zawar-Reza
c
,
Danang Sri Hadmoko
d
and Junun Sartohadi
e
a
Faculty of Geography, Universitas Muhammadiyah Surakarta, Central Java, Indonesia;
b
Graduate School of Maritime Sciences, Laboratory
of Volcanic Risks at Sea Kobe University, Kobe, Japan;
c
Geography Department, School of Science, University of Canterbury, Christchurch,
New Zealand;
d
Geography Faculty, Universitas Gadjah Mada, Yogyakarta, Indonesia;
e
Department of Soil Science, Faculty of Agriculture,
Universitas Gadjah Mada, Yogyakarta, Indonesia
ABSTRACT
Yogyakarta is one of the large cities in Central Java, located on Java Island, Indonesia. The city, and
the Pleret sub-district, where the study has taken place, is prone to earthquake hazards, because it is
close to several seismically active zones, such as the Sunda Megathrust and the active fault known as
the Opak Fault. Since a devastating earthquake of 2006, the population of the Pleret sub-district has
increased signicantly. Thus, the housing demand has increased, and so is the pace of low-cost
housing that does not meet earthquake-safety requirements, and furthermore are often located on
unstable slopes. The local alluvial material covering a jigsaw of unstable blocks and complex slope is
conditions that can amplify the negative impacts of earthquakes. Within this context, this study is
aiming to assess the multi-hazards and risks of earthquakes and related secondary hazards such as
ground liquefaction, and coseismic landslides. To achieve this, we used geographic information
systems and remote sensing methods supplemented with outcrop study and existing seismic data
to derive shear-strain parameters. The results have revealed the presence of numerous uncharted
active faults with movements visible from imagery and outcrops. show that the middle part of the
study area has a complex geological structure, indicated by many unchartered faults in the outcrops.
Using this newly mapped blocks combined with shear strain data, we reassessed the collapse
probability of buildings that reach level >0.75 near the Opak River, in central Pleret sub-district.
Classifying the buildings and from population distribution, we could determine that the highest risk
was during nighttime as the buildings susceptible to fall are predominantly housing buildings. The
secondary hazards follow a slightly dierent distribution with a concentration of risks in the West.
ARTICLE HISTORY
Received 28 August 2019
Accepted 23 July 2020
KEYWORDS
Earthquake multi-hazard and
risk; coseismic landslide;
outcrop study; liquefaction
1. Introduction
Cascading hazards have received relatively little atten-
tion in disaster and risk studies, as the majority of
hazard and disaster risk studies have concentrated on
one single hazard. The reality of scientific funding and
projects often leads researchers to pay less attention to
the possibility that one incident can cascade to other
secondary hazards. Consequently, the interactions
between hazards are still relatively ignored (Budimir,
Atkinson, and Lewis 2014), especially because as scien-
tists we are often trained in one discipline and it is still
problematic to work across boundaries. As a particular
area may be exposed to more than one type of hazard,
even if each hazard can lead to a given type of disaster
with different magnitudes (Westen 2011), each single
incident can trigger secondary hazards that cannot be
encapsulated by the magnitude approach from the ori-
ginal hazard (Ren and Liu 2013). An example of this
such disaster chain-reaction occurred in Beichuan,
China, due to the 2008 Wenchuan earthquake. The
Wenchuan earthquake generated a complex disaster
chain that caused important damage to towns in the
Beichuan county. The earthquake had a shallow focal
depth of about 19 km, and the magnitude of the earth-
quake was 7.9–8.0 Ms. It caused ~90% of the buildings
in the area to collapse. They had also triggered coseis-
mic landslides upstream the township, generating in
turn a dammed lake, which collapsed with the heavy
rainfall a month after the earthquake. The water from
the Tangjiashan Lake mixed with the rubbles of the
earthquake and washed away the ruins of the collapsed
buildings. Fortunately, during the event, the Beichuan
county town was off-limit due to the expected cascad-
ing effects triggered by the earthquake.
Another similar situation occurred in
Christchurch, New Zealand between September 2010
and February 2011, Christchurch – the second largest
city in New Zealand – experienced two large series of
earthquakes. The first earthquake (6.3 Ms) took the
lives of 185 people, making this event the second dead-
liest disaster that ever occurred in New Zealand.
Amidst numerous aftershocks, a second earthquake
(7.3 Ms), just underneath the city, caused significant
damage in the central city of Christchurch. Afterward,
CONTACT Aditya Saputra as105@ums.ac.id
GEO-SPATIAL INFORMATION SCIENCE
https://doi.org/10.1080/10095020.2020.1801335
© 2020 Wuhan University. Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits
unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
in March 2014, Christchurch experienced more flood-
ing due to the impact of the Canterbury Earthquake
and the general lowering of the ground level (Gomez
et al. 2019). Some researchers concluded that this
extraordinary flood occurred not just due to the
heavy and prolonged rainfall but due to ground defor-
mation, liquefaction, subsidence, narrowing of chan-
nels, and uplifting of river beds after the earthquake.
Coastal plains are therefore particularly at risk,
especially for earthquake-prone island coastal plains
of New Zealand, Japan or Indonesia. In Indonesia, the
study location is situated approximately 300 km north
of the Java Mega thrust. The study area, in Central Java
(the west flank of Baturagung Escarpment) is particu-
larly prone to earthquakes. Having a total subduction
segment of 840 km, the Java Mega thrust can poten-
tially generate earthquakes with maximum recorded-
magnitudes up to 8.1 Mw (Figure 1) (Irsyam et al.
2012). Link to this activity, one of the main active
fault of Java, the Opak Fault, passes through the
research area along the west flank of the Baturagung
Escarpment. According to (PUSGEN 2017), which is
the National Center of the Earthquake studies, with
a total fault length of approximately 45 km and a slip
rate of about 2.4 mm per year, the Opak fault can
potentially generate shallow earthquakes with
a maximum magnitude of 6.2 Mw. According to the
same agency, such an earthquake would have
a devastating impact on the study area and the south-
east part of Yogyakarta City.
Administratively, the west flank of Baturagung con-
sists of several sub-districts, including Kretek,
Pundong, Imogiri, Dlingo, Pleret, and Piyungan. As
a pilot study, the present work has focused on the
Pleret sub-district, Bantul Regency, of Yogyakarta
Province. Pleret sub-district is located approximately
10 km southeast of Yogyakarta Central City. Located
near to subduction zone and over the active Opak
Fault makes this area very prone to earthquake
hazards. Furthermore, high seismicity is expected to
be amplified due to unconsolidated river and marine
sediments, which could lead to liquefaction in the
center lowland, while to the west, amplified movement
is expected due to the unstable slope of the Baturagung
Escarpment.
Probabilistic Seismic Hazard Analysis (PSHA)
(Ashadi and Kaka 2015), Deterministic Seismic Hazard
Analysis (DSHA) (Chopra et al. 2012), liquefaction
(Civico et al. 2015), ground motion, landslide
(Hadmoko et al. 2010), and tsunami studies have been
developed worldwide. However, earthquakes can trigger
other related hazards that increase the impact on society
(Marano et al. 2010) and studies of multi-hazards and
risks in this area are still limited. Thus, the study of
earthquake triggers and other related secondary hazards
is needed in Pleret Sub-District, Southeast Yogyakarta
Province.
The primary objective of this study was to con-
struct a multi-hazard risk assessment in Pleret Sub-
District using remote sensing and a Geographic
Figure 1. The segmentation model of subduction earthquake source (Megathrust) in Indonesia. Source: Irsyam et al. (2012)
2A. SAPUTRA ET AL.
Information System (GIS). The primary objective
can be divided into several sub-studies with the
objectives of (1) identifying the potential area of
coseismic landslide, (2) conducting an outcrop
study in order to better understand the fault con-
figuration, (3) identifying the liquefaction zonation,
(4) assessing the vulnerability level of some ele-
ments at risk, and (5) describing the multi-
hazards and risks in the study area.
2. Overview of the study area
Pleret sub-district is located 10 km southeast part of
Yogyakarta City, Indonesia. The study area is the part
of Bantul Graben. This graben was formed by two con-
vergent normal faults, which generated two horst zones
in the east and the west and a graben zone in the middle.
