Landslide Susceptibility Mapping Using Machine Learning: A Danish Case Study

Mapping of landslides, conducted in 2021 by the Geological Survey Denmark and Greenland (GEUS), revealed 3202 landslides Denmark, indicating that they might pose a bigger problem than previously acknowledged. Moreover, changing climate is assumed to have an impact on landslide occurrences future. The aim this study conduct first susceptibility mapping (LSM) reducing geographical bias existing LSM studies, identify areas prone future following representative concentration pathway RCP8.5, based set explanatory variables area interest located around Vejle Fjord, Jutland, Denmark. A subset from inventory provided GEUS used as ground truth data. Three well-established machine learning (ML) algorithms—Random Forest, Support Vector Machine, Logistic Regression—were trained classify data samples or non-landslide, treating ML task binary classification expressing results form probability order produce maps. were validated through test external for outside region interest. While high predictive performance varied slightly among three models data, LR SVM demonstrated inferior accuracy area. show RF model has robustness potential applicability low-lying landscapes present. can become step forward towards planning mitigative protective measures landslide-prone providing policy-makers with necessary decision support. However, map change scenario shows reduction susceptible areas, raising question choice analysis.