Controlled-release validation of a UAV-based mass balance approach for quantifying methane emissions at two sites.

Quantifying methane emissions from diffuse sources, including landfills and agricultural systems, is essential for improving emission inventories and assessing the effectiveness of mitigation measures in near-real time. Unmanned aerial vehicles (UAVs) provide a flexible and cost-efficient platform for atmospheric methane measurements, particularly in complex or difficult-to-access environments. However, confidence in UAV-derived emission estimates depends on robust validation and transparent uncertainty characterization. Despite the growing use of UAV-based quantification methods, systematic validation remains limited, and the lack of standardized validation procedures and consistent uncertainty reporting continues to hinder comparability across studies and limits confidence in reported emission rates. 

Here we evaluate a UAV-based mass balance approach for methane emission quantification using controlled-release experiments operated by the National Physical Laboratory at two UK sites: an isolated aerodrome providing an idealised test environment, and an operational agricultural facility with measurement transects positioned downwind of the controlled release to avoid interference from background sources. Controlled methane releases spanned a wide range of emission rates (0.02–40 kg h⁻¹) and included both point and extended source configurations representative of agricultural (manure) and landfill emission scenarios. Release rates were blind to the researchers prior to flux calculation. Methane concentrations were measured in situ using a Los Gatos Research GLA-133 analyser mounted on a DJI M600 UAV, with emissions quantified using downwind horizontal transects within a mass balance framework. We also present wind measurements from an onboard 2D sonic anemometer, which were compared with an on-site high-precision anemometer mast after accounting for UAV motion/orientation and compass calibration. Together, these data were used in a mass balance framework to assess the accuracy and operational robustness of the approach. Overall, comparison between known and estimated fluxes showed very good agreement (slope = 0.998; Pearson’s r = 0.98), with a mean bias of −24.5%.

This study supports the development and validation of UAV-based techniques for methane monitoring and highlights their potential for use in regulatory contexts and emission inventory verification. We further examine how environmental conditions, source geometry, and release characteristics influence agreement between estimated and controlled emission rates.