PSI - Issue 84

Alessandro Brunetti et al. / Procedia Structural Integrity 84 (2026) 97–102

98

Keywords: Infrastructure resilience; landslide hazard; EO data; AI.

1. Introduction Infrastructure systems worldwide are becoming increasingly vulnerable to natural hazards, particularly landslides, resulting in significant economic and social impacts. Prior studies estimate that around 0.5% of global assets are exposed to such hazards annually (Koks et al., 2019). These threats often disrupt vital services such as transport and logistics, emphasizing the need for resilient infrastructure planning. In the literature, several frameworks have been developed to support multi-hazard risk management, particularly in relation to linear assets. These approaches typically aim to develop integrated and quantitative frameworks capable of modelling multi-hazard scenarios, infrastructure vulnerability, and resilience. This type of analysis is inherently multidisciplinary and typically requires high-resolution input data, including detailed fragility curves and comprehensive ancillary datasets (Argyroudis et al., 2019, 2020). To overcome the challenges associated with data availability, Joshi et al. (2024) and Arvin et al. (2023) adopted index-based methodologies, offering a more qualitative approach that emphasizes the exposure and vulnerability components of risk rather than detailed hazard modelling. In this context, SGAM (Smart Geotechnical Asset Management) framework was introduced as a semi-automated decision support system integrating EO data, geohazard databases, morphometric data, geotechnical monitoring, and data fusion algorithms (Di Renzo et al., 2024). The original SGAM methodology, laid the groundwork for a multi hazard approach to infrastructure risk analysis. This paper advances that framework by implementing the standardized Italian guidelines for landslide susceptibility assessment of bridges (“Italian guidelines” in what follows) (https://cslp.mit.gov.it/circolari-e-linee-guida/linee-guida la-classificazione-e-gestione-del-rischio-la-valutazione-della), introducing the systematic P Factor calculation methodology, improving the integration of InSAR-derived movement data with hazard assessments, and streamlining the generation of prioritized summary layers for infrastructure management. 2. Methodology The present study builds upon the SGAM framework previously introduced in Di Renzo et al., 2024, refining its methodology for the hazard assessment of linear infrastructure through integration with the Italian guidelines. In this paper, we present methodological advancements with specific focus on the landslide hazard assessment. This version implements a three-phase approach for landslide hazard assessment, aiming to improve standardization, and accuracy at the asset level. While the geodatabase architecture and structure have already been described in detail in Di Renzo et al., 2024, here it is referenced as a resource for hazard data management with extensions to support Italian guidelines compliance (Figure 1). SGAM employs a multi-component workflow, integrating ground motion data from satellite InSAR with thematic layers (e.g., topography, geology, land use, landslide maps and databases) through supervised learning algorithms and systematic parameter calculation. The spatialized outputs are then segmented and intersected with infrastructure elements to enable the classification of asset segments into risk levels. Key enhancements include the implementation of the P Factor methodology incorporating activity state assessment (P a /P c ), velocity-based scoring (P v ), and magnitude evaluation (P m ), followed by model reliability and mitigation measures assessment for final CdA (Attention Class) classification. This development enables direct application of landslide hazard assessment to linear infrastructure, providing risk informed prioritization of intervention areas along the road network.