Professional Documents
Culture Documents
Soil Dynamics and Earthquake Engineering
Soil Dynamics and Earthquake Engineering
A R T I C L E I N F O A B S T R A C T
Keywords: Assessment of potential seismic risk and losses in urban environments is necessary for several purposes ranging
Seismic hazard from risk mitigation to city and regional planning. It is well known that loss estimation should be performed
Regional velocity model within an interdisciplinary setting involving earth sciences and engineering. Field experience from recent events
Ground motion simulation
worldwide shows that the spatial variability of seismic damage is due to the combined effects of earthquake
Building vulnerability
Seismic damage
source properties, local site conditions and structural characteristics. In this study, a scenario-based multi-input
Gaziantep damage estimation framework in an urban region was utilized for the Gaziantep city center (southeastern
Turkey Turkey) that is located in a region of high seismic hazard with no large events in the instrumental era. Initially, a
thorough geological and seismo-tectonic assessment of the area was performed followed by estimation of two
critical scenario events with moment magnitude (Mw) of 6.5 and 6.6 on nearby active faults. Then, a regional
velocity model was compiled from regressions of existing regional geotechnical and seismic data in terms of the
VS30 parameter. As the next step, field surveys for the assessment and classification of buildings in the study area
were performed followed by vulnerability analyses. As the last step, the mean damage ratios were computed at
198 neighborhoods within the city center. The results indicate not only a high hazard but also high risk in the
Gaziantep area due to the combination of close proximity to the faults along with local site effects and building
fragilities.
* Corresponding author.
E-mail addresses: ararzu@metu.edu.tr (A. Arslan Kelam), shaghkn@civil.uminho.pt (S. Karimzadeh), yousefibavil.karim@metu.edu.tr (K. Yousefibavil), hakgun@
metu.edu.tr (H. Akgün), aaskan@metu.edu.tr (A. Askan), altug@metu.edu.tr (M.A. Erberik), mustafakockar@hacettepe.edu.tr (M.K. Koçkar), opekcan@metu.edu.tr
(O. Pekcan), ciftcihcr@gmail.com (H. Ciftci).
https://doi.org/10.1016/j.soildyn.2021.107129
Received 31 May 2021; Received in revised form 18 December 2021; Accepted 19 December 2021
Available online 28 December 2021
0267-7261/© 2021 Published by Elsevier Ltd.
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
The objective of this study is to assess the seismic hazard and po
tential losses anticipated in the existing building stock in Gaziantep city
of Turkey based on a regional evaluation ranging from seismic sources to
soil models; from hazard to structural response and damage. The scope
includes preparation of a future earthquake master plan for the evalu
ation of seismic hazard and vulnerability of the building stock within the
Gaziantep Central Districts. For this purpose, an interdisciplinary
approach has been proposed for the Gaziantep region, which has not
been studied in detail before. The fundamental steps of the approach
used herein are characterization of the faults, development of a regional
soil model, simulation of critical scenario earthquakes on nearby faults,
on-site evaluation and classification of the existing building stock and
estimation of potential seismic losses in the form of mean damage ratios,
respectively. Fig. 1 presents a flowchart that summarizes the approach
used in this study and its main components mentioned previously.
2
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 2. a) Location map of Gaziantep city, b) provinces of Gaziantep, c) central provinces of Gaziantep and the study area.
Fig. 3. Mw > 4 earthquakes that have been recorded between 1900 and 2020 in the vicinity of the study area that is located in the Gaziantep region.
3. Assessment of seismic source characteristics and critical source map was prepared. However, it should be noted that these maps
scenario earthquakes of the study area display information that lead to inconsistencies in the fault geometry,
locations, fault models, identification and fault segmentation due to the
For seismic source characterization, areas were extracted from the differences in the scales utilized. Due to this scaling problem, where the
zones that revealed major seismic activities for the period ranging from scale varies depending on whether the study performed is local, regional
1900 to 2020 that is presented in Fig. 3. According to these seismic or country-wide, there tends to be differences in the fault geometry,
source zones, the available fault maps from the literature and the location, fault mechanism and fault segmentation as compiled and
authorized government institutions (i.e., the General Directorate of presented as an output. Therefore, for this study, several fault maps that
Mineral Research and Exploration of Turkey) were digitized, and a new exist within the study area [10,22–29] were considered and a database
3
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 4. Active fault map that displays the major faults and their sub-segments (modified from Refs. [10,22–27].
Fig. 5. Developed seismic source zones of Gaziantep region and its surroundings.
4
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Table 2
Expected moment magnitudes of the Maximum Credible Earthquake (MCE) per each fault section that are calculated by the Wells and Coppersmith [35] relation for
each fault segment (N: Normal Fault; LS: Left-Lateral Strike-Slip Fault RS: Right-Lateral Strike-Slip Fault) and the shortest distances of the fault zones to the study area.
Seismic Source Zone Fault or Segment Length (km) Mechanism (Orientation) Mw of MCE (Average) Distance (km)
(Wells and Coopersmith, 94)
earthquake scenarios are developed for each seismic source zone. Coppersmith [35] relationships. The assigned magnitudes and the
However, the seismological data available regarding DSSZ is not suffi shortest distances to the identified faults are presented in Table 2.
cient to characterize the entire fault zone. Therefore, the seismic source Considering the probable earthquakes with largest Mw within the
characteristics along with the tectonic deformation of these faults were shortest distances to the fault zones, it is inferred that DSSZ and EAFSZ
considered [23,26] and the lengths of the fault segments were delin are the most critical seismic zones in the study area. The Dead Sea Fault
eated based on the suggestions of Slemmons [33] and dePolo and is an active N-S striking left-lateral transform fault that runs from the
Slemmons [34]. Then, the expected magnitudes of the maximum cred Gulf of Aqaba to Antakya in the northern part of the study area and
ible earthquakes were calculated for each segment with the Wells and extends approximately 1000 km [36–38] (Fig. 4). According to
5
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Kirkpinar Faults which have the same orientation with the Afrin Fault,
are critical in terms of performing a seismic hazard assessment of the
study area. Although the historical activity of the DSSZ was not well
documented, particularly in the Gaziantep region, the evidence of the
earthquakes can be seen in several studies. Ambraseys and Barazangi
[43] have reported the historical destructive earthquakes in the Bekaa
valley in 1202 and 1759, along the Ghab Fault in 1170, 1404, 1796, and
1872, in the Karasu valley that have been created by the Amanos Fault in
1822.
The East Anatolian Fault Zone, which is the second critical zone in
the region, has a length of 580 km and has played a significant role on
the geodynamic evolution and seismicity of the area [8,22,43–46]
(Fig. 4). The fault zone runs from the east of Karlıova, where it crosses
the NAFZ, continues through Göynük Valley and reaches Antakya.