Two normal faults, Progo and Opak Faults, formed the
Bantul Graben. Based on the difference in gravity, Progo
fault is located near the Progo River in the west, and
Opak Fault is located in the border area of the Southern
Mountain (Baturagung Escarpment) to the east
(Nurwihastuti et al. 2014). The study area is located in
the transition zone between a flat area and the escarp-
ment zone in the Eastern part of Bantul Graben (Figure
2). Saputra et al. (2018) stated that Pleret sub-district is
located in the Eastern Horts of Bantul Graben, which has
experienced step faults due to the complexity of specific
geological structure.
Geographically, the study area can be divided into
three major zones: east, middle, and west. The west
and the middle part zones are dominated by extensive
flat areas, while the Eastern zone is dominated by
undulating and mountainous areas. The extensive
flat area mainly consists of Young Volcanic Deposits
of Merapi Volcano (Qmi) and few of them consist of
alluvium (from the denudational process in the
Eastern mountainous areas), which is located in the
foothills, near the undulating areas. The Eastern part
consists of mainly tertiary deposits of Semilir
Formation (Tmse). Only a few areas, including the
summit of the Baturagung Escarpment, belong to the
Nglanggran Formation (Tmn). Tmse formed approxi-
mately 27.82–23.03 million years ago. Tmse consists of
mainly interbedded layers of breccia pumice, tuff-
breccia, dacite tuff, andesite tuff, and tuffaceous clay,
whereas the Tmn was deposited in early Miocene
approximately 23.03 to 11.608 million years ago.
Tmn was deposited in parallel on the top of Tmse.
Tmn is mainly distributed in the summit of
Baturagung Escarpment in the Eastern part of the
study area. Both formations were generated from the
eruption of a nearby ancient volcano (Bronto et al.
2008; Winarti and Hartono 2015; Pandita, Sukartono,
and Isjudarto 2016). Bronto et al. (2018) found that
both Tmse and Tmn can be classified as pyroclastic
density flow from the ancient volcano that might be
located in the east part of the study area.
In terms of geomorphology, three major groups of
landforms can be identified in the study area: structural,
fluvial, and denudation. The structural landforms are
indicated by the existence of the Baturagung
Figure 2. The location of study area and relationship with the east horst of Bantul Graben.
GEO-SPATIAL INFORMATION SCIENCE 3
Escarpment in the east part. The intensive denudation
process occurred in the middle to upper slope of the
escarpment. Some triangle facets formed due to the
strong erosion processes in the middle to upper slope
of the escarpment. The fluvial process occurs in the
middle part of the study area along the Opak River and
in the narrow flat area near the undulating area and
Baturagung Escarpment. The fluvial process along the
Opak River and narrow flat area produced an extensive
alluvial plain and a colluvial plain, respectively. The
alluvial plain consists of Qmi, whereas the colluvium
consists of Qa (denudation material from Semilir and
Nglanggran Formation). Figure 3 depicts the geomor-
phological aspect of the study area.
3. Method
3.1. Scope of the analysis
Situated in a complex geological and geomorphological
zone, the study area is prone to several hazards, with
frequent and strong seismic activity. Pleret Sub-District is
located approximately 250 km north of the active Sunda
Megathrust and intersected by an active inland fault
namely Opak Fault. The middle part of the study area
is prone to soil amplification because this area is domi-
nated by the dense quaternary material from the Merapi
Volcano. The abundance of shallow groundwater in this
area can also potentially generate liquefaction.
Furthermore, the mountainous area in the east part of
the study area is also prone to slope stability issues and
coseismic landslides due to the type of lithology and
intensive erosion. Based on these considerations, the
multi-risk study several types of hazards such as earth-
quake, soil amplification, liquefaction, and coseismic
landslide aspects.
3.2. Data
The study mainly used the historic earthquake data from
the United States Geological Survey (USGS) between
1900 and 2019 to support the probabilistic seismic
hazard analysis and liquefaction analysis. The fieldwork
data of the outcrop study (Saputra et al. 2018) and
seismic vulnerability index (kg) from micro tremor mea-
surement (Daryono 2011) were used in the earthquake
and liquefaction analysis, respectively. The direct mea-
surement data of groundwater were obtained from
household well measurements. Some boreholes and geo-
electric data were used to support the analysis of ground-
water condition in the middle part of the study area. The
rainfall data from surrounding weather stations,
a 1:25,000 topographic map, and detailed geology map
based on remote sensing analysis were used to determine
the coseismic landslide-prone area (Saputra et al. 2015,
2016). Furthermore, detailed land use data and building
damage were used to analyze the vulnerability level. The
land use data were generated using the visual interpreta-
tion of the latest QuickBird image (Saputra et al. 2017).
The data for the building damage caused by the 2006
Yogyakarta earthquake were obtained from previous
Figure 3. The general landform of study area and surroundings.
4A. SAPUTRA ET AL.
research conducted by Kerle (2010). The list of data used
in this study is provided in Table 1.
3.3. Peak ground acceleration (PGA)
As much as 4,593 earthquake data from USGS were
used to obtain the PGA of 320 observation points
which were distributed entire study area. Kanai
(1966) (Douglas 2019) attenuation was used to gener-
ate the peak ground acceleration. Kanai attenuation
was used because this formula considers the funda-
mental period of the site, which is closely related to the
local site effect that can amplify earthquake shaking.
We calculated the Peak Ground Acceleration (PGA) in
the study area as follows:
α¼α1
ffiffiffiffi
T
p10α2MPlogRþQ(1)
whereα is peak ground acceleration, α1 is the first
constant value (5), α2 is the second constant value
(0.61), T is the fundamental period of the site, M is
the earthquake magnitude, R is the hypocenter (km),
and P and Q are the values from equations 2 and 3,
respectively.
P¼1:66 þ3:6
R
(2)
Q¼0:167 þ1:83
R
(3)
Thus if the equations 2 and 3 are substituted in equa-
tion 1, the Peak Ground Acceleration (PGA) can be
calculated using equation 4 below.
a¼5
ffiffiffiffi
T
p10 0:61Mð Þ 1:66þ3:60
R
ð ÞLog10Rþ0:1671:83
R
ð Þ
½(4)
3.4. The outcrop study
The outcrop study was conducted to get better under-
standing of geological condition in study area as only
geological map of Yogyakarta (scale 1:100,000) was
available in study area. The outcrop study was divided
into three main stage: pre-fieldwork analysis, fieldwork
activities, and post-fieldwork analysis. The pre-
fieldwork analysis included geological data (lithological
and geological structure) extraction from mainly
Yogyakarta Geology map scale 1:100,000 and Landsat
8 interpretation (Saputra et al. 2018). The visual inter-
pretation of QuickBird imagery was conducted in the
pre-fieldwork analysis of the outcrop study. The pri-
mary purpose of the interpretation was to identify the
location of the outcrop and determine the location for
outcrop observation. The aims of the fieldwork activ-
ities were to characterize the outcrop (identify the litho-
facies and qualitative grain size of each rock layer),
identify the micro fault, and to record the three-
dimensional (3D) surface model of the outcrop using
the structure from motion technique. The brief work-
flow of the outcrop study stage is provided in Figure 4.
3.5. Coseismic landslide assessment
The method proposed by Mora and Vahrson (1999)
was adopted to generate the coseismic landslide sus-
ceptibility (Saputra et al. 2016). There were two main
parameters, site characteristic factors and trigger fac-
tors, that could cause coseismic landslide occurrence.
The site characteristics consist of physical parameters
which are related to slope stability analysis such as
relief, geology, and soil humidity. Meanwhile the trig-
ger factors consist of seismic and rainfall intensity.
As shown in Table 1, other spatial data, such as
relief and lithology, were derived from contour and
geology maps, respectively. The contour map resolu-
tion was about 12.5 m. The geology map scale was
1:100,000. Saputra et al. (2016) interpreted some
remote sensing data and completed fieldwork observa-
tions to increase the scale of the geology map and
provided more detailed geology and lithology infor-
mation. The other parameters, such as natural humid-
ity of soil and rainfall intensity, were derived from
monthly average rainfall and annual rainfall intensity,
respectively. The detailed work flow of the coseismic
landslide assessment that was conducted by Saputra
et al. (2016) is provided in Figure 5.
