Evaluating the impact of gridded population datasets variability on flood exposure estimates across South Asia

Abstract

Floods pose a significant threat to human life and property in South Asia, necessitating accurate population exposure assessments. Most studies simply overlay flood hazard maps with gridded population data, often overlooking population datasets variability. This study examines the influence of different element-at-risk datasets on flood exposure estimates by integrating a 1-in-100-year flood hazard map with seven gridded population. Results show substantial variations in flood-exposed population estimates with variations ranging from –10.1% to 38.1% compared to the High-Resolution Settlement Layer (HRSL) dataset. The spatial distribution of the datasets exhibits 78.4% to 91.3% agreement with HRSL-derived population estimates at the basin scale, as quantified using Geodetector. The Global Human Settlement Population Layer (GHS-POP) dataset, with 3-arc-second resolution, captures over 90% of spatial heterogeneity in flood-exposed populations at both basin and municipal scales. Additionally, demographic analysis attributes highlights vulnerable populations, such as the elderly and children, in regions like Maharashtra, Odisha, Sindh, and Sri Lanka’s western coast, underscoring the need for targeted disaster mitigation. These findings emphasize the importance of selecting appropriate element-at-risk datasets for flood risk assessments, particularly in data-scarce regions.