Geophysical Journal International https://doi.org/10.1093/gji/ggac477

The Neglected İstanbul Earthquakes in the North Anatolian Shear Zone: Tectonic Implications and Broadband Ground Motion Simulations for a Future Moderate Event

Onur Tan^1*, Özlem Karagöz², Semih Ergintav³, Kemal Duran⁴

¹Department of Geophysical Engineering, Faculty of Engineering, İstanbul University – Cerrahpaşa, Istanbul, Turkey. Email: onur.tan@iuc.edu.tr

²Department of Geophysical Engineering, Faculty of Engineering, Çanakkale Onsekiz Mart University, Çanakkale, Turkey. E-mail: ozlemkaragoz@comu.edu.tr

³Department of Geodesy, Kandilli Observatory and Earthquake Research Institute, Boğaziçi University, Istanbul, Turkey. E-mail: semih.ergintav@boun.edu.tr

⁴Department of Soil and Earthquake Research, Istanbul Metropolitan Municipality, İstanbul, Turkey. E-mail: kemal.duran@ibb.gov.tr

* Corresponding author

SUMMARY

İstanbul (Marmara Region, NW Turkey) is one of the megacities in the world and suffered from destructive earthquakes on the North Anatolian Fault, a member of the North Anatolian Shear Zone, throughout history. The 1999 Kocaeli and Düzce earthquakes emphasize the earthquake potential of the fault, crossing the Sea of Marmara, and the importance of seismic hazards in the region. The studies in the last 20 years have concentrated on the main fault and its future destructive earthquake potential. In this study, unlike the previous ones, we focus on the two main topics about the earthquakes not interested previously in İstanbul: (1) Investigating recent earthquake activity masked by the blasts in the metropolitan area and its tectonic implications, (2) revealing their effects in İstanbul utilizing numerical ground motion simulations for a future moderate event (M_w 5). Firstly, the 386 earthquakes from 2006 to 2016 are relocated with the double-difference method using the dense seismic network operated in the same period. The source mechanisms of the events (M_L ≥ 3), including the most recent 2021 Kartal-İstanbul earthquake (M_L 4.1), are determined. In addition to the analysis of the recent seismic activity, the location of the two moderate and pre-instrumental-period İstanbul earthquakes, which occurred in 1923 (M_w 5.5) and 1929 (M_w 5.1), are revised. Using the relocated epicenters outside of the principal deformation zone and the fault plane solutions, the roles of the earthquakes in the stress regime of the Marmara region are explained. The epicenters on the Cenozoic or Paleozoic formation in the Istanbul-Zonguldak Zone are interpreted as the re-activation of the paleo-structures under the recent tectonic stresses, and their fault plane solutions agree with the synthetic/antithetic shears of a transtensional regime corresponding to the right lateral strike-slip system with mainly N-S extension in the Marmara Region. In the second part, we investigate the effects of moderate scenario events (M_w 5) considering the current earthquake epicenters in the İstanbul metropolitan area, using characterized earthquake source model and 1D velocity structure verified with the broadband (0.1 – 10 Hz) numerical ground motion simulation of the 2021 Kartal-İstanbul earthquake. The simulated PGAs agree with the ground-motion prediction equations for short epicentral distances (< 30 km). Furthermore, according to the empirical relation for Turkish earthquakes, the maximum PGA value of the synthetic models (~0.3 g) corresponds to the felt intensity of MMI IX. The simulated spectral accelerations for the M_w 5 earthquake scenarios may exceed the design spectrum between 0.2 and 0.6 s given in the Turkish Building Earthquake Code (2018). In addition, certain models also generate spectral accelerations close to the design-level spectrum between 0.4 and 1 s, leading to resonance phenomena. The results indicate that a moderate event (M_w 5) in the İstanbul metropolitan area is capable of damage potential for the mid-rise buildings (4 - 10 stories) because of the site condition with resonance phenomena and poor construction quality.

1 INTRODUCTION

The North Anatolian Fault (NAF), as one of the largest plate-bounding transform faults, separates the Anatolian and Eurasian plates and extends for ~1600 km between eastern Anatolia and northern Aegean. Sengor et al. (2005) mentioned that the NAF is a member of the E-W North Anatolian Shear Zone (NASZ). Anatolia moves westward with respect to the collision zone between the Eurasian and Arabian Plates at a rate of ~25 mm/yr (Reilinger et al. 2006), activating major strike-slip and also N-S extensional normal faulting earthquakes south of the Marmara region (Ambraseys 2009). The seismic activity is mainly on the NAF, the principal deformation zone (PDZ). A series of large earthquakes started in eastern Anatolia in 1939. They propagated westward, along the northern branch of NAF in the Marmara Sea, toward the Istanbul-Marmara region in north-western Turkey. This migration ended in 1999 with the devasting Kocaeli earthquake (M_w 7.4). West of the 1999 Kocaeli rupture, a "seismic gap" exists along a ~100 km-long segment below the Sea of Marmara which connects the Mürefte (1912, M_w 7.3) and Kocaeli (1999, M_w 7.4) ruptures (Toksoz et al. 1979; Altunel et al. 2004; Aksoy et al. 2010; Bohnhoff et al. 2013). The region has very high seismic activity, and several destructive events (M_w > 7) on the segments of the NAF are reported in both historical and instrumental period earthquake catalogs (Figure 1). The instrumental period seismological observations indicate high seismic activity in a narrow zone in the Sea of Marmara with a seismic gap on the central part of the northern branch (Kumburgaz segment in Figure 2). There are also seismological and geodetical evidences of another seismic gap on the Princes' Island segment (Bohnhoff et al. 2013; Ergintav et al. 2014; Wollin et al. 2018). These two seismic gaps are the potential destructive event source considering the recurrence of the destructive Marmara earthquakes. In the western part of the Marmara Region, the Ganos segment ruptured by the 1912 Mürefte earthquake (M_w 7.3) accumulated strain due to an M_w 7 earthquake (Ergintav et al. 2014). The other branch of the NAF follows the southern part of the Marmara with scattered seismic activity, and geodetic data shows that they have lower strain accumulation with respect to the northern branch (< ~5 mm/yr) (Ergintav et al. 2014).

Contrary to the central and southern Marmara Region, the north of the NAF has very low seismic activity in the last century (Figure 1). However, according to the Ottoman archives, at least 20 felt-events in the 15^th – 19^th centuries are addressed to Edirne by Ambraseys (2009). The 29 July 1752 Edirne earthquake (MMI IX-X) is one of the most destructive events in the region. In the east, the destructive historical earthquakes addressed to İstanbul are located in the Sea of Marmara. In contrast, a few earthquakes are recorded in İstanbul throughout the instrumental period (after 1900). Due to the lack of a seismic network around İstanbul before the 1990s, information about the seismic activity is limited. In the earlier part of the instrumental period, two significant events (M_w > 5) in the west of İstanbul on 26 October 1923 and 10 October 1929 are reported in the International Seismological Summary (ISS), and the magnitudes were determined as 5.5 and 5.1, respectively, by Kondorskaya & Ulomov (1999).

The metropolitan area of İstanbul, the interest of this study, covers both sides of the Bosporus and is mainly located on Palaeozoic sedimentary rocks which are intruded or overlain by Mesozoic and Cenozoic magmatic and sedimentary deposits, which we simplified as pre-Miocene rocks and Miocene-Pliocene deposits in Figure 2. The Quaternary alluvium is only observed in the northern part of the province and the creek beds in the city center. Although there is no Holocene active fault in the metropolitan area, the Phanerozoic units are cut by palaeofaults in various orientations and sizes, especially on the Asian side. The previous studies also mention two remarkable old fault zones on the European side. The West Black Sea Fault (WBSF) is a dextral transform fault zone covered by undeformed Eocene sediments, between the Strandja Massif and İstanbul-Zonguldak Paleozoic Zone, in the west of İstanbul (Okay et al. 1994). The recent geophysical observations also show a sharp resistive-conductive boundary in the upper-most crust interpreted as the WBSF (Karcıoğlu et al. 2013). The NW-SE oriented dextral Çatalca Fault is the other fault in the area that is the boundary between the Strandja Massif and Thrace Basin (Yılmaz et al. 1997).

The geodetical observations indicate slip partitioning between the northern and southern branches of the NAF (Armijo et al. 2002; Ergintav et al. 2014), and the deformation mainly occurs along the northern branch (Meade et al. 2002; Reilienger et al. 2006). The rate of secular deformations of the northern branch of the NAF is around 10-15 mm/yr in the southern part of the Istanbul province. After the devastating 1999 Kocaeli earthquake, İstanbul was affected by its postseismic deformations. Its order was cm-level after immediately the earthquake (Ergintav et al. 2009) and reduced to the mm-level (Diao et al. 2016) in time, as the logarithmic nature of the postseismic deformations (Hearn et al. 2002; Ergintav et al. 2007, 2009). Hence, as expected, the stress change increased on the active and passive fault systems in the eastern Marmara Region. Correspondingly, the seismic hazard potential of İstanbul increases and hazard assessment studies for utilizing probabilistic and hybrid simulations based on a destructive earthquake on the NAF are performed (i.e., Sorensen et al. 2009, Yalcinkaya 2014; Aochi & Ulrich 2015; Douglas & Aochi 2016).

