UAV-Satellite Integration Reveals Continuous Methane Emission Variability from Chinese Coal Mine Ventilation Shafts

Abstract Satellite quantification of coal mine methane lacks validation for continuously operating ventilation systems with high temporal variability. We integrate 75 unmanned aerial vehicle (UAV) flights (30 successful, 13 days) with 6.5 years of satellite observations at a Chinese coal mine representing modern deep operations. Ventilation systems emit continuously with extreme variability (28-fold across days; 66–1,849 kg·h–1), challenging simple ON-OFF emission assumptions. Satellite sensor-dependent detection limits systematically exclude low-emission states directly documented by UAV measurements, introducing biases of −4% to +14% relative to our UAV-constrained Markov estimate across six statistical models. Our UAV-constrained Markov model integrates satellite detection frequencies with ground-truth subthreshold distributions, yielding 0.024 ± 0.015 Mt·a–1 for the target ventilation shaft and reconciling the model-dependent spread. At the facility scale, our estimate (0.075 ± 0.022 Mt·a–1) agrees with GCM2024 (−6%) and EDGAR v2024 (+15%); our regional estimate (0.145 ± 0.034 Mt·a–1) closely matches GFEI (0.147 Mt·a–1; −1%), and a sensitivity test applying UAV-derived low-emission parameters to all point sources yields estimates consistent with EDGAR (0.063 Mt·a–1 vs 0.065 Mt·a–1; −3%). Emission variability (28-fold range) exceeds typical mitigation targets (20–30% reduction), precluding verification from single observations. This integrated approach enables cost-effective monitoring supporting China’s methane reduction commitments.