FIELD AND GEOSPATIAL BASED LANDSLIDE AND FLOOD HAZARD ASSESSMENT FOR DOMINICA, AN ISLAND IN THE EASTERN CARIBBEAN
The field survey was conducted from the West to the East coast of the island along main roadways and coastal areas of high poulation centers. For example, along the Grand Bay Main Road from west to east several slides were activated or reactivated during Maria in 2017. The communities of Bellevue_Chopin, Pichelin, Dubic, Bagatelle, Community of Good Hope, San_Sauveur, and Petite Soufriere were impacted by flooding and landslides from Tropical Storm Erika (2005), Hurricane Maria (2017) and from trough induced rainfall in Nov 22 affecting life and livelihood. Most landslides were classified as debris flows and rotational slides in soil slopes, and most rivers were flooded with water-laden debris. Flooding affected houses and infrastructures in low lying communities along the entire west coast of the island. Using the field data and previous literature, soil landslide susceptibility models were prepared using geospatial methods, including heuristic, logistic regression and random forest approach. Factors like elevation, slope angle and aspect, slope curvature, distance from the ridges and streams, bedrock geology and soils, land cover, and distance from the roads were used to prepare the models. The landslide susceptibility model created using logistic regression provided most realistic results indicating areas of future potential landslides. The study showed landslides were concentrated in the higher elevation, steeper slopes, close to the ridges. Flood hazard assessment using geospatial methods showed that the major controls were lithology, proximity to streams, presence of fractures/faults and heavy rainfall.