TY - JOUR A1 - Bindi, Dino A1 - Kotha, Sreeram Reddy A1 - Weatherill, Graeme A1 - Lanzano, Giovanni A1 - Luzi, Lucia A1 - Cotton, Fabrice T1 - The pan-European engineering strong motion (ESM) flatfile BT - consistency check via residual analysis JF - Bulletin of earthquake engineering : official publication of the European Association for Earthquake Engineering N2 - We present the results of a consistency check performed over the flatfile extracted from the engineering strong motion (ESM) database. The flatfile includes 23,014 recordings from 2179 earthquakes in the magnitude range from 3.5 to 7.8 that occurred since the 1970s in Europe and Middle East, as presented in the companion article by Lanzano et al. (Bull Earthq Eng, 2018a). The consistency check is developed by analyzing different residual distributions obtained from ad-hoc ground motion prediction equations for the absolute spectral acceleration (SA), displacement and Fourier amplitude spectra (FAS). Only recordings from earthquakes shallower than 40 km are considered in the analysis. The between-event, between-station and event-and-station corrected residuals are computed by applying a mixed-effect regression. We identified those earthquakes, stations, and recordings showing the largest deviations from the GMPE median predictions, and also evaluated the statistical uncertainty on the median model to get insights on the applicable magnitude–distance ranges and the usable period (or frequency) range. We observed that robust median predictions are obtained up to 8 s for SA and up to 20 Hz for FAS, although median predictions for Mw ≥ 7 show significantly larger uncertainties with ‘bumps’ starting above 5 s for SA and below 0.3 Hz for FAS. The between-station variance dominates over the other residual variances, and the dependence of the between-station residuals on logarithm of Vs30 is well-described by a piece-wise linear function with period-dependent slopes and hinge velocity around 580 m/s. Finally, we compared the between-event residuals obtained by considering two different sources of moment magnitude. The results show that, at long periods, the between-event terms from the two regressions have a weak correlation and the overall between-event variability is dissimilar, highlighting the importance of magnitude source in the regression results. KW - Ground motion prediction equation KW - Residual analysis KW - European strong motion data Y1 - 2018 U6 - https://doi.org/10.1007/s10518-018-0466-x SN - 1570-761X SN - 1573-1456 VL - 17 IS - 2 SP - 583 EP - 602 PB - Springer CY - Dordrecht ER - TY - JOUR A1 - Kotha, Sreeram Reddy A1 - Bindi, Dino A1 - Cotton, Fabrice T1 - Site-Corrected Magnitude- and Region-Dependent Correlations of Horizontal Peak Spectral Amplitudes JF - Earthquake spectra : the professional journal of the Earthquake Engineering Research Institute N2 - Empirical correlations of horizontal peak spectral amplitudes (PSA) are modeled using the total-residuals obtained in a ground motion prediction equation (GMPE) regression. Recent GMPEs moved toward partially non-ergodic region-and site-specific predictions, while the residual correlation models remained largely ergodic. Using mixed-effects regression, we decompose the total-residuals of a pan-European GMPE into between-event, between-site, and event-and-site corrected residuals to investigate the ergodicity in empirical PSA correlations. We first observed that the between-event correlations are magnitude-dependent, partially due to the differences in source spectra, and influence of stress-drop parameter on small and large events. Next, removing the between-site residuals from within-event residuals yields the event-and-site corrected residuals which are found to be region-dependent, possibly due to the regional differences in distance-decay of short period PSAs. Using our site-corrected magnitude- and region-dependent correlations, and the between-site residuals as empirical site-specific ground motion adjustments, we compute partially non-ergodic conditional