Proceedings of the 7th International Conference on Earthquake Engineering and Seismology

Inhaltsverzeichnis

1. Frontmatter

2. Seismology and Strong Ground Motion

   1. Frontmatter

   2. Earthquake Epicenter Forecast Studies with Single-Station and Multi-Station Algorithms Using 30.10.2020 Sesame Earthquake Records

      H. Turan, K. Peker, B. Taskin

      Abstract

      After an earthquake, it is possible to apply numerical analysis procedures to the data captured by the earthquake monitoring network. With the use of advanced algorithms integrated into the seismographs at the recording stations, the epicenter of the earthquake and the intensity distribution around it can be estimated within seconds, even while the earthquake tremors are still happening. Forecasting the intensity distribution around the epicenter and potential structural damages and losses in residential areas enables us to anticipate all of these scenarios. To anticipate all the potential scenarios, it is necessary to analyze many complex data parameters. However, accurate calculation of the coordinates of the earthquake’s epicenter and its distance to surrounding settlements is the primary requirement. Although it is possible to make an approximate estimation of the earthquake epicenter when the earthquake waves reach a single station capable of recording three components, the margin of error decreases as the number of recording stations increases. The earthquake at Samos on October 30, 2020, with a moment magnitude of Mw 6.6, was used in this study. The algorithms designed for approximating the earthquake epicenter using data from single and multiple stations were utilized to compare the estimated epicenter with the actual coordinates, and the effectiveness of the approach was assessed. The study concluded with remarks regarding the utilization of the data gathered by the earthquake early warning systems.

   3. Performance of Existing Fling Step Predictive Models with Fling Database of Earthquakes in Türkiye

      Emrecan Adanır, Gülüm Tanırcan

      Abstract

      Fling step (permanent displacement) has been recognized as one of the most damaging earthquake effects on structures located in the vicinity of the fault rupture trace. Nevertheless, ground motion records containing fling step are not identified on account of the fact that standard data processing methods, such as filtering, mask the permanent displacements (PDs). Thus, this effect is commonly ignored during the seismic design of engineering structures. Fling step prediction models has certain limitations since models are mainly derived from the synthetic strong motions of earthquakes. In the last 3 decades, an increase in near fault strong motion recordings made it possible to calculate fling amplitudes with special baseline correction procedures. With this emerging database, it is possible to develop fling amplitude prediction equations or to adjust existing equations. Hence, in this study, performance of the existing fling step predictive models of Kamai et al. (Kamai et al. in Bull Seismol Soc Am 104:1914–1929, 2014), Burks and Baker (Burks and Baker in Soil Dynam Earthq Eng 80:119–126, 2016) and Türkiye adjusted Burks and Baker (Burks and Baker in Soil Dynam Earthq Eng 80:119–126, 2016) are evaluated using the fling inventory of moderate to large magnitude earthquakes in Türkiye including the 2023 Kahramanmaraş (Mw7.7) earthquake. Evaluations are performed only with total residual values of fling amplitudes. Burks and Baker (Burks and Baker in Soil Dynam Earthq Eng 80:119–126, 2016) model is found successsful for earthquakes with magnitudes between Mw 6 and Mw 6.9 whereas Kamai et al. (Kamai et al. in Bull Seismol Soc Am 104:1914–1929, 2014) well captured the PDs originated from large Mw7 + earthquakes. In general, estimates of Türkiye adjusted Burks and Baker (Burks and Baker in Soil Dynam Earthq Eng 80:119–126, 2016) model are much closer to observed averaged PDs in the dataset.

