Agricultural methane plume detection using MethaneAIR: A targeted scene-based approach with wavelet denoising and divergence integral methods.

Abstract. Methane is a potent greenhouse gas, and accurate emission estimates are essential for effective climate mitigation. Agricultural sources, particularly concentrated animal feeding operations (CAFOs), are significant anthropogenic contributors, yet their emissions remain difficult to quantify, contributing to uncertainty in inventories.

MethaneAIR, an aircraft-based imaging spectrometer and precursor to MethaneSAT, was primarily developed to characterize methane emissions from oil and gas infrastructure. Between 2021 and 2024, MethaneAIR conducted 75 flights across the United States and Canada, producing orthorectified mosaics of column-averaged methane. These data were used to assess agricultural emissions at high resolution through a novel scene-based approach. Agricultural "scenes" were defined as spatial subsets of flight mosaics encompassing CAFOs and surrounding areas, enabling targeted plume detection and quantification. Wavelet denoising and a Gaussian-based Divergence Integral method were applied to 209 agricultural scenes coincident with 84 CAFOs. Of 200 detected plumes, 89 met our quantitative robustness criteria and were analysed further, with emphasis on northeast Colorado.

While limited on-farm data, such as the number of animals and waste management practices, constrained the ability to fully interpret emission drivers, the analysis revealed elevated emissions relative to inventories and high variability, likely influenced by interactions between wind and waste management systems. These findings both highlight variability not captured in annual inventories and inform the design of future satellite missions like MethaneSAT, which will improve global methane monitoring and climate models. With improved on-farm information, this approach could provide a scalable pathway for emission and mitigation verification.