PSI - Issue 84

Francesca Ceccato et al. / Procedia Structural Integrity 84 (2026) 599–606

602

The RF model produced landslide susceptibility scores ranging from 0 to 100, which were classified into five susceptibility levels: Low (0–20%), Moderate (20–40%), Moderate–High (40–60%), High (60–80%), and Very High (>80%). The final susceptibility map was generated using ArcGIS Pro. Further details on the methodology can be found in Qadri and Ceccato (2024). This study assesses landslide susceptibility and exposure of the bridge infrastructure within the Veneto region (Italy). The interaction between landslide susceptibility and infrastructure was quantified through a Combined Infrastructure Landslide Susceptibility Index (CILSI), accounting for susceptibility in the area of influence of the bridge (Figure 1). Landslide hazard information was derived from a raster-based susceptibility map with values ranging from 0 to 100, representing relative landslide likelihood. All datasets were harmonized spatially and clipped to the study area. Landslide susceptibility (LS) was normalized to the [0, 1] interval (Figure 1a) to preserve its continuous nature and avoid information loss associated with early categorical classification.

Fig. 1. (a) Landslide Susceptibility at the pixel scale; (b) Infrastructure spatial index; (c) CILSI given by the composition of LS and ISI.

Bridge locations, represented as points, were rasterized and processed using a circular kernel density function with an 800 m influence radius, representing the area of interest of the bridge infrastructure (Figure 1b). The resulting bridge density surface was min–max normalized to produce the Infrastructure Spatial Index (ISI) on a [0, 1] scale. In this study, the kernel function is constant over the circle, but different functions can be used in the future to account, for example, a decreasing density with the increasing distance from the infrastructure. A Combined Infrastructure Landslide Susceptibility Index (CILSI) was computed at the pixel level as the multiplicative interaction between landslide susceptibility (LS) and infrastructure Spatial Index (ISI) by following relation (Figure 1c): CILSI = LS×ISI (1) The final CILSI of the bridge is the average CILSI over the zone of influence of the bridge and it represents the weighted landslide hazard around the bridge. Average landslide susceptibility around bridges was evaluated using a buffer-based approach. For each bridge, the mean susceptibility was calculated within an 800 m zone of influence as a circular buffer and uniformly assigned to the entire buffer area, creating a representative exposure surface. Where zone of influence overlapped, values were combined using their average, resulting in a spatially explicit raster of mean susceptibility conditions surrounding bridge infrastructure. The size of the buffer zone is an assumption adopted in this study for regional-scale screening, future studies will address the optimization of this value. For representation, the CILSI of the bridge was classified into five risk categories: Low, Moderate, Moderate–High, High, and Very High, using progressively increasing thresholds to support spatial comparison and decision-making. Bridge counts per susceptibility range and corresponding class were derived directly from bridge attributes, ensuring each bridge contributed exactly once to the statistical summaries. The methodology produces continuous and classified landslide susceptibility maps and tabular summaries, providing a reproducible and robust framework for regional scale assessment of landslide risk to bridge infrastructure.