   4. Prediction of Peak Ground Velocity (PGV) and Cumulative Absolute Velocity (CAV) of Earthquakes Using Machine Learning Techniques

      F. Kuran, G. Tanırcan, E. Pashaei

      Abstract

      This study presents the prediction of cumulative absolute velocity (CAV) and peak ground velocity (PGV) using machine learning (ML) algorithms, which are relatively new compared to ground motion models with fixed functional forms. The performance of three ML algorithms, namely Linear Regression, Artificial Neural Network, and Gradient Boosting are evaluated and compared. The New Turkish Strong Motion Database (N-TSMD), containing over 23,000 recordings of 743 earthquakes that occurred in Turkiye between 1983 and 2020, is used to build ML models. In addition to N-TSMD, new recordings, including the recent M_w 7.7 and M_w 7.6 (Kahramanmaraş), M_w 6.6 (Gaziantep), and M_w 6.4 (Hatay) earthquakes, are added. In developing ML models, the moment magnitude (M_w), Joyner-Boore distance (R_JB), shear-wave velocity averaged in the top 30 m of soil (Vs₃₀), and style-of-faulting (SoF) are used as estimator parameters to characterize the source, path, site, and tectonic environment. Mean square error (MSE), root mean squared error (RMSE), and correlation coefficient (R) metrics are used to evaluate models. Results indicated that the Gradient Boosting algorithm demonstrates the best performance in predicting CAV and PGV according to all performance metrics. This is followed by Artificial Neural Network and Linear Regression, respectively. Residual analyses with predictions of the Gradient Boosting model indicated that there is almost no trend in the distribution of the total residuals of both PGV and CAV. The GB model’s prediction skill can be considered fair in all M_w, R_JB, and Vs₃₀ ranges.

   5. Nonparametric Ground Motion Models of Cumulative Absolute Velocity and Peak Ground Velocity for the Italian Dataset

      A. H. Mohammadi, S. M. S. Hussaini, D. Caicedo, S. Karimzadeh, P. B. Lourenço

      Abstract

      One crucial aspect that needs to be considered when studying the behaviour of historic masonry structures under earthquake ground excitation is the selection of appropriate intensity measures that can accurately capture the characteristics of seismic shaking. Among various features of ground motion, the cumulative absolute velocity (CAV) and peak ground velocity (PGV) have been identified as crucial factors affecting the response of masonry structures. These measures account for the amplitude, frequency content, and duration of ground motion, which can induce significant damage to masonry structures. These parameters can be assessed using either parametric linear or nonlinear regression techniques or non-parametric methods such as sophisticated machine-learning algorithms. However, parametric models, which are based on specific mathematical formulations, are prone to significant bias. In this study, we employ machine learning techniques that are more flexible and can capture the complex behaviour of ground motions, making them suitable for regions with high seismicity or geological complexity. Particularly, we employ the backpropagation neural network (BPNN) to develop models of CAV and PGV by considering an extensive dataset of Italian strong motions. A bias-variance trade-off was considered by tunning hyper parameters of the machine to ensure accurate predictions for future unseen data. Overall, the developed models demonstrate promising results compared to the prior models in the literature.

   6. Determination of the Preliminary Velocity Model Using Receiver Function Analysis Method with DEUNET Seismological Observation Network Data

      B. Kalkar, E. Gök

      Abstract

      The Western Anatolia and Aegean Region, renowned for its history of potent earthquakes, necessitates a specialized seismic velocity model for precise hazard assessments. This research leverages DEUNET Seismological Observation Network data to establish a region-specific velocity model. Fifteen teleseismic earthquakes (2019 to January 2023) were carefully selected, ranging from magnitudes 7.5–8.2, based on signal-to-noise ratios exceeding three. Epicentral distances spanned 30° to 90°. Station noise analysis involved choosing suitable earthquakes for spectral assessment. Steps included instrument response-based restitution and horizontal-to-radial/tangential component transformation (SAC software). Z components largely reflected direct P phases, while R components dominated Ps phases. ZRT to LQT rotation enabled Ps and P wave separation. The focus lay on P-SV-SH wave phases. L components conveyed strong P wave signals, while Q and T components featured converted S wave energy. Deconvolution effectively removed source and ray path contributions from L component P wave signals. The inversion performed subsequently determined the subsurface S-wave velocity structure based on depth. The station with the lowest signal-to-noise ratio, Çeşme, emerged from noise analysis. Receiver function analysis determined crust-mantle boundaries for DEUNET stations, indicating depths between 13 and 39 km.

