Neo-deterministic seismic hazard assessment of Corsica-Sardinia block

The Corsica-Sardinia lithospheric block is commonly considered as a region of very low seismicity and the scarce reported seismicity for the area has till now precluded the reliable assessment of its seismic hazard. The time-honored assumption has been recently questioned and the historical seismicity of Sardinia has been reevaluated. Even more, several seismogenic nodes capable of M5 + have been recognized in the Corsica-Sardinia block exploiting the morphostructural zonation technique, calibrated to earlier results obtained for the Iberian peninsula, which has structural lithospheric affinities with the Corsica-Sardinia block. All this allows now for the computation of reliable earthquake hazard maps at bedrock conditions exploiting the power of Neo Deterministic Seismic Hazard Assessment (NDSHA) evaluation. NDSHA relies upon the fundamental physics of wave generation and propagation in complex geologic structures and generates realistic time series from which several earthquake ground motion parameters can be readily extracted. NDSHA exploits in an optimized way all the available knowledge about lithospheric mechanical parameters, seismic history, seismogenic zones and nodes. In accordance with continuum mechanics, the tensor nature of earthquake ground motion is preserved computing realistic signals using structural models obtained by tomographic inversion and earthquake source information readily available in literature. The way to this approach has been open by studies focused on continental Italy and Sicily, where the agreement between hazard maps obtained using seismogenic zones, informed by earthquake catalog data, and the maps obtained using only seismogenic nodes are very good.


Introduction
In contrast to the neighboring Apennine Peninsula, the seismic history of the Corsica-Sardinia region, reported by earthquake catalogs, is characterized by sporadic and weak shocks, magnitudes of which, as a rule, do not exceed 5 (Rovida et al. , 2021Postpischl 1985). Meletti et al (2021) concluded that Sardinia island is "not earthquakeless" and contradict the time-honored assumption. The recent instrumental observations evidence the ongoing low seismic activity in the region: specifically, during 2020 the INGV seismic network recorded four small shocks in Sardinia (http:// terre moti. ingv. it). The very scarce seismicity reported in official documents (parametric catalogs and historical records, Table 2) prevented till now a reliable assessment of earthquake hazard of the single Sardinia-Corsica lithospheric block, characterized by a common geodynamic evolution (Minzoni 1991). In a recent paper Gorshkov et al. (2021) evaluated the seismic potential of the region employing the morphostructural zoning (MSZ) that allows for the identification of the location of potential earthquakes independently from the seismic history just relying on the systematic and quantitative analysis of the available geological and morphostructural data. The MSZ delineates the hierarchical system of morphostructural blocks, the network of lineaments bounding the blocks, and the loci of nodes formed around intersections of lineaments. Nodes are treated as earthquake controlling structures (Gorshkov et al. 2003). Their seismic potential is evaluated exploiting the existing criteria of seismicity defined earlier for seismogenic (M5 +) nodes in Iberia, whose structural setting is grossly similar to the one of the Corsica-Sardinia block Panza et al. 2007;Carminati et al. 2012). 24 out of the 81 nodes delineated by MSZ meet the characteristic features (criteria of seismicity) of the seismogenic nodes prone to M5 + that have been defined by pattern recognition in the Iberian Peninsula (Gorshkov et al. 2010). The study by Gorshkov et al. (2021) suggests that the seismic potential of the Corsica-Sardinia block is not low and provides the key information for long-term reliable seismic hazard assessment of the region and motivates the application of the Neo Deterministic Seismic Hazard Assessment (NDSHA) to reliably evaluate the earthquake hazard of the study area. NDSHA (Panza et al. 2001a, b, 2012Panza and Bela 2020;Bela and Panza 2021;Panza et al. 2021) is a now well established technique based on the physical description of the earthquake rupture process and of the seismic propagation paths aimed to obtain synthetic seismograms from which reliable ground motion parameters relevant to engineering applications can be extracted even when the available records are very poor. NDSHA exploits all the known seismicity and other potential seismic sources and the available knowledge of the Earth structure (i.e. the propagation medium, the ray-path), preserving the tensor nature of earthquake ground motion (Aki and Richards 2002). While ground motion parameters at the regional scale are given, as a rule, for bedrock conditions, more detailed studies can be performed at specific sites, taking into account the nature of local soil and the rupture complexity at the source. We refer the reader to previous works for a detailed description of the theoretical and computational workflow of NDSHA (e.g. Panza et al. 2001aPanza et al. , b, 2012Panza and Bela 2020;Bela and Panza 2021;Panza et al. 2021), while in the following we describe the input data and the procedural steps applied in this study.