mean spectra at four well-recorded sites in Europe and Middle Eastern regions. Y1 - 2017 U6 - https://doi.org/10.1193/091416EQS150M SN - 8755-2930 SN - 1944-8201 VL - 33 SP - 1415 EP - 1432 PB - Earthquake Engineering Research Institute CY - Oakland ER - TY - JOUR A1 - Weatherill, Graeme A1 - Kotha, Sreeram Reddy A1 - Cotton, Fabrice T1 - Re-thinking site amplification in regional seismic risk assessment JF - Earthquake spectra : the professional journal of the Earthquake Engineering Research Institute N2 - Probabilistic assessment of seismic hazard and risk over a geographical region presents the modeler with challenges in the characterization of the site amplification that are not present in site-specific assessment. Using site-to-site residuals from a ground motion model fit to observations from the Japanese KiK-net database, correlations between measured local amplifications and mappable proxies such as topographic slope and geology are explored. These are used subsequently to develop empirical models describing amplification as a direct function of slope, conditional upon geological period. These correlations also demonstrate the limitations of inferring 30-m shearwave velocity from slope and applying them directly into ground motion models. Instead, they illustrate the feasibility of deriving spectral acceleration amplification factors directly from sets of observed records, which are calibrated to parameters that can be mapped uniformly on a regional scale. The result is a geologically calibrated amplification model that can be incorporated into national and regional seismic hazard and risk assessment, ensuring that the corresponding total aleatory variability reflects the predictive capability of the mapped site proxy. KW - earthquake hazard analysis KW - ground motion KW - seismic risk KW - site effects KW - regional mapping Y1 - 2020 U6 - https://doi.org/10.1177/8755293019899956 SN - 8755-2930 SN - 1944-8201 VL - 36 IS - 1_SUPPL SP - 274 EP - 297 PB - Sage Publishing CY - Thousand Oaks, CA ER - TY - THES A1 - Kotha, Sreeram Reddy T1 - Quantification of uncertainties in seismic ground-motion prediction T1 - Quantifizierung von Unsicherheiten bei der seismischen Bodenbewegungsvorhersage N2 - The purpose of Probabilistic Seismic Hazard Assessment (PSHA) at a construction site is to provide the engineers with a probabilistic estimate of ground-motion level that could be equaled or exceeded at least once in the structure’s design lifetime. A certainty on the predicted ground-motion allows the engineers to confidently optimize structural design and mitigate the risk of extensive damage, or in worst case, a collapse. It is therefore in interest of engineering, insurance, disaster mitigation, and security of society at large, to reduce uncertainties in prediction of design ground-motion levels. In this study, I am concerned with quantifying and reducing the prediction uncertainty of regression-based Ground-Motion Prediction Equations (GMPEs). Essentially, GMPEs are regressed best-fit formulae relating event, path, and site parameters (predictor variables) to observed ground-motion values at the site (prediction variable). GMPEs are characterized by a parametric median (μ) and a non-parametric variance (σ) of prediction. μ captures the known ground-motion physics i.e., scaling with earthquake rupture properties (event), attenuation with distance from source (region/path), and amplification due to local soil conditions (site); while σ quantifies the natural variability of data that eludes μ. In a broad sense, the GMPE prediction uncertainty is cumulative of 1) uncertainty on estimated regression coefficients (uncertainty on μ,σ_μ), and 2) the inherent natural randomness of data (σ). The extent of μ parametrization, the quantity, and quality of ground-motion data used in a regression, govern the size of its prediction uncertainty: σ_μ and σ. In the first step, I present the impact of μ parametrization on the size of σ_μ and σ. Over-parametrization appears to increase the σ_μ, because