The last decade-earthquakes show that most of the buildings in Turkey are still not resistant to a nearby moderate or distant large earthquake. For example, the significant 2011 Van-Erciş (eastern Turkey, M_w 7.1) devastated more than 8,000 buildings. On the other hand, the moderate 2017 Çanakkale-Ayvacık earthquake sequence (western Turkey, M_w 5.3) damaged adobe and masonry buildings in Ayvacık city and the nearby villages. Similarly, buildings were damaged, 70 people were injured, and three died in the 2011 Kütahya-Simav earthquake (Mw 5.8). The 2020 Elazığ-Sivrice earthquake (M_w 6.8, eastern Turkey) heavily damaged more than 650 buildings in the cities of Elazığ, Malatya and Diyarbakır cities, with the epicentral distance between 40 and 110 km. In the 2020 Samos earthquake (M_w 7.0), buildings with poor construction quality in Bornova, İzmir, ~70 km from the epicenter, collapsed. Likewise, the 26 September 2019 İstanbul-Silivri earthquake (M_w 5.7, Figure 1) on the Kumburgaz segment in the Marmara Sea damaged a mosque in Avcılar and some residential and public buildings on the European side of İstanbul, 40-60 km far from the epicenter. These examples show that moderate earthquakes (M_w ~5) close to or beneath the İstanbul metropolitan area can be destructive.

The effects of smaller events on buildings, such as the 2021 İstanbul-Kartal earthquake (M_L 4.1), are ignored due to their low intensities. However, they should be considered because an unbroken asperity at the same focal area may be capable of generating a moderate event (M_w 5.0-5.5). In this case, the radiating waves from a seismic source close to a densely populated city, such as İstanbul, may affect the residential and industrial structures with low seismic resistance.

In this study, unlike the previous ones, we focus on the moderate earthquakes beneath the İstanbul metropolitan area and their possible seismic hazard potential in the future. For this aim, the activity and source properties of the earthquakes in the period of 2006-2016, including the recent 19 June 2021 Kartal earthquake, are investigated using the data from the available seismic networks. The foci of the events are relocated with the double-difference algorithm utilizing the waveform similarity. The focal mechanism solutions with low uncertainties are determined by employing numerous P-wave first-motion polarities. The 1923 (M_w 5.5) and 1929 (M_w 5.1) İstanbul earthquakes, which are never discussed before, are also relocated with the body wave arrival times reported in the bulletins and their roles in the regional stress regime are estimated. In addition to the tectonic implication of the source mechanisms of the İstanbul earthquakes, we numerically simulate scenario earthquakes in regard to ground motion prediction. The numerical simulation methodology utilizing the Discrete Wave Number Method (DWNM) and including the effects of propagating source, path and the site amplification of shallow 1D soil layers over the engineering bedrock are verified with the 2021 Kartal earthquake records. The fundamental soil frequencies at the AFAD strong motion stations are also estimated using the Horizontal-to-Vertical Spectral Ratio (HVSR, Nakamura 1989) method with at least ten earthquake records to verify the previously reported values. After identifying the source locations for possible future M_w 5 events in the metropolitan area, the scenario earthquakes with different source models are numerically simulated at the strong motion stations. Then the simulated peak ground accelerations (PGAs) and pseudo-acceleration response spectra (Sa) are compared with two attenuation relationships (AC2010: Akkar & Çağnan 2010; B2014: Boore et al. 2014) and the 2018 Turkish Building Earthquake Code (TBEC), respectively. Consequently, the possible effects of a moderate earthquake, depending on the source and the site condition in the İstanbul metropolitan area, are evaluated.

2 SEISMIC ACTIVITY IN THE ISTANBUL METROPOLITAN AREA

2.1 Data analysis

The parametric and waveform data of the earthquakes used in this study are obtained from regional and national seismological networks around the Marmara Sea. The regional weak motion networks for micro-earthquake observations are operated in the projects between 2006 and 2016, such as MARsite (www.marsite.eu). Additional weak and strong motion data are retrieved from the national network stations of the Kandilli Observatory and Earthquake Research Center (KOERI, 1971) and the Disaster and Emergency Management Authority (AFAD, 1990).

The Hypocenter location algorithm (Lienert & Havskov 1995) is used to determine the parameters of the events in the Istanbul metropolitan area with the velocity model proposed by Karabulut et al. (2011). The total 386 events between 2006 and 2016 in the study area shown in Figure 1 are located using the 3,038 P- and 2,581 S-arrivals. The minimum and the average number of stations used in the relocation procedure are four and eight, respectively. The average uncertainties of the absolute event hypocenters are about ±2 km.

The hypocenters are improved with double-difference (DD) inversion, one of the methods to reduce location uncertainties. For this aim, we re-analyze the events utilizing the hypoDD algorithm developed by Waldhauser & Ellsworth (2000). The algorithm assumes that the hypocentral separation between two earthquakes is small compared to the event-station distance and the scale length of velocity heterogeneity, so the ray paths are similar. The travel time difference between two events observed at one station can be accurately attributed to the spatial offset between the events (Fréchet 1985; Got et al. 1994; Waldhauser & Ellsworth 2000). We use absolute location parameters of the events from the catalog and the P/S travel time differences between the event pairs as the inversion inputs. P/S-wave cross-correlation (CC) data is also used to obtain travel-time lag with a 0.01 s resolution. The differential travel-time data, with a relative timing precision of approximately tens of milliseconds, allows for calculating the relative location between earthquakes with errors of only a few hundred meters. A 10 km-search radius is chosen to select neighboring earthquakes, and a minimum of 8 P/S arrival times, the lower limit used to solve unknown parameters of pairs (6 for space and 2 for time), are chosen as a threshold for an event pair. The waveform similarities of the events are determined with the coherence algorithm in the MTSPEC package of Prieto et al. (2009). Two waveforms recorded at a common station are considered similar when all squared coherency values (C_xy²) exceed 0.5 in the frequency range from 1 to 10 Hz. Figure 3 shows two examples of the waveform similarities of event pairs in the study area. The average coherency in the first example is low (C_xy² = 0.65) for two neighboring events (07.05.2010 02:41 M_L 1.7, 11.05.2010 22:40 M_L 1.5) in Figure 3a. The CC function indicates the P-waves' arrival time difference (Tcc) of 0.03 s. The second event's waveform is shifted according to the Tcc value to show the waveform similarity in the bottom row of the figure. A high coherency example (C_xy² = 0.92) of P-waveforms is given in Figure 3b for the neighboring events with M_L 1.3 (20.09.2009 22:22) and M_L 2.0 (08.05.2011 04:49). The spiky CC function with 0.07 s time-lag indicates a good waveform match of the two events. In the hypoDD inversion scheme, the C_xy² values of the waveforms are used for weighting. The location uncertainties of the relocated events are estimated using the statistical approach described by Tan et al. (2010) and Tan (2013). The random numbers (distances) between -3.0 and +3.0 km that agreed with the average hypocentral uncertainties (±2 km) were added to the initial locations of the events in the X, Y and Z directions. The randomly shifted event pairs are relocated with repeated ~500 well-conditioned inversions. The outliers in the data set are removed using the interquartile range (IQR) method. The average horizontal and vertical location uncertainties for the earthquakes in Istanbul city are ±400 m and ±1200 m, respectively.

M_L magnitudes

To standardize the calculation of the magnitudes from the different instruments, we use local magnitudes (M_L). A methodology was introduced into the Seismic Analysis Code (Goldstein et al. 2003) to calculate the M_L magnitudes of an event. First, the sensor and digitizer responses are removed from the velocity records, and a bandpass Butterworth filter between 0.1 and 20 Hz is used. Then, each waveform is convolved with the Wood-Anderson seismometer response to generate a displacement record in units of nanometers. Next, the maximum zero-to-peak amplitude is selected from the three components recorded at each station, and M_L is calculated using the equation given by Hutton & Boore (1987). Finally, station-related low and high magnitude values (larger than one standard deviation) of the events are removed, and then the remaining magnitudes are averaged for that event.

Fault plane solutions

The earthquake fault plane solutions (FPS) are determined from the P-wave first motion (FM) polarities using the focmec algorithm (Snoke et al. 2003). All available polarities at the local and regional weak and strong motion stations are read to constrain the nodal planes. The P-wave incidence angles for each station are calculated using the focal depth and 1D velocity model. If the incidence angle is larger than 90° for a local station in İstanbul, the polarity is located to the antipode on the focal sphere. Therefore, the stations close to the epicenter are shown in the opposite azimuthal direction on the lower hemisphere projection.

2.2 Blasts

While studying the seismicity of İstanbul, the artificial seismic events in the city, masking the earthquakes, must be classified. There are several large quarry areas and stripe coal mines on both Asian and European sides, shown with white diamonds in Figure 4a. The blast catalog of KOERI (2021) indicates that the majority of the blasts have a local magnitude between 1.0 and 2.0 (Figure 4b). The magnitudes of blasts can reach up to M_L 2.0-2.5, and the records at the distant stations are very similar to earthquake waveforms. The blast activities generally begin in the early hours and show two picks in the daytime around the daily lunch break (Figure 4c). Although there are a few studies about blast identification (i.e., Horasan et al. 2009; Yıldırım et al. 2011), which are applied to limited parts of the catalogs, and a published quarry-free catalog is not available for a long period. Hence, international (i.e., ISC, EMSC) and Turkish national earthquake catalogs are contaminated by artificial seismic events. Unfortunately, it is impossible to view a clear long-term earthquake activity for İstanbul. Therefore, following our range estimation in Figure 4b, applying a magnitude threshold of 3.0 for the small events in the land area is an effective way to clean the catalogs and interpret the Istanbul seismicity.

2.3 Recent earthquake activity in Istanbul

The earthquake activity in the İstanbul metropolitan area is very low compared to the Marmara Sea. There are eight events with M_w* ≥ 3.0 on both sides of İstanbul between Büyükçekmece Lake and Tuzla in the years 2006 - 2016 (Figure 5). The largest two events in the land area are on the coastline between Kartal and Tuzla districts in eastern İstanbul. The 29 September 1999 Tuzla earthquake was the largest event (M_w 5.2) and occurred in Tuzla-İçmeler geothermal area, shown in Figure 2. Its Global Centroid Moment Tensor (GCMT) solution indicates oblique normal faulting (Figure 5, Table 1). We cannot analyze this event due to the lack of nearby seismic stations in the region. The second event (M_w 4.5) occurred on 7 July 2000, northwest of the previous. There are no reported source parameters for the event. Both events occur after the 1999 Kocaeli mainshock (M_w 7.4) and are in the positive lobe of Coulomb stress change by Çakır et al. (2003).

