Constraining Annual Coal Mine Methane Emissions with Multitemporal Hyperspectral Satellite Observations

Methane emissions from coal mining are difficult to constrain because of temporal variability and sparse observations. We used multitemporal hyperspectral observations from GF5A, GF5B, ZY1E, and ZY1F to quantify facility-scale emissions and evaluate sampling requirements for annualized estimates. Of 301 scenes acquired from January 2024 to 15 April 2025, 117 cloud-free scenes yielded 237 confirmed plume events from 16 sources at 10 underground coal mines. Event-scale emission rates varied substantially and reached 22.1 ± 4.7 t h-1. The highest-emitting source had an annualized mean rate of 10.2 ± 2.3 t h-1. Source-specific bootstrap analysis showed decreasing estimation error with increasing sampling, although convergence varied among sources; approximately 20 usable plume-level observations per year provided an empirical guideline for the studied source population. We then simulated potential clear-sky observation opportunities for the world's 100 largest active coal mines by production using six satellite missions, orbital and illumination constraints, and cloud cover. These mines received an average of approximately 18 opportunities per year, and 40% received more than 20. Coordinated multisatellite observations can strengthen methane measurement, reporting, and verification, provided that plume detectability, source attribution, and quantification requirements are met.