Answers are grounded in this paper and cite their sources.
Rapid vegetation damage assessment following chemical accidents is critical for emergency response, yet current methods lack the spatial coverage and temporal resolution needed for effective incident management. This study develops a physiologically validated hyperspectral uncrewed aerial vehicle (UAV) framework to detect vegetation exposed to toxic chemical gases. Controlled exposure experiments using hydrochloric acid and xylene on three tree species (Zelkova serrata, Chionanthus retusus, Syringa vulgaris) were monitored over 37 days through hyperspectral imaging, chlorophyll analysis, and scanning electron microscopy. Near-infrared reflectance reduction consistently indicated physiological stress across all species and chemicals, with damage peaking within 7 days postexposure. A logistic regression model using eight physicochemically significant wavelengths (502, 510, 538, 598, 650, 698, 858, and 902 nm) achieved 96.1% validation accuracy, with mechanistic validation through chlorophyll and scanning electron microscopy analyses. At a real-world ammonia leak site with different plant species, the model achieved 91.6% accuracy at 80 m altitude, demonstrating cross-chemical generalizability and altitude scalability beyond the 30 m training conditions, substantially outperforming normalized difference vegetation index-based classification (52.3%). This study establishes a field-deployable UAV protocol with broad applicability across diverse chemical types, plant species, and operational altitudes for emergency response scenarios.
No takes yet. Share an insight, caveat, or question.
Park et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: