Empirical ground-motion models for point- and extended-source crustal earthquake scenarios in Europe and the Middle East

This article presents the latest generation of ground-motion models for the prediction of elastic response (pseudo-) spectral accelerations, as well as peak ground acceleration and velocity, derived using pan-European databases. The models present a number of novelties with respect to previous generations of models (Ambraseys et al. in Earthq Eng Struct Dyn 25:371–400, 1996, Bull Earthq Eng 3:1–53, 2005; Bommer et al. in Bull Earthq Eng 1:171–203, 2003; Akkar and Bommer in Seismol Res Lett 81:195–206, 2010), namely: inclusion of a nonlinear site amplification function that is a function of $\text{ V }_\mathrm{S30}$ and reference peak ground acceleration on rock; extension of the magnitude range of applicability of the model down to $\text{ M }_\mathrm{w}$ 4; extension of the distance range of applicability out to 200 km; extension to shorter and longer periods (down to 0.01 s and up to 4 s); and consistent models for both point-source (epicentral, $\text{ R }_\mathrm{epi}$ , and hypocentral distance, $\text{ R }_\mathrm{hyp}$ ) and finite-fault (distance to the surface projection of the rupture, $\text{ R }_\mathrm{JB}$ ) distance metrics. In addition, data from more than 1.5 times as many earthquakes, compared to previous pan-European models, have been used, leading to regressions based on approximately twice as many records in total. The metadata of these records have been carefully compiled and reappraised in recent European projects. These improvements lead to more robust ground-motion prediction equations than have previously been published for shallow (focal depths less than 30 km) crustal earthquakes in Europe and the Middle East. We conclude with suggestions for the application of the equations to seismic hazard assessments in Europe and the Middle East within a logic-tree framework to capture epistemic uncertainty.     

Towards fully data driven ground-motion prediction models for Europe

We have used the Artificial Neural Network method (ANN) for the derivation of physically sound, easy-to-handle, predictive ground-motion models from a subset of the Reference database for Seismic ground-motion prediction in Europe (RESORCE). Only shallow earthquakes (depth smaller than 25 km) and recordings corresponding to stations with measured $V_{s30}$ properties have been selected. Five input parameters were selected: the moment magnitude $M_{W}$ , the Joyner–Boore distance $R_{JB}$ , the focal mechanism, the hypocentral depth, and the site proxy $V_{S30}$ . A feed-forward ANN type is used, with one 5-neuron hidden layer, and an output layer grouping all the considered ground motion parameters, i.e., peak ground acceleration (PGA), peak ground velocity (PGV) and 5 %-damped pseudo-spectral acceleration (PSA) at 62 periods from 0.01 to 4 s. A procedure similar to the random-effects approach was developed to provide between and within event standard deviations. The total standard deviation ( $\sigma $ ) varies between 0.298 and 0.378 (log $_{10}$ unit) depending on the period, with between-event and within-event variabilities in the range 0.149–0.190 and 0.258–0.327, respectively. Those values prove comparable to those of conventional GMPEs. Despite the absence of any a priori assumption on the functional dependence, our results exhibit a number of physically sound features: magnitude scaling of the distance dependency, near-fault saturation distance increasing with magnitude, amplification on soft soils and even indications for nonlinear effects in softer soils.     

Comparisons among the five ground-motion models developed using RESORCE for the prediction of response spectral accelerations due to earthquakes in Europe and the Middle East

This article presents comparisons among the five ground-motion models described in other articles within this special issue, in terms of data selection criteria, characteristics of the models and predicted peak ground and response spectral accelerations. Comparisons are also made with predictions from the Next Generation Attenuation (NGA) models to which the models presented here have similarities (e.g. a common master database has been used) but also differences (e.g. some models in this issue are nonparametric). As a result of the differing data selection criteria and derivation techniques the predicted median ground motions show considerable differences (up to a factor of two for certain scenarios), particularly for magnitudes and distances close to or beyond the range of the available observations. The predicted influence of style-of-faulting shows much variation among models whereas site amplification factors are more similar, with peak amplification at around 1s. These differences are greater than those among predictions from the NGA models. The models for aleatory variability (sigma), however, are similar and suggest that ground-motion variability from this region is slightly higher than that predicted by the NGA models, based primarily on data from California and Taiwan.     