of the large number of regression coefficients (in μ) to be estimated with insufficient data. Under-parametrization mitigates σ_μ, but the reduced explanatory strength of μ is reflected in inflated σ. For an optimally parametrized GMPE, a ~10% reduction in σ is attained by discarding the low-quality data from pan-European events with incorrect parametric values (of predictor variables). In case of regions with scarce ground-motion recordings, without under-parametrization, the only way to mitigate σ_μ is to substitute long-term earthquake data at a location with short-term samples of data across several locations – the Ergodic Assumption. However, the price of ergodic assumption is an increased σ, due to the region-to-region and site-to-site differences in ground-motion physics. σ of an ergodic GMPE developed from generic ergodic dataset is much larger than that of non-ergodic GMPEs developed from region- and site-specific non-ergodic subsets - which were too sparse to produce their specific GMPEs. Fortunately, with the dramatic increase in recorded ground-motion data at several sites across Europe and Middle-East, I could quantify the region- and site-specific differences in ground-motion scaling and upgrade the GMPEs with 1) substantially more accurate region- and site-specific μ for sites in Italy and Turkey, and 2) significantly smaller prediction variance σ. The benefit of such enhancements to GMPEs is quite evident in my comparison of PSHA estimates from ergodic versus region- and site-specific GMPEs; where the differences in predicted design ground-motion levels, at several sites in Europe and Middle-Eastern regions, are as large as ~50%. Resolving the ergodic assumption with mixed-effects regressions is feasible when the quantified region- and site-specific effects are physically meaningful, and the non-ergodic subsets (regions and sites) are defined a priori through expert knowledge. In absence of expert definitions, I demonstrate the potential of machine learning techniques in identifying efficient clusters of site-specific non-ergodic subsets, based on latent similarities in their ground-motion data. Clustered site-specific GMPEs bridge the gap between site-specific and fully ergodic GMPEs, with their partially non-ergodic μ and, σ ~15% smaller than the ergodic variance. The methodological refinements to GMPE development produced in this study are applicable to new ground-motion datasets, to further enhance certainty of ground-motion prediction and thereby, seismic hazard assessment. Advanced statistical tools show great potential in improving the predictive capabilities of GMPEs, but the fundamental requirement remains: large quantity of high-quality ground-motion data from several sites for an extended time-period. N2 - Der Zweck der probabilistischen seismischen Gefährdungsbeurteilung (PSHA) auf einer Baustelle besteht darin, den Ingenieuren eine probabilistische Schätzung des Bodenbewegungspegels zu liefern, die mindestens einmal in der Entwurfslebensdauer der Struktur erreicht oder überschritten werden könnte. Eine Gewissheit über die vorhergesagte Bodenbewegung erlaubt es den Ingenieuren, das strukturelle Design sicher zu optimieren und das Risiko von weitreichenden Schäden oder im schlimmsten Fall eines Zusammenbruchs zu minimieren. Es liegt daher im Interesse des Ingenieurwesens, der Versicherung, der Katastrophenvorsorge und der Sicherheit der Gesellschaft insgesamt, die Unsicherheiten bei der Vorhersage der Bodenbewegungsebenen des Entwurfs zu reduzieren. In dieser Studie, beschäftige ich mich mit der Quantifizierung und Reduzierung der Vorhersageunsicherheit von Regressions-basierten Bodenbewegungsvorhersage-Gleichungen (GMPEs). Im Wesentlichen sind GMPEs am besten angepasste Formeln, die Ereignis-, Pfad- und Standortparameter (Prädiktorvariablen) auf beobachtete Bodenbewegungswerte an der Stelle (Vorhersagevariable) beziehen. GMPEs sind gekennzeichnet durch einen parametrischen Median (μ) und eine nichtparametrische Varianz (σ) der Vorhersage. μ erfasst