3.6. The liquefaction assessment
Liquefaction is a secondary earthquake hazard that
often causes the worst damage to cities around the
world. In 1999, a 7.6 Mw earthquake occurred in
Chi-Chi and Taiwan, China. This earthquake gener-
ated massive liquefaction throughout Taiwan, China
and especially in Nantou, Wufeng, and Yuanlin pre-
fectures (Chu et al. 2004). In New Zealand, two strong
sequential earthquakes in September 2010 and
December 2011 in Christchurch caused massive lique-
faction across Christchurch (Morgenroth, Hughes,
and Curbinovski 2016). On 28 September 2018,
a strong earthquake struck Palu City, Central
Sulawesi, Indonesia. The Palu-Koro strike-slip fault
generated inland earthquake with a magnitude of 7.4
on the Richter scale at an earthquake depth of 10 km.
This earthquake mainly affected Palu Central City.
A 1.5-m tsunami occurred in Palu and Donggala
coastal areas. Massive liquefaction also occurred in
some parts of Palu, causing severe damage.
Liquefaction is closely related to geological and
geomorphological characteristics, as liquefaction
often reoccurs in the same locations (Youd and
Perkins 1987; Yasuda and Tohno 1988; Tatsuoka
et al. 1980). For instance, for the case of the 2006
GEO-SPATIAL INFORMATION SCIENCE 5
Table 1. The data used in this study.
Data Scale Format Source Use
The earthquake data history year 1900–2019
Magnitude greater than 5
Richter
Earthquake surrounding
Java
Tabular data .csv
format
United Stated Geological Survey (USGS) To generate PGA in earthquake hazard assessment and liquefaction
analysis
Predominant frequency of soil Bantul Regency Tabular data .xml
format
(Daryono 2011) To generate PGA and support liquefaction analysis
Rainfall data* Yogyakarta Province Tabular data .xml
format
Weather stations To determine coseismic landslide hazard
Geology map of Yogyakarta 1:100,000 Raster .jpg format Geological Research and Development Center,
Bandung Indonesia
To determine coseismic landslide hazard, outcrop distribution, and support
outcrop study
Contour map 1:25,000
Contour interval 12.5 m
Vector shapefile
format
Indonesian Geospatial information agency To generate coseismic landslide hazard
Outcrop distribution 1:35,000 Vector shapefile
format
Quick bird imagery interpretation Support outcrop study and earthquake susceptibility based on the
proximity of faults.
Micro faults and rock-layer characteristics of the
outcrop
1:25,000 Vector shapefile
format
Fieldwork Support outcrop study and earthquake susceptibility based on the
proximity of faults.
3d surface of the outcrop Unit outcrop Raster, format geo
tiff
Structure from motion Support outcrop study
Microtremor data Regency level Tabular .xml format (Daryono 2011) Liquefaction analysis
Groundwater table depth Sub-district level Vector shapefile
format
Household well survey
Geo-electric survey
Borehole
Liquefaction analysis
Building damage data due to 2006 Yogyakarta
earthquake
Provincial level Vector shapefile
format
(Kerle 2010) Deterministic model of building damage
Type of occupation Sub-district level Tabular report form Statistics Indonesia Population modeling.
Note: *Derived from the rainfall data (1983–2013) of 10 rainfall stations surrounds the study area (Barongan, Dogongan, Jatingarang, Karangploso, Piring, Tanjungtirto, Terong, Umbulharjo, Wates, and DPU Yogyakarta station).
6A. SAPUTRA ET AL.
Yogyakarta, Indonesia earthquake, the liquefaction
occurred in a flat area, with shallow ground water
table, not too far from the earthquake source, and
has a high seismic vulnerability index (Kg). Thus, an
area with these same characteristics are highly suscep-
tible to liquefaction. Using a similar approach (Yasuda
and Tohno 1988; Tatsuoka et al. 1980), the liquefac-
tion assessment was conducted in the study area. We
also analyzed the liquefaction from the share-wave
velocity and seismic vulnerability index point of
view. By knowing the share-wave velocity and seismic
index vulnerability, the sites exposed to liquefaction
risk can be identified (Daryono 2011; Yasuda and
Tohno 1988). The higher the value of ground share-
strain, the more easily the ground deformation occurs,
leading to a crack, liquefaction, and coseismic land-
slide (Huang and Tseng 2002; Ishihara 1982). The
relationship of the shear-strain value and the potential
liquefaction is outlined in Table 2.
Based on this concept, we combined Probabilistic
Seismic Hazard Assessment (PSHA) and the Ishihara
concept to analyze the potential liquefaction area in the
study area. Thus, the first step of this analysis was to
determine the Peak Ground Acceleration (PGA) based
on the historical earthquake data. The PGA value was
used to assess the ground shear strain based on the
earthquake vulnerability index (Kg). The ground shear
strain value in the study area was calculated using follow-
ing equation (Daryono 2011; Ishihara 1982):
Ground Shear strain ¼Kg106PGA (4)
The complete workflow used to assess the potential lique-
faction area in the study area is provided in Figure 6.
Figure 4. The general work flow of outcrop study.
Figure 5. The general workflow of coseismic landslide susceptibility zone [19].
GEO-SPATIAL INFORMATION SCIENCE 7
3.7. Multi-vulnerability assessment
The multi-vulnerability assessment used in this study
followed the previous research conducted by Saputra
et al. (2017). We applied logistic regression analysis to
predict the damage level of the residential buildings
based on the 27 May 2006 Yogyakarta earthquake
damage data. The main data used in that research were
earthquake damage data, QuickBird images, geologic
maps, and building footprint data. The earthquake
damage data were obtained from Kerle (2010), which
included building damage data (low, medium, and col-
lapsed) in impacted area (Jetis, Pleret, Imogiri, and
Bantul Sub-Districts). The QuickBird imagery was used
to obtain the land use data in more detail (the year 2012,
scale 1:25,000) based on the modified Anderson system
2002. Saputra et al. (2017) used the geology map of
Yogyakarta to extract additional data such as the lithol-
ogy, type of material, and distance from the epicenter of
Yogyakarta earthquake (27 May 2006).
In general, there were two main elements at risk
used in the analysis: population and building collapse
probability. The population data were obtained from
the local statistics agency (Badan Pusat Statistik (BPS)
or Statistics Indonesia), and the building collapse
probability was generated from the logistic regression
analysis. We followed four steps to conduct the multi-
vulnerability assessment. The first step was to extract
the land use data via the visual interpretation of
QuickBird imagery. The second step was to conduct
the probability of building collapse. The third step
involved creating a population distribution model
and the last step involved combining the building
collapse probability and population distribution into
multi-vulnerability analysis. The steps of the multi-
vulnerability analysis are provided in Figure 7.
We used 15 parameters to calculate the probability
of building collapse based on the 2006 earthquake
damage data. The parameters consist of dependent
variable (Y), which refers to the building damage,
and independent X variables (X1–X15) (Table 3). We
applied logistic regression based on the binary model
(0 and 1 for No or Yes, respectively). The input data
(the spatial building damage data) were converted into
the binary system. For instance, the building ID 505
has the characteristics of wood structure, asbestos or
zinc roof material, located within 8 km of the earth-
quake epicenter, has a Semilir geological condition,
and experienced moderate damage, was converted
into binary code as shown in Table 4.
3.8. Multi-hazard and risk assessment
The approach to generate the multi-hazard and risk
assessment followed the general concept of risk assess-
ment and reduction. Risk is the function of hazard and
vulnerability, which can be expressed as follows:
R¼HV(5)
Table 2. The shear-strain value and soil dynamic.
Size of strain 10
–6
10
−5
10
−4
10
−3
10
−2
10
−1
Phenomena Wave, Vibration Crack, Diff Settlement Landslide, Soil Compaction, and liquefaction
Dynamic Properties Elasticity Elasto Plasticity Repeat-effect Speed-effect of Loading
Source: Ishihara (1982)
Figure 6. The potential liquefaction assessment.
8A. SAPUTRA ET AL.
where R is risk, H is hazard, and V is vulnerability.
For the multi-hazard and risk scheme of this study,
we referred to Alkema et al. (2018) who considered
hazards as the combination of ground motion, soil
amplification, liquefaction, and coseismic landslide.
In terms of vulnerability, we focused only on building
block (building collapse probability) and the popula-
tion (the population distribution). The multi-hazard
Figure 7. The general workflow of multi-vulnerability analysis.
Table 3. Variables used.
