**2.2 Earthquake early warning and rapid response system (EEWRRS) in Istanbul**

Ground motion estimation without earthquake source parameters aims to provide near-real-time (within a few seconds of the event origin time) assessments of earthquake-induced ground motions and associated building damage in Istanbul. It is based on the fact that a dense accelerometric network (approximately 2–3 kilometers between adjacent instruments) is installed in the southern part of the city, where both population density and earthquake hazards are highest. No source parameters (magnitude, rupture length, or mechanism) are required to compute the earthquake-generated ground motion distribution. PGAs and spectral accelerations (SAs) at various periods are interpolated using two-dimensional splines to derive the ground motion values at the center of each 0.01×0.01° geo-cell (1120×830 m grid size). Spectral displacements are used to calculate the seismic demand in the center of each geo-cell. Coherency functions are being used to assess shaking maps created by interpolating discrete ground motion measurements. The ELER hazard module executes the following tasks: (1) using a regional seismotectonic database, determine the most likely site of the earthquake's origins, (2) the spatial distribution of chosen ground motion parameters at engineering bedrock is estimated using region-specific ground motion prediction models, (3) a bias correction of ground motion estimations using strong motion data where available, (4) estimation of the spatial distribution of chosen site-specific ground motion parameters by the use of a regional geology database and appropriate amplification models, (5) alternatively, ground motion parameters can be estimated site-specifically using Next Generation Attenuation (NGA) models.

Due to the complicated segmentation of the Marmara fault line and its proximity to the city, the Istanbul earthquake rapid response system (IEEWS) designed a robust and straightforward earthquake early warning algorithm based on the exceeding threshold values. To trigger, at least three stations must surpass the threshold level within a 5-second interval. Böse [35] stochastically simulated 280 earthquake scenarios in the Marmara Sea ranging from *Mw* 4.5 to *Mw* 7.5 and determined that the average early warning time spans between 8 and 15 seconds, depending on the event's source location.

The data transmission between the remote stations and the KOERI processing hub is accomplished via fiber optic cable with redundancy provided by a satellite system. The time required to transmit data from remote stations to the KOERI data center is a few milliseconds via fiber optic connections and less than a second via satellites. The continuous online data from these stations is processed at the hub, and subsequent alerts of emerging potentially disastrous ground motions provide real-time warning to vital infrastructures, allowing for the activation of shut-off mechanisms before the damaging waves reach the location.

At the moment, the IEEWS has not issued a public alert. Only the Istanbul Natural Gas Distribution Company (IGDAS) and the Marmaray Tube Tunnel actively utilize the EEW alert to activate automated shut-off systems in these facilities [36]. Both end users manage their own networks, with strong motion stations co-located at highpressure district gas regulators in the case of IGDAS and spaced throughout the tunnel in the case of Marmaray. Regarding the IGDAS, gas flow is automatically halted at the district regulator level in response to IEEWS alarms and the local site exceeding ground motion parameter threshold values. Local threshold values are determined on a case-by-case basis based on the local building stock.

The Marmaray Tube Tunnel, completed in 2013, is operated by the Turkish State Railways (TCDD). Train operations within a 1.4-kilometer-long tunnel beneath the Bosphorus, which connects the city's European and Asian sides, can be halted based on a combination of IEEWS alerts and a local threshold exceedance detected by its 26 tunnel sensors. Although IEEWS signals were sent to these vital facilities in recent seismic events such as the *Mw* 3.8 Yalova on 13 August 2015 and the *Mw* 4.2 Marmara Sea on 16 November 2015, *Mw* 4.5 Silivri on 24 September 2019, and the *Mw* 5.8 on 26 September 2019, no action was taken because the local threshold values were not surpassed.

#### *Urban Damage Assessment after the* Mw *5.8 Silivri Earthquake: The Case of Istanbul City DOI: http://dx.doi.org/10.5772/intechopen.109758*