die bekannte Bodenbewegungs-Physik, d. h. Skalierung mit Erdbebenbrucheigenschaften (Ereignis), Dämpfung mit Abstand von der Quelle (Region/Pfad) und Verstärkung aufgrund lokaler Bodenbedingungen (Standort); während σ die natürliche Variabilität von Daten quantifiziert, die sich dem μ entziehen. In einem weiten Sinne ist die GMPE-Vorhersageunsicherheit kumulativ von 1) Unsicherheit bezüglich der geschätzten Regressionskoeffizienten (Unsicherheit auf μ; σ_μ) und 2) der inhärenten natürlichen Zufälligkeit von Daten (σ). Das Ausmaß der μ-Parametrisierung, die Menge und die Qualität der Bodenbewegungsdaten, die in einer Regression verwendet werden, bestimmen die Größe der Vorhersageunsicherheit: σ_μ und σ. Im ersten Schritt stelle ich den Einfluss der μ-Parametrisierung auf die Größe von σ_μ und σ vor. Überparametrisierung scheint die σ_μ zu erhöhen, da die große Anzahl von Regressionskoeffizienten (in μ) mit unzureichenden Daten geschätzt werden muss. Unterparametrisierung mindert σ_μ, aber die reduzierte Erklärungsstärke von μ spiegelt sich in aufgeblähtem σ wider. Für eine optimal parametrisierte GMPE wird eine ~ 10% ige Verringerung von σ erreicht, indem die Daten niedriger Qualität aus paneuropäischen Ereignissen mit inkorrekten Parameterwerten (von Prädiktorvariablen) verworfen werden. In Regionen mit wenigen Bodenbewegungsaufzeichnungen, ohne Unterparametrisierung, besteht die einzige Möglichkeit, σ_μ abzuschwächen, darin, langfristige Erdbebendaten an einem Ort durch kurzzeitige Datenproben an mehreren Orten zu ersetzen - die Ergodische Annahme. Der Preis der ergodischen Annahme ist jedoch aufgrund der Unterschiede in der Bodenbewegungsphysik von Region-zu-Region und von Ort-zu-Ort ein erhöhter σ. σ einer ergodischen GMPE, die aus einem generischen ergodischen Datensatz entwickelt wurde, ist viel größer als die von nicht-ergodischen GMPEs, die aus regions- und ortsspezifischen nicht-ergodischen Teilmengen entwickelt wurden - die zu dünn waren, um ihre spezifischen GMPEs zu erzeugen. Glücklicherweise konnte ich mit dem dramatischen Anstieg der erfassten Bodenbewegungsdaten an mehreren Standorten in Europa und im Nahen Osten die regions- und standortspezifischen Unterschiede bei der Bodenbewegungsskalierung quantifizieren und die GMPE mit 1) wesentlich genauerer Regionalität verbessern, und ortspezifische μ für Standorte in Italien und der Türkei, und 2) signifikant kleinere Vorhersage Varianz σ. Der Vorteil solcher Verbesserungen für GMPEs ist ziemlich offensichtlich in meinem Vergleich von PSHA-Schätzungen von ergodischen gegenüber regions- und ortsspezifischen GMPEs; wo die Unterschiede in den prognostizierten Bodenbewegungsebenen an verschiedenen Standorten in Europa und im Nahen Osten bis zu ~ 50% betragen. Die Lösung der ergodischen Annahme mit gemischten Regressionen ist machbar, wenn die quantifizierten bereichs- und ortsspezifischen Effekte physikalisch sinnvoll sind und die nicht-ergodischen Teilmengen (Regionen und Standorte) a priori durch Expertenwissen definiert werden. In Ermangelung von Expertendefinitionen demonstriere ich das Potential von maschinellen Lerntechniken bei der Identifizierung effizienter Cluster von ortsspezifischen nicht-ergodischen Untergruppen, basierend auf latenten Ähnlichkeiten in ihren Bodenbewegungsdaten. Geclusterte ortsspezifische GMPEs überbrücken die Lücke zwischen ortsspezifischen und vollständig ergodischen GMPEs mit ihrem teilweise nicht-ergodischen μ und ~ 15% kleiner als die ergodische Varianz. Die methodischen Verbesserungen der GMPE-Entwicklung, die in dieser Studie entwickelt wurden, sind auf neue Bodenbewegungsdatensätze anwendbar, um die Sicherheit der Bodenbewegungsvorhersage und damit die Bewertung der seismischen Gefährdung weiter zu verbessern. Fortgeschrittene statistische Werkzeuge zeigen ein großes Potenzial bei der Verbesserung der Vorhersagefähigkeiten von GMPEs, aber die grundlegende Anforderung bleibt: eine große Menge an hochwertigen Bodenbewegungsdaten von mehreren Standorten für einen längeren Zeitraum. KW - ground-motion variability KW - predictive modeling KW - mixed-effect analysis KW - Probabilistic Seismic Hazard and Risk Assessment KW - machine learning Y1 - 2018 U6 - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:kobv:517-opus4-415743 ER - TY - JOUR A1 - Kotha, Sreeram Reddy A1 - Weatherill, Graeme A1 - Bindi, Dino A1 - Cotton, Fabrice T1 - Near-source magnitude scaling of spectral accelerations BT - analysis and update of Kotha et al. (2020) model JF - Bulletin of earthquake engineering : official publication of the European Association for Earthquake Engineering N2 - Ground-motion models (GMMs) are often used to predict the random distribution of Spectral accelerations (SAs) at a site due to a nearby earthquake. In probabilistic seismic hazard and risk assessment, large earthquakes occurring close to a site are considered as critical scenarios. GMMs are expected to predict realistic SAs with low within-model uncertainty (sigma(mu)) for such rare scenarios. However, the datasets used to regress GMMs are usually deficient of data from critical scenarios. The (Kotha et al., A Regionally Adaptable Ground-Motion Model for Shallow Crustal Earthquakes in Europe Bulletin of Earthquake Engineering 18:4091-4125, 2020) GMM developed from the Engineering strong motion (ESM) dataset was found to predict decreasing short-period SAs with increasing M-W >= M-h = 6.2, and with large sigma(mu) at near-source distances <= 30km. In this study, we updated the parametrisation of the GMM based on analyses of ESM and the Near source strong motion (NESS) datasets. With M-h = 5.7, we could rectify the M-W scaling issue, while also reducing sigma(mu). at M-W >= M-h. We then evaluated the GMM against NESS data, and found that the SAs from a few large, thrust-faulting events in California, New Zealand, Japan, and Mexico are significantly higher than GMM median predictions. However, recordings from these events were mostly made on soft-soil geology, and contain anisotropic pulse-like effects. A more thorough non-ergodic treatment of NESS was not possible because most sites sampled unique events in very diverse tectonic environments. We provide an updated set of GMM coefficients,sigma(mu), and heteroscedastic variance models; while also cautioning against its application for M-W <= 4 in low-moderate seismicity regions without evaluating the homogeneity of M-W estimates between pan-European ESM and regional datasets. KW - Ground-motion model KW - Spectral accelerations KW - Magnitude scalin KW - Near-source saturation KW - Within-model uncertainty KW - Heteroscedastic KW - variability Y1 - 2022 U6 - https://doi.org/10.1007/s10518-021-01308-5 SN - 1570-761X SN - 1573-1456 VL - 20 IS - 3 SP - 1343 EP - 1370 PB - Springer CY - Dordrecht ER - TY - JOUR A1 - Bindi, Dino A1 - Picozzi, Matteo A1 - Spallarossa, Daniele A1 - Cotton, Fabrice A1 - Kotha, Sreeram Reddy T1 - Impact of Magnitude Selection on Aleatory Variability Associated with Ground-Motion Prediction Equations BT - Part II-Analysis of the Between-Event Distribution in Central Italy JF - Bulletin of the Seismological Society of America N2 - We derive a set of regional ground-motion prediction equations (GMPEs) in the Fourier amplitude spectra (FAS-GMPE) and in the spectral acceleration (SA-GMPE) domains for the purpose of interpreting the between-event residuals in terms of source parameter variability. We analyze a dataset of about 65,000 recordings generated by 1400 earthquakes (moment magnitude 2: 5 <= M-w <= 6: 5, hypocentral distance R-hypo <= 150 km) that occurred in central Italy between January 2008 and October 2017. In a companion article (Bindi, Spallarossa, et al., 2018), the nonparametric acceleration source spectra were interpreted in terms of omega-square models modified to account for deviations from a high-frequency flat plateau through a parameter named k(source). Here, the GMPEs are derived considering the moment (M-w), the local (M-L), and the energy (M-e) magnitude scales, and the between-event residuals are computed as random effects. We