Cluster-Based Analysis of Drought Dynamics and Socioeconomic Exposure Across Continents

Drought events exert profound impacts on both ecosystems and human societies, making it essential to develop precise and scalable methodologies for their identification and impact assessment. In this study, we propose a dynamic clustering framework to detect and track spatio-temporal drought objects using gridded drought indices. The method accounts for spatial proximity through Haversine distance calculations and handles longitudinal periodicity, enabling robust identification of coherent drought clusters. These clusters are further refined by applying a land-sea mask and by excluding or merging small clusters. The approach is designed to be adaptable across different spatial scales and drought indices, ensuring broad applicability across regions and datasets.

Beyond identifying drought regions, we integrate socioeconomic exposure to assess their real-world implications. Using high-resolution gridded population and GDP datasets, we estimate both total exposure and severity- and frequency-weighted exposure to droughts, linking physical drought characteristics with population density and economic productivity. This coupling allows us to quantify not just where droughts occur, but also the potential number of people affected and the potential GDP loss. The framework is applied to historical drought events from the Geocoded Disasters (GDIS) database to analyze spatial correlations with reported disaster impacts such as the number of people affected and economic damage. Additionally, the approach enables more informed risk assessment and supports the design of region-specific adaptation and mitigation strategies.

This integrated approach offers a reproducible, data-driven method to bridge drought characteristics with socioeconomic vulnerability. The outputs—including high-resolution drought cluster maps, exposure estimates, and statistical summaries—are valuable for disaster risk reduction, climate adaptation planning, and future drought risk modeling.