Along with IEEWS, KOERI has implemented the regional EEW algorithms VS(SC3) and Probabilistic and Evolutionary Early Warning System (PRESTo) as part of the REAKT project. The Virtual Seismologist (VS) algorithm [37] is a networkbased Bayesian approach to EEW, and the SED group integrated an operational VS into the evolving Californian EEW prototype system [38] with support from the USGS ShakeAlert project. In 2013, VS was integrated as a set of self-contained modules into the open-source and widely distributed SeisComP3 (SC3) [39, 40] earthquake monitoring software, incorporating an EEW algorithm in the same system that many seismic networks use on a regular basis [41, 42]. This solution is referred to as "VS" (SC3). Further detail can be found in Clinton, Zollo [43]. These applications take advantage of KOERI's Marmara regional seismic network, which consists of 40 broadband and 30 strong motion seismic stations. PRESTo is currently monitoring 18 of the regional network's strong motion stations. Scenarios for various seismic events, including the 1999 *Mw* 7.4 Kocaeli Earthquake, show that a recurrence of this event would provide Istanbul with around 11 seconds of early warning. It is intended to expand the number of stations, including those utilized by PRESTo. Regional EEW algorithms VS (SC3) and PRESTo are not integrated with the existing IEEWS. Therefore, in their current configuration, the VS (SC3) and PRESTo systems would not provide warning in advance of strong motions associated with near-source seismic occurrences such as those in the Marmara Sea. On the other hand, the regional early warning system (EWS) is meant to be used in conjunction with the threshold-based IEEWS to provide warning for remote events that could be catastrophic for tall buildings and long-span bridge structures. On May 24, 2014, the *Mw* 6.9 Northern Aegean earthquake was intensely felt across Istanbul's high-rise buildings and was accurately characterized by VS (SC3) within 36 seconds of the origin time. The *Mw* 6.9 earthquake in Northern Aegean illustrated the importance of merging regional and threshold-based techniques. With the IEEWS and regional EWS algorithms, Istanbul has on-site structural monitoring efforts for historical buildings, high-rise buildings, and suspension bridges. These initiatives are not currently integrated into early warning systems.

#### **2.3 Structural health monitoring systems**

Instrumentation of major structures using strong ground motion recorders monitors the vibrations during an earthquake. Comparing these recordings with those made prior to the earthquake can reveal differences in structural response that have relevance for structural damage and a loss in seismic resistance. Therefore, these systems can be integrated into EEW systems to take critical measures for mitigating seismic risk. The structural health of significant structures in the Marmara region, particularly in Istanbul, has begun to be examined due to the region's high seismicity. Especially after the strong earthquakes in 1999, the public began to demand SHM systems, and the relevant government officials began to meet these demands.

SHM systems are currently monitoring five different long-span cable-supported bridges in the Marmara region. These are the 15 July Martyrs Bridge (the First Bosphorus Bridge), the Fatih Sultan Mehmet Suspension Bridge (the Second Bosphorus Bridge), the Yavuz Sultan Selim Bridge (the Third Bosphorus Bridge), the Osman Gazi Bridge (Izmit Bay Bridge), and the 1915 Çanakkale Bridge (Dardanelles). Regarding their SHM systems, design considerations and further information can be found in [44]. The following are some examples of currently monitored facilities in the Marmara region: Hagia Sophia, The Maltepe Mosque, Sapphire Tower, Kanyon Building, Isbank Tower, Polat Tower, and the Marmaray Tube Tunnel [45].

### **2.4 Next generation attenuation and ground motion prediction equations models (NGA GMPE)**

The initial stage is to mitigate the potential seismic damage by developing earthquake-resistant structures. Additionally, seismic hazard assessments of a region are critical for preventing earthquake-related destruction. Establishing an efficient seismic hazard and a comprehensive seismic risk assessment is crucial for the nation's sustainable growth.

Seismic hazard analyses demand the application of region-specific attenuation relations. Ground motion prediction equations (GMPEs) are used to determine ground motion parameters required for designing and evaluating vital structures [46]. Since a large number of GMPEs can be used to assess a region's seismic hazards and risks, selecting proper GMPEs can significantly impact design and safety evaluation [14]. Attenuation relationships allow for the accurate estimation of ground motion parameters (macroseismic intensity, PGA, PGV, and SAs) in terms of magnitude and source-to-site distance, and site features [47].

GMPEs for shallow crustal earthquakes were recently created in the Next Generation Attenuation (NGA). PEER created a number of GMPE models based on the NGA-West1 database (original NGA project). Following that, GMPEs were updated using the NGA-West2 database. The average horizontal component of shallow crustal earthquakes in active tectonic zones is estimated empirically using the PEER NGA-West2 database [48]. While it was designed for tectonically active locations, it is also applicable to other regions. NGA-West2 investigated the regionalization of ground motion properties, including an elastic attenuation, site response, and within-event standard deviation. GMPEs are frequently used to predict ground motions using deterministic and probabilistic seismic hazard analyses. The results of the seismic hazard analysis are used to conduct site-specific seismic analysis and design of facilities, to estimate social and financial losses, and to generate regional seismic hazard maps for use in building standards and financial estimation, among other purposes [47].

