PSI - Issue 84

N. Kheirkhahan et al. / Procedia Structural Integrity 84 (2026) 33–40 N. Kheirkhahan et al./ Structural Integrity Procedia 00 (2026) 000–000

35 3

network are assessed through comparison with a random Erdős –Rényi network to analyze if it is a random network or a small world network (Erdős et al., 1959; Newman et al., 2003). 3. Multi-hazard analysis Given that most of the Italian territory is highly prone to natural hazards, this study focuses on those posing significant threats to road infrastructure, namely earthquakes, floods, landslides, and tsunamis. The proposed multi hazard analysis follows the framework for critical infrastructures introduced by Arduin et al., 2025, which integrates GIS-based methods with a Multi-Criteria Evaluation (MCE) approach. First, the data related to the hazards were structured and organized using Quantum GIS software (QGIS), as detailed in Table 1. The flood hazard dataset classifies areas into low, medium, and high flood-affected probability. The landslide hazard dataset identifies five classes: attention areas, moderate, medium, high, and very high. The tsunami dataset identifies areas potentially exposed to tsunami inundation. For earthquake hazard assessment, the seismic hazard parameters (Meletti et al., 2006), the average shear-wave velocity in the top 30 meters (VS30; Mori et al., 2020), and a 20-meter resolution Digital Elevation Model (DEM; MASE, WCS) were used to derive a map of horizontal accelerations accounting for topographic and stratigraphic amplification, including the spectral acceleration amplification factor. Table 1. Data sources, version and reference for earthquake, flood, landslide and tsunami hazard assessment. In the table, the seismic hazard parameters are a g (maximum acceleration on rigid soil ) and F o (spectral acceleration amplification). Hazard Parameter Version/Year Reference Flood Probability of flood (low, medium, and high) 2020 ISPRA, Idrogeo Landslide Hazard classes (attention areas, moderate, medium, high, and very high) 2024 ISPRA, Idrogeo Tsunami Area exposed to tsunami 2017 ISPRA Tsunami Then, each individual hazard map was spatially intersected in QGIS with the road network layer of Messina, and a discrete single-hazard value x h,e (based on five levels, spanning from 0-absence to 3-high) was assigned to each edge of the network (Fig. 1) according to the hazard classification adopted for each map, as summarized in Table 2. Considering earthquakes, hazard values are predominantly equal to 2, with only 12 edges at 1 and 87 at 3. For floods, 641 edges have a hazard value of 3, due to the overlap of low-, medium-, and high-probability flood areas, while all others are 0. For landslides, 198 edges have a value of 3, 43 a value of 2, 1570 a value of 1, and the remaining edges are unaffected by landslides. Finally, 2,570 edges lie in tsunami-affected areas. Based on the derived single-hazard values, considering the assumption of independence of events, the MCE (Malczewski and Rinner, 2015) analysis is performed, and the multi-hazard value for each edge of the network MH e is derived by adopting a Weighted Linear Combination (Modica et al., 2025) following the equation: Earthquake a g ; F o Vs30 DEM 2006 2020 2001 Meletti et al., 2006 Mori et al., 2020 MASE, WCS

4

1 h MH w x = =  e

(1)

, h h e

where e is the e -th edge of the network, and w h is the weight assigned to the h -th hazard considered. The weights are assigned a value of 0.25 for each hazard based on the authors’ expertise, assuming that, in terms of road network functionality, any of the considered events results in the loss of edge functionality. As shown in Fig. 2, many edges (3012, approximately 41%) have a multi-hazard value at most 0.5, indicating low overall hazard level. 3899 edges exhibit moderate values between 0.75 and 1.75, reflecting intermediate exposure hazard levels from one or more hazards. Only 331 edges (around 4.5%) show high multi-hazard values between 2 and 2.75, highlighting localized areas where multiple hazards contribute significantly to the combined hazard score.