Exploring interactions between socioeconomic context and natural hazards on human population displacement

This publication shows explainable machine learning techniques can be employed to model and understand internal displacement flows and patterns from observational data alone. For this purpose, a large, harmonized, global database of disaster-induced movements in the presence of floods, storms, and landslides during 2016–2021 is presented. The publication accounts for environmental, societal, and economic factors to predict the number of displaced persons per event in the affected regions. This shows that displacements can be primarily attributed to the combination of poor household conditions and intense precipitation, as revealed through the interpretation of the trained models using both Shapley values and causality-based methods. Lastly, empirical evidence is provided that differential or uneven vulnerability exists and provide a means for its quantification, which could help advance evidence-based mitigation and adaptation planning efforts.

To detect linkages between human mobility and weather hazards, a variety of indicators of population movements have been suggested, but their suitability has been challenged. Aggregation of NDP at a continent level shows that Asia is by far the most impacted continent regardless of the type of disaster. This is particularly evident in the most densely populated coastal regions, often affected by severe storms. North America is also affected mainly by storms, but the impact of total movements of people is about a factor of 102 lower compared to Asia. The other continents have more NDP associated with floods. Landslides represent a marginal component of displacements in all continents with respect to floods and storms.