Ground Motion Characterization Research Articles

Beta functions adaptation on empirical damage for a fragility model of Italian buildings

In this study, heuristic vulnerability model is presented providing damage distribution as a function of ground motion severity, through the use of given distribution function.Firstly, observational damage distribution are obtained from data collected after the recent earthquakes occurred in Italy and made available by DPC through Da.D.O. platform. The latter collect information on the main characteristics about building position, metrical data (e.g., number of storeys, interstorey height, storey area and construction age), structural types, severity and extension of observed damage on different building components. Moreover, the ground motion characterization was described via the ShakeMap published by the National Institute of Geophysics and Volcanology.Then, the two statistical models mostly used in the literature, i.e. binomial and beta, are used to determine the one most suitable to reproduce the observational damage distributions by means of godness-of-fit test. An introductive section properly examine the pro and cons in their use (number of parameters, flexibility). The observation of mean value and variance couple for more than 2000 damage samples obtained based on the considered database is used to obtain a general relationship which allows to reduce the number of parameters required by beta distribution, preserving its renomate flexibility.This enables the formulation of a novel heuristic vulnerability model for derivation of fragility curves. The model relies on the knoledge of mean damage trend with intensity measure and employs a probability distribution function. In this study, specifically, binomial and beta distributions are utilized. The generality of the method is then customized to suit the characteristics of Italian residential building stock, leveraging the information gathered after the 2009 L'Aquila Earthquake. This database presented an unprecedented opportunity due to the extensive data available on building geometrical and typological characteristics, seismic vulnerability, to the high level of representativeness of samples at municipality level, and to the thorough representation of damage patterns observed on building components.The fragility curves obtained through the presented vulnerability model are compared with those empirically derived from the observational data and are used to fully derive seismic risk scenario after 2009 L'Aquila Earthquake, in terms of economic losses, number of long term unusable buildings, the number of collapse, the number of deaths and severely injuries. The ensuing consequences deriving from the adoption of different fragility curves, i.e. empirical or with beta and binomial vulnerability models are deeply investigated.

The 2023 US National Seismic Hazard Model: Ground-motion characterization for the conterminous United States

We update the ground-motion characterization for the 2023 National Seismic Hazard Model (NSHM) for the conterminous United States. The update includes the use of new ground-motion models (GMMs) in the Cascadia subduction zone; an adjustment to the central and eastern United States (CEUS) GMMs to reduce misfits with observed data; an updated boundary for the application of GMMs for shallow, crustal earthquakes in active tectonic regions (i.e. western United States (WUS)) and stable continental regions (i.e. CEUS); and the use of improved models for the site response of deep sedimentary basins in the WUS and CEUS. Site response updates include basin models for the California Great Valley and for the Portland and Tualatin basins, Oregon, as well as long-period basin effects from three-dimensional simulations in the Greater Los Angeles region and in the Seattle basin; in the CEUS, we introduce a broadband (0.01- to 10-s period) amplification model for the effects of the passive-margin basins of the Atlantic and Gulf Coastal Plains. In addition, we summarize progress on implementing rupture directivity models into seismic hazard models, although they are not incorporated in the 2023 NSHM. We implement the ground-motion characterization for the 2023 NSHM in the US Geological Survey’s code for probabilistic seismic hazard analysis, nshmp-haz-v2, and present the sensitivity of hazard to these changes. Hazard calculations indicate widespread effects from adjustments to the CEUS GMMs, from the incorporation of Coastal Plain amplification effects, and from the treatment of shallow-basin and out-of-basin sites in the San Francisco Bay Area and Los Angeles region, as well as locally important changes from subduction-zone GMMs, and from updated and new WUS basins.

