Evaluating spatial upscaling strategies for Arctic carbon fluxes: high-density mobile chamber measurements at Stordalen Mire

Estimating the carbon balance of Arctic ecosystems is challenging because of their high spatial heterogeneity, which is difficult to account for using traditional methods: static chambers linked to fixed points allow for tracking the seasonal dynamics of processes but are limited in spatial coverage, whereas the Eddy Covariance method provides only an integral assessment of fluxes from large areas, averaging the contribution of various microlandscapes. 

In this work, we present the results of a STORDALENX25 field campaign conducted during the 2025 growing season at Stordalen Mire (Abisko, Sweden), during which over 650 measurements of CH₄ and CO₂ (NEE) fluxes were obtained using the mobile chamber technique, quasi-randomly distributed within and beyond the Eddy Covariance footprint, covering a total area of approximately 0.1 km². The unique density of this spatial dataset allows it to be used not only for calculating the regional budget but also as a testbed for evaluating various spatial upscaling strategies. 

As a first key methodological task, we compare the effectiveness of different base maps describing the study domain: we contrast classical upscaling based on land cover types (Palsa, Fen, Bog) with the use of data-driven functional zonation. Another research objective is to determine the factors contributing most to model accuracy: we conduct a comparative analysis of predictors, assessing the "value added" by remote sensing data (Sentinel-2, UAV) compared to direct field measurements such as soil temperature and moisture. Furthermore, we analyze the "performance plateau" to identify the minimum necessary number of measurement points and compare the efficiency of classical vegetation-based scaling against clustering based on environmental response functions. The results, validated by data from static chambers and the Eddy Covariance tower, allow for the optimization of future field campaign designs by determining the balance between labor effort and the accuracy of spatial estimates.