Fine-scale spatial variability of winter CO₂ and CH₄ fluxes in Arctic tundra derived from snowpack gradient measurements

Abstract. Winter carbon dioxide (CO₂) and methane (CH₄) fluxes from soils under seasonal snowpacks make non-negligible, yet poorly constrained contributions to annual carbon budgets across Arctic regions. Quantifying these fluxes and their spatial variance will better constrain uncertainties in simulations of winter carbon fluxes from terrestrial biosphere models. We address this gap by measuring and identifying patterns and spatial variability in CO₂ and CH₄ soil-atmosphere fluxes through late winter snowpacks at an upland tundra site in the western Canadian Arctic. Instantaneous fluxes were calculated from CO₂ and CH₄ microsite concentration gradients at 10 cm to 20 cm vertical resolution (n = 119) through the snowpack across five homogeneous surface covers, representing dominant vegetation types. We measured consistent soil-to-atmosphere CO₂ fluxes but with significantly different rates across surface covers (0.8 to 100 mgC m^-2 day^-1), which were strongly influenced by snow depth and soil surface temperature, exhibiting higher emissions under deeper snowpacks and warmer soil surfaces. CH₄ fluxes were also coupled to soil surface temperature and varied between −0.04 and 0.08 mgC m^-2 day^-1. Persistent CH₄ uptake was observed in warmer soils (-6.0 to -0.5 °C) in a sparsely populated black spruce and shrub dominated area with deep snow, indicating active methane oxidation during winter. Microsite scale CO₂ and CH₄ fluxes were statistically independent of vertical snow microstructure, indicating that winter fluxes could be reliably calculated from single gas concentration at the soil-snow interface, negating the need for additional snowpack gas measurements. These results open new avenues for quantifying fine-scale spatial variability of wintertime CO₂ and CH₄ fluxes in Arctic tundra, which can constrain biogeochemical process representations in terrestrial biosphere models and inform spatial upscaling methodologies.