Landslide susceptibility maps – Dominica, 2025

Science Case Name Multi-hazards in the Caribbean SIDS Dataset Name/Title Landslide susceptibility maps – Dominica, 2025 Dataset Description Landslide susceptibility maps Key Methodologies As the basis for the landslide susceptibility assessment a data layer was generated that represents homogeneous terrain units. Initially an attempt was made to generate these automatically, using the r.slopeunits application, which generates Terrain Units using a set of parameters from a Digital Elevation Model. To compute the susceptibility, we opted for a terrain unit partition and for the implementation of statistical models, in combination with expert-based mapping. These models learn from past events (and specifically from past landslide occurrences) to find patterns with respect to a set of predisposing factors. On the basis of these patterns a prediction is then made on the expected unstable locations in the future. To assess the susceptibility, we used a statistical model known as binomial Generalized Linear Model or Logistic Regression. Temporal Domain The maps are generated for the current situation (2025). Spatial Domain The maps cover the country of Dominica. Key indicators Landslide susceptibility classes (Very low, Low, Moderate, High, Very high). Data format GeoTIF Source data The maps were generated by the University of Twente Accessibility Zenodo, https://doi.org/10.5281/zenodo.15182981 Stakeholder Relevance Identifying area affected by landslide and expected unstable area for future planning of landuse and mitigation measures Limitations/Assumptions The maps does not provide information on the actual landslide runout, and focuses on the initiation zones, based on the Geomorphological units Additional Output/Information NA Contact Information University of Twente, Cees van Westen & Luigi Lombardo