A few earthquake clusters were on the European side of Istanbul from 2006 to 2016 (Figure 5). The micro-earthquake cluster labeled with c1 is observed on the shelf area in east Silivri. A total of 22 events are identified in the cluster on 27-28 March 2014. The focal depths are between 10 and 15 km, and their local magnitudes (M_L) range from 1.5 to 2.8. The offshore cluster c2, in Selimpaşa town, contains 15 events occurring at a depth of ~10 km in different years. The other cluster (c3) in the westernmost of the study area has events with M_L between 2.0 and 2.5. There is insufficient P-wave first motion polarity to obtain reliable focal mechanism solutions for these three clusters. Besides, the cluster c4 is observed on the shelf area between the lakes of Büyükçekmece and Küçükçekmece. The largest event (M_L 3.0) in the cluster occurred on 19 January 2015 at a depth of 10 km. There are 26 reliable polarity readings in the three quadrants on the focal sphere (Figure 6). The best nodal planes that divide the quadrants show a strike-slip mechanism with a thrust component.

The two events with a high number of first motion polarities show strike-slip events on the southern and northern coastline of the European side. The 19 October 2012 earthquake (M_L 3.6) occurred beneath the Esenyurt district (pop. ~960,000) between the Büyükçekmece and Küçükçekmece lakes and has 39 polarity readings with good azimuthal coverage (Figure 6, Table 1). The polarity changes at the NNW and NE stations constrain both nodal planes. The NNW-SSE (strike 335°) nodal plane indicates left-lateral strike-slip faulting with a dip angle of 85°. The second event with strike-slip faulting occurred on 5 February 2014 (M_L 3.9) on the Black Sea coastline, northwest of the new İstanbul International Airport. The nodal planes are controlled by the 45 first motion polarities around the epicenter. Both 2012 and 2014 events have a focal depth of about 10 km. The third event (M_L 3.2) on the European side occurred in 2008 in the Sultangazi district, with a population of ~540,000. Its epicenter is very close to the coal mines in the region. Although its origin time is 19:57 (local) and the focal depth is 5.9 km, it is cataloged as a quarry blast by KOERI. The revised focal depth of the event is 12 km, and the P-wave polarities indicate normal faulting with a strike-slip component. The magnitudes, focal depths, and fault plane solutions prove that the three events (M_L > 3) are not blasts. However, there is no event to interpret as subsequent aftershocks one or two years later.

The earthquakes are clustered in two localities on the Asian side. The most active cluster is in the Tuzla district (pop. 273,000), and its distance to Sabiha Gökçen International Airport is about 5 km. After the 1999 Tuzla earthquake (M_w 5.2), the seismic activity in Tuzla becomes significant. The epicentral coordinates and focal depths of the two events in 2009 (M_L 2.5) and 2010 (M_L 3.6) are the same (Figure 5, Table 1). Their first motion polarities are in good agreement, and their joint solution shows a strike-slip fault with a normal component. The 9 May 2011 event (M_L 3.4) also occurred at the same location, and its source has similar faulting parameters. These three events originate at 6-7 km depth on the same fault surface. A detailed map and the cross-section of the Tuzla cluster are presented in Figure 7. The events dip to the north-northeast and agree with the dip angles of the fault plane solutions (62-67°). This consistency may indicate a mainly E-W-oriented right-lateral fault plane in Tuzla. The micro-earthquake activity extends to the northeastern Marmara shelf and joints with seismicity of the NAF in the Çınarcık Basin (Figure 5).

The latest earthquake in the metropolitan area (Kartal district, pop. 474,000) occurred on 19 July 2021 and was felt around the city. The local magnitude of the event is calculated as 4.1. Clear 32 P-wave first motion polarities are grouped in all quadrants on the focal lower hemisphere (Figure 6). The nodal planes are well bounded by the local stations around the epicenter. The fault plane solution shows oblique faulting with both strike-slip and thrust components.

2.4 Relocation of the 1923 and 1929 İstanbul earthquakes

The 26 October 1923 and 10 October 1929 earthquakes that occurred at the beginning of the instrumental seismology period are not mentioned in the previous studies. However, their locations and sizes are important to understand the seismicity of İstanbul. They have occurred in the inactivate zone of the WBSF. The moment magnitudes (M_w) were calculated as 5.5 and 5.1, respectively, by Kondorskaya & Ulomov (1999) using the observed amplitudes in the stations' periodic bulletins. Because the original records of both events are unavailable, the reported observations in the ISS catalogs are used for relocation (Figure A1). For this aim, the scanned version of the periodic ISS bulletins by Villasenor et al. (1997) is obtained from the ISC Seismological Dataset Repository (ISC 2021a).

The P- and S-wave arrival times are preferred and used in the hypocenter location algorithm. The doubtful phases labeled with letters such as "?S" and "?L" are ignored to reduce the complexity of the location problem (Figure A1). In addition, the event depths are fixed at 10 km to reduce the unknown parameters due to the limited observations. Then, the origin time is estimated by fixing the reported epicenter. After that, the epicenter coordinates are determined using the new origin time. Finally, both parameters are obtained freely using the previous estimations as initial model parameters.

We determined that the origin time of the 1923 earthquake is 12:13:27, using the station arrivals in Table 2, and the epicenter is close to the Black Sea coastline (41.328°N 28.517°E, ±30 km). The 1929 earthquake origin time is 23:01:16, and the epicenter (41.093°N 28.583°E, ±20 km) is relocated 12 km south of the previously reported location. The new parameters of both earthquakes are given in Table 3. The updated 1923 and 1929 earthquake epicenters shown with stars in Figure 5 are close to the recent events in 2012 (near Durusu Lake in the north) and 2014 (near Büyükçekmece Lake in the south), respectively. The revised locations, with their inherently high uncertainties, disclose that they are on the European side of İstanbul and are away from the NAF, which is the only active fault in the study area. Consequently, the epicenters are most likely in the tectonic boundary between the Strandja and İstanbul-Zonguldak Zone and can be correlated with inactive fault zones such as the WBSF (Figure 2).

2.5 The tectonic implication of the fault plane solutions

The earthquakes in the İstanbul metropolitan area occurred outside the PDZ (NAF) and are evidence of the crustal deformation in the İstanbul-Zonguldak Zone. One of the possible explanations for the focal mechanism solutions mentioned above is that the Marmara Sea and its surroundings are under the control of the NASZ, which significantly widens in the Marmara Region (Şengör et al. 2005, 2014). Therefore, the Marmara Region is not a pure dextral strike-slip regime and is characterized by a transtensional system. The sketch in Figure 8 shows the structures associated with the transtensional regime and corresponding focal mechanism solutions obtained in this study. In pure dextral strike-slip deformation, the maximum principal axis (s₁, pressure) has an azimuthal direction of 135° from the north for an ideal case. If the region has a transtensional characteristic, the azimuth of s₁ decreases to ~120° given in Figure 8a (Şengör et al. 2014). Wollin et al. (2019) also presented a similar s₁ direction of ~125° analyzing the Marmara earthquakes. The P-axis azimuth angles of the recent İstanbul earthquake in Figure 8b indicate that the earthquakes occur under the WNW-ESE compressional force. The average azimuth of P-axes is 110°, which is compatible with s₁ of a transtensional regime.

The secondary structures such as Riedel, anti-Riedel, P shears and normal/thrust faults are well oriented according to the principal stresses of this dextral shear system. The findings (Figure 5, Table 1) show that the E-W right-lateral nodal planes with normal components of the 2010 and 2011 events are compatible with the synthetic Riedel shears (R) of the transtensional character of the Marmara. The 29 September 1999 Tuzla event occurred in the same region after the 1999 Kocaeli earthquake (M_w 7.4) and has a SW-NE nodal plane with a right-lateral motion like a P shear. The high-resolution foci of the micro-earthquakes in the Tuzla cluster in Figure 7 support the NNE dipping fault surface. The Tuzla cluster is under the Tuzla-İçmeler geothermal area used for medical treatments. Therefore, this buried fault surface might be the source of thermal water.

The NW-SE nodal plane with the right-lateral strike-slip motion of the 19 January 2015 earthquake on the northern shelf of the Marmara Sea agrees with the NW-SE lineaments between Büyükçekmece and Küçükçekmece lakes (Gökaşan et al. 2003). Ergintav et al. (2011) interpret these lineaments as a series of right-lateral faults using high-resolution seismic data and local GPS campaigns. Although these faults extend to the NAF at an angle of ~70°, they are not related to an anti-Riedel (R') shear. They may relate to the activation of the old fault systems, which represents the pre-times of initiations of the NAF in the region. On the other hand, the NNW-SSE left-lateral mechanisms of the 2012, 2014, and most recent 2021 Kartal earthquakes are likely on R' shear. The clear normal fault solution of the 2008 event and its NW-SE nodal plane agrees with the orientation of a tensional structure.

The 1923, 1929, 2012 and 2014 earthquake epicenters are in the zone of the buried right-lateral West Black Sea Fault between the Strandja Massif and İstanbul-Zonguldak Paleozoic (Figure 2, Figure 5). The NW-SE left-lateral nodal planes of the recent two earthquakes also agree with the orientation of the fault. These four events occur in the deformation zone of the paleo-transform fault that also has the same orientation as anti-Riedel shears (R'). On the Asian side, the possible fault plane strikes (Table 1) are similar to the faults between the Paleozoic units in the Kartal-Tuzla region. Considering the orientation consistency between the possible Riedel/anti-Riedel shears of the current transtensional regime in Marmara and the Paleozoic faults in İstanbul, we can interpret the earthquakes mentioned above may occur on the old fault surface under the current stress loading.