Westaway [26] calculated the slip rate of the EAFSZ as 8 mm/year.
However, according to McClusky et al. [40]; the slip rate is 9 ± 1
mm/year [24]. The most important earthquakes recorded along this
fault zone in the last several centuries are the 1513 Pazarcık Earthquake,
the 1822 Kahramanmaraş Earthquake, the 1866 Karlıova Earthquake,
the 1872 Antakya Earthquake, the 1874 Gezin Earthquake, the 1875
Sivrice Earthquake, the 1893 Çelikhan Earthquake, the 1905 Pötürge
Earthquake, the 1971 Bingöl Earthquake, the 1977 Palu Earthquake, the
June–July 1986 Sürgü Earthquakes [44], the 1998 Ceyhan Earthquake
and the 2020 Sivrice Earthquake [47].
Based on the fault segment characteristics of the DSSZ and the EAFSZ
that are presented in Table 2, it is observed that there are fault segments
capable of generating earthquakes greater than 7.0 in the proximity of
the study area. However, as far as the critical scenario earthquakes on
the entirety of the seismic source zones are concerned, these fault seg
ments are deemed to be distant (i.e., >50 km) and hence, the effects of
earthquakes which can occur on these distant segments is expected to be
relatively low in the Gaziantep city center. On the other hand, fault
Fig. 6. Active fault zones along with the fault segments that were used in the segments located at a closer distance (i.e., <10 km) that are capable of
scenario simulations in the study area. generating moderate size earthquakes with magnitudes of Mw = 6.5–6.6
have to be considered in this study since they can induce destructive
Garfunkel et al. [39] and McClusky et al. [40]; the average slip rate is effects due to their proximity to the study area. Along this line, the most
1.5 cm/year. In the south, the Dead Sea Fault has a border with the Red critical fault segment is the Logan-02 Fault Segment of DSSZ that has the
Sea [41]. In the north, it borders the southeastern part of Anatolia to the closest distance (around 1 km) and is capable of generating an earth
west and passes through the Amik plain that is located between Kırıkhan quake of Mw = 6.6. In addition, in the Dead Sea Seismic Region, the
and Hassa. The Dead Sea Fault evolves into a complex fault system with Kirkpinar-01 Fault Section which is only 9 km away from the study area
three parallel systems starting from the Turkish-Syrian border and that is capable of generating an earthquake with a magnitude of Mw =
merges with the EAFZ in the north. These parallel systems are named as 6.5 can also cause devastating seismic effects in Gaziantep (Fig. 6). For
the Amanos Fault, East Hatay Fault and Afrin Fault [24,26,42]. The Kilis, these reasons, the worst possible scenarios with large magnitudes on two
Berent, Gaziantep and Logan Faults were formed following the Afrin nearby faults mentioned above, which have the highest potential in
Fault, where DSSZ branches to the east [23,26]. According to Coskun creating earthquake hazards, have been used to assess the potential
and Coskun [23]; DSSZ is the fault that has affected the structural damage in Gaziantep city. Based on two scenario earthquakes, the
development of the Gaziantep Basin, where especially the Logan and damage levels were evaluated depending on the spatial distribution in
the city, using local sources, soil models, ground motion simulations,
Fig. 7. Flowchart showing the successive steps followed to develop a regional VS30 model.
6
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
7
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 8. The VS30 velocity model of the 198 neighborhoods with an intense building stock. The model was prepared to serve as a basis for the central districts
of Gaziantep.
VS30 less than 0.25 indicate that the geologic units are constrained and the existing boring studies were particularly concentrated in certain
geographically. Therefore, the COV values estimated in the Gaziantep regions (i.e., Metro and light railway lines, parcel studies from large
Basin are consistent with the smaller values reported by Wills and Silva private companies, school and public building studies, highway studies,
[66]; Wills and Clahan [65] and Kockar et al. [48] because the meth etc.). In particular, the general distribution or deposition of lithology
odology of this study also focuses on geologic units that are geograph within the boring logs appeared to be relatively variable due to the
ically constrained. Finally, the calculated VS30 results have been presence of different units ranging from recent alluvial sedimentary
combined together with the directly determined seismic measurements units to competent rock in the central districts of Gaziantep. Because of
of the field survey to develop an extended, consistent and these reasons, after the digitizing process, to clarify the availability of
well-distributed database to prepare a regional VS30 model. Then, all of data, a grid system of 500 m × 500 m in size was introduced into a GIS
these VS30 results of the study area that have been combined together environment in the study area (Fig. 7 – Step 6). Initially, each gridded
were digitized (Fig. 7 – Step 5). area has been considered based on the available VS30 data points. The
In the evaluation of these digitized data in a large area of this scale, it most reliable and descriptive results have been selected and this process
should be stressed that the spatial distribution of VS30 values was not as was used in the development of the seismic velocity model. It should be
homogeneous as desired in the project area. Moreover, the available noted that the thickness of alluvium varies between 5 and 15 m within
data points of the VS30 results were mostly located in different rock units the boring logs that are 30 m deep. Therefore, contrary to the expected
8
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
results, the calculated Vs results possess relatively higher values based Table 5
on the boreholes put down in these younger alluvium units due to the Input parameters for the simulation of scenario earthquakes on the Logan and
coarser particle size and the shallow depositional characteristics of this Kirkpinar faults.
