Testing the Significance of Vegetational Parameters on Shallow Landslide Occurrence: For Landslides triggered in August 2023 during the Extreme Weather Event Hans

Landslides are widespread geohazards that have consistently impacted infrastructure and human lives in mountainous regions (Kirschbaum et al., 2015; Petley, 2012). In Norway, landslides are mainly triggered by high-intensity and/or long-duration rainfall events (Høeg et al., 2014). The extreme rainstorm Hans in August 2023 was one of these events that triggered shallow landslides in Southern Norway. With climate change expected to increase the frequency of extreme rainfall events (Høeg et al., 2014), predicting the likelihood of landslides and developing mitigation strategies is needed. While existing studies often focus on geological, climatic and meteorological factors, this thesis aims to assess the often-overlooked role of vegetation in shallow landslide occurrence on a regional scale. In this study, I used the Random Forest method to assess the role of vegetation, alongside geological, climatic, and meteorological factors, in shallow landslides triggered by Storm Hans. To address variations in the environmental factors across the study area, and assess the importance of space in landslide probability models, spatial autocorrelation was incorporated into the modelling process by using both non-spatial and spatial Random Forest models. Unusually high accumulated rainfall, elevation, and the lack of underground biomass were found to be important factors influencing landslide occurrence during Storm Hans. Forested terrain had higher rainfall thresholds before landslides occurred, leading to a lower landslide probability in forests than in non-forested areas. Furthermore, model performance was higher for the spatial Random Forest models compared to the non-spatial Random Forest models The results suggest that incorporating vegetational factors can improve landslide occurrence assessment. Further research is needed to improve model performance and predicting capacity, for example by using high-quality terrain and vegetation data, and including soil characteristics. Regional variation plays an important role in predicting landslides in the Hans study area. Localized landslide susceptibility models could prepare stakeholders to mitigate the consequences like those caused by Storm Hans, thereby improving resilience to future extreme weather events and reducing the risk that shallow landslides pose to Norwegian society.

GE4-304