Current empirical ground-motion prediction equations for Europe and their application to Eurocode 8

The first ground-motion prediction equation derived from European and Middle Eastern strong-motion data was published more than 30 years ago; since then strong-motion networks and the resulting databank of accelerograms in the region have expanded significantly. Many equations for the prediction of peak ground-motion parameters and response spectral ordinates have been published in recent years both for the entire Euro-Mediterranean and Middle Eastern region as well as for individual countries within this region. Comparisons among empirical ground-motion models for these parameters, developed using large regional datasets, do not support the hypothesis of there being significant differences in earthquake ground-motions from one area of crustal seismicity to another. However, there are certain regions within Europe—affected by different tectonic regimes—for which the existing pan-European equations may not be applicable. The most recent European equations make it possible to now implement overdue modifications to the presentation of seismic design actions in Eurocode 8 that allow an improved approximation to the target uniform hazard spectrum (UHS). Using these recent equations, this study outlines a new approach via which an approximation to the UHS may be constructed using hazard maps calculated for peak ground velocity and the corner period T _D in addition to the maps for peak ground acceleration that underpin the current stipulations of Eurocode 8.     

Partially non-ergodic region specific GMPE for Europe and Middle-East

The ergodic assumption considers the time sampling of ground shaking generated in a given region by successive earthquakes as equivalent to a spatial sampling of observed ground motion across different regions. In such cases the estimated aleatory variability in source, propagation, and site seismic processes in ground motion prediction equations (GMPEs) is usually larger than with a non-ergodic approach. With the recently published datasets such as RESORCE for Europe and Middle-East regions, and exploiting algorithms like the non-linear mixed effects regression it became possible to introduce statistically well-constrained regional adjustments to a GMPE, thus ‘partially’ mitigating the impact of the assumption on regional ergodicity. In this study, we quantify the regional differences in the apparent attenuation of high frequency ground motion with distance and in linear site amplification with V_s30, between Italy, Turkey, and rest of the Europe-Middle-East region. With respect to a GMPE without regional adjustments, we obtain up to 10 % reduction in the aleatory variability σ, primarily contributed by a 20 % reduction in the between-station variability. The reduced aleatory variability is translated into an epistemic uncertainty, i.e. a standard error on the regional adjustments which can be accounted for in the hazard assessment through logic-tree branches properly weighted. Furthermore, the between-event variability is reduced by up to 30 % by disregarding in regression the events with empirically estimated moment magnitude. Therefore, we conclude that a further refinement of the aleatory variability could be achieved by choosing a combination of proxies for the site response, and through the homogenization of the magnitude scales across regions.     

Compatible ground-motion prediction equations for damping scaling factors and vertical-to-horizontal spectral amplitude ratios for the broader Europe region

In a companion article Akkar et al. (Bull Earthq Eng, doi:10.1007/s10518-013-9461-4, 2013a; Bull Earthq Eng, doi:10.1007/s10518-013-9508-6, 2013b) present a new ground-motion prediction equation (GMPE) for estimating 5 %-damped horizontal pseudo-acceleration spectral (PSA) ordinates for shallow active crustal regions in Europe and the Middle East. This study provides a supplementary viscous damping model to modify 5 %-damped horizontal spectral ordinates of Akkar et al. (Bull Earthq Eng, doi:10.1007/s10518-013-9461-4 2013a; Bull Earthq Eng, doi:10.1007/s10518-013-9508-6, 2013b) for damping ratios ranging from 1 to 50 %. The paper also presents another damping model for scaling 5 %-damped vertical spectral ordinates that can be estimated from the vertical-to-horizontal (V/H) spectral ratio GMPE that is also developed within the context of this study. For consistency in engineering applications, the horizontal and vertical damping models cover the same damping ratios as noted above. The article concludes by introducing period-dependent correlation coefficients to compute horizontal and vertical conditional mean spectra (Baker in J Struct Eng 137:322–331, 2011). The applicability range of the presented models is the same as of the horizontal GMPE proposed by Akkar et al. (Bull Earthq Eng, doi:10.1007/s10518-013-9461-4 2013a; Bull Earthq Eng, doi:10.1007/s10518-013-9508-6, 2013b): as for spectral periods $0.01 \hbox { s}\le \,\hbox {T}\le \,4\hbox { s}$ as well as PGA and PGV for V/H model; and in terms of seismological estimator parameters $4\le \hbox {M}_\mathrm{w} \le 8, \hbox { R} \le 200 \hbox { km}, 150\hbox { m/s}\le \hbox { V}_\mathrm{S30}\le $ 1,200 m/s, for reverse, normal and strike-slip faults. The source-to-site distance measures that can be used in the computations are epicentral $(\hbox {R}_\mathrm{epi})$ , hypocentral $(\hbox {R}_\mathrm{hyp})$ and Joyner–Boore $(\hbox {R}_\mathrm{JB})$ distances. The implementation of the proposed GMPEs will facilitate site-specific adjustments of the spectral amplitudes predicted from probabilistic seismic hazard assessment in Europe and the Middle East region. They can also help expressing the site-specific design ground motion in several formats. The consistency of the proposed models together with the Akkar et al. (Bull Earthq Eng, doi:10.1007/s10518-013-9461-4 2013a; Bull Earthq Eng, doi:10.1007/s10518-013-9508-6, 2013b) GMPE may be advantageous for future modifications in the ground-motion definition in Eurocode 8 (CEN in Eurocode 8, Design of structures for earthquake resistance—part 1: general rules, seismic actions and rules for buildings. European Standard NF EN 1998-1, Brussels, 2004).     