unit. In addition, the COV values of VS30 for Quaternary deposits support Model parameter Logan-02 scenario Kirkpinar-01 scenario
the argument that these deposits are geographically constrained to the Moment 6.6 6.5
Gaziantep Basin. magnitude (Mw)
In the final step, based on the distribution of the available data into Upper edge of the 36.9929 N, 37.1818 E 37.1731 N, 37.2187 E
grids, a hierarchical VS30 mapping strategy has been used (Fig. 7 – Step fault plane
Distance to the top 2 km 2 km
7). By combining different constraints with the VS30 such as geological
of the fault
unit, topography, and slope gradient, a more consistent and well- Fault geometry 43 and 90
◦ ◦
11◦ and 90◦
distributed database for the seismic zonation map has been developed. (Strike and dip,
In the literature, similar assessments have been mentioned by various respectively)
researchers [48,50,51,67,68]. Regarding these studies, it can be inferred Fault dimensions 18 km × 10 km 16 km × 10 km
(Length x width)
that systematic geomorphological (i.e., topography and slope gradient), Crustal S-wave 3500 m/s 3500 m/s
geological, and geospatial data along with logical prediction variables velocity
can be used as a reasonable method for the estimation of the VS30. These Rupture velocity 3000 m/s 3000 m/s
variables are referred herein as the “geology-based” variables which rely Crustal density 2800 kg/m3 2800 kg/m3
Stress drop 56 bars 56 bars
on the premise that the mapped near-surface geologic units can be
Pulsing percentage 40% 40%
reliable indicators of VS30 values within that unit wherever they are Quality factor Q = 59.7f 0.87 Q = 59.7f 0.87
mapped. Additionally, by considering the gradient of the topography Geometric R− 1.1 , R ≤ 30 km R− 1.1 , R ≤ 30 km
and choosing ranges for the slope maximizes the correlation with spreading factor R− 1 , R > 30 km R− 1 , R > 30 km
shallow shear-velocity observations. In other words, topography and Duration model T = T0 + 0.05R T = T0 + 0.05R
slope gradient should be a diagnostic indicator of VS30 since more Site filters Kappa model (κ0 = 0.035) Kappa model (κ0 = 0.035)
competent (i.e., high-velocity) materials are more likely to maintain a Amplification factors of Amplification factors of
steep slope whereas deep basin sediments are deposited primarily in Boore and Joyner [75] as a Boore and Joyner [75] as a
environments with very low gradients [53]. In this respect, Wald et al. function of Vs30 values function of Vs30 values
obtained from the regional obtained from the regional
[53] have proposed a hierarchical VS30 mapping strategy, combining
velocity model developed in velocity model developed in
different constraints on VS30 values such as geological unit and topo this study this study
graphic slope. Thompson et al. [69] have also created a similar
geo-statistical method that defines an average VS30 for each geological
unit. Consequently, the methodology followed in the central districts of the region in order to compute potential damage rates. These faults are
Gaziantep is a hierarchical model similar to these studies. Then, a displayed in the regional map of the study area in Fig. 6.
regional zonation map for local seismic velocities was prepared by In this study, a stochastic finite fault method using a dynamic corner
applying the Kriging method through evaluating the VS30 results that frequency [70] was used to obtain ground motions from the scenario
was measured in the central districts of Gaziantep (Fig. 7 – Step 8). This events. This method has been previously verified against records of past
methodology may be helpful in differentiating the characteristics of the earthquakes in various regions in the world both in time and frequency
generalized local seismic velocity that is almost always consistent with domains (e.g., Refs. [71–74].
the near-surface geological age, soil type, and depositional character. Finite fault models are composed of smaller subfaults which are all
As a consequence of the required model steps that were discussed in assumed to be point sources with ω-2 displacement spectrum repre
this section, a local seismic velocity model of the central districts of senting the earthquake source. Among various available source models,
Gaziantep that possesses a footprint area of 193 km2 and intense the stochastic point-source assumes ω-2 model where the source spec
building stock has been prepared. In order to evaluate this building stock trum at high frequencies falls off proportional to ω-2. One of these point
on a neighborhood scale, the VS30 values of 198 neighborhoods in the sources on the fault plane is considered to be the hypocenter from which
central districts was determined, and a regional shear wave velocity map waves are radiating. While the point sources are modeled in the fre
was created (Fig. 8). All these results were used as the main parameters quency domain as a function of source, path and site filters, their
in the seismic hazard analysis and scenario earthquake simulations since contribution in an acceleration time history recorded at an observation
local site conditions have a direct impact on the distribution of damage point is computed in the time domain kinematically with appropriate
rates that are likely to occur as a result of possible earthquakes in the delays.
structures located in the study area. It should be noted that the regional In this study, due to lack of strong motion data from large events in
seismic velocity model prepared for this study is a spatial distribution the region, initially alternative generic and regional quality factor and
(seismic zonation) map prepared as a base map for this study which geometrical spreading models from the literature were tested in the
indicates the general situation. The methodology for the integrated simulation of ground motions. Then, the parameters which provide the
analysis presented is suitable for seismic hazard analysis on a regional closest fit to the ground motion models (GMMs) were selected as
scale to present the general estimates of the potential damage distribu preferred parameters. Since there is no instrumental-era earthquake
tions and to indicate areas that require more detailed investigations. with finite-fault effects and a slip model in the region, we have preferred
to use random slips to encounter the slip variability on the fault in our
5. Ground motion simulations of selected scenario earthquakes simulations. The amplification factors of Boore and Joyner [75] were
used as a function of VS30 values obtained from the regional velocity
It is important to assess the ground motions for past and potential model developed in this study. Next, the simulation parameters used for
events in a hazardous region. Since, there is no large earthquake in the the scenario events on the Logan and Kirkpinar faults that are presented
instrumental era in Gaziantep, in this study simulations to obtain ground in Table 5 were determined.
motions which could occur during potential earthquakes have been Stochastic finite fault method produces a random single horizontal
utilized. In this study, critical scenario events of Mw = 6.6 and Mw = 6.5 component at each observation point. For each selected scenario, one set
have been modeled on two segments of the Dead Sea Fault zone which of simulation is performed. A horizontal ground motion record was
are the Logan-02 and Kirkpinar-01 faults, respectively. These simulated computed at each of the 198 nodes in the study area for each of the
records were then used to estimate the seismic forces on the structures in Logan and Kirkpinar scenarios. It should be noted that the spatial scale
9
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 9. A comparison of the attenuation of the simulated peaks with the ground motion models for the Logan-02 (Mw = 6.6) scenario earthquake (To be consistent
with the governing site class in the study region, Vs30 = 620 m/s is used in GMMs).
for the ground motions was the neighborhood level. The soil model was data including long-period motions. Finally, overall the simulated data
initially constructed at a 500-m grids but then to be consistent and to with the GMMs reveal that the simulation models yield peak motions,
provide input to the ground motion and building models, the VS30 data which remain within ±1 standard deviation of the median values ob
was lumped at the neighborhood centers. tained from the GMMs. It is also observed that the uncertainty of
It is necessary to validate simulated motions against observed re simulated PGA and PGV is lower than those of the GMMs. This is mostly
cords. However, there are no recorded earthquakes in the region during attributed to the region-specific character of simulated ground motions
the instrumental era. Thus, the simulated peak ground motions are which are well-constrained with accurate source, path and site
compared against empirical ground motion relations in this study. Since modeling.
the simulated motions are single horizontal component per node, their Next, the simulated waveforms were determined at 5 dummy sta
peaks are compared against the median from the GMMs. For this pur tions in the region with varying source-to-site distances and local site
pose, one regional and one global prediction equation which are the conditions as presented by Fig. 11. These dummy stations are located in
models by Boore and Atkinson [76] and Akkar and Cagnan [77] were the Ogumsogut, Beykent, Yaprak, Kucukkizilhisar, and Osmanli neigh
selected for both scenarios as presented in Figs. 9 and 10. The compar borhoods. For both scenarios, it was observed that the ground motion
isons are made in terms of Peak Ground Acceleration (PGA), Peak amplitudes were larger at softer soil sites located at closer distances to
Ground Velocity (PGV) and also Spectral Acceleration (SA) at T = 0.33 the fault planes. However, the Logan scenario yields larger amplitudes
as well as T = 1 s. The last spectral parameter represents the demand due to its proximity to the site.
