OOS 33-5 - Understanding patterns in CH4 emissions across natural ecosystems

The exchange of methane (CH4) between the surface and atmosphere is influenced by a complex interplay of biological and environmental processes such as ebullition and aerenchyma pumping. However, our understanding of these processes and their impact on CH4 sources and sinks remains limited, particularly in terms of spatial and temporal variability, including interannual variations. Vast areas with relatively small uptake and emission rates (e.g., polar deserts, grasslands, forests) have been largely understudied but could contribute significantly to regional and global budgets. These unknowns hinder our ability to predict future climate change due to the complex feedbacks between biological processes, climate change, and land cover change.

Evaluating surface–atmosphere exchange of CH4 from biogenic sources and sinks across a broad diversity of ecosystem types is essential for understanding how natural ecosystems influence atmospheric composition. While biogenic CH4 emissions are of a similar magnitude as anthropogenic emissions, biogenic emissions remain the most uncertain source within the global CH4 budget. As a result, current eddy covariance (EC) towers measuring CH4 fluxes are biased toward wetlands, and other areas where we expect to observe large fluxes. The Flux Gradient Project aims to enhance our understanding of biogenic CH4 fluxes by utilizing existing infrastructure and data streams from the National Ecological Observatory Network (NEON) and the AmeriFlux EC tower network. The project seeks to improve representativeness of networked CH4 flux data by 400-800-fold through the development of a novel method that allows us to calculate CH4 fluxes at tower sites with vertical CH4 concentration profiles.

We are developing an improved gradient-flux technique to calculate CH4 fluxes from concentration profiles at NEON/AmeriFlux towers where CH4 concentration measurement frequency is insufficient for the EC method. Our approach utilizes functional relationships between flux observations (responses) and corresponding environmental (meteorological and land surface) drivers, which complement each other to explain spatio-temporal variation of the flux responses and provide lower and upper boundary conditions, respectively, for a subsequent gradient-diffusion machine learning structure. By accounting for surface heterogeneity and changing sample characteristics over time, we can incorporate these characteristics to develop better flux data products. Our aim is to produce decameter and sub-hourly resolution CH4 flux maps alongside process benchmarks. This will enable differentiation of hot spots from hot moments and underlying processes, and improve interoperability with EC, remote sensing, and plot-based observation.