Figure 1 - uploaded by Ezio Faccioli
Content may be subject to copyright.
Map of northern Italy (in background), where Po Plain extends roughly between Torino and Venezia, with earthquake epicenters ( M w ≥ 4 : 0 ) from working catalog (1005 – 2013) discussed in Faccioli (2013). Triangles show the accelerometer sites considered in this study. Polygons with continuous lines are the surface projections of the area sources ( AS s) taken from the ZS9 model of Meletti et al. (2008), with corresponding number, and polygons with dashed lines show changes introduced in Faccioli (2013). SPP is the southern Po Plain AS , grouping together the ZS9 912, 913, and 914 AS s. The polygon labeled “ slab ” is the surface projection of an inclined slab zone, to which the deep events have been associated in Faccioli (2013). Stars denote the most recent regional events with M w > 4 .
Source publication
Site-specific ground-motion hazard, as assessed by a probabilistic seismic- hazard analysis one-step approach that handles a single-site sigma and its uncertainties and uses a simple logic tree, is compared with a two-step approach that includes bedrock motion evaluation and wave propagation through a local soil profile with consideration of the ma...
Contexts in source publication
Context 1
... sites, whereas at T0821 the spectra derived from the AS model are more conservative. Besides the AS model, a gridded seismicity representation has also been used in separate branches of the SH analysis, by applying a map of occurrence probabilities to a regular grid of point sources. For this purpose, the time-independent, Poisso- nian HAZGRID model (Akinci, 2010) was selected, in its latest version with the 2011 Catalogo Parametrico dei Terremoti Ital- iani (Rovida et al. , 2011) as its reference catalog for Italian earthquakes (updated with 2012 events). This model uses gridded historical seismicity that is spatially smoothed to different length scales. The number of earthquakes with magnitude M w ≥ M min 4 : 7 in each cell i of the grid is converted from cumulative to incremental values (i.e., number of events with magnitude M w ). For each cell of the grid ( 0 : 1 ° × 0 : 1 °), the model estimates the Gutenberg – Richter a and b parameters using the declustered catalog. Gridded values of occurrence rates 10 a (earthquakes/cell/year) were computed and smoothed spatially by a 2D Gaussian function with 25 km correlation length (Console and Murru, 2001). The empirical attenuation models used in the present analyses were the regional ones developed by Bindi et al. (2011), herein ITA10, and its updated version ITA13, specifically developed by Pacor et al. (2013) within the SIGMA project (see Data and Resources). ITA13 was derived from a northern Italy dataset (called DBN2_B), the bulk of which comes from the 2012 Emilia seismic sequence, recorded by national accelerometer networks and temporary arrays in- stalled after the 20 May 2012 mainshock. The considered northern Italy dataset features, among other things, relatively low spectral amplitudes at short periods, amplification of spectral ordinates in the 80 – 100 km dis- tance range (likely due to Moho reflections), and notable low-frequency amplification at stations lying on the Po Plain deep sediments. To capture these features, apart from the standard A, B, and C site categories of Eurocode 8 (2003), an additional category C1 was introduced that includes the stations on deep sediments within the Po Plain and, hence, should account for basin effects. The subsequent residual analysis by Pacor et al. (2014; see Data and Resources) shows that the site term is variable even for C1 sites, although these tend to have mostly similar geological features. Because the northern Italy dataset contains predominantly records from thrust-fault events, mainly from category A and B sites at far distance and from C1 sites at short distance, we adopted the ITA13 for source zones with a predominant focal mechanism of the thrust type; the study sites are all of the C1 category. For shallow source zones with other styles of faulting, the ITA10 GMPE was adopted, following recommendations by Pacor et al. (2013). For ITA13, we used the formulation with the Joyner – Boore distance metric R JB , without conver- sion of M L into M w in the metadata, which provided the best score from log-likelihood tests (see Pacor et al. , 2013). For the passive subduction AS labeled “ slab ” in Figure 1, we used the Zhao et al. (2006) GMPE , with version standard and option interface selected, in ergodic mode. One-Step Hazard Estimation at Study Sites We first consider the data-based direct hazard evaluation at the soil sites by a single-station sigma approach, skipping the intermediate bedrock ground-motion ...