around the fundamental period of the low rise residential buildings Finally, the spatial distribution of the peak ground motion values in
which constituent the common typology of the building stock in the terms of PGA, PGV, 5% damping SA (T = 0.33 s) and SA (T = 1 s) were
region. Comparison of simulated data against the GMMs show that for determined for both scenarios in the study region as displayed in Figs. 12
both scenarios, the simulated data in terms of PGA, PGV and SA (T = and 13. It was observed that the scenario event with Mw = 6.6 on the
0.33 s) are slightly above the median GMM values. For SA (T = 1 s), the Logan fault yielded large ground motion amplitudes that reached 0.8 g
simulations are below the median. This is mostly due to the fact that the and 60 cm/s in terms of PGA and PGV, respectively. This was more
region is not laid on very soft soils with lower VS30 values. Thus, the site evident for the northwestern residential districts. On the other hand, the
amplifications factors used in the simulations do not exhibit long period spectral accelerations at T = 0.33 s and T = 1 s in the eastern region were
amplifications. Another reason is that EXSIM models only the S-waves around 0.3 g and 0.1 g, respectively. For the western region that is close
and not surface waves, which are rich in terms of long-period content. to the Logan fault, these values were 1 g and 0.4 g, respectively. For the
Thus, the simulated spectral amplitudes at longer periods, SA (T = 1 s), Kirkpinar fault with Mw = 6.5, the peak ground motion amplitudes were
are lower in amplitude than the median GMM which employ empirical lower due to the larger distance from the causative fault and relatively
10
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 10. A comparison of the attenuation of the simulated peaks with the ground motion models for the Kirkpinar-01 (Mw = 6.5) scenario earthquake (To be
consistent with the governing site class in the study region, Vs30 = 620 m/s is used in GMMs).
lower Mw value. Hence, the maximum simulated PGA, PGV, SA (T = Detailed information from the districts of interest reveals the
0.33 s) and SA (T = 1 s) reached 0.4 g, 25 cm/s, 0.5 g and 0.2 g, following statistical data: Percentages of masonry and RC residential
respectively. buildings in the first central district are 73% and 27%, respectively. Most
of the masonry buildings are 1–2 story structures whereas RC buildings
6. Characteristics of building stock in the study region generally have 1-4 stories. Most of these buildings that were observed to
be built between 1980 and 2000 may be regarded as none or low-
One of the most important ingredients of seismic risk assessment is engineered structures. The characteristics of the buildings in the sec
the collection of building inventory data. It is especially important in ond central district are not much different: The percentages are 67% and
generating new fragility functions or employing existing ones that are 33% for masonry and RC buildings, respectively. However, the number
suitable for local building classes. The accuracy of regional seismic risk of mid-rise RC buildings (i.e., with 4–8 stories) are higher in the second
estimation directly depends on the resolution of the collected data. Thus, district. There are also few high-rise buildings in both districts which
a hybrid approach was pursued in this study to obtain the characteristics were constructed within the last decade.
of the building stock in the study region. The building data has been Next, building information was collected in the field. The TSI data
gathered from two different sources: data from the Turkish Statistical base yielded general information about the type of construction, year of
Institute (TSI) in the city and district levels as well as data from a field construction, number of stories and the construction material but it did
study. not give any information about the spatial distribution of these buildings
The residential and non-residential building data for the two districts within the region. Therefore, it was necessary to conduct a walk-down
that has been considered in this study is based on the values from the survey type rapid screening procedure to observe the district-wise dis
2000 Building Census [78] report and the extracted data from the TSI tribution of the building characteristics.
digital database after 2001 [79]. According to this data, 78% of the The field survey was conducted as follows: First, the satellite maps of
buildings in Gaziantep (i.e. 115,941 buildings) reside in two central the case study districts were examined to detect the distribution of the
districts. The number of buildings in the first central district is nearly building stock and to divide the study area that contained 198 neigh
twice of those in the second central district. In addition, 91% of the borhoods into 12 grids. An ID number was assigned to each neighbor
buildings in the first district and 86% of the buildings in the second are hood and this number was referred during the field study. Fig. 14 shows
used for residential purposes. Hence, giving consideration to the resi the street map along with a labelled version with neighborhood ID
dential buildings in these two central districts has provided invaluable numbers of a typical grid in the study region. The existing buildings in
information for the regional seismic risk assessment of Gaziantep City. the labelled districts were investigated using the prepared survey path.
11
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 11. Simulated waveforms at 5 dummy stations in Gaziantep due to the Logan (Mw = 6.6) and Kirkpinar (Mw = 6.5) scenario earthquakes.
At some locations, the TSI data was verified and at some other locations, earthquake codes in use during the construction era, whereas no
brand new building information was collected. After the street survey, a compliance stands for the case in which the building had been con
general picture of the building stock was obtained. It should be noted structed in a traditional manner without engineering skills. Then, the
that, similar to the ground motion simulations, the spatial scale of the percentages of building subclasses within each district were determined
building analyses is the neighborhood level. using the TSI data, satellite maps and field data. All the building classes
Accordingly, the central part of the study region, that is close to the with their abbreviations, properties and percentages in the total building
boundary between the two districts, is the old urban center of Gaziantep stock are presented in Table 6.
City. It is attractive with many touristic spots, but the buildings in this
area are rather old and in poor conditions. They are generally low-rise 7. Seismic fragility identification of the building stock in the
brick masonry structures, which have been constructed side by side study region
(Fig. 15a). The city seems to be expanding towards the south, where
there are mid-rise and high-rise RC frame structures with structural The final phase was to identify the neighborhood-level fragility of the
walls. The buildings in this region are new and are in good condition building stock in the study region by matching the building sub-classes
(Fig. 15b). The western side of the city is also composed of rather new, with the corresponding fragility sets. The fragility functions used herein
isolated, mid-rise apartment buildings. It is observed that both con are based on Karimzadeh et al. [3] where simulated full waveforms were
struction types (i.e., RC frame and masonry) have been used in this part used in equivalent single degree of freedom (ESDOF) analyses for the
of the city (Fig. 15c). The industrial center of the city is in the eastern derivation of regional fragility curves for Erzincan, Turkey. A validation
part. The buildings in this region have characteristics that are very of the simulated motions against records was possible in Karimzadeh
similar to the ones in the central part of Gaziantep. They are generally et al. [3] since damaging events occurred during the instrumental era
low-rise masonry buildings with ground floors for commercial use and with available strong motion data. In this study, since there were no data
upper floors for residential use (Fig. 15d). from an instrumental-era earthquake in Gaziantep, the simulated full
The next phase after the field survey was to classify the buildings waveforms were not used to derive fragility curves. Herein, only the
according to some major construction types. The common building simulated peak ground motion parameters (PGA and PGV) which were
construction typologies in Turkey have already been obtained in pre validated against ground motion models were used in existing fragility
vious studies [3,80–82]. The field survey in the Gaziantep provinces curves to obtain damage probabilities per each ground motion level. It
revealed that the properties of the residential buildings in the region needs to be noted that both regions show similar faulting type and
were in accordance with some of these predefined building sub-classes. building type characteristics.
