Source, and its hard limits
Source: Sepin, Vashold & Kuschnig (2025, Nature Sustainability) — an annual panel of machine-learning-segmented mine polygons (Sentinel-2, 87.7% validation accuracy) with a year and country on every polygon, so mapped mine area per country per year falls straight out.
Hard limits: tropics only (no China, Australia, Canada or temperate belt — not a global total); all-commodity (no per-mineral label, so expansion in single-commodity regions is indicative, not attributed); area is mapped surface footprint, not tonnage; and 2024 is under-mapped (it dips below 2023), so 2023 is the robust recent endpoint. Computed by build_mining_expansion.py.
The expanding critical-material hubs
Tropical countries with the most mine-footprint growth, and the critical materials they’re known for. “CM” = materials for which the country holds ≥10% of world mine production in the atlas.
| Country | 2016 km² | 2024 km² | growth | critical materials / note |
|---|
Sepin et al. 2025, Zenodo (precise polygon set) → mining_expansion.json. Tropical belt only; all-commodity footprint; ML-segmented (87.7% accuracy). Pairs with the snapshot on the satellite footprint page.