On-site Earthquake Early Warning: Predictive Models for Acceleration Response Spectra Considering Site-Effects

Abstract

<p>Including site specific amplification factors in ground motion prediction models represented an advance for PSHA (Atkinson 2006; Rodríguez-Marek et al. 2013; Kotha et al. 2017) that has become nowadays a standard. However, this issue has only recently received attention by the seismological community of earthquake early warning (EEW) (Spallarossa et al., 2019; Zhao and Zhao, 2019), which applications require a real-time prediction of ground motion and the delivery of alert messages to users for mitigating their exposure to seismic risk. Indeed, all EEW systems are high-technological infrastructures devoted to the real-time and automatic detection of earthquakes, rapid assessment of the associated seismic hazard for targets and the prompt delivery of alerts trough fast telecommunication networks. Among them, the on-site approaches are based on seismic networks placed near to the target, indifferently by the location of seismic threats and they issue the alert predicting the ground motion at the target from P-wave parameter. This configuration cause that On-Site EEWS are generally highly affected by site conditions.</p><p>In this work, we calibrated ground motion prediction models for on-site EEW considering acceleration response spectra (RSA) and the P-waves EEW parameters Iv2 and Pd, and we investigated the role of site-effects. We considered a dataset of nearly 60 earthquakes belonging to the Central Italy 2016-17 sequence. The high density of stations near to the sequence has allowed us to use a non-ergodic random-effect regression approach to explore and to reduce the contribution of site-effects to the uncertainty of the On-site laws predictions. We grouped the records in two ways: by stations and by EC8 classification. Then, we validated the estimated models by the Leave One Out (L1Out) technique and applied a K-means analysis to assess the performance of the EC8 classification.</p><p>The residuals analysis proved that grouping by station provides a set of relations that improves the predictions at many stations. On the contrary, L1Out cross-validation proved that the regressions retrieved grouping by EC8 classification produce higher uncertainties on the predictions than the others. Furthermore, the cross-validation proved that Iv2 is more correlated to RSA than Pd. Finally, the analysis of the random effect vs period curves confirmed that EC8 classification is unrelated to the site effect on RSA even looking only at the trend of these curves.</p><p>In conclusion, non-ergodic random-effect regression can be used also in the EEW applications to predict site-specific ground motion. EEWS that use this approach are less dependent by site-effect and able to provide more precise and reliable alerts.</p>

Earthquake Early Warning System for Structural Drift Prediction Using Machine Learning and Linear Regressors

“…Nevertheless, the full investigation and explanation of the causes of these site conditions is beyond the aim of this paper. In our opinion, the significant variation in residuals shown in Figure 5A is not surprising, being in agreement with other studies (Spallarossa et al, 2019;Iaccarino et al, 2020); which have recently discussed how to reduce the prediction variability considering site-effect terms in EEW model using the mixed-effect regression approach (Pinheiro and Bates, 2000).…”

“…In other words, models calibrated combining datasets from different regions are exported to further areas assuming that regional effects do not play role in the model uncertainty (Stafford, 2014). However, results of recent EEW studies (e.g., among others Spallarossa et al, 2019;Iaccarino et al, 2020) have shown the opposite; that is to say, regional characteristics can play an important role in the robustness and accuracy of the EEW predictions, leading to increase the epistemic uncertainty (Al Atik et al, 2010). For this reason, we proceeded setting four different case studies using datasets of increasing order of heterogeneity.…”

“…Brondi et al (2015) used the Pd and the Integral of squared Velocity, IV 2 to predict the PGV and the Housner Intensity, I H . Spallarossa et al (2019) and Iaccarino et al (2020) explored the use of Pd and the IV 2 and for predicting PGV and the Response Spectra of Acceleration, RSA, amplitudes at nine periods, respectively, using a mixedeffect regression approach aiming to account for site-effects.…”

Local and Moment Magnitude Analysis in the Ridgecrest Region, California: Impact on Interevent Ground-Motion Variability

“…The decomposition of the ground motion residuals into source, path and site specific terms is of interest for a wide spectrum of applications (Al Atik and Youngs, 2014;Baltay et al, 2017;, such as the develpment of partially-or non-ergodic ground motion models for probabilistic hazard assessment, e.g. (Anderson and Brune, 1999;Atkinson, 2006;Rodriguez-Marek et al, 2013;Villani and Abrahamson, 2015;Kuehn and Scherbaum, 2016;Sahakian et al, 2019;Kotha et al, 2020); on-site early warning applications (Spallarossa et al, 2019;Iaccarino et al, 2020); characterization of source process such as monitoring the preparation phase of earthquake nucleation (Piña-Valdès et al, 2 2018;Picozzi et al, 2019) and fault healing (Bindi et al, 2018). Among the different components, the distribution of event-specific deviations from the median predictions (the so-called inter-event or between-event residuals) are connected to source processes which are only partially captured, or not described at all, by the explanatory variables used to develop the ground motion model.…”

A deep neural network framework for real‐time on‐site estimation of acceleration response spectra of seismic ground motions

“…An accurate estimation of the groundmotion S a (T) values in real time can help stakeholders better design engineering applications of EEW, including those relying on risk-informed decision support systems (Cremen & Galasso, 2021). Recently, Iaccarino et al (2020) and Jozinović et al (2020) proposed data-driven methods (such as mixed-effect model and convolutional neural networks, respectively) to obtain S a (T) values for few periods using initial seconds of the arriving ground motions; however, based on the results presented in their work, the frameworks' accuracy and robustness to work efficiently in real time could be further improved.…”