Methodology and data
No seismogenic zones (Meletti et al. 2008) have been defined for the study area, thus we use only the seismogenic nodes identified by MSZ as seismic sources (Fig. 1), such approach being proven valid by Rugarli et al. (2019) for continental Italy and Sicily. The nodes are prone to M5 + earthquakes (Gorshkov et al. 2021), thus we assign M W = 5 to all sources that are uniformly distributed within the nodes of 25 km of radius ( Fig. 2), at a fixed depth of 10 km.
Very little information about the focal mechanisms of earthquake sources in the study area is available and the few mechanisms that have been determined mostly fall outside the nodes. Therefore, we associate to each node a representative average focal mechanism determined by Pondrelli et al. (2020) for the Sardinia-Corsica Block. Of this mechanism we retain dip and rake angles, while the strike is chosen according to the orientation of the higher rank lineament intercepted by each node (Fig. 2). Anyway, using the average strike for all the nodes brings typically consistent results, so the latter will not be presented in the following section.
The regional structure that is used as propagation medium is described by a set of 1D an elastic seismic cellular models ( Fig. 3 and Table 2) with a resolution of 1° × 1° (Brandmayr et al. 2010;Brandmayr 2012) obtained from surface wave tomography down to 350 of depth, while a standard reference structure (Kennett and Engdahl 1991) was used at larger depths.
Regional hazard maps are computed using the Size Scaled Point Source model (SSPS, after Parvez et al. 2011) approximation. The Design Ground Acceleration (DGA) is obtained considering appropriate response spectra (Panza et al. 2001a, b), specifically the Eurocode 8 (CEN 2004) response spectrum for soil type A.

Results and discussion
Vertical displacement map show widespread low values on both Corsica and Sardinia (< 0.1 cm). Similarly, vertical velocities are everywhere < 0.5 cm/s and vertical DGAs are < 0.005 g (Fig. 4, top). Horizontal components are slightly higher: horizontal displacement tops 0.5 cm in most of Corsica and eastern Sardinia; horizontal velocities reach 1 cm/s on the western coast of Corsica and the Cagliari area; horizontal DGAs reach 0.01 g in Corsica and eastern Sardinia (Fig. 4, bottom).
These values are consistent with I = V (MCS, applying the regression after Panza et al. (2001a, b) reported for most of the historical events felt in the area (Meletti et al. 2021). However, Meletti et al. (2021) also report I = VI caused by the 1948-1949 seismic sequence with epicentral area in NW Sardinia.
M W = 5 is the lower bound of the magnitude threshold identified by the pattern recognition methodology. Thus, following Rugarli et al. (2019), in a second run we increase the source magnitude to M W = 5.5, applying an increment defined as ΔM = γ EM σ M , where γ EM (here set to 2.0) is a Maps obtained in such a way (second run) show vertical displacement up to 0.5 cm in the whole study area, vertical velocities up to 1 cm/s in most of Corsica and interior of Sardinia and vertical DGAs up to 0.02 g in SW Corsica and, locally, in western Sardinia (Fig. 5, top).
Horizontal displacement reaches 2 cm in central-western Corsica; horizontal velocities reach 4 cm/s in W Corsica and Cagliari area, while are generally over 1 cm/s in most of the study area; horizontal DGAs can be as high as 0.04 g in most of the study area excepted W coast of Sardinia (Fig. 5, bottom). These values correspond to I = VII on the MCS scale (Panza et al. 2001a, b).
In our opinion, the second run supplies a reliable and conservative description of the actual seismic hazard of Corsica-Sardinia block, also on account of the scarce empirical knowledge of the seismicity of the area, including major uncertainties on the focal depth and mechanisms of the earthquakes.
Rugarli et al (2019) applied M design to the rest of the Italian territory considering only seismogenic nodes and obtained hazard maps in agreement with those based on seismogenic zones and the CPTI04 catalog. Brandmayr et al. (2021) showed that the application of M design (at 1 Hz cut-off frequency) in Kosovo gives hazard levels consistent with 10 Hz cut-off frequency computation, which needs a knowledge of sources and structures with a detail not available for the study area. Therefore, we consider the ground motion maps obtained applying M design the most, now possible, reliable advancement in the understanding of the seismic hazard of the area. M design is a readily available tool for assessing and mitigating seismic risk, especially where critical infrastructures (e.g. rad-waste surface disposal facility, natural-gas pipelines, etc.) may be planned.

Conclusions
NDSHA relies upon the fundamental physics of wave generation and propagation in complex geologic structures to generate realistic synthetic seismograms, used as input for the computation of several ground motion parameters, integrating all the available knowledge about seismic sources and medium structure. Seismogenic nodes identified in the Corsica-Sardinia block by Gorshkov et al. (2021) supply information about the seismic potential of the study area filling the gap due to the scarce and poorly known seismic history of the area and allow for the assessment of seismic hazard using NDSHA.
Considering seismogenic nodes in Italy, Rugarli et al. (2019), showed that adding a safety increment ΔM is equivalent to applying a partial safety factor γ q to the seismic moment at the fault M 0 , as normally asked by Eurocodes and other international standards for typical structural design. Ground motion estimates so obtained envelope the ones obtained using the seismic catalog. Similarly, we suggest that the application of seismogenic nodes and M design = 5.5 to Corsica-Sardinia block provides the most reliable assessment of seismic hazard of the area so far.
In particular, we show that large portions of Corsica and Sardinia could experience ground motion levels compatible with I = VII on the MCS macroseismic intensity scale. Such macroseismic intensity envelopes the values felt and documented so far, pointing out that some further actions of prevention and mitigation of seismic risk should be enforced, especially in the siting of critical infrastructures.  Coordinates refer to the center of 1° × 1° cells. Q S after Martinez et al. (2009Martinez et al. ( , 2010. Q P = 2.2*Q S