Thus far, we evaluate the neglected seismic activity and seismotectonics in the dense settlement areas in İstanbul by conducting current transtensional tectonics and Paleozoic faults. The continuous stress loading due to the westward motion of Anatolia may cause an earthquake with a higher magnitude than the previous ones in the preexisting weak zones in the study area. Such a moderate event (Mw ~5) can damage the old and won multistory buildings in the city. The epicenter, depth and source mechanism of such a future event in the populated area cannot be estimated. However, the seismological and geological evidence discussed above allows us to foresee the seismic hazard/risk of the metropolitan area. Apart from the previous studies on a destructive (Mw ≥ 7) earthquake on the NAF beneath the Marmara Sea (i.e., Sorensen et al. 2006; Douglas & Aochi 2016; Aochi et al. 2017), the effects of a future moderate earthquake in the İstanbul metropolitan area must be investigated in light of the newly presented data. For this aim, we perform and discuss numerical ground motion simulations of an Mw 5 earthquake utilizing the source and 1D velocity model properties in the next section.

3. BROADBAND GROUND MOTION SIMULATIONS FOR A MODERATE EARTHQUAKE IN ISTANBUL

Based on the past and present activity levels in the metropolitan area (Figure 1, Figure 5), ground motions of a moderate scenario earthquake and their effect on weak constructions can be estimated. For this aim, we conduct broadband ground motion simulations to generate synthetic seismograms in the metropolitan area, including the source, path, and site effects. A hypothetical event source in the simulation is characterized by faulting parameters and rupture propagation. The path effect is defined with the appropriate 1D crustal velocity structure of the region utilizing the previous studies. The amplification of the subsurface soil calculated from the 1D shallow S-wave velocity structure is also included as the site effect.

We use the epicenter locations in Figure 5 for the hypothetic source areas (A-E). Two of them are on the Asian side: Kartal (A) and Pendik (B). The other three sources are on the European side of the city: Durusu Lake (C), Esenyurt (D), and Sultangazi (E). Simulating earthquake ruptures for these five source areas at 11 strong motion stations (Figure 5, Table 4) allows us to overview the impact of a possible medium-sized earthquake in the metropolitan area.

3.1 Determining 1D shallow S-wave velocity structures and site amplifications

Defining a proper subsurface velocity model, which controls soil amplification, is a key point in obtaining a reliable synthetic waveform after deterministic numerical simulation on the bedrock. The 1D shallow soil velocity structures beneath the strong motion sites in the Istanbul metropolitan area are retrieved from the AFAD station reports (tadas.afad.gov.tr) based on MASW (Multichannel Analysis of Surface Waves), REMI (Refraction Microtremor) and single station microtremor measurements. The stations' information and 1D shallow soil Vs structures used in this study are given in Table 4 and Figure 9, respectively. The MASW observation depths are limited in the reports, and there is no velocity information of the soil layers deeper than 20 m. Therefore, the REMI observation results that contain information down to engineering bedrock with Vs of 700-800 m/s are preferred. Because there is no site report for stations 3415, 3427, and 3428 in Figure 5, the soil velocity models for these three sites are obtained from the nearby microtremor study results given in the İBB microzonation reports (İBB 2007, 2009). On the other hand, there is no engineering bedrock observation at the sites 3411, 3412, 3415, and 3428 in Figure 5. Because an approximate depth of the engineering bedrock at these sites is needed to calculate soil amplification, we utilize the empirical relation by Karabulut & Özel (2018) for the study area. The engineering bedrock velocity is used as 780 m/s, considering the AFAD site reports and previous studies (Birgoren 2009; Karabulut & Ozel 2018; Karagoz et al. 2019).

Because the amplification of the subsurface soil layers is hard to determine, it is assumed that SH-wave is incident vertically to the horizontal layers on the engineering bedrock, and linear site amplification for S-wave is calculated using Haskell's (1960) 1D multiple reflections. The amplitude ratio between the waves at the surface and the incident on the engineering bedrock is used as a soil amplification factor. Kramer (1996) indicates that the incident motion amplitude is half of the surface motion amplitude due to the free stress condition. Therefore, the amplification factor is calculated as two at low frequencies (Figure 9, top row). The linear site amplification factor for the sites in İstanbul is presumed as the effects of the soil layers over the engineering bedrock. The quality factors (Q) of the layers are assumed to be constant at 1/15 of Vs (Q = Vs/15) in this study (Iida et al. 2005; Karagoz et al. 2015).

The rock sites (ZA, ZB) show no amplification at low frequencies (< 10 Hz). The other sites have significant S-wave amplifications. While the peaks of the calculated amplification (4-5) for the site class ZC are between 3 and 10 Hz, the ZD class sites have high amplification factors (5-7) between 0.3 and 3 Hz. Although site 3416 (Yeşilköy) is in the ZC class, its fundamental frequency (f₀) is ~1 Hz and stands out from the other sites in the class because the observed engineering bedrock depth is about 110 m (Figure 9 top).

The horizontal-to-vertical spectral ratio from earthquake data

The fundamental frequencies, f₀, obtained from the horizontal-to-vertical spectral ratio (HVSR) are presented in the AFAD site reports. However, we notice that most f₀ are not reliable values according to the HVSR analyses criteria given in SESAME (2004). The HVSR peaks for all sites are unclear, and the amplitudes (H₀) are lower than the minimum limit of 2, except the station 3412 (Büyükçekmece), located on the Quaternary alluvium. The HVSR curves from the single station observations at the selected sites (3417, 3411, and 3416) in the reports are shown in Figure 10. Although the HVSRs are approximately flat below the ratio of 2, the f₀ values are reported at 7.3 and 2.1 Hz for 3417 and 3411, respectively. Site 3416 has a clear peak at ~6.4 Hz, but the H₀ is 0.9. On the contrary, a proper amplification is obtained at 3412 at 1 Hz with an amplitude of ~4.

We re-calculate the HVSR using the raw strong-motion records of the AFAD stations to verify the reported fundamental frequencies (Figure 11). A minimum of ten earthquakes that occurred in the Marmara Region are selected for each site considering high PGA values rather than event magnitudes. Although the selected records have high PGAs, it is impossible to define P, S and coda parts for all records because of background noise, low signal/noise ratio and soil effects. On the other hand, Kawase et al. (2018) mentioned that the HVSR from S-wave and coda parts are similar. Therefore, the HVSR amplitudes are calculated using the baseline-corrected full waveforms (at least 60 s-length). Before calculating the ratios, the Fourier spectra of three components are smoothed with Konno & Ohmachi's (1998) approach. The individual HVSRs are averaged to obtain the main characteristics of the soil structure beneath the sites. The results of HVSR analyses are given in Figure 11 and show a good correlation with the site amplifications in Figure 9 (see Figure A2 for goodness-of-fit scores). The stations with site class ZA and ZB have no amplification. Only station 3417 has high HVSR values (~5) between the frequency 0.2 and 2.0 Hz for the 19 June 2021 Kartal earthquake that occurred beneath the site. Also, the estimated theoretical amplification of the defined 1D shallow soil Vs model has a good agreement with the HVSR from the earthquakes in Figure 9. The HVSR peaks are clear at high and low frequencies for ZC and ZD site classes, respectively. Site 3416 shows a peak at low frequencies (<1 Hz) that is compatible with the calculated amplification factor in Figure 9. Site 3411 in the governor's office (class ZD) has a peak value at frequencies higher than 1 Hz, similar to site class ZC. The HVSR characteristics of these two sites are incompatible with their reported site class information and further site investigations must be needed. On the contrary, the HVSR curve of 3412 (class ZD) in the AFAD report also agrees with the average curve of the earthquake data.

3.2 Deterministic numerical waveform simulations

Characteristic source model definition and seismogram calculation

We follow Karagoz et al. (2018) to evaluate the effects of different scenario earthquake sources in İstanbul. Takeo's (1985) algorithm, based on the DWNM by Bouchon & Aki (1977) and Bouchon (1981), is used to calculate synthetic velocity seismograms at the selected sites. The method generates a wavefield in space, and a horizontally stratified velocity model is utilized. Event sources are parameterized according to the recipe by Irikura & Miyake (2011). Although the recipe is proposed for large earthquake simulations, it could be successfully applied to moderate events (Karagöz 2022).

Based on the characterization of the rupture area by Sommerville et al. (1999), we define a single asperity (ASP) with a high-stress drop and a background (BG) area with less stress drop. We adopt area (S) and slip (D) partitioning between asperity and background areas, considering S_a / S_b = 0.22 and D_a / D_b = 2.3 (Somerville et al. 1999; Dalguer et al. 2004). The BG, which is responsible for the low-frequency part of seismograms, has a low-stress drop (∆s_b = 1 MPa), while the ASP that generates strong ground motion has a high-stress drop (Ds_a = 10 MPa). Rather than using a point source, propagating rupture with V_r = 2.7 km/s is implemented to the defined characterized source model (CSM). The BG area is divided into subfaults of 150 m x 150 m. On the other hand, the smaller subfaults (100 m x 100 m) for the ASP are constructed to prevent artificial frequencies and amplification in seismograms (Figure 12). The ASP subfault size related to rupture velocity and waveform sampling (∆s_max = V_r * ∆t = 2700*0.05 = 135 m) ensures the limit of the largest subfault size proposed by Panza & Suhadolc (1987). To generate a realistic rupture front and increase the high-frequency content of seismograms, incoherent random fluctuation is applied to the theoretical rupture time of each subfault (see Karagoz et al. 2018; Karagoz 2022 for details). Each subfault is assumed to be a point source and has the same faulting parameters (strike, dip, rake).