   7. Source Parameters of the 2023 Kahramanmaraş Earthquake Aftershocks in the East Anatolian Fault Zone

      O. Batman, S. Akbaşak, B. S. Demirtaş, G. Tanırcan

      Abstract

      Abundant strong motion recordings from the 6 February 2023 Kahramanmaraş and Elbistan Earthquake aftershocks provide invaluable information on source characteristics of events on the East Anatolian Fault Zone (EAFZ). In this study, basic source parameters of 61 moderate-size aftershock have been deduced from more than 200 strong motion recordings. Three approaches are employed to estimate the source corner frequency (f_c) of events from S-wave spectra following the Brune’s source model fitting: visual inspection, numerical (RMSE), the spectral ratio of empirical Green’s functions (EGF). Power spectral ratio calculation of Andrews is utilized as the fourth approach. Source areas and stress drops are calculated following the circular static crack model and constant rupture velocity. Results indicate that the numerical and Andrews’ approaches provide the highest and lowest fc, respectively. All approaches are strongly sensitive to filter cut offs. Andrews’ approach underestimates the fc of small magnitude events (Mw < 4.9). Nevertheless, fc values converge to similar values for (Mw ≥ 5) for all approaches except the numerical fitting. The Mw calculated from seismic recordings are consistently lower than those reported by institutions. The stress drop results found in this study match with those of previous studies for the same region.

   8. Determination of Earthquake Focal Mechanism Using Artificial Intelligence

      Ilknur Kaftan, Yavuz Şenol, Berkay Kalkar, Elçin Gök

      Abstract

      Focal mechanism solution is important to analyze faulting type, fault geometry, and to evaluate aftershock patterns after a devastating earthquake occurs. Focal mechanisms were obtained by S/P amplitude ratios and P wave polarities, are fundamental data to investigate the geometry of fault zone, slip size, and the crustal stress field. Methods for solving the focal mechanism, have categories such as full waveforms, first motions of P waves, amplitudes of P or S waves, based on the waveform information used. In this study, we applied artificial intelligence to solve the focal mechanisms of small earthquakes in Izmir and its surroundings, where there is high microearthquake activity. The data set required for artificial intelligence applications was generated by using the records of earthquake stations located in western Turkey, especially in Izmir and its surroundings. Focal mechanism solution for Buca earthquake was determined by using Artificial neural networks are a branch of artificial intelligence. We compare the results obtained with artificial intelligence and traditional methods, and it is seen that artificial intelligence can be applicable for the focal mechanism solutions.

   9. Çanakkale Basin Strong Ground Motion Network

      H. A. Alçık, O. Eyisüren, A. Korkmaz, G. Tanırcan, S. Öztoprak

      Abstract

      Çanakkale region is under the influence of the western extension of the strike-slip North Anatolian Fault in the north and active faults with different mechanisms in the south and east. Çanakkale city center is located on a Quaternary alluvial deposit with a depth of up to 100 m. Shear wave velocities are quite low (<350 m/s) and shallow drillings could not reach to bedrock. Although there are no active faults that will cause major earthquakes in the immediate vicinity of the city center, it is possible that seismic motions from distant earthquakes may increase the earthquake demand of structures on the Çanakkale basin. Hence, understanding the behavior of seismic waves in the Çanakkale basin is an important step in the assessment of earthquake hazard. In this context, a 9-station accelerometric network has been deployed in the city. 4 out of 9 stations are located on the drilled boreholes which are conducted within the TÜBITAK-1001 project “Development of corresponding methods, coefficients and maps for including the bedrock depth and basin effect into the design response spectra obtained from Earthquake Design Codes and seismic site response analyses”. Within the scope of this study, recordings of the network have been used to calculate site characteristics underneath the stations. Possible basin effect is sought.

