A monthly sub-national Harmonized Food Insecurity Dataset for comprehensive analysis and predictive modeling
Existing subnational food insecurity data lack a unified, high-frequency, comparable, and open-access database, impeding early warning and modeling of global food crises. Method: We construct HFID—the first monthly-updated, subnational food insecurity dataset—integrating four authoritative sources: IPC/CH, FEWS NET, WFP-FCS, and rCSI. We propose a novel spatiotemporal alignment and dynamically weighted fusion framework for heterogeneous multi-source indicators, leveraging a standardized administrative unit system and an open-source geospatial-temporal framework to ensure consistent boundary harmonization and cross-source data mapping. Contribution/Results: HFID covers high-risk regions globally, filling a critical gap in high-frequency, comparable, and openly accessible subnational food insecurity data. It significantly enhances situational awareness timeliness and cross-regional comparability. The dataset has already been adopted to develop and validate early-warning models by multiple international organizations.