The velocity seismogram of each subfault on the bedrock is calculated by the DWNM algorithm developed by Takeo (1985). Because of the maximum and minimum wave number related to the seismic velocities in the velocity model, the maximum wavelength criteria by Bouchon (1981) are considered in the simulations. The calculated seismograms of the subfaults are summed according to their rupture start times utilizing the point source summation technique by Spudich & Archuleta (1987) to obtain the overall synthetic velocity time series. Then, the surface ground motions are calculated using 1D linear amplifications of the shallow soils in the frequency domain. The amplitude spectra of bandpass filtered (0.1 – 10 Hz) bedrock motions are divided by two to remove the free-stress effect and multiplied by the site amplifications in the frequency domain. The surface motion time series are obtained with the inverse Fourier transform in the final stage.

Besides the felt event magnitudes in the İstanbul metropolitan area ranging from 3.0 to 5.0 (Table 1), there is a possibility of a moderate earthquake at the same locations in the future. Therefore, we numerically generate the seismograms utilizing the CSM of a moderate event with M_w 5.0. According to the previous earthquake self-similarity studies (Wells & Coppersmith 1994; Mai & Beroza 2000; Tan & Taymaz 2005), the model event can rupture about 4 x 4 km² area in the upper crust. The ASP area is 1.9 x 1.9 km² at the center of the BG, agreeing with Irikura & Miyake's (2011) recipe. The rupture propagation with random fluctuation of the CSM used in the simulations is given in Figure 12. A circular rupture stars at the center of both areas, and there is no time delay between them. The different rupture propagation geometries and starting points are not tested as in larger earthquake (Mw ≥ 6) simulations because the CSM has a smaller area. The smoothed-ramp type source time function is used for the subfaults. The rise time (t_r) is 2.5 s for BG to generate low-frequency content around 0.4 Hz and 0.2 s for ASP to obtain high frequency components up to f_max = 5 Hz.

Estimation of the broadband ground motions for scenario earthquakes

We construct different faulting parameters at five locations, considering the results in the previous section, to investigate peak ground motions and spectral accelerations for an event in the Istanbul metropolitan area. The possible future moderate events in the previous epicenter areas are assumed, and their source mechanisms are derived from the solutions given in Table 1 and Figure 5. The five epicenters (Kartal, Pendik, Durusu Lake, Esenyurt, and Sultangazi) with approximate coordinates and different faulting mechanisms at each location are given in Table 6. Because the event focal depths are between 6 and 15 km, the same depth (10 km) is used for all models (A-E). The fault planes are chosen according to the interpretations based on the right-lateral transtensional stress regime, summarized in Figure 8. The first source models (#1) in Table 6 are the same as the observed focal mechanisms. The successive nine models (#2-7) represent the uncertainties of the strike, dip and rake angles given in Table 1. Models #8 and #9 have similar faulting orientation and dip angles but different rake angles (± 45°). The last models (#10) have a low dip angle (45°). These different source models allow obtaining the effect of radiation patterns for S-waves.

The ten scenarios (M_w 5) at each epicenter location are done for the 11 strong motion sites to sample different source models and site classes in the İstanbul metropolitan area. The distance between the source and site ranges from ~2 to ~80 km. A total of 1100 horizontal velocity seismograms (NS, EW) of the 550 simulations are generated. Selected waveforms calculated on the different site classes are shown in Figure 14. The scenario events with a short epicentral distance (< 20 km) contain high frequency waveforms (f > 1 Hz) on the rock sites (3405, 3418). The seismograms calculated at the stiff soil sites (especially at 3415-Küçükçekmece and 3428-Avcılar) with larger epicentral distances (>20 km) show low-frequency contents (f ≤ 1 Hz). The peak ground velocities (PGV) of the synthetics reach up to 13 cm/s. While the waveform durations increase at the distant sites (>50 km), the amplitudes decrease due to attenuation. Therefore, a moderate event with a moment magnitude of ~5 may strongly affect the structures on the same side of İstanbul as the epicenter.

The velocitograms are derived in the frequency domain to obtain accelerograms, peak ground accelerations (PGAs) and spectral accelerations for the scenarios. The maximum simulated PGA value from the synthetic models is about 0.3 g and corresponds to the felt intensity of MMI IX (great damage) according to the relationship by Bilal & Askan (2014). The simulated PGAs are compared with the ground motion prediction equations (GMPE) to generate an overall image between the prediction curves and a set of different possible event sources. We implement two prediction models developed by (1) Akkar & Çağnan (2010, AC2010) derived from the local Turkish strong ground motion database and (2) Boore et al. (2014, B2014) using the global strong-motion observations in the seismically active areas including Turkey (Figure 15). The GMPEs are calculated for an M5 event with oblique faulting considering our source models. The site response of the engineering bedrock (Vs = 760 m/s) is assumed for a generic relation because different Vs30 values for soil classes do not show a remarkable difference in interpretation (Figure A3 in Appendix). The PGAs from the simulations are in good agreement with the B2014 model for the distance between 5 and 30 km. The values are less than predicted at the rock sites with shorter distances (R_jb < 5 km). However, the B2014 curve is higher than the simulated PGAs for all site classes at larger distances (> 30 km). The AC2010 model predicts lower PGAs than the B2014 and shows a better fit for the same distances. Some extremely high values are also obtained from the Model-D simulations for the stiff soil sites 3412 and 3415 (V_S30 < 300 m/s) located near Büyükçekmece and Küçükçekmece lakes, respectively. These high values relate to the site amplification of stiff soil class (ZD) for these source-site pairs. Similar high PGA values above the empirical GMPE are observed at the stiff soil sites close to the shoreline in the Bornova Basin, İzmir, during the 30 October 2020 Samos earthquake (Akinci et al. 2021). The PGA differences among the different focal mechanisms at a source indicate the radiation pattern effects on the waveforms. Two clear examples of the rock sites are seen in Figure 15 for the scenario models A and B. The simulated PGAs at station 3417 for the Model-A range from 0.02 to 0.07 g. The second example is 3418, the nearest station for source B, and its PGA values change from 0.07 to 0.18 g for different faulting types. Depending on the radiation pattern, the PGA value may vary 3-4 times. It can be concluded that the faulting type is important for short epicentral distances in hazard estimation for urban areas. On the other hand, the PGA difference because of the radiation pattern decreases with the source-station distance. Additionally, it can be noticed that the PGAs decrease faster with distance in the simulations than in the GMPEs in Figure 15. One reason for this discrepancy may be the common average velocity structure with assumed quality factors used in the simulation. The other reason may be the performance of the GMPEs in the study area. Similar differences were also observed in İzmir city for the 2020 Samos event (i.e., Akinci et al. 2021; Askan et al. 2021; Gulerce et al. 2022).

The observed PGAs from the latest 2021 Kartal earthquake (M_L 4.1) are also compared with both empirical models in Figure 15. The values are classified according to the soil conditions at the recording sites. A good correlation with the M 4.0 curve is seen for site classes A and C (no observation for B-class). However, the observed PGA values for the D-class sites are higher than the empirical relationships (R_jb > 20 km). All models are also used with a focal depth of 5 km to understand shallow source effects (Figure A4). While the focal distance does not change significantly, the PGA values slightly decrease or increase at different sites because of changing the station location on the radiation pattern.

Pseudo acceleration response spectra for the scenarios

The response spectrum is used to evaluate the potential damage of seismic waves on buildings. The spectral amplitudes are the measurement tools for building design that resists earthquake forces lower than the spectral values as a function of period. In this section, the design spectra of TBEC (2018) are compared with the spectra calculated from the simulated ground motions to understand the effect of an M_w 5.0 event beneath the İstanbul metropolitan area. Figure 16 shows pseudo-spectral accelerations (Sa) from the horizontal components (NS, EW) and their averages (H_ave) at 11 sites. In addition, the response spectra for the return periods of 43 and 72 years are also presented. The simulation results indicate that the pseudo-spectral accelerations can reach up to 0.4-0.6 g according to the source parameters and soil class.

The rock sites (3405, 3417, 3418) close to the sources have high spectral values. The Sa amplitudes at these sites exceed the design spectrum around 0.2-0.6 s periods, which are the natural vibration periods of the buildings up to 6 stories. The simulated spectral accelerations of certain models are very close to the upper boundaries of the design-level spectrum between 0.4 and 1 s periods, leading to resonance phenomena. The results for site 3416 also show that model-E generates acceleration values close to the design level. The highest Sa amplitudes between 0.2 and 1 s are calculated at 3415 near Küçükçekmece Lake. The event sources D and E with a magnitude of M_w 5 have damage potential for the mid-rise buildings designed according to the recent TBEC (2018). The simulation results of a focal depth of 5 km indicate that the shallow sources can generate high Sa amplitudes at higher periods of 0.6 – 0.8 s (Figure A5). The Sa curves of 3418-NS for model-A, 3413-EW and 3415-NS/EW for model-E are good examples of increasing spectral amplitudes. Thus, the hazard risk on the recent mid-rise buildings increases and the older buildings are much more fragile under shallow earthquakes.

The calculated spectral acceleration results show that the source parameters and the epicentral distance of a possible moderate event beneath the İstanbul metropolitan area have an important role in seismic hazard assessment. Considering that 69% of the buildings (~1.2 million) in Istanbul were constructed before 2000 and mid-rise buildings consist of 98% of the total (İBB 2022), an M_w ~5 event may seriously damage the near-source buildings that were not constructed according to the engineering applications and recent design codes. As seen in İzmir city, after the 30 October 2020 Samos earthquake (M_w 7.0), the buildings with poor construction quality are easily damaged although the observed acceleration spectra at the sites are below the previous and current Turkish seismic design codes (Akinci et al. 2021; Askan et al. 2021).

