A Bayesian inference approach to determine experimental Typha latifolia paludiculture greenhouse gas exchange measured with eddy covariance

Measurements of greenhouse gas exchange (GHG) using the eddy covariance method are crucial for identifying strategies to achieve emission reductions and carbon sequestration. There are many sites that have heterogeneous land covers where it would be useful to have balances of particular land areas, such as field trials of emission mitigation strategies, but the flux footprint infrequently covers only the area of interest. Filtering the data based on a footprint area threshold can be done but may result in the loss of a high proportion of observations that contain valuable information. Here, we present a study that uses a single eddy covariance tower on the border of two land uses to compare GHG exchange from a Typha latifolia paludiculture experiment and the surrounding area (SA) which is primarily a dairy meadow. We used a Bayesian inference approach to predict carbon dioxide (CO2) and methane (CH4) fluxes where the relative contribution of the two source areas, derived from a two-dimensional footprint for each timestep, was used to weight and parameterise equations. Distinct differences in flux behaviour were observed when contributions of the two land areas changed and that resulted in clearly different parameter distributions. The annual totals (posterior mean ± 95% confidence interval) from the simulations showed that Typha was a net sink of CO2 for both simulation years (−18.5 ± 2.9 and −17.8 ± 2.9 t CO2ha−1yr−1) while SA was a net source (16.8 ± 2.9 and 17.4 ± 2.9 t CO2ha−1yr−1). Using the 100-year global warming potential of CH4, even though CH4 emissions were higher for paludiculture in both years (13.6 ± 0.6 and 15.9 ± 1.0 t CO2-eqha−1yr−1) than SA (7.1 ± 0.6 and 6.8 ± 1.2 t CO2-eqha−1yr−1), the net GHG balance indicates that Typha paludiculture is a viable strategy to limit GHG emissions from drained peatlands.