UAV-Based Hyperspectral Imaging Bridges the Spatial Gap in Pollution Source Identification

Accurate localization and quantification of pollution sources are critical for effective air quality management, yet existing monitoring technologies struggle to resolve emission hotspots beneath clouds or bridge the spatial gap between satellite and ground-based observations. Here, we present an unmanned aerial vehicle (UAV)-based hyperspectral imaging system that overcomes these limitations by combining meter-scale spatial resolution and 20-30 min temporal resolution to directly pinpoint and quantify pollutant outfalls from industrial facilities. To address UAV remote sensing challenges─including payload constraints, low spectral signal-to-noise ratios (SNR), surface albedo variability, and turbulent-induced observation shifts─we developed: (1) a lightweight hyperspectral payload optimized for UAV deployment; (2) enhanced SNR module for ground-level pollutant detection; (3) a robust air mass factor correction accounting for surface heterogeneity; (4) stabilized observation angles under atmospheric turbulence. Validation against mobile differential optical absorption spectroscopy (DOAS) demonstrated strong agreement (R = 0.94 for NO₂). Unlike satellites, our technology captures subcloud pollution plumes and identifies high-emission facilities, effectively bridging the observational gap between orbital and ground monitoring. This technology will facilitate the refinement of regional emission models, leveraging its strengths in scalability and precise source identification to drive a paradigm shift in pollution control.