Knowing seismic wave propagation and ground motion is critical for urban hazard mitigation and urban planning in metropolitan areas in active seismic zones worldwide. Estimates of the expected ground motion are essential for designing, assessing, and decommissioning different infrastructures in urban and rural areas. An essential step when reassessing the seismic hazard is estimating the ground shaking level generated by strong earthquakes. One method for estimating the ground motion intensities is through equations/models based on strong ground motion recorded during previous earthquakes. To build up robust GMMs for accurate prediction of the ground motion, it is recommended to use a large strong-motion dataset that contains the entire ranges of small-to-large magnitudes, distances from the source to the site, and local conditions. However, the available datasets used to build GMMs are limited since they do not cover the entire magnitude range and usually sample sparsely observed data near the source.
Nevertheless, Abrahamson et al. (2019) showed that the expanded number of strong-motion records over the past decade exhibits noteworthy differences in the scaling of the ground motions within relatively small provinces. Most of the variability generally treated as aleatory is actually due to systematic source, path, and site effects. Kuehn et al. (2019) showed the importance of capturing variations in quality factor (Q) over small spatial scales (30 km) in California. It showed that accounting for path effects leads to less aleatory variability and results in different median predictions, counting on source and site location. The innovative approach to this purpose is to relax the ergodic hypothesis. A fully non-ergodic system is feasible to determine the systematic contributions of variability into the event-, source-, site- and path- effects through a statistical decomposition technique wherever numerous records from multiple earthquakes are available at each station (Kotha et al. 2020; Sgobba et al. 2021).
On the other hand, the recent development of numerical simulations of earthquakes based on the physics of the causative source rupture and wave propagation has contributed to considerable progress in predicting the variability of ground motion, especially at the near source, affected by the source and wave propagation effects. For example, Graves et al. (2011) showed that the combination of rupture directivity and basin response effects could increase hazards in particular sites relative to that calculated by GMMs. Pitarka et al. (2022) found that rupture propagation effects with the amplification due to local topography can result in large ground motion amplifications with complex spatial variability. Advances in 3-D wave-propagation models and high-performance computing have enabled the capability to generate ground-motion simulations for almost all plausible scenarios (e.g., event magnitude, faulting mechanism, site conditions, geology structure, etc.) and provide site-specific input for infrastructure analysis (Paolucci et al. 2014). Therefore, GMMs could count on synthetic earthquake ground motion data for noteworthy extrapolation in the near-source boundary and places with incomplete observations. A hybrid GMM, based on empirical and synthetic ground motion databases, is expected to better constrain the predictions, especially in the epicentral area, thus improving the evaluation of the epistemic uncertainties and the aleatory variabilities (Paolucci et al. 2021; Infantino et al. 2021).
This project will examine the use of physics-based numerical models and a non-ergodic ground-motion prediction in the Central Apennines. We will particularly focus on three target areas; Amatrice-Visso-Norcia (AVN), L’Aquila, and Fucino basin constructing detailed crustal velocity models (including the complex earth structures (basins, topography), earthquake rupturing, and the impact of the physical state of media on wave propagation) to generate broad-band ground motion simulations (Figure 1). We have selected these target areas because:
a) The study area offers vast datasets collected during the L’Aquila 2009 and AVN 2016-2017 earthquake sequences that may help us verify and test our ground motion estimations. For example, the unprecedented density of near-fault seismic stations and the high quality of the recorded data made the 30 October 2016 Norcia event one of the best-recorded earthquakes in Italy. The highest PGA observed in the earthquake’s epicentral area is the largest recorded during an Italian earthquake. In addition, the 3D geological structure of the seismogenic volume of the epicentral area of the AVN 2016-2017 earthquake sequence was defined during the RETRACE-3D Project (centRal Italy EarThquakes integRAted Crustal modEl, Di Bucci et al. 2021) offering thus the possibility of knowing in detail its velocity model.
b) The 1915 Avezzano earthquake (Ms7.0, according to Margottini et al. 1993), which struck the Fucino Plain (central Italy), is one of the major seismic events in Italy over the last few centuries. However, the instrumental recordings are missing, while the macroseismic intensities (Molin et al. 1999) are well documented. There is no consensus on the epicentral location while its surface rupture is well-mapped, and overall the deep geometry of the seismogenic source is well-constrained. The Fucino Plain, one of the major basins of the central Apennines, is filled by more than 1000 m of sediments and surrounded by mountains. This circumstance poses the fundamental problem of relating the presence of the seismic fault with soft sedimentary basins and high topography, thus enhancing the hazard. High-quality datasets of P and S wave travel times will be derived from the temporary stations deployed on the alluvial plain during the 2008-2009 and 2012 DPC (Italian Civil Defence Department) projects dedicated to site effect studies (Cara et al. 2011; Famiani et al. 2015). These new data, used jointly with the available dataset recorded by the operating seismic stations to monitor the seismic activity, will allow local earthquake tomography investigations to be performed with a sensible increase of resolution relative to the published velocity models. For the Fucino area, our investigation will be carried out in collaboration with the ongoing project “Pianeta Dinamico”, Task (S2) aimed at the reconstruction of the crustal structure of Italy by the joint use of seismic, magnetic, and gravimetric investigations.