No Variable Type data Value Other name
Dependent variable (Y)
1 Building damage
●Not damaged (Y1)
●Moderately damaged (Y2)
●Collapsed (Y3)
Ordinal 1
2
3
Damage
Independent variable (X)
1 Wood structure (X1) Binary (0 or 1) Structure = 0
2 Unreinforced masonry (X2) Binary (0 or 1) Structure = 1
3 Reinforced masonry (X3) Binary (0 or 1) Structure = 2
4 Asbestos or zinc roof (X4) Binary (0 or 1) Roof = 0
5 Cement tile roof (X5) Binary (0 or 1) Roof = 1
6 Clay tile roof (X6) Binary (0 or 1) Roof = 2
7 Concrete slap roof (X7) Binary (0 or 1) Roof = 3
8 Within 8 km of the epicenter (X8) Binary (0 or 1) Distance = 1
9 Between 8.1 and 10 km (X9) Binary (0 or 1) Distance = 2
10 Between 10.1 and 15 km (X10) Binary (0 or 1) Distance = 3
11 Greater than 15 km (X11) Binary (0 or 1) Distance = 4
12 Semilir Formation (Tmse) (X12) Binary (0 or 1) Geology = 0
13 Alluvium (Qa) (X13) Binary (0 or 1) Geology = 1
14 Young Merapi Volcanic deposit (Qmi) (X14) Binary (0 or 1) Geology = 2
15 Nglanggran Formation (Tmn) (X15) Binary (0 or 1) Geology = 3
GEO-SPATIAL INFORMATION SCIENCE 9
and risk assessment of earthquakes and other related
secondary hazards applied in this study are provided
in Figure 8.
Thus, based on Figure 8, equation 5 was modified to
accommodate the multi-hazard aspect in this study.
Equation 6 shows how the multi-hazard and risk were
generated in this study:
Multi Risk ¼Multihazard PþSA þLþCLð Þ:
Multivulnerability BV þPVð Þ (6)
where P is PGA value, SA is soil amplification, L is
liquefaction, CL is a coseismic landslide, BV is build-
ing vulnerability, and PV is population vulnerability.
4. Results and discussion
4.1. Peak ground acceleration (PGA) results
Based on the Kanai attenuation, we used the historical
data of over 3,481 earthquakes with a magnitude
greater than 5 on the Richter Scale, which occurred
between 1900 and 2017, to determine that the peak
ground acceleration of study area ranges from 531.04
to 967.66 cm/s
2
. Theoretically, the higher the PGA
value in a particular area, the higher the degree of
damage probability when an earthquake occurs
(Yamazaki and Matsuoka 2018; Walter et al. 2008).
Based on this attenuation, we found higher PGA
values in the middle part of study area, extending to
northeast and southeast along the Opak River, which
is closely associated with the location of the Opak
Fault. Lower PGA values were found in the northwest
part in the Yogyakarta City direction. This distribu-
tion of PGA values is inversely related to the dominant
period of the soil. The direct microtremor measure-
ments that were recorded by Daryono (2011)
explained that the highest (0.84) predominant period
of soil in the study area is located in the northwest
area. The predominant soil period and PGA are pro-
vided in Figure 9(a,b), respectively.
4.2. Results of outcrop study
Based on the outcrop study that was conducted by
Saputra et al. (2018), the study area has a complex
geological structure condition. We found significant
evidence of fault displacement including the great
normal fault in this area – the Opak Fault. Saputra
et al. (2018) found the same rock of the Nglanggran
Formation (Tmn is described as a volcanic breccia,
lava flow containing breccia agglomerate rock and
tuff). The main location of the Nglanggran formation
is at the summit of Baturagung Escarpment. However,
based on the field observation, the authors found an
isolated hill of Nglanggran Formation located in the
center of study, separated by approximately 4.24 km
from the main location of the Nglanggran Formation.
Saputra et al. (2018) found that the middle parts of the
study area are more vulnerable to ground amplifica-
tion. At least 30 fault displacements were found in the
middle part of the study area, with a maximum dis-
placement of 2.39 m. Most of the faults are typically
normal faults, and only a few of them are strike-slip
faults. The direction of most of the micro faults is
similar to the direction of the Opak Fault. The outcrop
study conducted by Saputra et al. (2018) revealed that
the Segoroyoso Village, Srumbung Sub Village, the
middle part of Bawuran Village, the middle part of
Pleret Village, and the middle part of Wonolelo
Village are vulnerable due to the complex geological
structure and ground amplification. The map of the
fault evidence derived from the outcrop study and an
example of maximum fault displacement in Srumbung
Sub-Village are provided in Figures 10 and 11,
respectively.
4.3. Liquefaction results
4.3.1. The groundwater condition
Based on the integrated direct measurement of house-
hold wells during the 2015 rainy season, some geo-
Table 4. An example of the binary model for the variables. Building ID 505 = wood structure, asbestos roof, within 8 km, and
Semilir Formation.
ID X1 X2 X3 X4 X5 X6 X7 X8 X9 X10 X11 X12 X13 X14 X15 Y
505 √ - - √ - - - √ - - - √ - - - 2
505 1 0 0 1 0 0 0 1 0 0 0 1 0 0 0 2
Figure 8. The multi-hazards and risk assessment. Source .Alkema et al. (2018)
10 A. SAPUTRA ET AL.
electricity observations, especially fluvial plains and local
people drilling activities in South Segoroyoso, the study
area is dominated by shallow groundwater areas (less
than 10 m). This abundance of groundwater is concen-
trated in the middle part of the study area (Wonokromo,
Pleret, middle Segoroyoso, middle Bawuran, and middle
Wonolelo). Geomorphologically, this area is dominated
by fluvial and colluvial plains, which consists of Alluvium
from the Young Merapi Volcanic Deposit (Qmi) and
Colluvium from the denudational material of Semilir
and Nglangran mountainous area (Qa), respectively
(Figure 12).
In line with the household wells observations, the
geo-electricity observation showed that the study area
has abundant groundwater. Shallow groundwater can
be found at depths of 2.5–15 m in the colluvial plain in
the narrow plain in the east part of the study area.
From the borehole data, we determined a similar var-
iation in groundwater depth. In alluvial and colluvial
plains, shallow groundwater can be found at depths of
1–15 m. This shallow aquifer is known as a confined
aquifer. Deep aquifers can be found at depths of
approximately 80–110 m.
4.3.2. The liquefaction
Based on the ground-shear strain and the potential lique-
faction by using Ishihara method, we determined that the
areas more prone to liquefaction are the middle part of
study area, especially in Kerto, Keputren, and Kanggotan
Villages (Figure 13). This area has the highest ground-
Figure 10. Fault reconstruction and lineament in study area.
Figure 9. (a) The predominant period of soil and (b) PGA in the study area.
GEO-SPATIAL INFORMATION SCIENCE 11
shear strain of 9,221.43 × 10
−6
, found on the alluvial plain
of young Merapi volcanic deposits with a sediment thick-
ness of 130 m and groundwater depth of around 2.31 m.
4.4. Coseismic landslide
Based on previous studies (Saputra et al. 2015, 2016),
the study area could be classified into four Coseismic
Landslide Susceptibility Levels (CLSLs): negligible,
low, moderate, and medium. The negligible zone is
the safest area in the study area. This zone is mainly
distributed in the west to the middle part of the study
area. This zone is located in a flat to gentle slope area
and in the alluvial plain zone, colluvial plain, and
natural levee of the Opak River. The low CLSL zone
is associated with the narrow plain that is located in
the border area between the flat and mountainous area
in the east part study area. The moderate zones are
mainly located in the middle slope of Baturagung
Escarpment, which features weakly to strongly eroded
denudational hills of Semilir Formation. The medium
zone – the most unstable areas – are located along the
upper slope of Baturagung Escarpment, which consists
of Semilir Formation. This result is in line with that
reported by Samodra et al. (2016), who found that the
middle slope a greater probability of rock fall occur-
rence in the Sewu Mountainous Area, which is in the
west part of Yogyakarta. The complete coseismic land-
slide susceptibility map is shown in Figure 14.
4.5. Multi-vulnerability results
The first element at risk used in this study was resi-
dential buildings. The vulnerability of residential
buildings was determined from the logistic regression.
There were two datasets used in this study. The first
dataset was the building damage data due to the
Yogyakarta earthquake. These data include some
Figure 11. Fault displacement in outcrop number 15, Srumbung Sub-Village.