show that the between-event residuals for the FAS-GMPE implementing M-w are correlated with stress drop, with correlation coefficients increasing with increasing frequency up to about 10 Hz. Contrariwise, the correlation is weak for the FAS-GMPEs implementing M-L and M-e, in particular between 2 and 5 Hz, where most of the corner frequencies lie. At higher frequencies, all models show a strong correlation with k(source). The correlation with the source parameters reflects in a different behavior of the standard deviation tau of the between-event residuals with frequency. Although tau is smaller for the FAS-GMPE using M-w below 1.5 Hz, at higher frequencies, the model implementing either M-L or M-e shows smaller values, with a reduction of about 30% at 3 Hz (i.e., from 0.3 for M-w to 0.1 for M-L). We conclude that considering magnitude scales informative for the stress-drop variability allows to reduce the between-event variability with a significant impact on the hazard assessment, in particular for studies in which the ergodic assumption on site is removed. Y1 - 2019 U6 - https://doi.org/10.1785/0120180239 SN - 0037-1106 SN - 1943-3573 VL - 109 IS - 1 SP - 251 EP - 262 PB - Seismological Society of America CY - Albany ER - TY - JOUR A1 - Kotha, Sreeram Reddy A1 - Bindi, Dino A1 - Cotton, Fabrice T1 - From Ergodic to Region- and Site-Specific Probabilistic Seismic Hazard Assessment: Method Development and Application at European and Middle Eastern Sites JF - Earthquake spectra : the professional journal of the Earthquake Engineering Research Institute N2 - The increasing numbers of recordings at individual sites allows quantification of empirical linear site-response adjustment factors (delta S2S(s)) from the ground motion prediction equation (GMPE) residuals. The delta S2S(s) are then used to linearly scale the ergodic GMPE predictions to obtain site-specific ground motion predictions in a partially non-ergodic Probabilistic Seismic Hazard Assessment (PSHA). To address key statistical and conceptual issues in the current practice, we introduce a novel empirical region-and site-specific PSHA methodology wherein, (1) site-to-site variability (phi(S2S)) is first estimated as a random-variance in a mixed-effects GMPE regression, (2) delta S2S(s) at new sites with strong motion are estimated using the a priori phi(S2S), and (3) the GMPE site-specific single-site aleatory variability sigma(ss,s) is replaced with a generic site-corrected aleatory variability sigma(0). Comparison of region- and site-specific hazard curves from our method against the traditional ergodic estimates at 225 sites in Europe and Middle East shows an approximate 50% difference in predicted ground motions over a range of hazard levels-a strong motivation to increase seismological monitoring of critical facilities and enrich regional ground motion data sets. Y1 - 2017 U6 - https://doi.org/10.1193/081016EQS130M SN - 8755-2930 SN - 1944-8201 VL - 33 SP - 1433 EP - 1453 PB - Earthquake Engineering Research Institute CY - Oakland ER - TY - JOUR A1 - Kotha, Sreeram Reddy A1 - Cotton, Fabrice A1 - Bindi, Dino T1 - Empirical models of shear-wave radiation pattern derived from large datasets of ground-shaking observations JF - Scientific reports N2 - Shear-waves are the most energetic body-waves radiated from an earthquake, and are responsible for the destruction of engineered structures. In both short-term emergency response and long-term risk forecasting of disaster-resilient built environment, it is critical to predict spatially accurate distribution of shear-wave amplitudes. Although decades’ old theory proposes a deterministic, highly anisotropic, four-lobed shear-wave radiation pattern, from lack of convincing evidence, most empirical ground-shaking prediction models settled for an oversimplified stochastic radiation pattern that is isotropic on average. Today, using the large datasets of uniformly processed seismograms from several strike, normal, reverse, and oblique-slip earthquakes across the globe, compiled