The 2023 New Zealand Ground-Motion Database

ABSTRACT This article summarizes the development of the 2023 New Zealand ground-motion database (NZGMDB). A preceding version was formally used as the central ground-motion database in the ground-motion characterization modeling for the 2022 New Zealand (NZ) National Seismic Hazard Model (NSHM) revision. The database contains ground motions for events with a moment magnitude greater than ∼3.0 from the years 2000 to the end of 2022. Several challenges associated with NZ earthquake source metadata are explained, including determination of earthquake location, magnitude, tectonic classification, and finite-fault geometry, among others. The site table leverages the site database developed as a part of the 2022 NZ NSHM revision, and several definitions of source-to-site distance are computed for the propagation path table. The ground-motion quality classification was initially assessed using a neural network. Subsequent waveform quality verification was conducted and additional quality criteria were enforced to ensure a sufficiently high-quality database. Standard processing techniques were applied to the ground motions before intensity measure (IM) calculation. IMs in the database include peak ground acceleration, 5%-damped pseudoacceleration response spectra, smoothed Fourier amplitude spectra, and other cumulative and duration-related metrics. The NZGMDB is publicly available and routinely updated as new and higher quality data become available.

The Ground-Motion Characterization Model for the 2022 New Zealand National Seismic Hazard Model

ABSTRACT This article summarizes the ground-motion characterization (GMC) model component of the 2022 New Zealand National Seismic Hazard Model (2022 NZ NSHM). The model development process included establishing a NZ-specific context through the creation of a new ground-motion database, and consideration of alternative ground-motion models (GMMs) that have been historically used in NZ or have been recently developed for global application with or without NZ-specific regionalizations. Explicit attention was given to models employing state-of-the-art approaches in terms of their ability to provide robust predictions when extrapolated beyond the predictor variable scenarios that are well constrained by empirical data alone. We adopted a “hybrid” logic tree that combined both a “weights-on-models” approach along with backbone models (i.e., metamodels), the former being the conventional approach to GMC logic tree modeling for NSHM applications using published models, and the latter being increasingly used in research literature and site-specific studies. In this vein, two NZ-specific GMMs were developed employing the backbone model construct. All of the adopted subduction GMMs in the logic tree were further modified from their published versions to include the effects of increased attenuation in the back-arc region; and, all but one model was modified to account for the reduction in ground-motion standard deviations as a result of nonlinear surficial site response. As well as being based on theoretical arguments, these adjustments were implemented as a result of hazard sensitivity analyses using models without these effects, which we consider gave unrealistically high hazard estimates.

Epistemic Uncertainty in Ground-Motion Characterization in the Indian Context: Evaluation of Ground-Motion Models (GMMs) for the Himalayan Region

Abstract One of the major challenges in probabilistic seismic hazard analysis (PSHA) studies, particularly for risk-based decision-making, is to constrain epistemic uncertainties. Epistemic uncertainty associated with ground-motion characterization (GMC) models exerts a strong influence on the hazard estimate for a given target level of ground shaking. In the Indian context (mainly along the Himalayan arc), constraining epistemic uncertainty is a significant challenge owing to the lack of recorded data. This study investigates the epistemic uncertainty associated with ground-motion models (GMMs) considered appropriate for the Himalayan region. First, a review of GMMs considered applicable to the Himalayan region is provided. Subsequently, a graphical comparison of median models is performed, followed by residual and statistical analysis. The evaluation utilizes observations from a recently compiled strong-motion dataset across the Himalayas and Indo-Gangetic plains of northern India. The dataset comprises 519 acceleration traces from 150 events in the moment magnitude (Mw) range Mw 3–7.4, recorded at epicentral distances in the range REpi<300 km. The analysis demonstrates significant between-model variability, particularly with regard to median magnitude and distance scaling. The residual analysis also indicates a large bias and aleatory uncertainty. Moreover, some of the GMMs exhibit trends with distance and magnitude. Overall, our evaluation analysis shows that there is clearly significant aleatory and epistemic uncertainty associated with the GMC modeling owing to the paucity of recorded data. The range of epistemic uncertainty represented by the GMMs (available in the literature) is much larger than that typically captured by the (multiple) global models often used in PSHA studies across India.

