Item

Abstract

On the eve of the new year, 2021, a single landslide claimed 70 souls in Ask, a village in Norway. This tragic event highlighted, once again, the need to understand whether research efforts to map landslide susceptible areas could help save lives and if these identified landslide-prone regions change with time. A landslide is a downslope gravitational mass wasting of earth materials. Hence, a classification model could estimate the likelihood of a landslide occurring under certain terrain conditions studying the landslide predisposing factors, such as hillslope inclination and land cover, of old landslides. Projecting these likelihoods in a landscape would be a landslide susceptibility map, which highlights areas that could potentially generate a landslide without any implication of an occurrence time. However, landslide predisposing factors change over time, resetting those susceptibility estimates—they are not static as traditionally assumed by most models. These changes could be evident, such as artificial alterations in land cover, or disguised, such as accumulated damage on hillslopes in the form of subsurface cracks due to a large earthquake. In times referred to as legacy effects, those latter hidden effects could be assessed by studying the spatial distribution of those landslides triggered by the same event. This perspective lists several potential biases of the time-invariant landslide susceptibility approach and offers hints to overcome these challenges using a more dynamic model that evolves.