Thus, these typologies could be safely used for building classification in The specific fragility generation approach is based on non-linear
this study. Accordingly, 21 different building sub-classes have been time history analyses of equivalent single-degree-of-freedom simula
considered for the existing RC and masonry buildings in the study region tions which are obtained from the idealized capacity curves of actual
in terms of three major parameters: construction type (RC frame, RC multi-degree-of-freedom (MDOF) models for that specific building sub-
shear wall, RC dual, i.e., frames with shear walls, masonry), number of class. In order to carry out the dynamic analyses of ESDOF systems, a
stories (low-rise, mid-rise and high-rise for RC and 1, 2, 3 for masonry) multi-parameter hysteretic model [83] has been employed, which was
and building status, i.e., compliance with earthquake resistant design capable of simulating a wide range of structural components and sys
and construction (full, partial or no compliance). Full compliance im tems. This model has been verified previously with various experimental
plies that the building had been fully designed and constructed ac data [84].
cording to the general principles of seismic design enforced by the In this study, the major parameters of the ESDOF model (i.e., period,
12
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 12. Spatial distribution of simulated PGA, PGV, SA (T = 0.33 s) and SA (T = 1 s) for the Logan (Mw = 6.6) scenario earthquake.
13
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 13. Spatial distribution of simulated PGA, PGV, SA (T = 0.33 s) and SA (T = 1 s) for the Kirkpinar (Mw = 6.5) scenario earthquake.
14
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 14. Satellite map (left) and labelled map (right) versions of a typical grid from the study region.
The fragility curve sets for various building types (RC frame, RC
shear-wall, RC dual and unreinforced masonry with prefixes RF, RW, RH
and MU, respectively) in Fig. 16 were observed to be very different from
each other. This was an anticipated observation since these building
types exhibit totally different behavior under seismic action with their
inherent superiorities and deficiencies. For instance, old and deficient
masonry buildings (MU3C in Fig. 16) are significantly more vulnerable
than earthquake resistant urban type masonry buildings (MU1A). In
addition, the limit states of masonry typologies are observed to be very
close to each other, indicating that their nonlinear behavior is limited
and these buildings reach failure state rapidly after they start to expe
rience damage. For RC frame buildings (RF1A and RF2C in Fig. 16), the
limit states are more distinct and separate from each other. These
fragility curves clearly reflect the importance of earthquake code con
formity. The fragility curves for RC shear wall (RW2A in Fig. 16) and RC
dual, i.e. frame + wall (RH2B in Fig. 16) building typologies seem to
reflect the inherent structural characteristics of these construction types.
15
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 15. Views of building stock in the a) central part (top left), b) southern side (top right), c) western side (bottom left), d) industrial part (bottom right), of the
study region.
Table 6
Building classes with their abbreviations, properties and percentages in the total building stock.
No Class Code Material Construction Type No. of stories Building Status Percent in total population (%)
DPMs are PGA and PGV since the fragilities are defined in terms of these neighborhood.
parameters. Each DPM corresponds to a building class as a function of Figs. 17 and 18 demonstrate the spatial distribution of MDR values of
various ground motion intensity levels. In order to express the DPM in scenario earthquakes on the Logan and Kirkpinar faults, respectively.
terms of a single value, MDR is defined as the weighted average of each The MDR distributions for the Logan-02 scenario indicate that medium
column of DS probabilities where CDR values are the weights. In this to severe damage states are anticipated at most locations in the city
study, initially, MDR values are computed for each existing building center, with MDR values around 50%–65%. These high values indicate
class at all neighborhoods. Then, a single MDR value is computed per anticipated high levels of damage at these locations in probable large
each neighborhood center by combining the individual MDR values earthquakes and point out that urgent assessment and intervention
according to the percentage of different building classes in that measures are necessary to avoid these potential losses. However, for the
16
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Fig. 16. Example fragility curve sets for the selected building sub-classes with their abbreviations: RF1A; high-code low-rise RC frame building, RF2C; low-code mid-
rise RC frame building, RH2B; moderate-code mid-rise RC dual building, RW2A; high-code mid-rise RC shear wall building, MU1A; high-code 1 story masonry
building, MU3C; low-code 3 story masonry building.
9. Conclusions
Table 7
A sample Damage Probability Matrix (DPM). In this study, seismic damage assessment for potential earthquakes
Damage State Central Damage Ground Motion Intensity Parameter (PGA, was performed in the central districts of Gaziantep using regional
(DS) Ratio (CDR %) PGV, Microseismic Intensity Values etc) seismic, geological and structural information. In this evaluation,
None 0 Damage Probabilities, Pr (DS,I) initially a regional velocity model was obtained from existing borehole
Light 5 data in the form of a VS30 map. Next, two potential scenario earthquakes
Moderate 30
with Mw = 6.6 and 6.5 were simulated, respectively on the Logan-02
Severe 70
Collapse 100 and Kirkpinar-01 segments of the Dead Sea Fault Zone. Then, the local
building stock was assessed in the field and categorized into various
seismic vulnerability classes. Finally, the mean damage ratios for the
same scenario event, MDR values of 0%–5% are obtained at some neighborhoods that are located in the two central districts were
neighborhoods despite the closest source-to-site distances. This obser computed by combining the simulated ground motion records of the
vation is attributed to the high-quality of construction at these locations scenario events and the building fragility information. The main find
with significantly less vulnerability despite the high ground motion ings of this study are summarized as follows:
levels. When the scenario event on the Kirkpinar-01 segment is
considered, at most neighborhoods, none to light damage states are • For the first scenario on Logan fault (about 1 km to the nearest site of
anticipated while the maximum MDR values indicate moderate damage interest) where most of the nodes have short source-to-site distances,
at the closest distances from the fault plane. The latter observation is uniform site conditions would yield a uniform distribution of ground
attributed to the relatively lower VS30 values and weaker structures in motions radiating from the rectangular fault. However, we observed
the northern parts of the city. irregularities in the spatial distributions of ground motion
17
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
18
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
Declaration of competing interest [23] Coskun B, Coskun B. The Dead Sea fault and related subsurface structures,
Gaziantep Basin, southeast Turkey. Geol Mag 2000;137:175–92.