specifically for engineering applications, we could reveal, quantify, and calibrate the frequency-, distance-, and style-of-faulting dependent transition of shear-wave radiation between a stochastic-isotropic and a deterministic-anisotropic phenomenon. Consequent recalibration of empirical ground-shaking models dramatically improved their predictions: with isodistant anisotropic variations of ±40%, and 8% reduction in uncertainty. The outcomes presented here can potentially trigger a reappraisal of several practical issues in engineering seismology, particularly in seismic ground-shaking studies and seismic hazard and risk assessment. Y1 - 2019 U6 - https://doi.org/10.1038/s41598-018-37524-4 SN - 2045-2322 VL - 9 PB - Nature Publ. Group CY - London ER - TY - JOUR A1 - Kotha, Sreeram Reddy A1 - Bazzurro, Paolo A1 - Pagani, Marco T1 - Effects of epistemic uncertainty in seismic hazard estimates on building portfolio losses JF - Earthquake spectra : the professional journal of the Earthquake Engineering Research Institute N2 - In catastrophe risk modeling, a defensible estimation of impact severity and its likelihood of occurrence to a portfolio of assets can only be made through a rigorous treatment of uncertainty and the consideration of multiple alternative models. This approach, however, requires repeating lengthy calculations multiple times. To limit the demand on computational time and resources, a frequent practice in the industry is to estimate the distribution of earthquake-induced portfolio losses using a simulated catalog of events from a single representative mean ground motion hazard model for the region. This simplified approach is faster but may provide biased estimates of the likelihood of occurrence of the large and infrequent losses that drive many risk mitigation decisions. Investigation through case studies of different portfolios of assets located in the San Francisco Bay Region shows the potential for both a bias in the mean loss estimates and an underestimation of their central 70% interpercentile. We propose a simplified and computationally practical approach that reduces the bias in the mean portfolio loss estimates. This approach does not improve the estimate of the interpercentile range, however, a quantity of no direct practical use. Y1 - 2018 U6 - https://doi.org/10.1193/020515EQS020M SN - 8755-2930 SN - 1944-8201 VL - 34 IS - 1 SP - 217 EP - 236 PB - Earthquake Engineering Research Institute CY - Oakland ER - TY - JOUR A1 - Kotha, Sreeram Reddy A1 - Cotton, Fabrice A1 - Bindi, Dino T1 - A new approach to site classification BT - Mixed-effects Ground Motion Prediction Equation with spectral clustering of site amplification functions JF - Soil Dynamics and Earthquake Engineering N2 - With increasing amount of strong motion data, Ground Motion Prediction Equation (GMPE) developers are able to quantify empirical site amplification functions (delta S2S(s)) from GMPE residuals, for use in site-specific Probabilistic Seismic Hazard Assessment. In this study, we first derive a GMPE for 5% damped Pseudo Spectral Acceleration (g) of Active Shallow Crustal earthquakes in Japan with 3.4 <= M-w <= 7.3 and 0 <= R-JB <= 600km. Using k-mean spectral clustering technique, we then classify our estimated delta S2S(s)(T = 0.01 - 2s) of 588 wellcharacterized sites, into 8 site clusters with distinct mean site amplification functions, and within-cluster site-tosite variability similar to 50% smaller than the overall dataset variability (phi(S2S)). Following an evaluation of existing schemes, we propose a revised data-driven site classification characterized by kernel density distributions of V-s30, V-s10, H-800, and predominant period (T-G) of the site clusters. KW - Mixed-effects regression KW - Ground Motion Prediction Equation KW - Site classification KW - Spectral clustering analysis KW - Empirical site amplification functions Y1 - 2018 U6 - https://doi.org/10.1016/j.soildyn.2018.01.051 SN - 0267-7261 SN - 1879-341X VL - 110 SP - 318 EP - 329 PB - Elsevier CY - Oxford ER -