Figure 12. The groundwater depth (left); Groundwater characteristic of location 2, 3, and 4 based on the drilling data (right).
12 A. SAPUTRA ET AL.
information about attributes such as location (x and
y), the owner, the building structure, the roof material,
geology, the distance from the epicenter, and the level
of damage. These data were obtained from field obser-
vations right after the disaster occurrence. The second
dataset was the building footprint data of existing
buildings after rehabilitation and reconstruction pro-
cess. These data were generated from a visual
interpretation process. These data included the same
attributes such as location (x and y), the type of build-
ings structure, roof materials, geology, and the dis-
tance from the epicenter.
The other examined element at risk was popula-
tion density. We used the dasymetric technique to
determine the population density per unit land use.
Several scenarios, such as the population of each
Figure 13. Potential liquefaction model in study area.
Figure 14. The potential Coseismic Landslide Susceptibility Level (CLSL) in study area.
GEO-SPATIAL INFORMATION SCIENCE 13
land use unit in daytime and nighttime both on
weekdays and holidays, were applied to determine
the real population condition. The land use units
were generated from visual interpretation based on
the Anderson classification. The vulnerability
results of each element at risk are explained
below. The results of multi-vulnerability are pro-
vided in Figure 20.
4.5.1. Land use based on visual interpretation of
Quick bird imagery
Based on the modified Anderson system, we found 29
land use units in the study area. This land use unit
covers the classification level III, which is suitable for
interpretation of QuickBird or medium-altitude data
captured between 3100 and 12,400 m or 1:20,000 to
1:80,000 map scales. Based on this classification, we
classified the study area into 29 land use units: aban-
doned mining sites, agricultural wetlands, canal, cem-
etery, cemetery on wetland, commercial strip
development, educational institution, government
centers, harvested cropland, health institution, inac-
tive cropland, light industrial, not built up, open areas,
other agricultural, other institutional (mosque), pas-
tureland, poultry farm, residential high density, resi-
dential low density, residential medium density, rural
single unit, road, shrub land, specialty farm, stone
quarries, stream, traditional market, and wetlands.
The distribution and the total area of the land use
units are provided in Figure 15.
4.5.2. Building unit collapse probability
There were three main results based on the logistic
regression: the model fitness results, pseudo R
2
, and
estimated parameter values. The model fitness shows
that the models (binary coding) for both independent
and dependent variables were significant, that the models
fit, and can be used for further analysis (Table 5). The
pseudo R
2
result provides information about how far the
model can explain the results. Based on the logistic
regression, the model can explain at least 33.20%,
which was indicated by the Nagelkerke value (Table 6).
This means 66.80% of the dependent variables cannot be
explained from this model. Thus, some additional para-
meters need to be added in future research.
The estimated parameter values (Table 7) show that
the threshold values of the damage categories are 1.529
and 2.426 for damage categories 1 and 2, respectively.
This means the predicted response value (Y*i) were
categorized as follows:
(1) Damage category 1 (low damage) if Y*i ≤ 1.529
(2) Damage category 2 (moderate damage) if
1.529 < Y*i < 2.246
(3) Damage category 3 (high damage or collapsed)
if Y*i ≥ 2.246
Figure 15. The land use map and its total area.
14 A. SAPUTRA ET AL.
The Y*i value can be calculated from the for-
mula that was derived from the value of each
parameter in Table 7. The formula followed the
general regression equation. Therefore, from Table
7, the general equation of damage category is
expressed as:
Yi¼ 0:255 X1ð Þ þ 0:685 X2ð Þ þ 0X3ð Þ
þ0:43 X4ð Þ þ 0:749 X5ð Þ þ 1:634 X6ð Þ
þ0X7ð Þ þ 2:265 X8ð Þ þ 0:949 X9ð Þ
þ0:744 X10ð Þ þ 0X11ð Þ 1:413 X12ð Þ
0:64 X13ð Þ þ 1:507 X14ð Þ þ 0X15ð Þ (7)
where the Y*i is the prediction value of damage cate-
gory, X1 is wood structure, X2 is unreinforced
masonry, X3 is reinforced masonry, X4 is asbestos
and zinc, X5 is cement tile, X6 is clay tile roof, X7 is
concrete slap roof, X8 is distance within 8 km of
epicenter, X9 is distance between 8.1 and 10 km of
epicenter; X10 is distance between 10.1 and 15 km of
epicenter, X11 is distance greater than 15 km from
epicenter, X12 is Semilir formation (Tmse), X13 is
alluvium (Qa), X14 is young volcanic deposits of
Merapi Volcano (Qmi), and X15 is Nglanggran
Formation (Tmn).
Based on the visual interpretation of 17,512 build-
ings, there were only 33 combinations of house char-
acteristics. Each combination has specific building
attributes and damage category. For instance, the
combination number 5 (Table 8) – wood structure,
clay tile roof, distance more than 15 km from the 2006
epicenter, and young volcanic deposit of Merapi
Volcano. Based on the equation 7, the combination 5
has damage category as follow:
Y*i = −0.255 (1) + 0.685 (0) + 0 (0) + 0.43 (0) +
0.749 (0) + 1.634 (1) + 0 (0) + 2.265 (0) +
0.949 (0) +0.744 (0) + 0 (1) – 1.413 (0) – 0.64
(0) + 1.507 (1) + 0 (0)
Y*i = −0.255 + 1.634 + 0 + 1.507
Y*i = 2.886
This means that combination of building attribute
number 5 has damage category 3 (high damage or
collapsed). The results of damage category calculation
using equation 7 of all building unit entire study area
can be seen in Figure 16.
The building damage probability of each building
unit was obtained by applying the damage category
resulted from equation 7 into equations 8, 9, and 10
below (Agresti 1990; Hosmer and Lemeshow 2000).
Probability of no damage Y1ð Þ ¼ 1
1þeYithreshold1ð Þ
(8)
Probability of moderate damage Y2ð Þ
¼1
1þeYithreshold2ð Þ Y1 (9)
Probability of high damage or collapsed Y 3ð Þ
¼11
1þeYithreshold2ð Þ ;(10)
Thus, for example, combination number 5 (wood
structure, clay tile roof, distance more than 15 km
from the 2006 epicenter, and young volcanic deposit
of Merapi Volcano) was categorized to damage cate-
gory 3 or collapsed. By using equations 8,9, and 10, the
building damage probability can be determined. For
Table 5. Model fit information.
Model
–2 Log
Likelihood
Chi-
Square
Degree of
freedom (df)
Significance prob-
ability (sig)
Intercept
Only
3055.003
Final 707.190 2347.813 11 0.000
Table 6. Pseudo R
2.
Model Value
Cox and Shell 0.264
Nagelkerke 0.332
McFadden 0.194
Table 7. Parameters estimation.
Estimate SE Wald df Sig.
95% Confidence Interval
Lower Bound Upper Bound
Threshold [Damage = 1] 1.529 0.478 10.247 1 0.001 0.593 2.465
[Damage = 2] 2.426 0.478 25.761 1 .000 1.489 3.363
Location [Structure = 0] (X1) –0.255 0.107 5.647 1 0.017 –0.465 –0.045
[Structure = 1] (X2) 0.685 0.070 95.306 1 0.000 0.547 0.822
[Structure = 2] (X3) 0 . . 0 . . .
[Roof = 0] (X4) 0.430 0.610 0.497 1 0.481 –0.766 1.627
[Roof = 1] (X5) 0.749 0.484 2.390 1 0.122 –0.201 1.698
[Roof = 2] (X6) 1.634 0.461 12.541 1 0.000 0.729 2.538
[Roof = 3] (X7) 0 . . 0 . . .
[Distance = 1] (X8) 2.265 0.236 91.712 1 0.000 1.801 2.728
[Distance = 2] (X9) 0.949 .106 80.173 1 0.000 0.742 1.157
[Distance = 3] (X10) 0.744 .084 78.154 1 0.000 0.579 0.909
[Distance = 4] (X11) 0 . . 0 . . .
[Geology = 0] (X12) –1.413 0.122 134.395 1 0.000 –1.652 –1.174
[Geology = 1] (X13) –0.640 0.145 19.402 1 0.000 –0.925 –0.355
[Geology = 2] (X14) 1.507 0.117 165.861 1 0.000 1.278 1.737
[Geology = 3] (X15) 0 . . 0 . . .