[24] Gursoy H, Tatar O, Piper JDA, Heimann A, Mesci L. Neotectonic deformation
The authors declare that they have no known competing financial linking the east Anatolian and Karataş-Osmaniye intracontinental transform fault
interests or personal relationships that could have appeared to influence zones in the Gulf of Iskenderun, Southern Turkey, deduced from paleomagnetic
the work reported in this paper. study of the Ceyhan-Osmaniye volcanics. Tectonics 2003;22:1067–79. https://doi.
org/10.1029/2003TC001524.
[25] Tatar O, Piper JDA, Gursoy H, Heimann A, Koçbulut F. Neotectonic deformation in
Acknowledgements the transition zone between the Dead Sea transform and the East Anatolian Fault
zone, Southern Turkey: a palaeomagnetic study of the Karasu rift volcanism.
Tectonophysics 2004;385:17–43. https://doi.org/10.1016/j.tecto.2004.04.005.
This project has been funded by the Gaziantep Municipality with a [26] Westaway R. Kinematic consistency between the Dead Sea fault zone and the
grant number 2018/495179. We acknowledge the financial aid and the Neogene and quaternary left-lateral faulting in SE Turkey. Tectonophysics 2004;
logistic support of the Gaziantep Municipality in the field. We also thank 391:203–37. https://doi.org/10.1016/j.tecto.2004.07.014.
[27] Koc A. Remote sensing study of Sürgü Fault Zone (Malatya, Turkey). M.S. Thesis,
Selim Cambazoglu for his invaluable contribution to the scenario-based Department of Geological Engineering. Ankara: Middle East Technical University;
seismic hazard analyses. 2005.
[28] Emre Ö, Duman TY, Olgun Ş, Elmacı H, Özalp S1. 250.000 scale active fault map
series of Turkey, Gaziantep (NJ 37-9) Quadrangle. Ankara: General Directorate of
References
Mineral Research and Exploration; 2012. Serial Number:38.
[29] Emre Ö, Duman TY, Olgun Ş. 1:250.000 scale active fault map series of Turkey,
[1] Ugurhan B, Askan A, Erberik MA. A methodology for seismic loss estimation in Antakya (NJ 37-13) Quadrangle. Ankara: General Directorate of Mineral Research
urban regions based on ground motion simulations. Bull Seismol Soc Am 2011;101 and Exploration; 2012. Serial Number:39.
(2):710–25. https://doi.org/10.1785/0120100159. [30] Erdik M, Biro YA, Onur T, Sesetyan K, Birgoren G. Assessment of earthquake
[2] Sørensen MB, Lang DH. Incorporating simulated ground motion in seismic risk hazard in Turkey and neighboring regions. Ann Geofisc 1999;42:1125–38. https://
assessment application to the lower Indian himalayas. Earthq Spectra 2014;31(1): doi.org/10.4401/ag-3773.
71–95. [31] Bommer J, Spence R, Erdik M, Tabuchi S, Aydınoglu N, Booth E, et al.
[3] Karimzadeh S, Askan A, Erberik MA, Yakut A. Seismic damage assessment based on Development of an earthquake loss model for Turkish catastrophe insurance.
regional synthetic ground motion dataset: a case study for Erzincan, Turkey. Nat J Seismol 2002;6:431–46. https://doi.org/10.1023/A:1020095711419.
Hazards 2018;92(3):1371–97. https://doi.org/10.1007/s11069-018-3255-6. [32] Deniz A. Estimation of earthquake insurance premium rates for Turkey. M.S.
[4] Chioccarelli E, Cito P, Iervolino I, Giorgio M. REASSESS V2.0: software for single- Thesis, Department of Civil Engineering. Ankara: Middle East Technical University;
and multi-site probabilistic seismic hazard analysis. Bull Earthq Eng 2019;17: 2006.
1769–93. [33] Slemmons DB. Determination of design earthquake magnitudes for microzonation.
[5] Pagani M, Garcia-Pelaez J, Gee R, Johnson K, Poggi V, Silva V, Simionato M, In: Proceedings of Third International Earthquake Microzonation Conference;
Styron R, Vigano D, Danciu L, Monelli D, Weatherill G. The 2018 version of the 1982. p. 119–30. USA: Seattle, Washington.
global Earthquake model: hazard component. Earthq Spectra 2020;36(S1):226–51. [34] dePolo CM, Slemmons DB. Estimation of earthquake size for seismic hazards. Rev
[6] Silva V, Amo-Oduro D, Calderon A, Costa C, Dabbeek J, Despotaki V, Martins L, Eng Geol 1990;8:1–28.
Pagani M, Rao A, Simionato M, Vigano D, Yepes-Estrada C, Acevedo A, Crowley H, [35] Wells DL, Coppersmith KJ. New empirical relationships among magnitude, rupture
Horspool N, Jaiswal K, Journeay M, Pittore M. Development of a global seismic risk length, rupture width, rupture area, and surface displacement. Bull Seismol Soc Am
model. Earthq Spectra 2020;36(S1):372–94. 1994;84:974–1002.
[7] Suzuki A. Iervolino I Intensity measure conversion of fragility curves. Earthq Eng [36] Jackson JA, Mckenzie D. Active tectonics of the Alpine-Himalayan belt between
Struct Dynam 2020;49(6):607–29. western Turkey and Pakistan. Geophys J Int 1984;77:185–264. https://doi.org/
[8] McKenzie DP. Active tectonics of the Alpine Himalayan Belt, the Aegean Sea and 10.1111/j.1365-246X.1984.tb01931.x.
surrounding regions. Geophys J Int 1978;55:217–52. https://doi.org/10.1111/ [37] Gulen L, Barka A, Toksoz MN. Continental collision and related complex
j.1365-246X.1978.tb04759.x. deformation: Maras triple junction and surrounding structures, SE Turkey. Bull
[9] Dewey JF, Şengör AMC. Aegean and surrounding regions: complex multiple and Earth Sci 1987;14:319–36 [in Turkish)].
continuum tectonics in a convergent zone. Geol Soc Am Bull 1979;90:84–92. [38] Lyberis N, Yürür T, Chorowicz J, Kasapoğlu E, Gündoğdu N. The East Anatolian
https://doi.org/10.1130/0016-7606(1979)90<84:AASRCM>2.0.CO;2. Fault: an oblique collisional belt. Tectonophysics 1992;204:1–15. https://doi.org/
[10] Bozkurt E. Neotectonics of Turkey-a synthesis. Geodin Acta 2001;14:3–30. https:// 10.1016/0040-1951(92)90265-8.
doi.org/10.1016/S0985-3111(01)01066-X. [39] Garfunkel Z, Zak I, Freund R. Active faulting in the Dead Sea rift. Tectonophysics
[11] Woessner J, Danciu L, Giardini D, Crowley H, Cotton F, Grünthal G, et al. The 2013 1981;80:1–26. https://doi.org/10.1016/0040-1951(81)90139-6.
