Search

The impact of anthropogenic disturbance on landslide occurrence has been clearly demonstrated for several case-studies worldwide. Deforestation has been identified as the main trigger for landslide activity. These studies are mainly based on landslide inventories from aerial photographs or remote sensing data, and often focus solely on the total number of landslides. Looking at the total number of landslides alone does not allow us to get a comprehensive insight into the overall anthropogenic impact on landslide denudation. In this study, we analysed the anthropogenic impact on landslide frequency-area distributions. Three small catchments of about 11-30 km2 were selected within the Ecuadorian Andes. They have a similar topographic setting but differ in their land cover dynamic. Our data show that human disturbances significantly alter the landslide frequency-area distributions. Human-induced land cover change is associated with an increase of the total number of landslides and a shift of the frequency-area distribution towards smaller landslides. However, the frequency of large landslides is not affected by anthropogenic disturbances. When analysing the geomorphic work realised by landslides in different environments, it becomes clear that the majority of landslide-induced sediment is coming from anthropogenic environments. The strong increase in the frequency of small landslides in anthropogenic environments thus plays an important role in enhancing the overall landslide denudation rate in tropical mountain regions.

Tropical mountain regions are prone to landslide hazards. Given the current land pressure with increasing occupation of steep uplands, landslide hazards are expected to increase in the near future. Understanding the factors that control landslide hazards is therefore essential. Rare event logistic regression allows us to perform a robust detection of landslide controlling factors. This technique is here applied to the tropical Andes to evaluate the impact of dynamic land cover changes on landslide occurrences. Land cover change trajectories (i.e. dynamic evolution of land cover through time) were specifically included in the probabilistic landslide analysis. While natural physical processes such as slope undercutting by rivers and failure of oversteepened slopes are important in this tropical mountainous site, landslides are increasingly associated with human activities. The data show that land cover trajectories are associated with landslide patterns. In this humid mountainous site, forest degradation does not lead to a measurable increase in landslide occurrence. However, few years after forests are converted to pastures, a rapid decline of slope stability is observed. Land cover conversion from forest to pasture permanently reduces slope stability. It is assumed that major changes in soil properties and hydrology induced by the vegetation conversion play a role in accelerating landslide hazards.

Many tropical mountain regions experience rapid land cover change and natural hazards like landslides. In this work, landslide controlling factors in two Andean catchments with different land cover dynamics were analysed using rare event logistic regression with replications. Our results show that topographical factors alone cannot explain the observed landsliding pattern. We suggest that major changes in soil properties and hydrology after deforestation play a role in accelerating landslide activity.

Statistical analysis of natural hazards needs particular attention, as most of these phenomena are rare events. This study shows that the ordinary rare event logistic regression, as it is now commonly used in geomorphologic studies, does not always lead to a robust detection of controlling factors, as the results can be strongly sample-dependent. In this paper, we introduce some concepts of Monte Carlo simulations in rare event logistic regression. This technique, so-called rare event logistic regression with replications, combines the strength of probabilistic and statistical methods, and allows overcoming some of the limitations of previous developments through robust variable selection. This technique was here developed for the analyses of landslide controlling factors, but the concept is widely applicable for statistical analyses of natural hazards.