Hazard Sensitivities Associated with Ground-Motion Characterization Modeling for the New Zealand National Seismic Hazard Model Revision 2022

ABSTRACT This article summarizes hazard sensitivities associated with the updated ground-motion characterization modeling (GMCM) scheme adopted in the recent revision of New Zealand National Seismic Hazard Model (NZ NSHM 2022). In terms of impact on ground-motion hazard, the current GMCM scheme (GMCM 2022) results in an overall, at times significant, increase in calculated mean hazard with respect to NZ NSHM 2010. With regard to relative impact, the update in GMCM accounts for the dominant change in high-hazard regions, whereas in low-hazard regions update in source characterization model dominate. Within GMCM 2022, the change in shallow crustal ground-motion models (GMMs) dominates the effect on calculated hazard, whereas change in subduction interface GMMs has a compounding effect for east coast of North Island and southwest of South Island. Impact of the two NZ-specific adjustments to some of the published GMMs is also discussed. The back-arc attenuation adjustment accounts for a 20%–30% reduction in calculated hazard for peak ground acceleration in northwest of North Island, whereas aleatory uncertainty adjustment accounts for 10%–20% reduction in high-hazard regions such as along the east coast of North Island and in the lower west of South Island.

Characterization and prediction of InSAR-derived ground motion with ICA-assisted LSTM model

Interferometric Synthetic Aperture Radar (InSAR) is a highly effective and widely used approach for monitoring large-scale ground deformation. The precise and timely prediction of deformation holds significant importance in mitigating and preventing geological hazards, particularly considering the long revisit cycle of satellites and the considerable time required for data processing. In this study, we propose a strategy that predicts spatiotemporal InSAR time series based on Independent Component Analysis (ICA) and the Long Short-Term Memory (LSTM) machine learning model. Unlike traditional methods that rely on physical or statistical models, the proposed strategy leverages the power of ICA and LSTM to achieve accurate predictions without such dependencies. ICA is employed to decompose and capture the InSAR displacement signals of interest caused by various natural or anthropogenic processes and to characterize each individual signal. The spatiotemporal unsupervised K-mean cluster method is then applied to partition large-scale deformation fields into homogeneous subregions, considering the spatial variations and temporal nonlinearities of time series. This process facilitates the refinement of the model, thereby enhancing the accuracy of large-scale predictions. The neural network models are then individually constructed for each cluster, and the optimal parameters are determined through a grid search strategy. Subsequently, the proposed framework is implemented and assessed using two datasets featuring distinct deformation patterns: Case I involves land subsidence in Willcox Basin, USA, while Case II focuses on post-seismic deformation following the 12 November 2017 Mw 7.3 Sarpol-e Zahab earthquake. The results demonstrate that our proposed ICA-assisted LSTM outperforms the original LSTM model on large-scale deformation prediction, with the average prediction accuracy for one-step prediction (12 days in our case) being improved by 34% and 17% for cases I and II, respectively. Furthermore, we perform iterative predictions on the spatiotemporal InSAR measurements with varying temporal characteristics for the subsequent five steps using Sentinel-1 data and evaluate its performance and limitations. The successful prediction of land subsidence and post-seismic deformation provides further evidence that the proposed prediction strategy can be effectively employed in monitoring other large-scale geohazards characterized by prolonged and gradual deformation. This capability enables expedited decision-making and timely implementation of risk mitigation measures.

Evaluation of Empirical Ground-Motion Models for the 2022 New Zealand National Seismic Hazard Model Revision

ABSTRACT This article presents an evaluation of empirical ground-motion models (GMMs) for active shallow crustal, subduction interface, and subduction slab earthquakes using a recently developed New Zealand (NZ) ground-motion database for the 2022 New Zealand National Seismic Hazard Model revision. This study considers both NZ-specific and global models, which require evaluation to inform of their applicability in an NZ context. A quantitative comparison between the models is conducted based on intensity measure residuals and a mixed-effects regression framework. The results are subsequently investigated to assess how the models are performing in terms of overall accuracy and precision, as well as to identify the presence of any biases in the model predictions when applied to NZ data. Many models showed reasonable performance and could be considered appropriate for inclusion within suites of models to properly represent ground-motion predictions and epistemic uncertainty. In general, the recent models that are NZ-specific or developed on large international databases performed the best. This evaluation of models helped inform suitable GMMs for the ground-motion characterization model logic tree. In addition, spatial trends in systematic site-to-site residuals to the west of the Taupō Volcanic Zone demonstrated the need for backarc attenuation modifications for slab earthquakes.

