Evaluation of effects on a viaduct affected by a snowmelt-induced landslide using AI-Driven Pattern Recognition

Over recent years, the worldwide increase in infrastructure failures due to natural hazards such as landslides and earthquakes has raised significant concern. Aggravated by the effects of climate change, this issue has had a considerable impact, particularly in Italy. This study focuses on the "Villa Ilii" Viaduct on the A24 highway to analyze landslide-structure interactions, emphasizing the significant deformation risks. The methodology includes comprehensive geological, morphometric, and geophysical analyses, with findings showing cumulative deformation rates up to +2.5 cm/year at critical points. Field surveys and high-resolution SAR data analysis have identified deformation patterns linked to seasonal snow melt and temperature variations, adding a quantitative dimension to the hazard assessment. A novel component of this study is the application of Gradient Boosting Machines (GBM) for automated pattern recognition within SAR data processed using the Differential Interferometry SAR (DInSAR) technique. By leveraging datasets on snow melting patterns and cumulative radar target deformation, GBMs uncover intricate, non-linear relationships between snowmelt dynamics and ground deformation. This integration of GBMs and DInSAR data advances landslide hazard assessment by providing a robust 87% accuracy predictive model, effectively capturing environmental triggers that impact infrastructure stability. These findings contribute to more effective hazard management strategies and inform resilient infrastructure design, highlighting the study's significance in addressing climate-driven geohazards.