GEO-SPATIAL INFORMATION SCIENCE 15
instance, the combination number 5 has a Y*i value of
2.886. Therefore, the building damage probability is:
Y1 (no damage)_ = 1
1þ2:718 2:8861:529ð Þ
Y1 (no damage) = 0.011
Y2 (moderate damage) = 1
1þ2:718 2:8862:246ð Þ 0:011
Y2 (moderate damage) = 0.334
Y3 (high damage or collapsed) = 1 1
1þ2:718 2:8862:246ð Þ
Y3 (high damage or collapsed) = 0.655
Based on the calculation above, the building com-
bination number 5 has a predicted damage category of
3 (collapsed) with a probability of high damage or
collapsed of 0.655. The safest building type in the
study area is the building with the attributes of rein-
forced masonry structure, asbestos or zinc roof mate-
rial, located between 10.1 and12 km from the
earthquake source, and located above the Semilir
Formation. This buildings unit have the probability
of collapse only 0.07. The most vulnerable buildings
are those with a combination of reinforced masonry
structure, clay tile roof material, located between 8.1
and 10 km from the epicenter, and located above the
young volcanic deposits of the Merapi Volcano. The
probability of collapse of this type reached 0.84.
Building block probability of collapse was deter-
mined from converting the probability of collapse
of a building unit to the block or land use scale.
The average value was applied to convert the col-
lapsed probability of building unit to the building
block. The illustration of how to convert the col-
lapse probability of building units to building block
is provided in Figure 17. Based on the results,
residential blocks located in the western part of
the Opak Fault are rated as high probability of
collapse between 0.60 and 0.73. The middle part
of a study area, to the left and right of the Opak
River, scored the highest values (very high) of
building collapse probability, ranging from 0.74 to
0.84. The Eastern part tended to have very low to
moderate probability of collapse (0.08–0.59)
Figure 16. Illustration of how to classify the buildings existence into several damage categories.
16 A. SAPUTRA ET AL.
because this area is mountainous areas consisting
of compact lithological characteristics. The distri-
bution of building collapsed probability in the
study area are provided in Figure 18.
4.5.3. Population distribution model
The distribution of the population in several scenarios
(day time, night time, both on weekends and week-
days) was defined based on dasymetric analysis
through the simple percentage of the occupation of
the local people. Based on the data, the people of Pleret
Sub-District are predominantly casual workers, stu-
dents, unemployed, entrepreneurs, and farmers, at
17.92%, 20.51%, 18.86%, 20.16%, and 23.99% of the
population, respectively. Thus, based on the dasy-
metric analysis, the population distribution on each
block of land use was estimated. For instance, during
work hours, the population in Wonokromo Village,
the densest village in the study area, tends to distribute
into three major land use units: settlements, schools,
and commercial areas. The population can be esti-
mated as 27.7% staying in commercial areas, 16.65%
in schools, and 16.47% in settlement areas. The rest of
the population distributes into other land use units.
Similar to this condition, the population in other vil-
lages, such as Pleret, Segoroyoso, Bawuran, and
Wonolelo, mainly distribute into commercial areas,
school, settlements, and agricultural areas. The popu-
lation distribution in all villages in the study areas
showed no significant difference among the scenarios
used. The population distribution under several sce-
narios is provided in Figures 19 and 20.
4.6. Multi-risk results
The multi-hazard zonation was generated by applying
the summation function of all hazard parameters (PGA,
the proximity of the faults, liquefaction, and coseismic
landslide). The summation function was determined
based on the consideration of the selected hazards
being secondary earthquake hazards that can amplify
the earthquake impact. Based on the analysis, we found
that the index of multi-hazard in the study area is
between 0.46 and 0.85, with minimum and maximum
values of 0 and 1, respectively. Based on this analysis,
the study area is dominated by the multi-hazard value
0.61, which can be classified as a moderate multi-hazard
zone. Most of this zone is located in the flat area in the
middle part of the study area. These areas are located
near Opak Fault, have a complex geological structure,
and soft surface sediment of alluvium or colluvium. In
terms of PGA, this zone has PGA values around
Figure 17. The example of converting the building unit collapse probability to block unit.
GEO-SPATIAL INFORMATION SCIENCE 17
676.59–822.12 gal and low ground shear strain, which
means vulnerable to ground motion and liquefaction
occurrence. The total area of this zone is 39.12% of the
study area (Figure 21).
The multi-risk value was obtained by adding the
multi-temporal information of the population distri-
bution. The results show that both in daytime or night
time on weekdays, the middle part has the highest
multi-risk value because this area has a higher multi-
hazard index value, high population density, and most
of the buildings inside the block have a higher prob-
ability of collapse. The holidays scenarios produced
different results. Based on the holiday scenarios,
during both daytime and night time, the population
tend to distribute within the residential blocks. For
example, Segoroyoso Village has a slightly higher
multi-hazard risk assessment value at night. The com-
plete multi-risk value picture is provided in Figure 22.
The multi-risk models produce slightly different
results between weekday and holiday multi-risk
index and produce distinctive results for daytime and
night time. On daytime, both on weekdays and holi-
days, the high multi-risk index is mainly distributed in
the southern part of Wonokromo village, in the mid-
dle part of Pleret village, and some areas of Wonolelo
village. During the night time both on weekdays and
Figure 19. Population distribution during (a) holiday day time, (b) night time, (c) work day day time, and (d) work day night time.
Figure 18. The collapse probability for building blocks in study area.
18 A. SAPUTRA ET AL.
holidays, high multi-risk index values are mainly dis-
tributed in most of the residential areas, as almost all
people stay in residential area at night both on holi-
days and weekdays. This model shows (Figure 22, left
picture) that the south part of Wonokromo village, the
middle part of Pleret village, the middle part of
Bawuran, Segoroyoso, and Wonolelo villages scored
high multi-risk index values.
Figure 20. Multi-vulnerability during (a) holiday daytime, (b) night time, (c) work day day time, and (d) work day night time.
Figure 21. Multi-hazards zonation in study area. Hazards information: PGA was derived based on Kanai attenuation by using USGS
earthquake historical data (1900–2019). Liquefaction was generated based on the ground shear strain value of USGS earthquake
historical data (1900–2019). Soil amplification was assessed based on the proximity to active faults which was generated from
outcrop study. Coseismic landslide resulted from the coseismic landslide model grade 2 proposed by Mora and Vahrson (1999) and
the multi-hazard index was generated from raster overlay function with benefits standardization method.
GEO-SPATIAL INFORMATION SCIENCE 19
5. Conclusions and recommendations
Based on the results above, the multi-risk index of the
study area for earthquakes and other secondary
hazards differs, being influenced by PGA, ground
motion, liquefaction, coseismic landslide, population
distribution, and building collapse probability. Based
on the CLSL analysis, the study area includes negligi-
ble, low, moderate, and medium zones. The negligible
and low CLSL areas are mainly distributed in the
extensive flat region in the west part of the study
area. The more vulnerable areas are mainly distributed
in the mountainous area in the middle part of the
study area. This study also shows that the middle
slope of Baturagung Escarpment is more prone to
coseismic landslide occurrence. These findings are in
line with the landslide study that was conducted in the
western part of Yogyakarta (Walter et al. 2008).
The study area has a complicated geological struc-
ture. Based on the lineament analysis, we found var-
ious lineament features of valleys, ridges, river, sudden
changes of river direction, ridge lines, scarp face, and
straight drainage segments. All these lineaments are
associated with the existence of faults (Samodra et al.
2016; Gannouni and Gabtni 2015). Based on the earth-
quake fracture displacements recorded on the outcrop
surface, the middle part of the study area, around the
north to the middle part of Segoroyoso and the middle
part of Pleret and Bawuran, has a complex fault con-
figuration as indicated by the dense area of faults and
highest displacement. Additionally, based on the
lithostratigraphic analysis, the outcrop deposits are
ignimbrite type that originated from the ancient vol-
cano (tertiary) in the east part of study area.
Based on the shear strain analysis that was derived
from the PGA analysis, the study area, and especially
the extensive flat area in from middle to west parts, is
dominated by shallow groundwater up to 10 m deep
with high ground shear-strain values. This means this
area is vulnerable to liquefaction occurrence.