On frequency‐domain estimation of increase in design yield force levels of SDOF structures due to multiplicity of earthquake events

AbstractThe current earthquake design philosophy based on only the most critical earthquake may prove inadequate when the smaller events occurring before this event cause significant damage to the structure and repairs become infeasible. The design force levels may therefore be raised using the modified design force ratio, defined as the ratio of the yield force level for a specified damage level under all the seismic events to the yield force level for the same damage level under only the most critical event. This study considers the updating of the existing frequency‐based approach, using the power spectral density function (PSDF)‐based characterization of design ground motion, for the estimation of modified design force ratio spectrum. First, an empirical expression developed in the companion paper for modifying the damping of the equivalent linear oscillator is used such that the use of the PSDF compatible with 5%‐damping response spectrum leads to correct peak response, even when the equivalent damping ratio exceeds 5%. Next, the frequency‐domain design force ratio estimates are benchmarked in the case of steel and reinforced concrete oscillators by comparing those with the time‐domain estimates of equivalent Bouc–Wen oscillators over various sequences of events and by suitably adjusting the stiffness degradation parameter. It is shown in the case of a hypothetical seismic scenario that the optimal values of this parameter as 2.0 and 1.7 work well for the steel structures and reinforced concrete structures with high stiffness degradation respectively. Also, the proposed approach needs to be improved further for applicability in the case of reinforced concrete structures exhibiting significant pinching behavior.

Testing and Evaluation of the New Zealand National Seismic Hazard Model 2022

ABSTRACTWe summarize the work that has been done within the New Zealand National Seismic Hazard Model 2022 (NZ NSHM 2022) to evaluate and test the updated hazard model and its components against observational data. We undertake a two-phase analysis to learn about the performance of the hazard model with respect to several limited databases. Phase 1 is the evaluation phase, involving multiple efforts to optimize various source rate model and ground-motion characterization model components against: (1) the New Zealand earthquake catalog for 1950–2020; (2) international catalogs (where relevant); and (3) New Zealand paleoseismic and geodetic data. Phase 2 involves testing the hazard results. We perform ground-motion-based testing of the NZ NSHM 2022 exceedance rates against the observed exceedance rates for strong-motion stations around New Zealand. To account for the modeled variability in rate, the comparisons are done by assuming a binomial distribution about the mean exceedance rate for 0.1g and 0.2g at each station location. We use a combined approach that considers the full epistemic uncertainty distribution for those exceedance rates by weighting the binomial for each branch in the logic tree. We find that, in general, the observed exceedance rates can be drawn from the NZ NSHM 2022 with probabilities greater than 0.05, and that the discrepancies are generally confined to areas close to major earthquake sequences (e.g., Christchurch). These sequences were not considered in the NZ NSHM 2022 forecast. This initial iteration of testing does not provide evidence to reject the NZ NSHM 2022 based on the New Zealand accelerograph record. Importantly, we can only draw limited conclusions from the testing due to the very short time frame of data available for testing.

Duration and damping for ground motion characterization in frequency‐domain damage estimation of SDOF structures

