Evaluating the interaction between snowmelt runoff and road in the occurrence of hillslope instabilities affecting a landslide-prone mountain basin: A multi-modeling approach

Abstract

• Roads influence the occurrence of land degradation processes in mountain landscapes. • Snowpack melting alters superficial water dynamics along mountain roads. • Novel multi-modeling approach can predict slope instabilities due to roads. • Advance DEM post-elaboration leads to detect road’s role in altering snowmelt runoff. The presence of roads in steep slope mountain areas is often linked with the occurrence of landslides and erosive dynamics. The use of Airborne Laser Scanning (ALS)-derived high-resolution topographic data increased the possibilities to better represent landscapes and related physical processes at the watershed scale. Additionally, the adoption of topographically-based hydrological models allows to simulating water overland flows and investigate the occurrence of specific degradative phenomena. Snowpack melting plays a primary role in altering superficial water dynamics in mountain landscapes, but accurate investigation of the interaction between snowmelt runoff and human infrastructures (such as roads and trails) in the occurrence of hillslope failures is still obscure. This research aims to assess the relationship between snowmelt runoff, road presence and terrain instabilities affecting a landslide-prone steep slope mountain meadow (northern Italy). An innovative multi-modeling approach was proposed to detect the alteration of snowmelt overflows due to the presence of the road and its relation with the activation of a shallow landslide. The road role in altering snowmelt runoff was investigated both considering its presence and assuming its absence. Different hydrological and slope stability models were interactively considered, starting from pre-event ALS-derived DEM to compute predictive basin-scale simulations. Results attested the key role played by the road in altering snowmelt runoff pathways, as well as their combined contribution to the foreseen activation of the observed shallow landslide. Starting from on-field observations conducted after the landslide triggering, the accuracy of instabilities predictions was finally tested through the statistical computation of the Area Under the Receiver Operating Characteristic curve (AUC-ROC) and the Cohen’s kappa -index, respectively resulted around 0.9 and 0.6. This work could be a useful tool for planning mitigation interventions able to reduce the occurrence of similar risk scenarios, also providing suggestions for developing and promoting efficient sustainable actions for mountain landscapes.