Reference database for seismic ground-motion in Europe (RESORCE)

This paper presents the overall procedure followed in order to assemble the most recent pan-European strong-motion databank: Reference Database for Seismic Ground-Motion in Europe (RESORCE). RESORCE is one of the products of the SeIsmic Ground Motion Assessment (SIGMA; projet-sigma.com) project. RESORCE is intended to be a single integrated accelerometric databank for Europe and surrounding areas for use in the development and testing of ground-motion models and for other engineering seismology and earthquake engineering applications. RESORCE aims to contribute to the improvement of earthquake risk studies in Europe and surrounding areas. RESORCE principally updates and extends the previous pan-European strong-motion databank (Ambraseys et al. in Bollettino di Geofisica Teorica ed Applicata 45:113–129, 2004a) with recently compiled Greek, Italian, Swiss and Turkish accelerometric archives. The updates also include earthquake-specific studies published in recent years. The current content of RESORCE includes 5,882 multi-component and uniformly processed accelerograms from 1,814 events and 1,540 strong-motion stations. The moment magnitude range covered by RESORCE is $2.8 \le \hbox {M}_{\mathrm{w}} \le 7.8$ . The source-to-site distance interval extends to 587 km and distance information is given by the common point- and extended-source distance measures. The paper presents the current features of RESORCE through simple statistics that also quantify the differences in metadata and strong-motion processing with respect to the previous version of the pan-European strong-motion databank.     

Prediction of elastic displacement response spectra in Europe and the Middle East

Empirical equations are presented for the prediction of displacement response ordinates for damping ratios of 2, 5, 10, 20 and 30% of critical and for response periods up to 4s, using 532 accelerograms from the strong‐motion databank from Europe and the Middle East. The records were all re‐processed and only employed for regressions at periods within the usable range, defined as a fraction of the filter cut‐off and depending on the instrument type (digital or analogue), earthquake magnitude and site class. The equations can be applied to predict the geometric mean displacement and pseudo‐acceleration spectra for earthquakes with moment magnitudes ( M ) between 5 and 7.6, and for distances up to 100km. The equations also include style‐of‐faulting and site class as explanatory variables. The predictions obtained from these new equations suggest that earlier European equations for spectral displacements underestimate the ordinates at longer periods as a result of severe filtering and the use of the spectral ordinates at periods too close to the filter cut‐off. The results also confirm that the period defining the start of the constant displacement plateau in the Eurocode 8 (EC8) spectrum is excessively short at 2s. The results not only show that the scaling factor defined in EC8 for estimating the spectral ordinates at damping ratios different from 5% of critical are a good general approximation, but also that this scaling varies with magnitude and distance (reflecting the influence of duration) and also displays a mild dependence on response period. Copyright © 2007 John Wiley & Sons, Ltd.     

A regionally adaptable ground-motion model for fourier amplitude spectra of shallow crustal earthquakes in Europe

Bulletin of Earthquake Engineering - Typical seismic ground-motion models predict the response spectral ordinates (GMM-SA), which are the damped responses of a suite of single-degree-of-freedom...     

