Integrating UAV-based CO2 and CH4 fluxes into an atmospheric transport model for surface flux attribution

Uncrewed aerial vehicles (UAVs) are becoming an essential tool to monitor greenhouse gases (GHGs) such as carbon dioxide (CO2) and methane (CH4), particularly over the known sources such as landfills and industrial sites. UAV-based flux quantification over these sources is generally practiced by flying vertical curtain patterns at certain downwind distances to constrain the whole plume originating from the source and applying either mass balance or Gaussian plume inversion techniques for constraining the source strength. Extending this technique over natural ecosystems, however, requires attribution of multiple sources and sinks that contribute to the observed GHG mixing ratios, as opposed to the single well-defined sources typical for single-plume applications.

The objective of this study is to develop a UAV-based approach for CO2 and CH4 flux estimations over natural ecosystems. In the context of the STORDALENX25 campaign, we conducted multiple vertical curtain pattern flights over a sub-Arctic wetland. Using a backward Langrangian stochastic model (bLSmodelR), we estimated the areal extent (i.e. footprint) of all vertical flux curtains. UAV-based CO2 and CH4 fluxes calculated by the mass balance technique were then normalized using the estimated footprint area to obtain flux values per square meter. A comparison was made against eddy-covariance-tower-based flux reference calculations whenever both platforms footprints were approximately overlapping. Subsequently, calculated areal fluxes were aggregated together with land cover classes using random forest regression to estimate surface fluxes across the mire. Overall, this study demonstrates a pathway towards UAV-based surface flux estimations over natural ecosystems, resolving patch-level variability and thus reducing uncertainties in flux upscaling to the ecosystem level.