   10. Field Survey-Based Macroseismic Intensity Map of the 6th of February 2023 Kahramanmaraş Earthquakes

       Kubilay Albayrak, Aysegul Askan, Murat Altug Erberik, Mustafa Kerem Kockar, Mehmet Firat Aydin

       Abstract

       Earthquakes are natural events that cause damage to the built environment. These damages can be classified as structural and non-structural and can vary regionally and spatially. In the immediate aftermath of an earthquake, it is crucial to obtain macroseismic intensity distributions in terms of Modified Mercalli Intensity (MMI). Given that macroseismic intensity levels can be derived from empirical relationships, an field survey-based macroseismic intensity distribution map is necessary to assess the alignment between empirical and field survey-based maps. To accomplish this, a team of experts, including geotechnical and structural engineers, investigated the areas affected by the earthquake sequence that occurred on the 6th of February in Kahramanmaraş. The damages were assessed using a region-based approach, employing contour mapping. Subsequently, a field survey-based macroseismic intensity distribution map was created, making it available for use in resilience assessments and disaster management planning.

   11. Probabilistic Seismic Hazard Analysis of the Izmir Bay Area

       Kubilay Albayrak, Aysegul Askan, Ozlem Karagoz, Onur Tan

       Abstract

       Probabilistic Seismic Hazard Analysis (PSHA) is an analysis that quantifies the probability of exceedance or the rate of various ground motion levels at a specific site or an area by using all the possible earthquakes. Since earthquakes are very intense in specific regions, it is important to identify the seismicity level of these areas. The Izmir Bay area is believed to be one of the most critical areas in Türkiye in terms of high seismicity. So, PSHA of the Izmir Bay area was constructed by using 34 SPAC locations which are used to obtain the average shear wave velocity of the upper 30 m (VS30). The diameter of the area to identify the seismic sources is selected as 200 km. Since PSHA mainly results in Peak Ground Acceleration (PGA) values as a quantifier of ground motion intensity measurement, PGA values are based on 15 different periods. Moreover, Response Spectrum results are based on 50-, 100-, 475-, 975-, and 2475-year recurrence intervals for mean in terms of 16%, 50%, and 84% quantiles. Finally, the seismicity contour maps based on 475- and 2475-year recurrence intervals are prepared based on PGA values to visualize the resulting seismicity of the specified region. Since the Izmir Bay area is mainly comprised of low VS₃₀ values, the results of this study are believed to be important to take precautions for disaster resilience considerations.

3. Ground Motion Simulation and Site Characterization Working Group Studies in COSMOS Session

   1. Frontmatter

   2. VS Profiles for Turkey

      Ali Cemil Şen, Zehra Çağnan

      Abstract

      The Disaster and Emergency Management Authority of Turkey is currently operating 850 strong-motion stations throughout the country. At 112 of 850 station sites, variation of shear-wave velocity (V_S) with depth was measured by the multi-channel analysis of surface waves (MASW) technique and also by the refraction microtremor (REMI) technique at another 256 station sites. For these 368 station sites, measured V_S profiles are grouped into five different site categories according to their corresponding V_S30 values, time-averaged shear-wave velocity over the top 30 m of the site profile. Parametric models for the median V_S profiles are proposed. Median, Median ± standard deviation V_S-depth trends derived for Turkey are critically compared with those obtained for California, Japan, and Taiwan by Kamai et al. [1].

   3. Stochastic Ground Motion Simulation for the 9th July 1998 Faial Earthquake Using Source-Based and Site-Based Stochastic Methods

      S. M. S. Hussaini, S. Karimzadeh, P. B. Lourenço

      Abstract

      Ground motion simulation techniques can be valuable tools, particularly in regions with limited recorded accelerograms, as they can provide region-specific ground motion time series for response history analyses. These region-specific ground motions can be critical in predicting and mitigating the impact of potentially catastrophic future earthquakes, which could result in massive destruction, loss of life, and significant economic damage worldwide. These simulation techniques require accurate modelling and calibration of various input parameters in terms of source, path, and site effects. Verification of these parameters can be accomplished by comparing real time series from past events against simulated data. This study aims to simulate the recorded time series of the 9th July 1998 Faial earthquake (M_w = 6.2) using two stochastic simulation approaches: (1) a source-based stochastic finite-fault approach based on a dynamic corner frequency concept and (2) a site-based stochastic ground motion simulation approach. Alternative models are employed and tested, and the best model is determined utilizing goodness of fit score. Using the complementary error function, the discrepancies between the real and simulated record sets are evaluated in terms of various seismological parameters, including peak ground acceleration, peak ground velocity, the ratio of peak ground velocity to peak ground acceleration, Arias intensity, cumulative absolute velocity, acceleration spectrum intensity, modified acceleration spectrum intensity for the period range of 0.1–2.5 s, velocity spectrum intensity, Housner intensity, significant duration, bracketed duration, Fourier amplitude spectra within the frequency range of 0.25–10 Hz and pseudo response spectra within the period range of 0.1–4 s. The results of the simulations from both approaches suggest satisfactory fits between the real and simulated time series, which imply that the input parameters are verified for the event of interest.