AbstractIn estimating the damage to a single‐degree‐of‐freedom (SDOF) structure due to multiple earthquake events, it may be convenient to use a frequency‐based approach, in which the anticipated ground motions are characterized through power spectral density functions (PSDFs) compatible with the specified design spectra for those motions. Since a spectrum‐compatible PSDF corresponds to a fictitious stationary ground motion process of a specified duration, the duration of this process should preferably be same as the length of the stationary segment in the actual ground acceleration process for a meaningful computation of the stochastic response. On assuming the strong‐motion segment of a given ground motion to represent its stationary segment, the nonstationarity of the strong‐motion segment is evaluated for the existing strong‐motion duration definitions. It is found that the Trifunac‐Brady definition is the most reasonable strong‐motion duration definition in respect of consistently giving the longest strong‐motion segment with least nonstationarity. Further, a spectrum‐compatible PSDF can be used for estimating the stochastic response of a linear system, provided the system has the same damping ratio as that of the design spectrum from which the PSDF has been computed. This causes errors in estimating the peak response of an equivalent linear system due to its damping ratio being higher than the initial damping ratio of the nonlinear system. To address this, an empirical expression is developed for the modified damping ratio of a linear system, which would lead to the correct peak response of the system on using the PSDF obtained from the 5%‐damping design spectrum.

Characterization and modeling of ground motion at depth in soft sedimentary rocks: Application to the Swiss Molasse Basin

The level of earthquake motion observed on the ground surface differs from the shaking at locations buried at a depth. Characterization of the underground shaking is important for the local seismic hazard assessment at depth which is an insufficiently investigated topic so far. Advancements in the prediction of the ground motion at depth are especially important for the design of deep geological disposals of nuclear waste, buildings with deep foundations, or studies focused on soil-structure interactions. In this paper, we analyze site-specific transfer functions that relate ground motions on the surface to that at a depth at sites located within soft sedimentary rocks of the Swiss Molasse Basin. We first evaluate empirical surface-to-borehole amplification functions at six borehole sites, compare them with predictions by a theoretical model, and characterize the high-frequency surface-to-borehole amplification. Next, we perform full-waveform prediction of ground motion at depth from surface recordings of seven significant regional earthquakes with ML > 4. The comparison of predicted and observed acceleration waveforms shows a high level of similarity in a broad frequency range and well-predicted values of PGA, PGV, and response spectra. Finally, we model a scenario earthquake for 9975 years return period at a test site, and we predict associated broadband waveforms at depth. By the latter, we demonstrate that our method to predict ground motions at depth can be used as a basis for the site-specific seismic hazard of deep geological disposals of nuclear waste.

EESD special issue: AI and data‐driven methods in earthquake engineering – (Part 1)

EESD special issue: AI and data‐driven methods in earthquake engineering – (Part 1)

The Rocks That Did Not Fall: A Multidisciplinary Analysis of Near‐Source Ground Motions From an Active Normal Fault

AbstractOn 8 July 2021 a M6.0 normal faulting earthquake rocked the community of Walker and the surrounding region near the California‐Nevada border. In the 1990s, field surveys of nearby Meadowcliff Canyon identified numerous precarious rocks deemed likely to topple in the event of strong shaking. Despite their proximity (∼6 km) to the 2021 earthquake, the precarious rocks still remain standing. In this work, we combine advanced source and ground motion characterization techniques to help unravel this mystery. High‐precision hypocentral locations reveal a clear north/south‐striking, east‐dipping rupture plane along the southern extension of the Slinkard Valley fault. The mainshock nucleated near the base of the fault, triggering thousands of aftershocks. Bayesian source spectral analyses indicate that the mainshock had a moderately‐high stress drop (∼17 MPa), and that aftershocks with deeper hypocenters have higher stress drops. Peak Ground Acceleration (PGA) recordings at regional stations agree well with existing ground motion models, predicting PGA of ∼0.3 g in Meadowcliff Canyon, a level sufficient to topple precarious rocks based on PGA‐derived stability criteria. We demonstrate that despite these large ground accelerations, the pulse duration in Meadowcliff Canyon is too short to supply the impulse necessary to damage these features, observations which support the application of dynamic toppling models that account for the joint effects of pulse amplitude and duration when assessing rock fragility. This study provides a unique vantage point from which to interpret rarely‐observed strong‐motion recordings from close to an active normal fault, one of many that dominate hazard along the eastern Sierra.

Exploring Australian hazard map exceedance using an Atlas of historical ShakeMaps

This article explores several area-based tests of long-term seismic hazard forecasts for the Australian continent. Using the observed seismicity, ShakeMaps are calculated for earthquakes that are expected to have generated moderate-to-high levels of ground shaking within continental Australia in the past 50 years (January 1972 through December 2021). A “composite ShakeMap” is generated that extracts the maximum peak ground acceleration “observed” in this 50-year period for any site within the continent. The fractional exceedance area of this composite map is compared with four generations of Australian seismic hazard maps for a 10% probability of exceedance in 50 years (~1/500 annual exceedance probability) developed since 1990. In general, all these models appear to forecast higher seismic hazard relative to the ground motions that are estimated to have occurred in the last 50 years, with the most recent hazard model yielding a fractional exceedance area most similar to the target 1/500 annual exceedance probability. The sensitivity of these results to various modeling assumptions was tested by exploring an alternative ground-motion characterization model that forecasts higher overall ground-shaking intensities. The sensitivity of these results is also tested with the interjection of a rare scenario earthquake with an expected regional recurrence of approximately 8700 years. While these area-based analyses do not provide a robust assessment of the performance of the candidate seismic hazard for any specific location given the limited independent data, they do provide—to the first order—a guide to the performance of the respective maps at a continental scale.

Peculiar characteristics of ground motions in Southern Italy: Insights from global and regional ground motion models

We investigate peculiar characteristics of ground motions in Southern Italy (e.g. apparent fast anelastic attenuation and trends of event terms at different periods) using a comprehensive dataset of earthquake recordings between 1969 and 2020. By doing so, we gained insights into the relative performance of eight selected region-specific, global, and global with regional adjustment ground motion models (GMMs). Our analysis is performed using a preliminary dataset (i.e. including all ground motions recorded in the area for the selected analysis period) and an independent dataset (i.e. comprising data not used to develop the models). We analyze total residuals, event terms, within-even residuals, and residuals standardized by model standard deviations (i.e. epsilon). The latter is performed to obtain a robust comparison of GMMs with different standard deviation types and levels. These approaches are employed to ground motion characterization studies for the first time in this region. Our results show that in Southern Italy, there is an apparent anelastic attenuation of the ground motion faster than in other seismic districts. Overall, regional models capture this feature better than global models. Regional adjustments to global models better capture the observed anelastic attenuation at large distances. Using the standardized residuals analysis, we observe that all selected GMMs systematically underestimate the observed ground motion for relatively high ground motion levels and its variability at any intensity levels in the study region. These outcomes may help improving future ground motion models and related engineering applications involving such models in performance-based frameworks.

Improving soil liquefaction prediction through an extensive database and innovative ground motion characterization: A case study of Port Island liquefied site

Liquefaction is a phenomenon in which soils lose their strength and stiffness significantly during earthquake shakings. It is very important to accurately predict the liquefaction triggering and permanent displacement of liquefied sites under different seismic conditions. This study systematically investigates the effect of ground motion characteristics on the evolution of important physical quantities during the liquefaction process, while the relative capacity of various ground motion intensity measures (IMs) in predicting the pore-pressure generation and the permanent deformation in liquefiable soils are studied. An extensive database containing 501 three-component ground motion records is established to improve soil liquefaction prediction under different earthquake scenarios. A significant number of fully nonlinear dynamic analyses are performed for the Port Island liquefied site, where advanced constitutive models are used to realistically characterize the nonlinear soil behaviors. Results from numerical simulations demonstrate that the energy-related IMs such as CAV5 have a superior predictive capacity than the peak ground acceleration in predicting the liquefaction initiation and permanent displacement, but the predictions based on scalar IMs are not sufficient because residuals are strongly depended on earthquake magnitude and rupture distance. On the other hand, vector IMs that incorporate both energy and peak value characteristics of ground motions can improve the predictive capabilities regarding pore pressure generation and permanent deformation in liquefiable soils. The study helps lay down the basis towards an improved procedure for liquefaction hazard evaluation and mitigation.