The NGA-West2 incorporates regionalization to account for regional changes in farsource distance attenuation and soil response. This is accomplished by using recordings from severe earthquakes in other active tectonic zones and California data [49]. As part of the NGA-West2 project, five GMPEs were developed: Abrahamson, Silva, and Kamai (ASK2014) [48], Boore, Stewart, Seyhan, and Atkinson (BSSA) [50], Campbell and Bozorgnia (CB) [51], Chiou and Youngs (CY2014) [52], and Idriss (I) [53].

The NGA-West2 project developed GMPEs for the purpose of calculating the medians and standard deviations of average horizontal component intensity measures (IMs) for shallow crustal earthquakes in active tectonic zones [50]. The equations were developed using data from a global database containing M 3.0-7.9 events. NGA-West2 GMPEs have a general limit of M8.5 for strike-slip faults, M8.0 for reverse faults, and M7.5 for normal faults, and a rupture distance Rrup or Joyner & Boore distance, Rjb of 0–300 km [53]. The ground motion IMs that comprise the GMPEs' dependent variables include the horizontal components PGA, PGV, and 5%-damped SAs [50]. These IMs were calculated using the RotD50 parameter [54], which represents the median horizontal single-component ground motion over all nonredundant azimuths. ASK2014 is applicable to magnitudes 3.0–8.5, distances from 0 to 300 km, and spectral periods ranging from 0 to 10 seconds. Regional differences were incorporated into the ASK2014 model based on distances. The ASK2014 model assumes that in active crustal zones, median ground motions at distances smaller than *Urban Damage Assessment after the* Mw *5.8 Silivri Earthquake: The Case of Istanbul City DOI: http://dx.doi.org/10.5772/intechopen.109758*

around 80 km are similar worldwide. The model confirmed that median stress dips are similar to those experienced during earthquakes in many active crustal locations such as California, Alaska, Taiwan, Japan, Turkey, Italy, Greece, New Zealand, and Northwest China [48]. BSSA2014 GMPE was developed to be generally applicable for earthquakes with magnitudes of M 3.0 to 8.5 (except for the lack of constraint for M > 7 normal slip events), at distances ranging from 0 to 400 km, at locations with Vs30 values ranging from 150 to 1500 m/s, and for spectral periods ranging from 0.01 to 10 sec [50]. The BSSA2014 model includes regional variation in source, path, and site effects. The CY2014 GMPE model is appropriate for predicting horizontal ground motion amplitudes associated with earthquakes in active tectonic zones that meet the following criteria [49]: (1) 3.5 ≤ M ≤ 8.5 for strike-slip earthquakes, (2) 3.5 ≤ M ≤ 8.0 for reverse and normal faulting earthquakes, (3) ZTOR ≤ 20 km, (4) 0 ≤ RRUP ≤ 300 km and 180 ≤ VS30 ≤ 1500 m/s.

Akcan et al. [55] evaluated the next generation attenuation (NGA) ground motion models that would be used in the region by comparing them to ground motion recordings from the Silivri earthquake in 2019. Among the NGA-West2 models, they found that ASK2014 and CY2014 provide the best fit for local PGA datasets. In conclusion, the high performance of these models proves that they can be used to estimate ground motions in metropolitan Istanbul.

### **3. Near-real-time strong motion network in Istanbul**

Istanbul has one of the most sophisticated strong motion networks in the world, integrated with Istanbul's Earthquake Rapid Response and Early Warning System (IERREWS), constructed in 2002 by Kandilli Observatory and the Earthquake Research Institute (KOERI) [56]. In this section, Istanbul's strong motion network is introduced. Later, the near-real-time hazard maps created by the officials using an earthquake parameter (PGA) recorded from the network are presented.

