Hybrid broadband strong-motion simulation to investigate the near-source characteristics of the M6.5, 30 October 2016 Norcia, Italy earthquake

Highlights

• We used a substantial data set and physics-based simulation to study the ground-motion features of the Norcia earthquake.

• A stochastic high-frequency method is used to generate -EW, NS, UD- seismograms instead of one generic horizontal component.

• We showed the value of choosing the region-specific source, attenuation and site factors for the ground-motion simulations.

Abstract

During the 2016–2017 Central Italy earthquake sequence, a series of moderate to large earthquakes M > 5 occurred near the Amatrice and Norcia towns. These events are recorded on a dense seismic network, providing relevant observational evidence of complex earthquakes in time and space. In this work, we used this substantial data set to study the ground-motion characteristics of the Norcia earthquake M6.5 on October 30, 2016, through a broadband ground-motion simulation. Three-component broadband seismograms are generated to cover the entire frequency band of engineering interest. Low and high frequencies are computed considering the heterogeneous slip rupture model of Scognamiglio et al. (2018) [1]. High frequencies are calculated using a stochastic approach including P, SV, and SH waves, while low frequencies are obtained through a forward simulation of the kinematic model at the various stations. To predict earthquake-induced ground-motions in the area, we adopted region-specific attenuation and source scaling parameters derived by Malagnini et al. (2011) [2]. Ground-motion parameters, including peak ground acceleration (PGA), peak ground velocity (PGV) and spectral amplitudes, are calculated at the selected sites adopting physics-based parameters to understand better the earthquake fault rupture, the wave propagation, and their impacts on the seismic hazard assessment in the region. We showed that combining the fault rupture history over the entire frequency spectrum of engineering interest, the attenuation characteristics of the seismic wave propagation and the properly defined site responses can improve the prediction of ground-motions and time histories, especially in near seismic sources.

Introduction

The Central Apennines belt is one of the most seismically active regions in Italy, with the frequent occurrence of strong (M > 6.0) destructive earthquakes. The region was formed in the Miocene to Pliocene under the ongoing subduction environment where the Adriatic Plate collides with and plunges beneath the Eurasian Plate [3]. As a consequence of this process thrust belts in the Adriatic coast and extension structures in the axial sector of the belt produced a broad and complex normal faults system elongated in the NW-SE direction [4]. Recently, a series of damaging earthquakes, M6.0 on August 24, M5.9 on October 26, and M6.5 on October 30, 2016, occurred near the town of Amatrice and Norcia in the Central Apennines. The Norcia earthquake, the largest event of the sequence, occurred just a few tens of kilometers north-west of the L'Aquila (Abruzzo) area, which was hit by the M6.3 earthquake on April 6, 2009. These earthquakes have increased attention to seismic hazard, engineering applications, and earthquake risk mitigation efforts in the region.

For these reasons, in such areas, simulated ground-motions may help to understand the earthquake characteristics for both seismological and earthquake engineering purposes. Physically-based ground-motion simulation and verification are the keys and basic tasks for establishing its predictive capabilities in seismic risk assessment and seismic engineering applications. Moreover, the complete time series of ground-motion simulations from 0.02 Hz up to frequencies of 10 Hz are needed for Civil engineers to design earthquake-resistant structures and retrofit vulnerable existing structures (e.g. Ref. [5]).

Earthquakes generate broadband ground-motions, which include both low (f < 1 Hz) and high frequencies (f > 1 Hz). However, the whole three-components waveform simulations are still challenging, especially at frequencies larger than 1 Hz. Ground-motion simulations have been refined along with developments on mathematical and computational tools and detailed regional seismological models (e.g. Refs. [[6], [7], [8], [9]]). Many methods have been proposed to simulate realistic ground-motion, one of them is a hybrid method, which combines low-frequency seismograms and high-frequency ones, and is currently popular in recent studies (e.g. Refs. [[10], [11], [12], [13], [14]]).

The low-frequency part of the ground-motions is mostly simulated using deterministic and numerical methods in time or frequency domain (e.g. Refs. [[15], [16], [17], [18]]) and most cases are limited to 1 Hz due to the site effects and the resolution of the adopted velocity structure. Stochastic methods are often used to account for independent random phases of motion on higher frequency bands (f > 1 Hz) [19]. The most commonly used approach proposed by Boore [20] is the stochastic point-source modeling, which was further modified by Beresnev and Atkinson [21] for finite-fault effects, where each sub-fault is considered as a stochastic point source. The stochastic finite-fault modeling has been upgraded by adding the dynamic corner frequency [22]. Otarola and Ruiz [23] further improved the stochastic generation of seismograms for a stratified velocity model, including more physical parameters for simulations such as the incident and azimuthal angles, free surface factors, and energy partition for the P and SV waves. These approaches allow us to simulate three-component seismograms instead of one generic horizontal seismogram by using only SH waves. Herein, we concentrated on the Norcia M6.5 earthquake and generated broadband seismograms by adopting a hybrid method. Low-frequency seismograms are calculated using a non-negative, least-squares inversion method with simultaneous smoothing and damping implemented by Dreger et al. [24]. The high-frequency seismograms including P, SV and SH waves, were produced using the improved stochastic approach of Otarola and Ruiz [23] and Ruiz et al. [25]. In these works, an average radiation pattern derived from analytical expressions was used, as is usual in this type of simulation. Here, we calculated the radiation coefficients for each sub-fault according to the formulation proposed by Aki and Richards [26]. The generic rock soil amplification curves [27], together with the soil amplification transfer functions for P, SV, and SH waves are implemented to improve the simulation of the three-components of accelerograms on the ground surface. Finally, broadband synthetic time series were composed by merging the low and high-frequency seismograms with a variable crossover frequency around 0.5–0.8 Hz depending on the plateau level of the spectral amplitudes [28].

In this study, we adopted region-specific physics-based input parameters related to the source and wave propagation to accurately build the earthquake rupture scenario, estimating three-components realistic time series, absolute ground-motion level, frequency content signal duration, and spatial distribution of simulated ground-motions. The simulated seismograms at the selected stations are compared against the corresponding observed ones in terms of complete ground-motion time histories, Fourier amplitude spectra (FAS), and Pseudo-spectral accelerations (PSA) with a 5% damping ratio to test and to improve the efficiency and performance of our simulations. The performances of the hybrid broadband ground-motion simulations and the predictive capability of the adopted method are discussed in terms of residuals between simulations and observations, together with empirical ground-motion models in the proximity of the epicenter. The velocity-depth profiles are considered when available from the surface to the hard rock to determine the site responses by the soil amplification transfer function approximation, including fundamental resonant peaks.