A data-driven approach to mapping global commodity-specific mining land-use

Mineral extraction is a key driver of environmental change globally, yet geospatial data on mining operations remains fragmented and incomplete across data sources. Datasets with complementary information, such as mining project inventories (points) and satellite-derived land use (polygons), are often disconnected due to spatial mismatches and the complex distribution of infrastructures, such as open pits, tailings, and processing facilities, which are frequently scattered. Integrating these geographic features is critical for enhancing mining data availability and leveraging data complementarity, thereby advancing the understanding of mining impacts globally. This study proposes a scalable approach to link heterogeneous mining datasets and demonstrates its applicability by quantifying the global mine land associated with specific commodities. The new approach introduces data-driven mine clusters , grouping geographic features through hierarchical clustering with locally optimised distance thresholds. This method enables associating information from inventory data with land-use polygons covering mine infrastructure derived from satellite data. To test the approach, data from various sources were integrated. The resulting integrated dataset covers over 145,000 km² and offers the most comprehensive overview of global mine land use linked to mineral commodities. Validation of the clusters against expert-labelled mines shows a high level of agreement, with 95 % of the clusters sharing at least one primary commodity. Results revealed that coal (22.5 %) and gold (21.1 %) dominate global mining land footprints. 26.8 % of the area could not be assigned to a commodity. This methodology provides a reproducible approach to enhancing the integration of spatial data on mining activities, supporting more robust global assessments of mining impacts.