Context 2
... densely inhabited and industrialized region of low-to- moderate seismicity, the Po Plain of northern Italy, chosen herein as the study area, is the central flat land in Figure 1. This also shows the historical and instrumental seismicity and a lay- out of a standard earthquake area sources ( AS s) model for Italy, denoted the ZS9 model (Meletti et al. , 2008), and is the basis of the SH maps linked to the current Italian seismic code (NTC08), with some modifications introduced by ...
Similar publications
One of the most challenging problems in the estimation of seismic hazard is the ability to quantify seismic activity. Empirical models based on the available earthquake catalogue are often used to obtain activity of source regions. The major limitation with this approach is the lack of sufficient data near a specified source. The non-availability o...
Seismic waves propagating through the Earth can be significantly affected by velocity structures such as sedimentary basins. We investigate the propagation characteristics of seismic waves across the Kanto basin, Japan, using Green's functions extracted from the ambient seismic field. We use two stations situated on the eastern and southern edges o...
We produce year-by-year probabilistic seismic hazard maps based on the seismicity from 2011 to 2018 for the western Canada sedimentary basin (WCSB) and compare them with corresponding maps of the baseline historical hazard levels through 2010. Rates of earthquakes across the WCSB, of moment magnitude (M) ≥ 3:0, have increased in the past decade bec...
The source mechanism of the ML 4.0 25 April 2016 Lacq earthquake (Aquitaine Basin, South-West France) is analyzed from the available public data and discussed with respect to the geometry of the nearby Lacq gas field. It is one of the biggest earthquakes in the area in the past few decades of gas extraction and the biggest after the end of gas expl...
Agartala, the capital town of state Tripura and a business corridor for South East Asia through Bangladesh which is situated in the highest vulnerable zone as per seismic zoning map of India. Past earthquakes (e.g., 1918 Srimangal earthquake) experienced damage of structures and some incidence of liquefaction cases in this area. The geology of cent...
Citations
... Incorporating AFs for soil seismic hazard determination is a recent approach, acknowledging some uncertainty in site-specific ground motion. Several methodologies are used for integrating site effects into PSHA (Barani and Spallarossa 2017;Bazzurro and Cornell 2004a, b;Cramer 2005;Faccioli et al. 2015;Jacobs 2002;Lee 2000;McGuire et al. 2001a, b;Papaspiliou et al. 2012a, b;Rodriguez-Marek et al. 2014). ...
Incorporating local site effects into seismic hazard analysis, especially for deep soil sites, is crucial. This study describes
probabilistic seismic hazard analysis (PSHA) incorporating site effects for a deep soil site in Pakistan. Bedrock hazard is
assessed in FRISK88M using a rigorous logic tree. Equivalent linear site response analysis (SRA) is performed in STRATA
and captured the soil variability properties via Monte Carlo simulations. We proposed a hybrid approach in this study which
utilizes the outcomes of Random Vibration Theory (RVT) and multiple Time Histories (TH) approach to assess the site-
specific amplifications
... The probabilistic estimation of seismic hazard at the soil surface is typically achieved by incorporating site effects into Probabilistic Seismic Hazard Analysis (PSHA) using a fully probabilistic approach. In recent years, various probabilistic methods have been proposed, including those developed by Bazzurro and Cornell (2004a), Barani et al. (2014a, b), Rodriguez-Marek et al. (2014), andFaccioli et al. (2015). The convolution method proposed by Bazzurro et al. (2004) is considered the most rigorous, although it is also the most complex. ...