4 DISCUSSION AND CONCLUSION

The Istanbul metropolitan area, where about 16 million people live, is one of the most populated megacities in the world. The seismic hazard studies for Istanbul city have been done based on a future destructive earthquake (M_w > 7) in the principal deformation zone, NAF, of the North Anatolian Shear Zone. Although the number of events in Istanbul is very low and the blasts mask seismicity, the felt events (M_w 3.0 – 4.5) occur on both sides of the city, and the focal areas of these events may be the source of a future earthquake. In this study, we focus on the two main topics about the earthquakes not interested previously in the metropolitan area: (1) Investigating earthquake activity beneath the İstanbul metropolitan area and its tectonic implications, (2) revealing the effect of a future moderate event in İstanbul.

The two moderate events in the 1920s and the recent activity in 2006-2016 are analyzed in detail to understand the faulting characteristics and their relations with regional tectonics. The earthquakes are outside the North Anatolian Fault and occurred on the secondary shear structures in the North Anatolian Shear Zone. The earthquakes with magnitudes of 3.0 – 4.5 in the metropolitan area indicate five significant event source areas on both sides of İstanbul. Furthermore, the fault plane solutions based on the P-wave first motion polarities imply that the strike-slip component is dominant. One of the source areas, in the Tuzla-İçmeler geothermal area, has the highest micro-earthquake activity, and the determined north dipping fault might be the thermal water source.

The fault plane solutions and properly selected nodal planes of the recent earthquakes agree with the synthetic/antithetic shears and normal structures in the NASZ. The P-axes of the focal mechanisms indicate a WNW-ESE compressional force (N110°E) that matches the maximum principal stress of the transtensional regime. We find out that the epicenter location and NW-SE nodal plane orientation of the recent earthquakes on the European side of İstanbul are in good agreement with the WBSF. Moreover, the focal mechanism solutions of the Asian side events, including the most recent 19 June 2021 Kartal earthquake (M_L 4.1), have a coherent strike angle with the mapped Paleozoic faults. Because of the orientation consistency between the possible Riedel/anti-Riedel shears of the current transtensional regime in Marmara and the Paleozoic faults in İstanbul, it is inferred that the earthquakes (M_w 3-4) may occur on the old fault surfaces.

The continuous stress loading due to the westward motion of Anatolia may cause a moderate earthquake (Mw 5.0-5.5) on the preexisting weak zones in the study area. Considering the five focal areas as the future moderate-event sources, we investigate earthquake effects utilizing deterministic numerical ground motion simulations with the source, path, and site effects at the 11 strong-motion stations in İstanbul. The simulated PGAs agree with the curves of the ground motion prediction equation by Boore et al. (2014) between 5 and 30 km of epicentral distances. The maximum PGA value of the synthetic models is about 0.3 g and corresponds to the felt intensity of MMI IX according to the empirical relation obtained from the Turkish earthquakes. On the other hand, we find evident PGA differences among the focal mechanisms for a source at the same site because of the radiation pattern effect. This result implies that the faulting type has an important role for short epicentral distances in seismic hazard estimation for urban areas.

The simulated spectral accelerations for the M_w 5 earthquake scenarios can exceed the design spectrum given in the Turkish Building Earthquake Code (2018) between 0.2 and 0.6 s. The focal source mechanism and focal distance control the level of spectral values for both rock and soil sites. Certain models also generate spectral accelerations close to the design-level spectrum between 0.4 and 1 s-periods, leading to resonance phenomena. The results indicate that a moderate event in the İstanbul metropolitan area has damage potential for the mid-rise buildings (4 - 10 stories) up to site condition with resonance phenomena and poor construction quality and also for recently constructed buildings according to the new seismic design codes.

Because some inconsistencies in the site parameters are determined in this study according to the different data sets, new site investigations are needed at strong-motion stations. Nevertheless, İstanbul Metropolitan Municipality and the universities started common multidisciplinary studies covering the metropolitan area for the earthquake hazard assessment in 2020-2021. Moreover, our results indicate that detailed knowledge about the regional crustal structure, rupture characteristics of earthquake source and site-specific soil properties are essential for a better understanding of the seismic hazard and predicting the earthquake ground motions in a region.

ACKNOWLEDGEMENTS

Onur Tan thanks Dr. Felix Waldhauser, who provided the HypoDD inversion program, and Dr. Hayrullah Karabulut for Bouchon's discrete wave number algorithm. Özlem Karagöz thanks Dr. Hiroaki Yamanaka for the numerical waveform simulation codes. We also thank Dr. Cengiz Zabcı and Dr. Aynur Dikbaş for their grateful comments on the tectonic interpretations. Finally, we would like to thank the two anonymous reviewers and Dr. Eiichi Fukuyama, the Editor, for the valuable and constructive comments. All maps and graphs, except Fig. 9 and 10, are plotted using the Generic Mapping Tools (GMT) by Wessel and Smith (1998).

We are very grateful to the Istanbul Metropolitan Municipality (İBB), Directorate of Earthquake and Geotechnical Investigation for supplying the earthquake data and different geological and tectonic maps of the study area. This study was not realized without their open data policy. "Investigation of Possible Active Faults in Istanbul Terrestrial Area and Development of Landslide Determination and Monitoring Methodologies by Carrying Out Multidisciplinary Researches in Primary Landslide Areas" (2009 - 2016), "Monitoring, Research and Investigation of Various Landslide Areas in Beylikdüzü and Büyükçekmece Counties" (2013 - 2016) supported by İBB; "New Directions in Seismic Hazard Assessment through Focused Earth Observation in the Marmara Supersite - MARsite" (2012 - 2016) supported by EU-FP7. Open database: GEOFON and EIDA Data Archives, 2015 (Bianchi et al., 2015).

APPENDIX

REFERENCES

[dataset]* AFAD, 1990. Turkish National Seismic Network [Data set]. Department of Earthquake, Disaster and Emergency Management Authority. https://doi.org/10.7914/SN/TU

Akkar, S. & Çağnan, Z., 2010. A local ground-motion predictive model for Turkey, and its comparison with other regional and global ground-motion models. Bulletin of the Seismological Society of America, 100(6), 2978-2995.

Akinci, A., Cheloni, D. & Dindar, A.A., 2021. The 30 October 2020, M7.0 Samos Island (Eastern Aegean Sea) Earthquake: effects of source rupture, path and local-site conditions on the observed and simulated ground motions. Bulletin of Earthquake Engineering, 19 (12), 4745-4771.

Aksoy, M.E., Meghraoui, M., Vallée, M., & Çakır, Z., 2010. Rupture characteristics of the AD 1912 Mürefte (Ganos) earthquake segment of the North Anatolian fault (western Turkey). Geology, 38(11), 991-994.

Altunel, E., Meghraoui, M., Akyüz, H. S., & Dikbas, A., 2004. Characteristics of the 1912 co‐seismic rupture along the North Anatolian fault zone (Turkey): Implications for the expected Marmara earthquake. Terra Nova, 16(4), 198-204.

Ambraseys, N., 2009. Earthquakes in the Mediterranean and Middle East: a multidisciplinary study of seismicity up to 1900. Cambridge University Press.

Anderson, J., 2004. Quantitative measure of the goodness-of-fit of synthetic seismograms. In Proceedings of the 13th World Conference on Earthquake Engineering, Vancouver, paper 243. Vancouver, B.C., Canada, 1–6 August 2004. Earthquake Engineering Research Institute.

Aochi, H. & Ulrich, T., 2015. A probable earthquake scenario near Istanbul determined from dynamic simulations. Bulletin of the Seismological Society of America, 105(3), 1468-1475.

Aochi, H., Douglas, J. & Ulrich, T., 2017. Stress accumulation in the Marmara Sea estimated through ground‐motion simulations from dynamic rupture scenarios. Journal of Geophysical Research: Solid Earth, 122(3), 2219-2235.

Armijo, R., Meyer, B., Navarro, S., King, G. & Barka, A., 2002. Asymmetric slip partitioning in the Sea of Marmara pull‐apart: A clue to propagation processes of the North Anatolian fault?. Terra Nova, 14(2), 80-86.

Armijo, R., Pondard, N., Meyer, B., Uçarkus, G., de Lépinay, B. M., Malavieille, J., ... & Sarikavak, K., 2005. Submarine fault scarps in the Sea of Marmara pull‐apart (North Anatolian Fault): Implications for seismic hazard in Istanbul. Geochemistry, Geophysics, Geosystems, 6(6).

Askan, A., Gülerce, Z., Roumelioti, Z., Sotiriadis, D., Melis, N. S., Altindal, A., Akbaş B., Sopaci E., Karimzadeh S., Kalogeras I., Theodoulidis N., Konstantinidou K., Özacar A.A., Kale, Ö. & Margaris, B., 2021. The Samos Island (Aegean Sea) M7.0 Earthquake: Analysis and engineering implications of strong motion data. Bulletin of Earthquake Engineering, 1-26.

Bilal, M. & Askan, A., 2014. Relationships between felt intensity and recorded ground‐motion parameters for Turkey. Bulletin of the Seismological Society of America, 104(1), 484-496.

Birgören, G., Özel, O. & Siyahi, B., 2009. Bedrock depth mapping of the coast south of Istanbul: comparison of analytical and experimental analyses. Turkish Journal of Earth Sciences, 18(2), 315-329.

Bohnhoff, M., Bulut, F., Dresen, G., Malin, P. E., Eken, T., & Aktar, M., 2013. An earthquake gap south of Istanbul. Nature communications, 4 (1), 1-6.

Boore, D.M., Stewart, J.P., Seyhan, E. & Atkinson, G.M., 2014. NGA-West2 equations for predicting PGA, PGV, and 5% damped PSA for shallow crustal earthquakes. Earthquake Spectra, 30(3), 1057-1085.

Bouchon, M., 1981. A simple method to calculate green's functions in elastic layered media. Bulletin of the Seismological Society of America, 71, 959–971.

Bouchon, M. & Aki, K., 1977. Discrete wave-number representation of seismic-source wave fields. Bulletin of the Seismological Society of America, 67(2), 259-277.