#### **3.1 Introduction to the network**

Apart from structures, a strong earthquake in Istanbul could damage infrastructures like the natural gas distribution network and subsequent damage from gas leaks from damaged pipelines (e.g., building damage, deaths, and injuries from fire and explosions), as well as production failures due to the natural gas cut-off for industrial subscribers. As a result, it is vital to determine the natural gas distribution system's reliability in the event of an earthquake to calculate the total economic losses and develop appropriate emergency response measures and plans. Following the loss and damage caused by the August 17th, and November 12th, 1999 earthquakes, the local administration, governmental agencies, non-governmental organizations, and scholars recognized the importance of developing elaborate earthquake response plans based on detailed earthquake analyses in Istanbul. This resulted in the Istanbul Metropolitan Municipality commissioning the research as "Updating the Possible Earthquake Losses in Istanbul" [57]. As a result of these investigations, IGDAS and the KOERI conducted a study to design a real-time risk mitigation system for the gas distribution network.

The strong motion network consists of 832 strong motion stations for use in automatic shut-off mechanisms when a ground motion parameter threshold is exceeded. The network's accelerometer stations were created in collaboration with the Turkish Scientific and Technological Research Center (TUBITAK). KOERI provides firmware for real-time ground motion parameter calculation. As shown in **Figure 1**, these accelerometers were installed at 832 selected district regulator stations to enable real-time shut-off of gas valves using the parametric data received online from the stations. The network was also built in the city's south, where population and seismic risk are highest. **Figure 2** depicts Istanbul's night-time population density.

The accelerometer stations were configured to compute the following real-time strong motion parameters to initiate the shut-offs: (1) Peak Ground Acceleration (PGA) and Peak Ground Velocity (PGV), (2) Arias intensity for ground motions in the x-direction (lax), (3) Cumulative Absolute Velocity integrated with a 5 cm/ s 2 lower threshold (CAV5) appears to adequately represent the ground motion's longer-period (lower-frequency) components, (4) As response spectral parameters, Pseudo-Spectral acceleration (PSA), Pseudo-Spectral velocity (PSV), and Spectral displacement (Sd), (5) As initially stated by [58], Spectral Intensity (SI) is defined as the average spectral velocity over the vibration period range [0.1, 2.5] s [59].

The KOERI earthquake rapid response system's real-time ground motion data and the strong motion network's accelerometric data are integrated to estimate earthquake hazards (real-time ground motion distribution) and risks (damage distribution). For rapid response purposes, a total of 220 strong motion stations (110 from KOERI and 110 from IGDAS) were first integrated in 2013 [60]. Data integration is carried out using the **E**arthquake **L**oss **E**stimation **R**outine (ELER)-earthquake shaking map algorithm. The data is merged at the KOERI and IGDAS Scada Centers via two servers. The data is exchanged via a fiber optic virtual private network (VPN).

The Istanbul Natural Gas Distribution Network Seismic Risk Reduction Project (IGRAS) earthquake risk mitigation system comprises three major modules: a scenario earthquake module, a real-time earthquake module, and a map management

**Figure 1.** *The location of the strong motion network in Istanbul.*

*Urban Damage Assessment after the* Mw *5.8 Silivri Earthquake: The Case of Istanbul City DOI: http://dx.doi.org/10.5772/intechopen.109758*

**Figure 2.** *Night-time population density distribution for metropolitan Istanbul.*

module. These primary modules are subdivided into the following: (a) an earthquake hazard map module; (b) an IGDAS pipeline infrastructure damage assessment module; and (c) a service box damage assessment module.

The earthquake hazard map module generates grid-based distributions of ground motion parameters such as macroseismic intensity, PGA, PGV, and spectral acceleration for various vibration periods based on the magnitude and location of the earthquake. The module makes use of GMPEs that are appropriate for Istanbul, as well as grid-based local site information such as Vs30 and a fault database. The recorded ground motion parameters from KOERI and IGDAS strong motion networks are used for GMPE bias correction, and integrated final earthquake hazard maps are delivered. The IGDAS pipeline infrastructure damage assessment module associates the earthquake hazard information generated by the earthquake hazard map module with the natural gas infrastructure elements and assesses the damage to the natural gas infrastructure elements analytically. A module for calculating natural gas building damage first connects the earthquake hazard and the building inventory and then calculates the building damages. Then, it correlates the findings of the building damage analysis with the locations of natural gas service boxes and determines the estimated number of damaged ones. Additionally, a GIS-based Web application for the IGRAS system is designed, which allows for online monitoring of scenario-based and real-time earthquake hazard, damage, and loss estimation findings.