This study presents probabilistic surface seismic hazard maps for Algiers city in terms of PGA and spectral acceleration for a return period of 475 years (corresponding to 10% in 50 years). The site effect is integrated through a fully probabilistic approach. Seismic hazard analysis for reference rock sites (Vs30 = 800 m/s) is conducted first based on an updated seismic source model, including area sources and active faults, and three selected ground motion prediction equations, Epistemic uncertainties related to source model (slip rate for faults, Gutenberg Richter’s parameters for area source) and ground motion equation selection are considered through logic tree approach. Then, 1D ground response analysis is performed to estimate soil amplification factors and their standard deviations considering uncertainties related to input motion variability. Finally, probabilistic site specific-PSHA using modified_GMPE with respect to amplification factors, is conducted for more than 4000 soil profiles. Surface seismic hazard maps in terms of PGA, and PSA(T) are shown for the study area, they highlight an important site to site variability within the same soil class (S2 and S3) according to the Algerian seismic design code (RPA). In addition, surface uniform hazard spectra (surface_UHS) for several selected sites belonging to different geotechnical units are compared with the current design spectra prescribed in seismic design codes (RPA, EC8). These findings emphasize the value of conducting a site-specific PSHA study, given the considerable discrepancy that can exist between a site-specific spectrum and a seismic design spectrum. While design spectra tend to be more conservative for longer periods (>0.3 s), the specific spectrum yields greater accelerations for shorter periods, particularly for soils of class S3.
... Thus, the parameters of the hyperbolic model for eg-unit groups and macro groups, respectively, are shown in Tables 2, 5. Neverthless, the outcomes of this study can be used in any code that simulates 1D propagating waves by using the parameters provided in Table 4 and the formulation in Eqs. (5,6), when Darendeli's model is not implemented. ...
Numerical simulations of seismic site response require the characterization of the nonlinear behaviour of shallow subsoil. When extensive evaluations are of concern, as in the case of seismic microzonation studies, funding problems prevent from the systematic use of laboratory tests to provide detailed evaluations. For this purpose, 485 shear modulus reduction, G\G 0 ( γ ) and damping ratio, D( γ ) curves were collected from multiple literature sources available in Italy. Each curve was associated with the related engineering geological units considered in seismic microzonation studies. A statistical analysis of the data was carried out with the aim of shedding light on the significant difference between the laboratory classification of samples and the macroscopic/engineering geological one, provided during seismic microzonation studies. Since the engineering geological classification plays a prominent role in extensive site response evaluations, the outcomes of the present work may be of help at least when preliminary seismic response estimates are of concern. The dataset provides reference information that can serve as key data for large-scale hazard assessments worldwide.
... In the recent years, site-specific seismic hazard assessments that account for seismic amplification in the Po Plain area were carried out by Faccioli et al. (2015) and Vanini et al. (2018), but extensive large-scale studies are still lacking. In this area, detailed hazard mapping inclusive of site effects is nowadays possible, thanks to the increasing number of seismic microzonation studies (e.g., Lai et al., 2020;Martelli and Ercolessi, 2020) and ground-response assessments (e.g., Mascandola et al., 2021). ...
... In the recent years, significant progress has been made in the field of site-specific PSHA. The works of Faccioli et al. (2015), Barani and Spallarossa (2017), and Aristizabal et al. (2022) provide a comprehensive review of the different approaches for the integration of seismic amplification into PSHA. The fully probabilistic approaches can be grouped into three levels of increased complexity. ...
... Conversely, the nonergodic approach separates those components that are known (or knowable) because of their repeatable nature, thus allowing for a better representation of the ground-motion sigma. In the present study, we apply the so-called partially nonergodic approach to avoid double counting of the uncertainty related to site response (e.g., Rodriguez-Marek et al., 2014;Faccioli et al., 2015;Mascandola et al., 2017;Barani et al., 2020). Compared to the fully nonergodic approach, the partially nonergodic one separates only the repeatable and systemic component related to the site behavior from the ergodic sigma, whereas ignoring those components that are related to earthquake source and wavepath (Lin et al., 2011;Villani and Abrahamson, 2015). ...
We present a probabilistic seismic hazard analysis (PSHA) for the entire Po Plain sedimentary basin (Italy)—one of the widest Quaternary alluvial basins of Europe, to evaluate the impact of site-response characterization on hazard estimates. A large-scale application of approach 3 of the U.S. Nuclear Regulatory Commission (NRC) to include seismic amplification in the hazard is presented. Both 1D amplification related to stratigraphic conditions and 3D amplification due to basin effects are considered with the associated uncertainties, and their impact on the hazard is analyzed through a sensitivity analysis. Whereas 3D basin effects are considered through the application of an empirical, spatial invariant correction term, 1D amplification was estimated throughout the study area by means of dynamic (equivalent linear) ground-response analysis. To separate aleatory variabilities and epistemic uncertainties related to site response, a partially nonergodic approach is used. The results provide a finer picture of the actual seismic hazard, highlighting those areas where the ground motion is affected by amplification effects due to local or regional geological features. We found that, for a return period of 475 yr, neglecting basin effects produces a 30% underestimation of the seismic hazard in the long-period (>1 s) range. Moreover, with reference to the hazard model adopted, such effects are responsible for most of the epistemic uncertainty (up to 80%) in the results. Therefore, such effects deserve special attention in future research related to PSHA in the Po Plain sedimentary basin.
... In PSHA, if sufficient earthquake recordings are available at the site of interest, site responses can be quantified using recording pair-or network-based techniques, that is, GIT, dS2Ss, or S(B)SR with a negligible epistemic uncertainty or standard error (e.g. Faccioli et al., 2015). Results from these techniques can capture the true but often unknown site responses at individual sites, and their corresponding f m S2S has no or an insignificant impact on the total sigma s. ...
In site-specific site response assessments, observation-based site-specific approaches requiring a target-reference recording pair or a regional recording network cannot be implemented at many sites of interest. Thus, various estimation techniques have to be utilized. How effective are these techniques in predicting site-specific site responses (average over many earthquakes)? To address this question, we conduct a systematic comparison using a large dataset which consists of detailed site metadata and Fourier outcrop linear site responses based on observations at 1725 K-NET and KiK-net sites. We first develop machine learning (i.e., random forest) amplification models on a training dataset (1580 sites). Then we test and compare their predictive powers at 145 independent testing sites with that of the one-dimensional (1D) ground response analysis (GRA). The standard deviation of residuals between observations and predictions, i.e., between-site (site-to-site or inter-site) variability, is used as the benchmark. Results show that the machine learning model using a few predictor variables, surface roughness, peak frequency fP,HV, VS30 and depth Z2.5, achieves better performance than the physics-based modelling (GRA) using detailed 1D velocity profiles. This implies that machine learning can be more effective in utilizing existing site information than 1D GRA which is inflicted by a high level of parametric and modelling uncertainties. This finding warrants the further exploration of machine learning in site effect characterization, especially on model transferability across different regions.
... Softer soils, with V S ≤ 800 m s −1 , are expected to amplify the seismic action, therefore the correct estimation of the site amplification should consider the entire soil profile from the ground surface down to the engineering bedrock. Nevertheless, in deep alluvial basins, a clear soft soil-bedrock interface may not be evident but buried under hundreds of metres of soft sediments (Guèguen et al. 2007;Poggi et al. 2012;Faccioli et al. 2015;Mascandola et al. 2019). In these geological contexts, thick and soft sediments are expected to amplify the seismic ground motion (Kramer 1996). ...
The average shear-wave velocity of the first 30 metres of subsoil and the depth of the engineering bedrock are considered the key parameters for simplified seismic site response modelling. However, a reliable estimate of the site amplification should consider the entire shear-wave velocity profile from the ground surface down to the engineering bedrock. In deep alluvial basins, a typical geological context where the soil–bedrock interface may lie below the penetration depth of most common prospecting methods, only the shallow velocity profile can be defined in detail, while the deeper structures are commonly extrapolated with linear equations. The choice of a realistic interpolation between the shallow and deep soil still remains an open issue. We compute the 1D seismic site response of two sectors of the Venetian Plain (Northern Italy) characterised by gravelly and sandy deep formations. We model the 1D soil columns using theoretical non-linear gradients proposed in literature for deep alluvial basins. The numerical modelling results, in terms of strong motion parameters, show variations in the seismic site response up to 20%. The effect of the velocity gradients is also evaluated comparing the numerical simulations with real accelerometers recorded by a deep borehole seismometer and a seismic station located at the top of the borehole. These results demonstrate that the selection of the velocity gradient is crucial for seismic site characterisation of deep alluvial basins. In particular, the study suggests which is the most conservative gradient among the ones tested in terms of ground motion hazard estimation.
... In PSHA, if sufficient earthquake recordings are available at the site of interest, site responses can be quantified using recording pair-or network-based techniques, i.e., GIT, δS2Ss, or S(B)SR with a negligible epistemic uncertainty or standard error (e.g., Faccioli et al., 2015). Results from these techniques can capture the true but often unknown site responses at individual sites, and their corresponding 2 has no or an insignificant impact on the total sigma σ. ...
In site response assessments, observation-based site-specific approaches requiring a target-reference recording pair or a recording network cannot be implemented at many sites of interest. Thus, various estimation techniques have to be utilized. How effective are these techniques in predicting site-specific site responses (average over many earthquakes)? To address this question, we conduct a systematic comparison using a large dataset which consists of detailed site metadata and Fourier outcrop linear site responses based on observations at 1725 K-NET and KiK-net sites. We first develop classic regression and machine learning (i.e., random forest) amplification models on a training dataset (1580 sites). Then we test and compare their predictive powers at 145 independent testing sites with those of the one-dimensional (1D) ground response analysis (GRA), and the empirical correction to the horizontal-to-vertical spectral ratio (eHVSR) of earthquakes (c-HVSR). The standard deviation of residuals between observations and predictions, i.e., between-site (site-to-site or inter-site) variability, is used as the benchmark. Results show that the machine learning amplification model using a few predictor variables, surface roughness, fP,HV, VS30 and Z2.5, achieves better performance than the 1D physics-based modelling (GRA) using detailed ground structures. This is because GRA is inflicted by a high level of parametric and modelling uncertainties. In addition, we propose a new machine learning amplification model using single-station eHVSR curve as a vector-valued predictor variable which is very effective in further lowering the between-site variability in the full frequency range 0.1-20 Hz. Meanwhile, GRA results can also be improved using eHVSR-consistent velocity profiles. These demonstrate the benefits of collecting on-site earthquake recordings and are particularly pertinent for cases where recording pair- or network-based empirical techniques remain a “luxury”. Future endeavors could further explore the modeling of more complex site effects using machine learning techniques, single-station recordings and geotechnical measurements, which could be an alternative to the physics-based simulation.
... Different methods can be used to evaluate site effects: studies based on recorded ground motions (i.e., data-based approach) or numerical simulations (i.e., simulation-based approach). In the absence of a sufficient number of available records at the site, the latter represents the only feasible option (Olsen 2000;Rodriguez-Marek et al. 2014;Faccioli et al. 2015). ...
The ground response to seismic waves is governed by the geometry and the mechanical properties of the site. A proper characterization of the soil behavior is thus a fundamental aspect, and it should account for the uncertainties associated with the model parameters. In particular, the quantification of the small-strain damping is a critical task, especially in low-to-moderate seismicity areas. In the present paper, the main issues related to the definition of the damping at small strains are firstly treated in the light of the biases affecting both laboratory and in situ tests. Higher values of damping are expected in field, where wave scattering phenomena take place. The influence of the parameter on the overall site response is subsequently assessed through a stochastic database of ground response analyses. The results highlight a reduction of the expected ground motion at the surface, especially for deep and soft sites when site-based small-strain damping is selected. Finally, the differences between site and laboratory values are analyzed regarding a specific case study. The influence of the damping at small strains resulted to be comparable or even higher with respect to the uncertainties related to the shear wave velocity profile and the modulus reduction and damping curves. Therefore, a proper evaluation of the uncertainties in the small-strain damping evaluation should not be neglected.