In term of vulnerability analysis, we concluded that,
statistically, buildings with reinforced masonry struc-
ture (RM), clay tile roof material, built above Young
Volcanic Deposits of Merapi Volcano (Qmi), and
8.1–10 km from the epicenter of the 2006 earthquake
have a higher probability of collapse. In terms of popu-
lation distribution, there is no significant difference in
population density on each land use unit between holi-
days and weekdays. However, we found that the popu-
lation density on each land use unit is different between
daytime and night time. Thus, population density
affects the multi-risk index, which demonstrated
a different pattern in daytime and night time.
The multi-hazard risk results show that residential
houses that are located in the center part of study area,
including the center of Wonokromo, Pleret, Bawuran,
and Wonolelo, and the north part of Segoroyoso
Villages, have a high multi-hazard and risk index
(>0.60) both during daytime and night time on both
weekdays and holidays. This means this area is more
Figure 22. Multi-risk value in study area.
20 A. SAPUTRA ET AL.
prone to earthquake hazards and other related second-
ary hazards.
This study shows the general characteristic of
multi-hazard and risk in study area. This study is
a preliminary assessment before conducting more
detailed investigation. This study has some limitations
and we have recommendations for future research.
Longer-term earthquake occurrence data are needed
to accommodate the longer reoccurrence period of
larger earthquake. The amplification due to the
micro fault was only based on the proximity analysis;
thus, further detailed investigations are required to
calculate the size the amplification for complex geolo-
gical structures when earthquakes occur from sur-
rounding sources. A good inventory of landslide and
coseismic landslide is needed to improve the coseismic
landslide hazard result. Additionally, several site-
specific observations need to be recorded to support
the coseismic landslide data. A deterministic seismic
hazard analysis model needs to be created to obtain
a better understanding of the type of fault and
mechanism that might trigger coseismic landslides
when the earthquakes occur. Some limitation of lique-
faction analysis was also found in this study such as
this study was still used Ishihara model to approach
the liquefaction which is only suitable for preliminary
assessment. We need to conduct some Standard
Penetration Tests (SPT) and cone Penetration Tests
(CPT) for verification and for detail site investigation.
Pore-water pressure, seasonal effects on groundwater,
and storm and heavy rainfall need to be considered as
parameters in future research. The statistical analysis
of building collapse vulnerability needs to be
improved by adding more parameters as independent
variables, such as soil type, the shape of buildings, and
the age of buildings. In terms of multi-hazard and risk
analysis, the cascading effect of earthquakes and other
related secondary hazards need to be examined to
better explain multi-hazard and risk aspects. The
effects of earthquakes also need to be considered on
the karst landform located in the southeast part of the
study area. Strong earthquakes can cause sinkhole
collapse, which leads to rock fall and block gliding.
Lastly, the further analysis such as loss estimation is
required to support the risk evaluation, management,
and reduction process.
Acknowledgments
The author is thankful to University of Canterbury New
Zealand and Universitas Muhammadiyah Surakarta,
Indonesia who provided adequate reference for this project.
Notes on contributors
Aditya Saputra is an assistant professor in Dept.
Geography, Universitas Muhammadiyah Surakarta,
Indonesia. He received the PhD degree from Dept. of
Geography, University of Canterbury, New Zealand. His
research interests are hazard modeling by using remote
sensing and GIS. He collected the data and carried out the
analysis in this paper, also he was drafted this manuscript.
Christopher Gomez is professor in Graduate School of
Maritime Sciences, Laboratory of Volcanic Risks at Sea
Kobe University, 5-1-1 Fukaeminami-machi, Higashinada-
ku, Kobe, Japan. His research interest are tsunami processes
and hazards at the coast, applied subsurface geophysics,
debris flows, lahars, and landslides, sediment transfers and
related hazards in mountain and volcanic terrain. He sup-
ported AS collecting the data and interpreting the results.
Ioannis Delikostidis is a senior lecturer in Dept. of
Geography, University of Canterbury, New Zealand. His
research interest are pedestrian context-aware navigation,
crowdsourced data applications, human geography, and
geographic information system. He supported AS to inter-
preting the results and revising the manuscript.
Peyman Zawar-Reza is professor in Dept. of Geography,
University of Canterbury, New Zealand. His research inter-
est are earth systems science, air pollution climatology/
meteorology, mountain meteorology. He supported AS to
revise the manuscript.
Danang Sri Hadmoko is associate professor in Dept. of
Environmental Geography, Faculty of Geography,
Universitas Gadjah Mada, Indonesia. His research interest
is Landslide Hazard, Natural Hazard Study, Volcanic
Hazard. He supported AS to interpret the results and revise
the manuscript.
Junun Sartohadi is professor in Department of Soil Science,
Faculty of Agriculture, Universitas Gadjah Mada, Indonesia.
He is expert on earth sciences, disaster mitigation, physical
geography and environmental geoscience. His research
interest are geohazard from soil science perspective, land
resource development planning, pedogeomorphology, soil
and water conservation. He supported AS to revise the
manuscript.
ORCID
Aditya Saputra http://orcid.org/0000-0003-1549-3832
Christopher Gomez http://orcid.org/0000-0002-1738-
2434
Ioannis Delikostidis http://orcid.org/0000-0002-1347-
4075
Peyman Zawar-Reza http://orcid.org/0000-0002-6120-
3317
Danang Sri Hadmoko http://orcid.org/0000-0003-0360-
5577
Junun Sartohadi http://orcid.org/0000-0002-0059-8335
References
Agresti, A. 1990. Categorical Data Analysis. New York:
JohnWiley & Sons..
Alkema, D., L. G. J. Boerboom, S. Ferlisi, and L. Cascini.
2018. Caribbean Handbook on Risk Information
Management ACE-EU, 93. GFDRR (Global Facility for
Disaster Reduction and Recovery. http://www.charim.
net/print/
Ashadi, A. L., S. I. Kaka, G. Yuliyanto, and S. I. Kaka. 2015.
“Probabilistic seismic-hazard for Central Java province,
GEO-SPATIAL INFORMATION SCIENCE 21
Indonesia.” Bulletin of the Seismological Society of
America 105 (3): 1711–1720. doi:10.1785/0120140277.
Bronto, S., S. Mulyaningsih, G. Hartono, and B. Astuti.
2008. “Gunungapi Purba Watuadeg: Sumber Erupsi
Dan Posisi Stratigrafi.” Jurnal Geologi Indonesia 3 (3):
117–128. doi:10.17014/ijog.3.3.117-128.
Budimir, M. E. A., P. M. Atkinson, and H. G. Lewis. 2014.
“Earthquake and landslide events are associated with
more fatalities than earthquake alone.” Natural Hazards
72: 895–914. doi:10.1007/s11069-014-1044-4.
Chopra, S., D. Kumar, B. K. Rastogi, P. Choudhury, and
R. B. S. Yadav. 2012. “Deterministic seismic scenario for
Gujarat region, India.” Natural Hazards 60: 517–540.
doi:10.1007/s11069-011-0027-y.
Chu, D. B., J. P. Stewart, S. Lee, J. S. Tsai, P. S. Lin, B. L. Chu,
R. B. Seed, S. C. Hsu, M. S. Yu, and M. C. H. Wang. 2004.
“Documentation of soil conditions at liquefaction and
non-liquefaction sites from 1999 Chi-Chi (Taiwan,
china) earthquake.” Soil Dynamic Earthquake
Engineering 24 (9–10): 647–657. doi:10.1016/j.
soildyn.2004.06.005.
Civico, R., C. A. Brunori, P. M. De Martini, S. Pucci,
F. R. Cinti, and D. Pantosti. 2015. “Liquefaction suscept-
ibility assessment in fluvial plains using airborne lidar:
the case of the 2012 Emilia earthquake sequence area
(Italy).” Natural Hazards 15 (11): 2473–2483.
doi:10.5194/nhessd-3-4527-2015.
Daryono. 2011. “Indeks Kerentanan Seismic Berdasarkan
Mikrotremor Pada Setiap Satuan Bentuklahan Di Zona
Graben Bantul Daerah Istimewa Yogyakarta.” PhD diss,
Universitas Gadjah Mada, Indonesia.
Douglas, J. 2019. “Ground motion prediction equation
1964-2019. Https://www.stratH.ac.uk/staff/douglas-
johndr/ Gannouni, S., And H. Gabtni. 2015. “structural
interpretation of lineaments by satellite image processing
(Landsat TM) in the region of Zahret Medien (Northern
Tunisia).” Journal of Geographic Information System 7:
119–127. doi:10.4236/jgis.2015.72011.