Toward a ground-motion logic tree for probabilistic seismic hazard assessment in Europe

The Seismic Hazard Harmonization in Europe (SHARE) project, which began in June 2009, aims at establishing new standards for probabilistic seismic hazard assessment in the Euro-Mediterranean region. In this context, a logic tree for ground-motion prediction in Europe has been constructed. Ground-motion prediction equations (GMPEs) and weights have been determined so that the logic tree captures epistemic uncertainty in ground-motion prediction for six different tectonic regimes in Europe. Here we present the strategy that we adopted to build such a logic tree. This strategy has the particularity of combining two complementary and independent approaches: expert judgment and data testing. A set of six experts was asked to weight pre-selected GMPEs while the ability of these GMPEs to predict available data was evaluated with the method of Scherbaum et al. (Bull Seismol Soc Am 99:3234?C3247, 2009). Results of both approaches were taken into account to commonly select the smallest set of GMPEs to capture the uncertainty in ground-motion prediction in Europe. For stable continental regions, two models, both from eastern North America, have been selected for shields, and three GMPEs from active shallow crustal regions have been added for continental crust. For subduction zones, four models, all non-European, have been chosen. Finally, for active shallow crustal regions, we selected four models, each of them from a different host region but only two of them were kept for long periods. In most cases, a common agreement has been also reached for the weights. In case of divergence, a sensitivity analysis of the weights on the seismic hazard has been conducted, showing that once the GMPEs have been selected, the associated set of weights has a smaller influence on the hazard.     

Extending ground-motion prediction equations for spectral accelerations to higher response frequencies

Ground-motion prediction equations (GMPEs) for spectral accelerations have traditionally focused on the range of response periods most closely associated with the dynamic characteristics of buildings. Providing predictions only in this period range (from 0.1 to 2 or 3 s) has also accommodated the assumed limitations on the usable period range resulting from the processing of accelerograms. There are, however, engineering applications for which estimates of spectral ordinates are required at shorter response periods. Recent work has demonstrated that high-frequency spectral ordinates are relatively insensitive to record processing, contrary to previous assumptions. In the light of this finding, additional regressions are performed to extend a recent pan-European GMPE to higher response frequencies. This model and others that also include coefficients for spectral ordinates at several high response frequencies are used to explore options for interpolating coefficients for equations that do not provide good coverage in this range. The challenges and uncertainties associated with such interpolations are discussed. The paper concludes that a set of standard response frequencies could be usefully established for future GMPEs.     

Update of likelihood-based ground-motion model selection for seismic hazard analysis in western central Europe

Scherbaum et al. [(2004) Bull Seismolo Soc Am 94(6): 2164–2185] proposed a likelihood-based approach to select and rank ground-motion models for seismic hazard analysis in regions of low-seismicity. The results of their analysis were first used within the PEGASOS project [Abrahamson et al. (2002), In Proceedings of the 12 ECEE, London, 2002, Paper no. 633] so far the only application of a probabilistic seismic hazard analysis (PSHA) in Europe which was based on a SSHAC Level 4 procedure [(Budnitz et al. 1997, Recommendations for PSHA: guidance on uncertainty and use of experts. No. NUREG/CR-6372-V1). The outcome of this project have generated considerable discussion (Klügel 2005, Eng Geol 78:285–307, 2005b) Eng Geol 78: 285–307, (2005c) Eng Geol 82: 79–85 Musson et al. (2005) Eng Geol 82(1): 43–55]; Budnitz et al. (2005), Eng Geol 78(3–4): 285–307], a central part of which is related to the issue of ground-motion model selection and ranking. Since at the time of the study by Scherbaum et al. [(2004.) Bull Seismolo Soc Am 94(6): 2164–2185], only records from one earthquake were available for the study area, here we test the stability of their results using more recent data. Increasing the data set from 12 records of one earthquake in Scherbaum et al. [(2004) Bull Seismolo Soc Am 94(6): 2164–2185] to 61 records of 5 earthquakes, which have mainly occurred since the publication of the original study, does not change the set of the three top-ranked ground-motion models [Abrahamson and Silva (1997) Seismolo Res Latt 68(1): 94–127; Lussou et al. (2001) J Earthquake Eng 5(1):13–33; Berge-Thierry et al. (2003) Bull Seismolog Soc Am 95(2): 377–389. Only for the lower-ranked models do we obtain modifications in the ranking order. Furthermore, the records from the Waldkirch earthquake (Dec, 5th, 2004, M _w = 4.9) enabled us to develop a new stochastic model parameter set for the application of Campbell’s [(2003) Bull Seismolo Soc Am 93(3): 1012–1033] hybrid empirical model to SW Germany and neighbouring regions.     

