Deforestation controls landslide susceptibility in Far-Western Nepal

Landslides are a major problem in the Far-Western region of Nepal, where one of the highest densities of shallow landslides within the country is observed. However, there is still a lack of understanding around the relative impact of deforestation as a predisposing factor. Here, we use remote-sensing techniques and freely accessible satellite data to quantify the effects of deforestation and related agriculture practices in the assessment of landslide susceptibility. We use a new regional landslide inventory of 8,778 events dated between 1993 to 2018 to train and validate bivariate landslide susceptibility models, in which preceding land cover/land use is included as a predisposition factor. Our results show that the use of land-cover /land-use maps enhances the average probability of model success by 7.9%. Besides, we assess land-cover changes for a period of 1 to 10 years prior to each landslide event to analyze the relative incidence of changes related to deforestation and agricultural uses. The analysis suggests that agricultural practices and deforestation that occur 5-7 years preceding the landslide event have a significant influence on the landslide occurrence (+16%). Our results reveal that negligent anthropological land use practices can significantly enhance landslide risk.