   4. Evaluation of Multivariate Adaptive Regression Splines for Prediction of Kappa Factor in Western Türkiye

      T. O. Kurtulmus, F. Yerlikaya-Ozkurt, A. Askan

      Abstract

      The recent seismic activity on the west coast of Türkiye, including the Aegean Sea region, indicates that a closer focus is necessary on this region. Located in an active tectonic regime of north–south extension with multiple basins on soft soil deposits, the region has a high seismic hazard. Recently, as a combination of basin effects and building vulnerability, the October 30, 2020, Samos event (Mw = 7.0) caused localized significant damage and collapse in İzmir city center despite the 70 km distance from the earthquake source. In spite of this activity, studies on site characterization and site response modeling, including local velocity models and kappa estimates, are still limited in this region. Kappa values exhibit regional characteristics, which necessitates local kappa estimates from past earthquake data for use in region-specific applications. To make the prediction, we used three-component strong ground motion records from accelerometer stations with known V_S30 values in western Türkiye that are a part of the Disaster and Emergency Management Presidency’s Turkish National Strong Ground Motion Observation Network. Multiple linear regression (MLR) and multivariate adaptive regression splines (MARS) algorithms have been implemented to build the prediction model. Three factors, such as distance, magnitude, and site class, are included in the kappa evaluation process. The performance of the models in kappa evaluation is calculated based on well-known accuracy measures. The MARS model showed better performance compared to MLR over the selected sites concerning all performance measures. This finding may challenge the most commonly assumed linear models of kappa in the literature.

   5. Estimating Site Amplifications for an Enhanced Understanding of Ground Motion Effects in Northwestern Türkiye

      Gamze Muratoğlu, Ayşegül Askan

      Abstract

      Earthquake ground motions are influenced by various factors, including source characteristics, propagation path, and local site properties. Among these factors, local site conditions play a crucial role, leading to variations in ground shaking and consequential structural damage during large earthquakes. The presence of soil layers over solid bedrock can significantly amplify ground motions, highlighting the need to accurately quantify and comprehend site amplifications. This study focuses on investigating site amplifications in Northwestern Türkiye, employing a comprehensive approach. Initially, S-wave velocity models are identified at multiple sites throughout Turkey, followed by the categorization of stations based on NEHRP site classes. Subsequently, suitable input motions are selected for analysis, and 1D site response analyses are performed using the DEEPSOIL software. A compilation of shear wave velocity profiles with diverse resolutions and depths is utilized to capture the site-specific characteristics. Generic site amplification factors are derived for NEHRP site classes of A, B, C, and D in the Northwestern region of Türkiye. These factors are then compared with existing literature values, revealing notable disparities. The obtained amplification functions provide valuable insights into the amplification effects induced by local site conditions. By quantifying and understanding the site amplifications, this study contributes to a deeper comprehension of the factors influencing ground shaking during large earthquakes. The findings can be applied to improve seismic hazard assessments and inform the design of earthquake-resistant structures in Northwestern Türkiye. Furthermore, the developed amplification factors can be utilized in stochastic ground motion simulations, contributing to more accurate predictions and assessments of ground motions in the region.

