Search for: All records

The Arctic is warming at twice the rate of the global mean. This warming could further stimulate methane (CH₄) emissions from northern wetlands and enhance the greenhouse impact of this region. Arctic wetlands are extremely heterogeneous in terms of geochemistry, vegetation, microtopography, and hydrology, and therefore CH₄fluxes can differ dramatically within the metre scale. Eddy covariance (EC) is one of the most useful methods for estimating CH₄fluxes in remote areas over long periods of time. However, when the areas sampled by these EC towers (i.e. tower footprints) are by definition very heterogeneous, due to encompassing a variety of environmental conditions and vegetation types, modelling environmental controls of CH₄emissions becomes even more challenging, confounding efforts to reduce uncertainty in baseline CH₄emissions from these landscapes. In this study, we evaluated the effect of footprint variability on CH₄fluxes from two EC towers located in wetlands on the North Slope of Alaska. The local domain of each of these sites contains well developed polygonal tundra as well as a drained thermokarst lake basin. We found that the spatiotemporal variability of the footprint, has a significant influence on the observed CH₄fluxes, contributing between 3% and 33% of the variance, depending on site, time period, and modelling method. Multiple indices were used to define spatial heterogeneity, and their explanatory power varied depending on site and season. Overall, the normalised difference water index had the most consistent explanatory power on CH₄fluxes, though generally only when used in concert with at least one other spatial index. The spatial bias (defined here as the difference between the mean for the 0.36 km²domain around the tower and the footprint-weighted mean) was between ∣51∣% and ∣18∣% depending on the index. This study highlights the need for footprint modelling to infer the representativeness of the carbon fluxes measured by EC towers in these highly heterogeneous tundra ecosystems, and the need to evaluate spatial variability when upscaling EC site-level data to a larger domain.