Çakir, Z., Barka, A. A., & Evren, E., 2003. Coulomb stress interactions and the 1999 Marmara earthquakes. Turkish Journal of Earth Sciences, 12 (1), 91-103.

Dalguer, L.A., Miyake, H. & Irikura, K., 2004. Characterization of dynamic asperity source models for simulating strong ground motion. In Proceedings of the 13th World Conference on Earthquake Engineering (Vol. 3286).

Diao, F., Walter, T. R., Solaro, G., Wang, R., Bonano, M., Manzo, M., Ergintav, S., Zheng, Y., Xiong, X. & Lanari, R., 2016. Fault locking near Istanbul: Indication of earthquake potential from InSAR and GPS observations. Geophysical Supplements to the Monthly Notices of the Royal Astronomical Society, 205(1), 490-498.

Douglas, J. & Aochi, H., 2016. Assessing components of ground‐motion variability from simulations for the Marmara Sea region (Turkey). Bulletin of the Seismological Society of America, 106(1), 300-306.

[dataset]* Bianchi, M., Evans, P. L.; Heinloo, A., Quinteros J., 2015. WebDC3 Web Interface. GFZ Data Services. doi:10.5880/GFZ.2.4/2016.001.

Emre, Ö., Duman, T.Y., Özalp, S., Şaroğlu, F., Olgun, Ş., Elmacı, H. & Çan, T., 2018. Active fault database of Turkey. Bulletin of Earthquake Engineering, 16(8), 3229-3275.

Ergintav, S., Reilinger, R.E., Çakmak, R., Floyd, M., Cakir, Z., Doğan, U., King, R.W., McClusky, S. & Özener, H., 2014. Istanbul's earthquake hot spots: Geodetic constraints on strain accumulation along faults in the Marmara seismic gap. Geophysical Research Letters, 41(16), 5783-5788.

Ergintav, S., Doğan, U., Gerstenecker, C., Cakmak, R., Belgen, A., Demirl, H., Aydın, R. & Reilinger, R., 2007. A snapshot (2003–2005) of the 3D postseismic deformation for the 1999, Mw= 7.4 Izmit earthquake in the Marmara Region, Turkey, by first results of joint gravity and GPS monitoring. Journal of Geodynamics, 44(1-2), 1-18.

Ergintav, S., Demirbağ, E., Ediger, V., Saatçılar, R., Inan, S., Cankurtaranlar, A., Dikbas, A. & Baş, M., 2011. Structural framework of onshore and offshore Avcılar, Istanbul under the influence of the North Anatolian fault. Geophysical Journal International, 185(1), 93-105.

Ergintav, S., McClusky, S., Hearn, E., Reilinger, R., Cakmak, R., Herring, T., Ozener, H., Lenk, O. & Tari, E. 2009. Seven years of postseismic deformation following the 1999, M= 7.4 and M= 7.2, Izmit‐Düzce, Turkey earthquake sequence. Journal of Geophysical Research: Solid Earth, 114(B7).

Fréchet, J., 1985. Sismogenèse et doublets sismiques, Thèse d’Etat, Université Scientifique et Médi- cale de Grenoble, 206 pp.

Gokaşan, E., Ustaömer, T., Gazioğlu, C., Yucel, Z.Y., Öztürk, K., Tur, H., Ecevitoğlu, B. & Tok, B., 2003. Morpho-tectonic evolution of the Marmara Sea inferred from multi-beam bathymetric and seismic data. Geo-Marine Letters, 23(1), 19-33.

Goldstein, P., Dodge, D., Firpo, M. & Minner, L., 2003. "SAC2000: Signal processing and analysis tools for seismologists and engineers, Invited contribution to "The IASPEI International Handbook of Earthquake and Engineering Seismology", Ed. by WHK Lee, H. Kanamori, P.C. Jennings, and C. Kisslinger, Academic Press, London.

Got, J.L., Fréchet, J. & Klein, F.W., 1994. Deep fault plane geometry inferred from multiplet relative relocation beneath the south flank of Kilauea. Journal of Geophysical Research: Solid Earth, 99(B8), 15375-15386.

Gulerce, G., Akbaş, B., Özacar, A., Sopacı, E. Önder, F., Uzel, B., Can, G., Cakir, E., Ilgaç, M., Söylemez, B., Saltoğlu, N., Askan, A., Cetin, K. Ö., Unutmaz, B., 2022. Predictive performance of current ground motion models for recorded strong motions in 2020 Samos Earthquake. Soil Dynamics and Earthquake Engineering, 152, 107053.

Haskell, N.A., 1960. Crustal reflection of plane SH waves. Journal of Geophysical Research, 65(12), 4147-4150.

Hearn, E.H., Bürgmann, R. & Reilinger, R.E., 2002. Dynamics of Izmit earthquake postseismic deformation and loading of the Duzce earthquake hypocenter. Bulletin of the Seismological Society of America, 92(1), 172-193.

Horasan, G., Güney, A. B., Küsmezer, A., Bekler, F., Öğütçü, Z., & Musaoğlu, N., 2009. Contamination of seismicity catalogs by quarry blasts: An example from Istanbul and its vicinity, northwestern Turkey. Journal of Asian Earth Sciences, 34 (1), 90-99.

Hutton, L.K. & Boore, D.M.,1987. The M_L scale in southern California. Bulletin of the Seismological Society of America, 77(6), 2074-2094.

İBB, 2007. İstanbul European side microzonation project report. İstanbul Metropolitan Municipality (İBB, İstanbul Büyükşehir Belediyesi), October, 2007. https://depremzemin.ibb.istanbul/calismalarimiz/tamamlanmis-calismalar-2 (last visit 23 November 2022).

İBB, 2009. İstanbul Asian side microzonation project report. İstanbul Metropolitan Municipality (İBB, İstanbul Büyükşehir Belediyesi), November, 2009. https://depremzemin.ibb.istanbul/calismalarimiz/tamamlanmis-calismalar-2 (last visit 23 November 2022).

İBB, 2022. Neighborhood-Based Building Numbers in 2017. İstanbul Metropolitan Municipality (İBB, İstanbul Büyükşehir Belediyesi). https://data.ibb.gov.tr/dataset/mahalle-bazli-bina-analiz-verisi (last visit 23 November 2022).

Iida, M., Yamanaka, H. & Yamada, N., 2005. Wave field estimated by borehole recordings in the reclaimed zone of Tokyo Bay. Bulletin of the Seismological Society of America, 95(3), 1101-1119.

Irikura, K. & Miyake, H., 2011. Recipe for predicting strong ground motion from crustal earthquake scenarios. Pure and Applied Geophysics, 168(1), 85-104.

[dataset]* ISC, 2021a. Seismological Dataset Repository, International Seismological Centre (ISC), https://doi.org/10.31905/6TJZECEY

[dataset]* ISC, 2021b. The ISC electronic archive of printed station/network bulletins, International Seismological Centre, https://doi.org/10.31905/GNLY467C

Karabulut, H., Schmittbuhl, J., Özalaybey, S., Lengline, O., Kömeç-Mutlu, A., Durand, V., Bouchon, M., Daniel, G., & Bouin, M.P., 2011. Evolution of the seismicity in the eastern Marmara Sea a decade before and after the 17 August 1999 Izmit earthquake. Tectonophysics, 510(1-2), 17-27.

Karabulut, S. & Özel, O., 2018. 3-D shear wave velocity structure beneath the European Side of Istanbul from seismic noise arrays analysis. Geophysical Journal International, 215(3), 1803-1823.

Karagöz, Ö, 2022. Estimation of 1D Deep Vs Models in Çanakkale and Surrounding Area (Biga Peninsula, NW Turkey) Verified with Numerical Ground Motion Simulation of Moderate-Sized Earthquakes. Pure and Applied Geophysics, 179, 709–745.

Karagoz, O., Chimoto, K., Citak, S., Ozel, O., Yamanaka, H. & Hatayama, K., 2015. Estimation of shallow S-wave velocity structure and site response characteristics by microtremor array measurements in Tekirdag region, NW Turkey. Earth, Planets and Space, 67(1), 1-17.

Karagoz, O., Chimoto, K., Yamanaka, H., Ozel, O. & Citak, S., 2019. Microtremor Array Explorations and Earthquake Simulation Studies in Western Marmara Region, Turkey. International Geological Congress of Turkey, 24 Jan – 1 Feb 2019, Ankara, Turkey.

Karagoz, O., Chimoto, K., Yamanaka, H., Ozel, O. & Citak, 2018. Broadband ground-motion simulation of the 24 May 24 Gokceada (North Aegean Sea) earthquake (Mw 6.9) in NW Turkey considering local soil effects. Bull Earthquake Eng., 16, 23–43.

Karcioğlu, G., Tank, S.B., Gürer, A., Çiftçi, E.T., Kaya, T. & Tunçer, M.K., 2013. Upper crustal electrical resistivity structures in the vicinity of the Çatalca Fault, Istanbul, Turkey by magnetotelluric data. Studia Geophysica et Geodaetica, 57(2), 292-308.

Kawase, H., Mori, Y. & Nagashima, F., 2018. Difference of horizontal-to-vertical spectral ratios of observed earthquakes and microtremors and its application to S-wave velocity inversion based on the diffuse field concept. Earth, Planets and Space, 70(1), 1-32.

[dataset]* KOERI, 1971. Bogazici University Kandilli Observatory And Earthquake Research Institute [Data set]. International Federation of Digital Seismograph Networks. https://doi.org/10.7914/SN/KO

Kondorskaya, N.V. & Ulomov, V.I., 1999. Special catalogue of earthquakes of the northern Eurasia (SECNE). http://static.seismo.ethz.ch/gshap/neurasia/nordasiacat.txt (last visited 15.05.2022)

Konno, K. & Ohmachi, T., 1998. Ground-motion characteristics estimated from spectral ratio between horizontal and vertical components of microtremor. Bulletin of the Seismological Society of America, 88(1), 228-241.