Regression analysis on ground motion parameters for the earthquakes (Mw≥4.0) in NE India with special emphasis on 3 Jan 2016 M6.7, Tamenglong earthquake

In the present study, ground motion parameters (GMPs) are estimated by processing 26 earthquake events containing 125 accelerogram records with 375 components that originated in NE, India, and its vicinity with special emphasis on the Jan 3, 2016, M6.7, Tamenglong Earthquake. Moreover, Ground Motion Prediction Equations were developed through multiple regression analysis on observed data of 8 GMPs namely, Peak Ground Acceleration, Peak Ground Velocity, Arias Intensity, Characteristics Intensity, Housner Intensity, Cumulative Absolute Velocity, Effective Design Acceleration, and Acceleration Spectrum Intensity. The developed equations are representative of statistics on changes in amplitudes of parameters with varying distances and magnitude in connection to Northeast India. Besides, newly developed GMPEs can be applied to 4–6.8 magnitude earthquakes and valid up to 525 km of distance. In the case of January 3, 2016, Mw = 6.7, Tamenglong earthquake, a total of 25 parameters are estimated from ground motion recordings. The highest Peak Ground Acceleration (PGA) of 0.05 (g) has been experienced by Kohima station along with Pseudo Spectral Acceleration of 0.2802 (g) at the frequency of 1.72 Hz. The peak pseudo-spectral accelerations for each recording site of the Tamenglong earthquake are perceived within the periods of 0.2s and 0.5s. These estimations help in the characterization of ground motion with varying distances and magnitude which is a prime input to rate damage potential of ground motion and seismic hazard analysis in Northeast India.

Fragility curves derivation for masonry buildings damaged after 2009 L'Aquila earthquake accounting for the effect of construction age

The aim of this study is the analysis of seismic fragility of residential masonry buildings, with particular emphasis to the evolution of seismic behaviour over the years. To this purpose, a detailed and comprehensive taxonomy has been established, considering a restricted selection of building's features, avoiding the ineffective fragmentation of the database. The fragility assessment has been performed based on the data collected after the L'Aquila 2009 earthquake and made available by the Italian Department of Civil Protection through the online platform Da.D.O. (Database of Observed Damage). The database has undergone a refinement process to guarantee the completeness of the information, avoiding possible bias in the subsequent fitting procedure for fragility analysis. Thus, PGA (peak ground acceleration) from the ShakeMap has been used for ground motion characterization and 5 + 1 damage levels defined according to the European Macroseismic Scale have been considered for damage classification. Different regression models have been adopted to determine the parameters of lognormal fragility curves, measuring their goodness of fit with the observed damage probability matrices.Starting from the unconstrained model, further regression constraints (i.e., a common value for logarithmic standard deviation and the respect of the hierarchy of median PGA with the construction age) have been introduced, thus leading to the definition of the constrained model. The benefits in the introduction of further regression constraints are counterposed to the effectiveness of constrained curves to model observational data through the comparison of the goodness of fit between the unconstrained and constrained models.

Variability of physics-based simulated ground motions in Thessaloniki urban area and its implications for seismic risk assessment

An accurate characterization of earthquake ground motion and its variability is crucial for seismic hazard and risk analysis of spatially distributed portfolios in urban areas. In this work, a 3D physics-based numerical approach, based on the high-performance spectral element code SPEED (http://speed.mox.polimi.it/), is adopted to generate ground shaking scenarios for strong earthquakes (moment magnitude MW=6.5–7) in the Thessaloniki area (Northern Greece). These simulations account for kinematic finite-fault rupture scenarios and a 3D seismic velocity including the two main geological structures present in the area (Thessaloniki and Mygdonia basins). The numerical model is successfully validated by comparing simulated motions, on the one hand, with the recordings of a real small-magnitude (MW4.4) earthquake and, on the other, with empirical Ground Motion Models for the historical MW6.5 1978 earthquake. The sensitivity of results to the velocity model, anelastic attenuation, and non-linear soil effects is evaluated. The variability of the ground motion intensity measures in Thessaloniki as a function of the finite-fault rupture realizations (causative fault, magnitude, hypocenter location) is explored to gain insight into its potential impact on seismic risk assessment in urban areas.