Intensifying Flood Extent and Human Displacement Risk Across Africa

Flood risk has intensified globally due to climate change and has become a major driver of human displacement, with Africa being particularly vulnerable. Limited access to high-resolution, long-term flood observations has constrained understanding of displacement dynamics across the continent, where adaptive capacity remains low. Here, we integrate four decades (1984–2024) of monthly satellite-derived flood observations from Landsat and Sentinel-2 with subnational displacement records from the Internal Displacement Monitoring Centre (IDMC) and socio-economic indicators such as GDP per capita and urbanization from the Global Human Settlement Layer (GHSL) across Africa. Results reveal a marked expansion of flooded areas across Western and Central Africa. In the Niger, Congo, and Benue basins, flood extent has increased by 4.02 km²yr^-1, while country-level trends are steepest in Mali (+6.08 km² yr^-1), Nigeria (+4.43 km² yr^-1), and the D.R. Congo (+4.11 km² yr^-1). To quantify the probability that floods trigger displacement and the magnitude of displacement conditional on occurrence, a hurdle modeling framework has been adopted. Using a hurdle modeling framework, we separately quantify the probability that floods trigger displacement and the magnitude of displacement conditional on occurrence. Displacement responses exhibit strong spatial heterogeneity. Conditional on displacement, a one standard deviation increase in flood severity is associated with an approximately 27% increase in displacement magnitude, with hotspots in the Sahel, Southern Africa, and the Horn of Africa. This flood–displacement sensitivity is amplified in more urbanized areas and dampened in higher-income areas. The expansion of flood extent across major African basins, coupled with socio-economic vulnerabilities, signals escalating displacement risk and underscores the need for locally tailored adaptation strategies that integrate flood preparedness with displacement-sensitive disaster risk management.

Acknowledgment: This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Ministry of Science and ICT (RS-2021-NR055516, RS-2025-02312954).