European seismic hazard model: key components and results. Bull Earthq Eng [40] McClusky SC, Balassanian S, Barka A, Ergintav S, Georgie I, Gurkan O, et al. Global
2015;13(12):3553–96. Positioning System constraints on plate kinematics and dynamics in the eastern
[12] Şeşetyan K, Danciu L, Demircioğlu Tümsa MB, et al. The 2014 seismic hazard Mediterranean Caucasus. J Geophys Res 2000;105:5695–719. https://doi.org/
model of the Middle East: overview and results. Bull Earthq Eng 2018;16:3535–66. 10.1029/1999JB900351.
https://doi.org/10.1007/s10518-018-0346-4. [41] Westaway R. Present-day kinematics of the Middle East and Eastern
[13] Gullu H, Ansal AM, Özbay A. Seismic hazard studies for Gaziantep city in South Mediterranean. J Geophys Res 1994;99:12071–90. https://doi.org/10.1029/
Anatolia of Turkey. Nat Hazards 2008;44(1):19–50. 94JB00335.
[14] Cambazoglu S, Eker AM, Koçkar MK, Akgün H. Development of seismic sources for [42] Erdik M, Oner M, Gülkan P. Malatya-Elazig Railway, Karakaya Bridge: assessment
Kilis region and a sample seismic hazard analysis for Sahinbey District. In: of design earthquake characteristics. Ankara: METU/EERI; 1980. Report No:8006.
Proceedings of 2nd Conference of Earthquake Engineering and Seismology. [43] Ambraseys NN, Barazangi M. The 1759 earthquake in the Bekaa Valley:
Turkey: Hatay; 2013. implications for earthquake hazard assessment in the Eastern Mediterranean
[15] Sertcelik F. Estimation of Coda wave attenuation in the East Anatolia fault zone, Region. J Geophys Res 1989;94:4007–13. https://doi.org/10.1029/
Turkey. Pure Appl Geophys 2012;169:1189–204. https://doi.org/10.1007/ JB094iB04p04007.
s00024-011-0368-1. [44] Saroglu F, Emre Ö, Boray A. Active faults of Turkey and their seismicity. Ankara:
[16] Gullu H, Iyisan R. A seismic hazard study through the comparison of ground General Directorate of Mineral Research and Exploration; 1987.
motion prediction equations using the weighting factor of logic tree. J Earthq Eng [45] Herece E, Akay E. East Anatolian fault between Karlıova and Çelikhan. In:
2016;20(6):861–84. https://doi.org/10.1080/13632469.2015.1104752. Proceedings of 9th Petroleum Congrees of Turkey. Ankara: Turkey; 1992.
[17] Gullu H, Karabekmez M. Investigation of seismic behaviour of Gaziantep Kurtulus p. 361–72.
Mosque. DUMF Eng J 2016;7(3):455–70 [in Turkish)]. [46] Nalbant S, Hubert A, King GCP. Stress coupling between earthquakes in northwest
[18] Terlemez HÇI, Şentürk K, Ateş Ş, Sümengen M, Oral A. Geology of Gaziantep Turkey and the north Aegean Sea. J Geophys Res 1998;103. https://doi.org/
region and Pazarcık-Sakçagöz-Kilis-Elbeyli-Oğuzeli area. Ankara: General 10.1029/98JB01491. 24.469-24.486.
Directorate of Mineral Research and Exploration of Turkey (MTA); 1992. Report [47] Disaster and Emergency Management Authority (AFAD)-Presidental of Earthquake
no:9526. Department. 2020 Sivrice (Elazığ) earthquake. Report of January 24, (in Turkish),
[19] Akbas B, Akdeniz N, Aksay A, Altun I, Balcı V, Bilginer E, et al. Geological map of https://deprem.afad.gov.tr/downloadDocument?id=1831. [Accessed 31 July
Turkey. Ankara: General Directorate of Mineral Research and Exploration 2020].
Publications; 2016. [48] Koçkar MK, Akgün H, Rathje EM. Evaluation of site conditions for the Ankara basin
[20] Soysal H, Sipahioğlu S, Kolçak D, Altınok Y. Historical earthquake catalogue of of Turkey based on seismic site characterization of near-surface geologic materials.
Turkey and surrounding area (2100 B.C.-1900 A.D.). Ankara: TUBITAK; 1981. Soil Dynam Earthq Eng 2010;30(1–2):8–20.
Technical Report No: TBAG341. [49] Tinsley JC, Fumal TE. Mapping quaternary sedimentary deposits for areal
[21] Kandilli Observatory and Earthquake Research Institute (KOERI)-Regional variations in shaking response. In: Evaluating Earthquake Hazards in the Los
Earthquake-Tsunami Monitoring Center. Earthquake Catalog Search System. 2020. Angeles Region, vol. 1360; 1985. p. 101–25. US Geological Survey Professional
http://www.koeri.boun.edu.tr/sismo/2/earthquake-catalog/. Paper.
[22] Saroglu F, Emre Ö, Kuşçu I. Active fault map of Turkey (1:2,000,000). Ankara: [50] Park S, Elrick S. Predictions of shear-wave velocities in southern California using
General Directorate of Mineral Research and Exploration; 1992. surface geology. Bull Seismol Soc Am 1998;88(3):677–85.
19
A. Arslan Kelam et al. Soil Dynamics and Earthquake Engineering 154 (2022) 107129
[51] Wills CJ, Petersen M, Bryant WA, Reichle M, Saucedo GJ, Tan S, et al. A site- [71] Ugurhan B, Askan A. Stochastic strong ground motion simulation of the 12
conditions map for California based on geology and shear wave velocity. Bull November 1999 Düzce (Turkey) earthquake using a dynamic corner frequency
Seismol Soc Am 2000;90(6B):187–208. approach. Bull Seismol Soc Am 2010;100(4):1–27.
[52] Holzer TL, Bennett MJ, Noce TE, Tinsley III JC. Shear-wave velocity of surficial [72] Roumelioti Z, Kiratzi A, Margaris B, et al. Simulation of strong ground motion on
geologic sediments in Northern California: statistical distributions and depth near-fault rock outcrop for engineering purposes: the case of the city of Xanthi
dependence. Earthq Spectra 2005;21(1):161–77. (northern Greece). Bull Earthq Eng 2017;15:25–49. https://doi.org/10.1007/
[53] Wald DJ, McWhirter L, Thompson EM, Hering AS. A new strategy for developing s10518-016-9949-9.