   6. BB-SPEEDset: A Validated Dataset of Simulated Earthquake Ground Motions for Engineering Aims

      Chiara Smerzini, Roberto Paolucci, Manuela Vanini, Chiara Amendola

      Abstract

      Physics-based numerical simulations (PBS) are recognized as one of the most promising tools for providing earthquake ground motions in the source and site conditions which are poorly sampled by records, such as in the near-source region of severe shocks and in complex geological conditions. This undersampling issue is likely to persist into the future, in spite of the ever increasing number of seismic networks providing high-quality recordings, and it often prevents a proper characterization of seismic input for earthquake engineering applications. The key to strengthen and broaden the engineering utilization of PBS is the availability and dissemination of datasets of simulated broadband accelerograms, which have passed rigorous validation tests from both seismological and engineering perspectives. This paper aims at presenting a validated dataset of broadband near-source earthquake ground motions from 3D PBS—named BB-SPEEDset—obtained by the spectral element code SPEED (http://speed.mox.polimi.it/). BB-SPEEDset (v1.0) includes around 12,000 broadband accelerograms obtained from the simulation of several earthquakes in a wide range of magnitude (M_W = 5.5–7.4), faulting styles and geological contexts. The main validation tests on BB-SPEEDset (v1.0) are illustrated in order to verify that: (a) the features and statistical distributions of earthquake ground motions are consistent with those obtained from a near-source records dataset, within the same magnitude and distance range; and (b) BB-SPEEDset accelerograms, when used to provide spectrum-compatible input motions for inelastic time-history analyses of structures, do not yield any systematic bias in terms of engineering demand parameters with respect to those obtained using recorded base motions.

   7. Machine Learning for Damage Classification, Risk Mitigation and Post-earthquake Management

      F. Di Michele, O. Giannopoulou, E. Stagnini, D. Pera, B. Rubino, R. Aloisio, A. Askan, P. Marcati

      Abstract

      In recent years, we have witnessed a steady increase in the amount of data collected and made available to the scientific community. Simultaneously, Artificial Intelligence (AI) shows great potential in transforming data to knowledge offering increasingly more accurate tools to analyze, interpret and extract information from the data. In this context, data-driven approaches, in the fields of seismology, geophysics, and earthquake engineering show enormous promise. In this work a Machine Learning (ML) study is presented, based on a dataset containing around 3000 buildings damaged by the 2009 L’Aquila earthquake. This event was the first in a series of strong earthquakes that hit central Italy, resulting in many casualties, and having enormous economic and social impact. Each building in the dataset is described by 22 characteristics. Among them the damage level, divided, into six classes, from D0, corresponding to no damage, to D5, corresponding to heavy damage or collapse. We employ a Random Forest based algorithm to predict the level of damage accounting for different combinations of damage levels. As a case study we consider the following binary target variables-D0-D1-D2-D3 (no to medium damage) and D4-D5 (serious to heavy damage)-D0-D1 (no to light damage) and D2-D3-D4-D5 (moderate to heavy damage)-D0-D1-D02 (light to moderate damage) and D3-D4-D5 (medium to heavy damage).

   8. Effectiveness of Synthetic Ground-Motions for Seismic Assessment of Unreinforced Masonry Structures: Application to a Case Study

      Matteo Salvalaggio, Shaghayegh Karimzadeh, Vasco Bernardo, Paulo B. Lourenço

      Abstract

      The conservation of architectural heritage in seismic prone areas is still challenging, despite the recent developments in numerical and analytical tools for seismic assessment of unreinforced masonry structures. The increase in computational capacity has permitted the management of complex numerical models and the simulation of their dynamic behavior during a seismic event. In common practice, engineers often rely on using real ground-motion records compatible with site seismic spectra to assess heritage structures. However, the database of real ground motions is not large enough to meet all requests. To fill this gap, ground motion simulations offer an alternative method for accessing the full time-series of earthquake events. Yet, to guarantee their accuracy and efficiency in engineering practice, additional investigation is necessary to validate them. In this paper, the ground motion records of the L’Aquila 2009 earthquake are simulated through the stochastic finite-fault simulation approach. The obtained results are then compared against the recorded data. A typical unreinforced masonry building of Central Italy was selected as a case study, and time-history analyses were performed in the framework of the Applied Element Method. i.e., by discretizing masonry components through rigid blocks and nonlinear deformable interfaces between them. The outcomes of analyses are subsequently discussed in terms of displacement capacity, base shear and amplification of base acceleration encompassing both real and simulated record-based time-history analyses.