... RESORCE dataset had about 150 stations with ≥ 3 records, while ESM has 1077. This increase is highly sought in empirical site-specific GMM, PSHA, and seismic risk applications (Faccioli et al. 2015;Kohrangi et al. 2020;Kotha et al. 2017;Rodriguez-Marek et al. 2013). ...
To complement the new European Strong-Motion dataset and the ongoing efforts to update the seismic hazard and risk assessment of Europe and Mediterranean regions, we propose a new regionally adaptable ground-motion model (GMM). We present here the GMM capable of predicting the 5% damped RotD50 of PGA, PGV, and SA(T = 0.01 − 8 s) from shallow crustal earthquakes of 3 ≤ MW ≤ 7.4 occurring 0 < RJB ≤ 545 km away from sites with 90 ≤ Vs30 ≤ 3000 m s−1 or 0.001 ≤ slope ≤ 1 m m−1. The extended applicability derived from thousands of new recordings, however, comes with an apparent increase in the aleatory variability (σ). Firstly, anticipating contaminations and peculiarities in the dataset, we employed robust mixed-effect regressions to down weigh only, and not eliminate entirely, the influence of outliers on the GMM median and σ. Secondly, we regionalised the attenuating path and localised the earthquake sources using the most recent models, to quantify region-specific anelastic attenuation and locality-specific earthquake characteristics as random-effects, respectively. Thirdly, using the mixed-effect variance– covariance structure, the GMM can be adapted to new regions, localities, and sites with specific datasets. Consequently, the σ is curtailed to a 7% increase at T < 0.3 s, and a substantial 15% decrease at T ≥ 0.3 s, compared to the RESORCE based partially non-ergodic GMM. We provide the 46 attenuating region-, 56 earthquake localities-, and 1829 site-specific adjustments, demonstrate their usage, and present their robustness through a 10-fold cross-validation exercise.
Open-access: https://rdcu.be/b4kWx
... The main advantage of relaxing the ergodic hypothesis in PSHA, for which the ground motion variability from a large data set of ground motions from various earthquakes recorded at multiple stations is an unbiased estimate of the variability at a single site, is that it avoids double counting of the uncertainty related to site amplification. Examples of partially non-ergodic PSHA are those presented in Rodriguez-Marek et al. (2014), Faccioli et al. (2015), Mascandola et al. (2017), Ameri et al. (2017), and Barani et al. (2019). ...
This study presents a first attempt to couple two of the most important components of seismic risk mitigation strategies: probabilistic seismic hazard analysis (PSHA) and seismic microzonation. We show an application in the westernmost sector of the Liguria region in northwestern Italy, an area where strong earthquakes (up to about magnitude 6.5) occurred in the past producing hundreds of casualties. The final goal is to produce probabilistic seismic hazard maps for peak ground acceleration and spectral acceleration that incorporate site amplification (by amending existing ground motion attenuation models for rock conditions with an additional amplification term provided by microzonation studies) on a municipal scale. To this end, a partially non-ergodic approach is applied according to modern practice. Compared to conventional, ergodic hazard assessments for rock conditions and flat topography, such maps provide at a glance a finer picture of the actual hazard of the area under study, highlighting those sites where the ground motion hazard is dominated by local amplification effects. At these sites, the expected ground motions are up to twice the values resulting from the traditional, ergodic PSHA on rock. Besides hazard maps, results are also presented in terms of uniform hazard spectra for a number of target sites. Comparison with the isoseismal maps of past earthquakes have revealed a good agreement between the spatial distribution of the assessed hazard and that of felt intensities, with greater damage observed in areas presenting significant amplification effects and, consequently, a higher seismic hazard.