Gomez, C., D. E. Hart, P. Wassmer, K. Imai, H. Matsui, and
M. Shimizu. 2019. “Coastal evolution, geomorphic pro-
cesses and sedimentary records in the Anthropocene.”
Forum Geografi 33: 1. doi:10.2391/forgeo.v33i1.7551.
Hadmoko, D. S., F. Lavigne, J. Sartohadi, P. Hadi, and
W. Winaryo. 2010. “Landslide hazard and risk assessment
and their application in risk management and landuse
planning in eastern flank of Menoreh mountains,
Yogyakarta Province, Indonesia.” Natural Hazards 54
(3): 623–642. doi:10.1007/s11069-009-9490-0.
Hosmer, D. W., and S. Lemeshow. 2000. Applied Logistic
Regression. 2nd ed. New York: Wiley and Sons.
Huang, H., and Y. Tseng. 2002. “Characteristics of soil
liquefaction using H/V of Microtremor in Yuan-Lin
Area, Taiwan,China.” Terrestrial, Atmospheric, and
Oceanic Sciences 13 (3): 325–338. doi:10.3319/
TAO.2002.13.3.325(CCE).
Irsyam, M., I. W. Sengara, F. Aldiamar, S. Widiyantoro,
W. Triyoso, D. H. Natawidjaja, E. Kertapati, I. Meilano,
S. M. Asrurifak, and M. Ridwan. 2012. “Ringkasan Hasil
Studi Tim Revisi Peta Gempa Indonesia.” http://www.
preventionweb.net/files/14654_AIFDR.pdf
Ishihara, K., eds. 1982. Evaluation of Soil Properties for Use
in Earthquake Response Analysis, in Geomechanical
Modelling and in Engineering Prectice. Rotterdam,
Netherland: Balkema.
Kerle, N. 2010. “Satellite-based damage mapping following
the 2006 Indonesia earthquake-how accurate was it?”
International Journal of Applied Earth Observation and
Geoinformation 12: 466–476. doi:10.1016/j.
jag.2010.07.004.
Marano, K. D., D. J. Wald, T. I. Allen, K. Marano,
D. J. Wald, and T. I. Allen. ed. 2010. “Global earthquake
casualties due to secondary effects: a quantitative analysis
for improving rapid loss analyses.” Natural Hazard 52
(2): 319–328. DOI:10.1007/s11069-009-9372-5.
Mora, S., and W. Vahrson. 1999. “Micro zonation for slope
instability in Costa Rica.” Revised ed. Manual for
Zonation on Seismic Geotechnical Hazards, 55–65.
Tokyo, Japan:Japanese Geotechnical Society.
Morgenroth, J., M. W. Hughes, and M. Curbinovski. 2016.
“Object-based Image analysis for mapping
earthquake-induced liquefaction ejecta in Christchurch,
New Zealand.” Natural Hazards 82: 763–775. doi:
doi:10.1007/s11069-016-2217-0.
Nurwihastuti, D. W., J. Sartohadi, D. Mardiatno, U. Nehren,
and R. Restu. 2014. “Understanding of earthquake
damage pattern through geomorphological approach:
a case study of 2006 earthquake in Bantul, Yogyakarta,
Indonesia.” World Journal Engineering and Technology 2:
61–70. doi:10.4236/wjet.2014.23B010.
Pandita, H., Sukartono, A. Isjudarto, and S. Sukartono.
2016. “Geological identification of seismic source at
Opak fault based on stratigraphic section of the southern
mountains.” Forum Geografi 30 (1): 77–85. doi:10.23917/
forgeo.v30i1.1508.
PUSGEN. (Pusat Studi Gempa Nasional/National Centre of
Earthquake Studies of Indonesia). edited by. 2017. “Peta
Sumber Dan Bahaya Gempa Indonesia Tahun. Pusat
Penelitian Dan Pengembangan Perumahan Dan
Permukiman, Badan Penelitian Dan Pengembangan,
Kementrian Pekerjaan Umum Dan Perumahan,” 19.
Rakyat:Bandung, Indonesia.
Ren, Z. L., and W. Liu. 2013. “Typical disaster chains and
cascading effect caused by 2012 Xinyuan Ms 6.6 earth-
quake and Ms 8.0 earthquake of China.” In Proceedings of
Sixth China-Japan-US Trilateral Symposium on Lifeline
Earthquake Engineering, Chengdu, China, May-1 June 28
May-1 2013: 66–72. doi:10.1061/9780784413234.009.
Samodra, G., G. Chen, J. Sartohadi, D. S. Hadmoko,
K. Kasama, and M. A. Setiawan. 2016. “Rockfall suscept-
ibility zoning based on back analysis of rockfall deposits
inventory in Gunung Kelir, Java.” Landslides 13: 805.
doi:10.1007/s10346-016-0713-7.
Saputra, A., C. Gomez, I. Delikostidis, P. Zawar-Reza,
D. S. Hadmoko, J. Sartohadi, and M. A. Setiawan. 2018.
“Determining earthquake susceptible areas southeast of
Yogyakarta, Indonesia-outcrop analysis from structure
from motion (Sfm) and Geographic Information System
(GIS).” Geosciences 8: 132. doi:10.3390/geosciences804132.
Saputra, A., C. Gomez, D. S. Hadmoko, and J. Sartohadi. 2016.
“Coseismic Landslide Susceptibility Assessment Using
Geographic Information System.” Geoenvironmental
Disaster 3: 27. doi:10.1186/s40677-016-0059-4.
Saputra, A., T. Rahardianto, M. D. Revindo, I. Delikostidis,
D. S. Hadmoko, J. Sartohadi, and C. Gomez. 2017.
“Seismic vulnerability assessment of residential buildings
using logistic regression and geographic information sys-
tem (Yogyakarta-indonesia).” Geoenvironmental Disaster
4: 11 (2017). doi:10.1186/s40677-017-0075-z.
Saputra, A., J. Sartohadi, D. S. Hadmoko, and C. Gomez.
2015. “Geospatial assessment of coseismic landslide in
Baturagung Area.” Forum Geografi 29 (2): 99–114.
doi:10.23917/forgeo.v29i2.693.
Tatsuoka, F., T. lwasaki, K. Tokida, S. Yasuda, M. Hirose,
T. lmai, and M. Kon-no. 1980. “Standard penetration
22 A. SAPUTRA ET AL.
tests and soil liquefaction potential evaluation.” Soils
Foundation 20 (4): 95–111. doi:10.3208/
sandf1972.20.4_95.
Walter, T. R., R. Wang, B. G. Luehr, J. Wassermann,
Y. Behr, S. Parolai, A. Anggraini, et al. 2008. “The
26 May 2006 magnitude 6.4 Yogyakarta earthquake
south of Mt. Merapi Volcano: did Lahar deposits amplify
ground shaking and thus lead to the disaster?”
Geochemistry, Geophysics, Geosystem 9 (5): Q05006. doi:
doi:10.1029/2007GC001810.
Westen, C. 2011. “Risk Analysis.” In Multi Hazard Risk
Assessment, Distance Education Course, Guide Book, edi-
ted by C. van Westen, D. Alkema, M. C. J. Damen,
N. Kerle, and N. C. Kingma, 6-1 – 6-33. Twente,
Netherland:United Nations University-ITC School on
Disaster Geo-Information Management (UNU-ITC
DGIM.
Winarti and G. H. Hartono. 2015. “Identification of volcanic
rocks in Imogiri Yogyakarta based on subsurface geologic
data.” In Proceedings of 2nd International Conference on
Engineering of Tarumanegara (ICET), 1–9. Jakarta,
Indonesia: Faculty of Engineering, Tarumanegara University.
Yasuda, S., and I. Tohno. 1988. “Sites of reliquefaction
caused by the 1983 Nihonkai-Chubu earthquake.” Soils
Foundation 28 (2): 61–72. doi:10.3208/
sandf1972.28.2_61.
Youd, T. L., and D. M. Perkins. 1987. “Mapping of liquefac-
tion severity index.” Journal of Geotechnical Engineering
113 (11): 1374–1392. doi:10.1061/(ASCE)0733-9410-
(1987)113:11(1374).
GEO-SPATIAL INFORMATION SCIENCE 23