A novel approach for classification of earthquake ground-motion records

This paper presents a new clustering procedure based on K-means and self-organizing map (SOM) network algorithms for classification of earthquake ground-motion records. Six scalar indicators are used in data analysis for describing the frequency content features of earthquake ground motions, named as the average spectral period (T _avg ), the mean period (T _m ), the smoothed spectral predominant period (T ₀), the characteristic period (T _4.3), the predominant period based on velocity spectrum (T _gSv ), and the shape factor (Ω). Different clustering validity indexes were applied to determine the best estimates of the number of clusters on real and synthetic data. Results showed the high performance of proposed procedure to reveal salient features of complex seismic data. The comparison between the results of clustering analyses recommend the smoothed spectral predominant period as an effective indicator to describe ground-motion classes. The results also showed that K-means algorithm has better performance than SOM algorithm in identification and classification procedure of ground-motion records.     

An evaluation of the applicability of the NGA models to ground-motion prediction in the Euro-Mediterranean region

The first phase of the Next Generation Attenuation (NGA) project has now finished, resulting in the publication of five new sets of empirical ground-motion models for PGA, PGV and response spectral ordinates. These models mark a significant advancement in the state-of-the-art in empirical ground-motion modelling and include many effects that are not accounted for in existing European equations. Under the assumption that the Euro-Mediterranean database from which the European relationships are derived is unlikely to drastically change in the near future, a prudent question to ask is: can the NGA models be applied in Europe? In order to answer this question, the NGA model of Boore and Atkinson (PEER Report 2007/01, Pacific Earthquake Engineering Research Center, Berkeley, CA, 234 pp., 2007), which is shown to be representative of the NGA models as a suite, is compared with the dataset used for the development of the most recent European empirical ground-motion models for response spectral ordinates and peak ground velocity. The comparisons are made using analyses of model residuals and the likelihood approach of Scherbaum et al. (Bull Seism Soc Am 94(6):2164–2185, 2004). The analyses indicate that for most engineering applications, and particularly for displacement-based approaches to seismic design, the NGA models may confidently be applied within Europe. Furthermore, it is recommended that they be used in conjunction with existing European models to provide constraint on finite-fault effects and non-linear site response within logic-tree frameworks. The findings also point to the potential benefits of merging the NGA and European datasets.     

Regional adjustment factors for three NGA-West2 ground-motion prediction equations to be applicable in northern Iran

Ground-motion prediction equations (GMPEs) are essential tools in seismic hazard studies to estimate ground motions generated by potential seismic sources. Global GMPEs which are based on well-compiled global strong-motion databanks, have certain advantages over local GMPEs, including more sophisticated parameters in terms of distance, faulting style, and site classification but cannot guarantee the local/region-specific propagation characteristics of shear wave (e.g., geometric spreading behavior, quality factor) for different seismic regions at larger distances (beyond about 80 km). Here, strong-motion records of northern Iran have been used to estimate the propagation characteristics of shear wave and determine the region-specific adjustment parameters for three of the NGA-West2 GMPEs to be applicable in northern Iran. The dataset consists of 260 three-component records from 28 earthquakes, recorded at 139 stations, with moment magnitudes between 4.9 and 7.4, horizontal distance to the surface projection of the rupture (R _JB) less than 200 km, and average shear-wave velocity over the top 30 m of the subsurface (V _S30) between 155 and 1500 m/s. The paper also presents the ranking results for three of the NGA-West2 GMPEs against strong motions recorded in northern Iran, before and after adjustment for region-dependent attenuation characteristics. The ranking is based on the likelihood and log-likelihood methods (LH and LLH) proposed by Scherbaum et al. (Bull Seismol Soc Am 94: 2164–2185, 2004, Bull Seismol Soc Am 99, 3234–3247, 2009, respectively), the Nash–Sutcliffe model efficiency coefficient (Nash and Sutcliffe, J Hydrol 10:282–290, 1970), and the EDR method of Kale and Akkar (Bull Seismol Soc Am 103:1069–1084, 2012). The best-fitting models over the whole frequency range are the ASK14 and BSSA14 models. Taking into account that the models’ performances were boosted after applying the adjustment factors, at least moderate regional variation of ground motions is highlighted. The regional adjustment based on the Iranian database reveals an upward trend (indicated as high Q factor) for the selected database. Further investigation to determine adjustment factors based on a much richer database of the Iranian strong-motion records is of utmost important for seismic hazard and risk analysis studies in northern Iran, containing major cities including the capital city of Tehran.