Kramer, S.L., 1996. Geotechnical earthquake engineering. Pearson Education India.

Lienert, B.R. & Havskov, J., 1995. A computer program for locating earthquakes both locally and globally. Seismological Research Letters, 66(5), 26-36.

Lom, N., Ülgen, S.C., Sakinç, M. & Şengör, A.C., 2016. Geology and stratigraphy of Istanbul region. Geodiversitas, 38(2), 175-195.

Mai, P. M. & Beroza, G.C., 2002. A spatial random field model to characterize complexity in earthquake slip. Journal of Geophysical Research: Solid Earth, 107(B11), ESE-10.

Meade, B. J., Hager, B. H., McClusky, S. C., Reilinger, R. E., Ergintav, S., Lenk, O., ... & Ozener, H., 2002. Estimates of seismic potential in the Marmara Sea region from block models of secular deformation constrained by Global Positioning System measurements. Bulletin of the Seismological Society of America, 92 (1), 208-215.

Mindevalli, O.Y. & Mitchell, B.J., 1989. Crustal structure and possible anisotropy in Turkey from seismic surface wave dispersion. Geophysical Journal International, 98(1), 93-106.

Nakamura, Y., 1989. A method for dynamic characteristics estimation of subsurface using microtremor on the ground surface. Railway Technical Research Institute, Quarterly Reports, 30 (1).

Okay, A. I., Sengor, A.M.C. & Görür, N, 1994. Kinematic history of the opening of the Black Sea and its effect on the surrounding regions. Geology, 22(3), 267-270.

Okay, A. I., Kaşlılar-Özcan, A., İmren, C., Boztepe-Güney, A., Demirbağ, E. & Kuşçu, İ., 2000. Active faults and evolving strike-slip basins in the Marmara Sea, northwest Turkey: a multichannel seismic reflection study. Tectonophysics, 321(2), 189-218.

Özgül, N., 2012. Stratigraphy and some structural features of the İstanbul Paleozoic. Turkish Journal of Earth Sciences, 21(6), 817-866.

Panza, G.F. & Suhadolc, P., 1987. Complete strong motion synthetics. In B.A. Bolt (Ed.), Seismic strong motion synthetics (pp. 205–265). Academic Press Inc.

Picozzi, M., Strollo, A., Parolai, S., Durukal, E., Özel, O., Karabulut, S., Zschau, J. & Erdik, M., 2009. Site characterization by seismic noise in Istanbul, Turkey. Soil Dynamics and Earthquake Engineering, 29(3), 469-482.

Prieto, G.A., Parker, R.L. & Vernon F.L., 2009. A Fortran 90 library for multitaper spectrum analysis. Computers & Geosciences, 35(8), 1701-1710.

Reilinger, R., et al. 2006. GPS constraints on continental deformation in the Africa-Arabia-Eurasia continental collision zone and implications for the dynamics of plate interactions, J. Geophys. Res., 111, B05411, doi:10.1029/2005JB004051.

Sekiguchi, H. & Iwata, T., 2002. Rupture process of the 1999 Kocaeli, Turkey, earthquake estimated from strong-motion waveforms. Bulletin of the Seismological Society of America, 92(1), 300-311.

Şengör, A.M.C., Grall, C., İmren, C., Le Pichon, X., Görür, N., Henry, P., Karabulut, H. & Siyako, M., 2014. The geometry of the North Anatolian transform fault in the Sea of Marmara and its temporal evolution: implications for the development of intracontinental transform faults. Canadian Journal of Earth Sciences, 51(3), 222-242.

Şengör, A.M.C., Tüysüz, O., Imren, C., Sakınç, M., Eyidoğan, H., Görür, N., Le Pichon, X. & Rangin, C., 2005. The North Anatolian fault: A new look. Annu. Rev. Earth Planet. Sci., 33, 37-112.

SESAME, 2004. Guidelines for the implementation of the H/V spectral ratio technique on ambient vibrations: Measurements, processing, and interpretations, SESAME European Research Project WP12 -D23.12, http://sesame.geopsy.org/Papers/HV_User_Guidelines.pdf (last accessed 30 January 2022).

Snoke, J.A., Lee, W.H.K., Kanamori, H., Jennings, P.C. & Kisslinger, C., 2003. FOCMEC: Focal mechanism determinations. International handbook of earthquake and engineering seismology, 85, 1629-1630.

Somerville, P., Irikura, K., Graves, R., Sawada, S., Wald, D., Abrahamson, N., Iwasaki, Y., Kagawa, T., Smith, N. & Kowada, A., 1999. Characterizing crustal earthquake slip models for the prediction of strong ground motion. Seismological Research Letters, 70(1), 59-80.

Sorensen, M. B., Oprsal, I., Bonnefoy-Claudet, S., Atakan, K., Mai, P.M., Pulido, N. & Yalciner, C., 2006. Local site effects in Ataköy, Istanbul, Turkey, due to a future large earthquake in the Marmara Sea. Geophysical Journal International, 167(3), 1413-1424.

Sorensen, M.B., Stromeyer, D. & Grünthal, G., 2009. Attenuation of macroseismic intensity: A new relation for the Marmara Sea region, northwest Turkey. Bulletin of the Seismological Society of America, 99(2A), 538-553.

Spudich, P. & Archuleta, R., 1987. Techniques for earthquake ground-motion calculation with application to source parameterization of finite faults, seismic strong motion synthetics. In B.A. Bolt (Ed.), Seismic strong motion synthetics (pp. 205–265). Academic Press Inc.

Takeo, M., 1985. Near-field synthetic seismograms taking into account of the effects of anelasticity: the effects of anelastic attenuation on seismograms caused by a sedimentary layer. Meteorol Geophys, 36, 245–257. (in Japanese with English abstract).

Tan, O., 2013. The dense micro-earthquake activity at the boundary between the Anatolian and South Aegean microplates. Journal of Geodynamics, 65, 199-217.

Tan, O., 2021. A homogeneous earthquake catalogue for Turkey. Natural Hazards and Earth System Sciences, 21(7), 2059-2073.

Tan, O. & Taymaz T., 2005. Self-similarity of the earthquakes occurred in the Eurasia-Arabia collision zone (in Turkish), İTÜ Dergisi, 4 (6), 105-115.

Tan, O., Tapirdamaz, M.C. & Yörük, A., 2008. The earthquake catalogues for Turkey. Turkish Journal of Earth Sciences, 17(2), 405-418.

Tan, O., Tapirdamaz, M.C., Ergintav, S., Inan, S., Iravul, Y., Saatcilar, R., Tuzel, B., Tarancioglu, A., Karakisa, S., Kartal R.F., Zunbul S., Yanik K., Kaplan, M., Saroglu F., Kocyigit A., Altunel E. & Özel, N.M., 2010. Bala (Ankara) earthquakes: Implications for shallow crustal deformation in central Anatolian section of the Anatolian platelet (Turkey). Turkish Journal of Earth Sciences, 19(4), 449-471.

Toksöz, M. N., Shakal, A. F. & Michael, A. J., 1979. Space-time migration of earthquakes along the North Anatolian fault zone and seismic gaps. PAGEOPH, 117, 1258–1270.

Villasenor, A., Bergman, E.A., Boyd, T.M., Engdahl, E.R., Frazier, D.W., Harden, M.M., Orth, J.L., Parkes, R.L., & Shedlock, K.M., 1997. Toward a comprehensive catalog of global historical seismicity. Eos Trans. AGU, vol. 78, 50.

Waldhauser, F. & Ellsworth, W.L., 2000. A double-difference earthquake location algorithm: Method and application to the northern Hayward fault, California. Bulletin of the seismological society of America, 90(6), 1353-1368.

Wells, D.L. & Coppersmith, K.J., 1994. New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement. Bulletin of the seismological Society of America, 84(4), 974-1002.

Wessel, P. & Smith, W.H.F., 1998. New, improved version of the Generic Mapping Tools released. Eos, 79, 579.

Wollin, C., Bohnhoff, M., Martínez-Garzón, P., Küperkoch, L. & Raub, C., 2018. A unified earthquake catalogue for the Sea of Marmara Region, Turkey, based on automatized phase picking and travel-time inversion: Seismotectonic implications. Tectonophysics, 747, 416-444.

Wollin, C., Bohnhoff, M., Vavryčuk, V. & Martínez-Garzón, P., 2019. Stress inversion of regional seismicity in the Sea of Marmara Region, Turkey. Pure and Applied Geophysics, 176(3), 1269-1291.

Yalcinkaya, E., 2014. The October 20, 2006 Manyas (ML= 5.2) and October 24, 2006 Gemlik (ML= 5.2) earthquakes in the Marmara region (NW Turkey): ground motion characteristics. Annals of Geophysics, 57(6).

Yilmaz, Y., Tüysüz, O., Yiğitbaş, E., Genç, Ş.C. & Şengör, A.M.C., 1997. Geology and tectonic evolution of the Pontides. In: Robinson, A.G. (ed), Regional and Petroleum Geology of the Black Sea and Surrounding Region. AAPG Memoir 68, 183−226.

Yıldırım, E., Gülbağ, A., Horasan, G., & Doğan, E., 2011. Discrimination of quarry blasts and earthquakes in the vicinity of Istanbul using soft computing techniques. Computers & Geosciences, 37 (9), 1209-1217.

Key words:

Earthquake ground motions, Earthquake hazards, Numerical modeling, Seismicity and tectonics

Turkey, Istanbul earthquake, Marmara earthquake, 1999 Kocaeli, 1912 Murefte, geophysics, seismology

İstanbul depremi, yıkıcı deprem, jeofizik, sismoloji