Vs30 maps. In: Proceedings of 4th IASPEI/IAEE International Symposium: Effects [73] Bajaj K, Anbazhagan P. Regional stochastic ground-motion model for low to
of Surface Geology on Seismic Motion. USA: Santa Barbara; 2011. moderate seismicity area with variable seismotectonic: application to Peninsular
[54] Forte G, Chioccarelli E, Cito P, De Falco M, Santo A, Iervolino I. Seismic soil India. Bull Earthq Eng 2019;17:3661–80. https://doi.org/10.1007/s10518-019-
classification of Italy based on surface geology and shear-wave velocity 00646-9.
measurements. Soil Dynam Earthq Eng 2019;122:79–93. [74] Tanırcan G, Yelkenci-Necmioğlu S. Simulation of the strong ground motion for the
[55] Turkish Building Earthquake Code-TBEC. Specifications for buildings to be built in 20 July 2017 (Mw. 6.6) Bodrum–Kos earthquake. Bull Earthq Eng 2020;18:
seismic areas. Ankara: Ministry of Public Works and Settlement; 2018. 5807–25. https://doi.org/10.1007/s10518-020-00892-2.
[56] EN 1998-1. Eurocode 8; Design of structures for earthquake resistance, Part 1: [75] Boore DM, Joyner WB. Site amplifications for generic rock sites. Bull Seismol Soc
general rules, seismic actions and rules for buildings. 2020. Am 1997;87:327–41.
[57] International Code Council-ICC. International Building Code. 2018. Whittier, CA. [76] Boore DM, Atkinson GM. Ground-motion prediction equations for the average
[58] Boore DM. Estimating Vs(30) (or NEHRP site classes) from shallow velocity models horizontal component of PGA, PGV, and 5%-damped PSA at spectral periods
(depths < 30 m). Bull Seismol Soc Am 2004;94(2):591–7. between 0.01s and 10.0s. Earthq Spectra 2008;24:99–138.
[59] Ohta Y, Goto N. Empirical shear wave velocity equations in terms of characteristic [77] Akkar S, Cagnan Z. A local ground-motion predictive model for Turkey and its
soil indexes. Earthquake Engineering & Structural. Dynamics 1978;6:167–87. comparison with other regional and global ground-motion models. Bull Seismol
[60] Imai T, Tonouchi K. Correlation of N value with S-wave velocity and shear Soc Am 2010;100:2978–95.
modulus. In: Proceedings of 2nd European Symposium on Penetration Testing. [78] Turkish Statistical Institute, TSI. Building Census 2000. Ankara: Turkish Statistical
Amsterdam: Netherlands; 1982. p. 67–72. 1982. Institute; 2000 (in Turkish).
[61] Pitilakis K, Raptakis D, Lontzetidis K, Tika-Vassilikou T, Jongmans D. Geotechnical [79] Ay BO, Eroglu Azak T, Erberik MA. Evaluation of changing building characteristics
and geophysical description of Euro-Seistests, using field, and laboratory tests and in Turkey. In: Proceedings of 12th International Congress on Advances in Civil
moderate strong ground motions. J Earthq Eng 1999;3(3):381–409. Engineering. Turkey: Istanbul; 2016.
[62] Sykora DE, Stokoe KH. Correlations of in-situ measurements in sands of shear wave [80] Erberik MA. Fragility-based assessment of typical mid-rise and low-rise RC
velocity. Soil Dynam Earthq Eng 1983;20:125–36. buildings in Turkey. Eng Struct 2008;30(5):1360–74.
[63] Hasancebi N, Ulusay R. Empirical correlations between shear wave velocity and [81] Erberik MA. Generation of fragility curves for Turkish masonry buildings
penetration resistance for ground shaking assessments. Bull Eng Geol Environ considering in-plane failure modes. Earthq Eng Struct Dynam 2008;37(3):387–405.
2007;66:203–13. [82] Askan A, Asten M, Erberik MA, Erkmen C, Karimzadeh S, Kilic N, et al. Seismic
[64] Koçkar MK, Akgün H. Development of a geotechnical and geophysical database for damage assessment of Erzincan. Project n. TUJJB-UDP-01-12. Turkish national
Seismic zonation of the Ankara Basin, Turkey. Environ Geol 2008;55(1):165–77. union of geodesy and geophysics project; 2015 (in Turkish).
[65] Wills CJ, Clahan KB. Developing a map of geologically defined site-condition [83] Ibarra LF, Medina RA, Krawinkler H. Hysteretic models that incorporate strength
categories for California. Bull Seismol Soc Am 2006;96(4A):1483–501. and stiffness deterioration. Earthq Eng Struct Dynam 2005;34:1489–511.
[66] Wills CJ, Silva W. Shear wave velocity characteristics of geologic units in [84] Ibarra LF, Krawinkler H. Global collapse of frame structures under seismic
California. Earthq Spectra 1998;14(3):533–56. excitations. Stanford, CA: The John A. Blume Earthquake Engineering Center,
[67] Wald DJ, Allen TI. Topographic slope as a proxy for seismic site conditions and Stanford University; 2005. Rep. No. TB 152.
amplification. Bull Seismol Soc Am 2007;97(5):1379–95. [85] Ay BO, Erberik MA. Vulnerability of Turkish low-rise and mid-rise reinforced
[68] Eker AM, Akgün H, Koçkar MK. Local site condition characterization and seismic concrete frame structures. J Earthq Eng 2008;12(S2):2–11.
zonation study by utilizing active and passive surface wave methods. Eng Geol [86] Iervolino I. Assessing uncertainty in estimation of seismic response for PBEE.
2012;151:64–81. Earthq Eng Struct Dynam 2017;46:1711–23.
[69] Thompson EM, Kayen RE, Carkin B, Tanaka H. Surface-wave site characterization [87] Askan A, Yucemen MS. Probabilistic methods for the estimation of potential
at 52 strong-motion recording stations affected by the Parkfield, California, M6.0 seismic damage: application to reinforced concrete buildings in Turkey. Struct Saf
earthquake of 28 September 2004. US Geological Survey Open-File; 2010. Report 2010;32(4):262–71.
2010-1168, http://pubs.usgs.gov/of/2010/1168. [88] Gurpinar A, Abali M, Yucemen MS, Yesilcay Y. Feasibility of obligatory earthquake
[70] Motazedian D, Atkinson GM. Stochastic finite-fault modeling based on a dynamic insurance in Turkey. 78-05. Ankara: Earthquake Engineering Research Center,
Corner frequency. Bull Seismol Soc Am 2005;95:995–1010. Civil Engineering Department, Middle East Technical University; 1978 [in
Turkish)].
20