Substantial health toll of oil-induced land contamination: Evidence from satellite observations and causal inference

This dataset contains the core public data products used to analyze oil-contaminated land (OLC), officially reported oil spills, and child-health exposure pathways in the Niger Delta, Nigeria. The dataset includes annual satellite-derived OLC polygon maps for 2016-2023, a cleaned record-level table of official oil-spill reports for 2016-2024, and DHS-derived model-ready health tables used for the child morbidity, infant mortality, and hemoglobin analyses. The OLC maps were produced from high-resolution satellite image time series using a deep-learning-based mapping workflow. The annual polygon shapefiles represent land areas classified as oil-contaminated in each year. These maps are intended for spatial statistics, state-level summaries, environmental exposure assessment, and comparison with official spill records. The official oil-spill table compiles reported incidents, harmonizes dates and locations, cleans estimated spill quantity and spill-area fields, and assigns records to administrative units when possible. The health tables link DHS child and household records to OLC exposure metrics, river-position classes, and spatial fixed-effect identifiers for regression analysis.