Small methane emissions from pasture observed with eddy covariance and flux-gradient methods: are they real or artefacts?

• Eddy covariance (EC) and flux-gradient (FG) data show net CH 4 emission from pasture. • This is at odds with chamber measurements usually showing net CH 4 uptake. • Dispersion modelling of cow emissions at >800 m upwind distance gives explanation. • Shapes of cospectra show EC fluxes of CH 4 are often non-local, i.e. from far upwind. • Advection overwhelms local ecosystem exchange in landscapes with strong CH 4 sources. Grassland soils are generally considered a small methane (CH 4 ) sink, based mostly on the findings from chamber experiments. However, micrometeorological studies using the eddy covariance (EC) and flux-gradient (FG) methods have often reported upward CH 4 fluxes over pasture, in the absence of grazing animals. Here, we collate available EC and FG data from New Zealand’s pasture systems finding a predominance of upwards fluxes, typically ≈ 5–10 mg m −2 d −1 . We then investigate whether these fluxes constitute real pasture emissions and conclude that they do not. Rather, these small upward fluxes are likely the result of strong CH 4 emissions from upwind sources, at distances on the order of 1 km or more, outside the area usually considered the flux footprint. In other words, horizontal advection is often a non-negligible term in the CH 4 mass budget and the homogeneity assumption for EC and FG is violated. This conclusion is based on the following points of evidence: 1) Exploratory surveys with chamber measurements, on three farms where EC and FG measurements operated, found small CH 4 uptake rates. 2) Simulations with a dispersion model of cow emissions 800–900 m upwind of a measurement point 2 m above ground showed systematically upward EC fluxes of 1–25 mg m −2 d −1 , and upward FG fluxes of similar magnitude for sufficiently stable or unstable stratification, demonstrating that fluxes of the observed magnitude can be entirely caused by advection. 3) Cospectra of CH 4 and vertical wind were compared with their carbon dioxide (CO 2 ) counterparts, because CO 2 is known to be taken up or emitted locally and spatially homogeneously, and the CH 4 cospectra were substantially different, being dominated at small wavenumbers indicating transport by large, organised structures. Based on our analyses, we urge caution when interpreting micrometeorological CH 4 flux data in landscapes where large, transient emissions sources are present.