The release of organic carbon from thawing permafrost is both a consequence and a driver of global warming, making knowledge of the Arctic carbon cycle a prerequisite for accurate climate prediction. Typically, remote sensing and modeling studies extrapolate results from process-based studies and local findings to entire regions or globally. We investigate two methods and their uncertainties for upscaling carbon cycle components, specifically the eddy covariance for ecosystem CO2 fluxes and X-ray microtomography for organic carbon content derived from permafrost cores. We used an open path eddy covariance system to measure CO2 fluxes at the Bayelva site (close to Ny-Ålesund, Svalbard) and X-ray microtomography (μCT) and supplementary laboratory determinations of the volumetric and gravimetric composition of Yedoma permafrost from Kurungnakh Island (Lena River Delta, Siberia). These methods operate on very different scales, from the local/regional (CO2 fluxes) to the micro scale (size of a soil pore). The CO2 exchange at the Svalbard site during winter 2015 using eddy covariance showed strong uptake and emission events. We show that events of strong CO2 emission amount to 16 g C m-2, while uptake events contribute -22 g C m-2 to the net annual CO2 uptake of -21 g C m-2 yr-1. While technical limitations likely cause the apparent wintertime CO2 uptake, emission events can be related to advection of frontal air masses, as any other known local physical processes cannot explain them. Following standard approaches (for example, by the circumarctic CO2 flux community), that means excluding all negative CO2 flux measurements during winter as well as positive fluxes exceeding a threshold value, the Bayelva site becomes a net CO2 source rather than a sink. Taking the strong wintertime CO2 exchange at face value can introduce a significant bias into long-term carbon budgets and thus to the upscaled Arctic CO2 exchange. X-ray computerized tomography is a nondestructive technique that allows three-dimensional imaging of internal structures of samples, determined by variations in their density and atomic composition. The X-ray tomography of the one meter frozen permafrost core of the Yedoma complex resulted in 22393 images with a resolution of 50 micrometers. Spatial distributions of ice, organic material, minerals and air were determined using several image classification algorithms. Variation of the bulk composition of the soil over depth was determined using destructive laboratory analyses. The results of the imaging and laboratory methods vary (between 2 and 25% for ice content) and thus directly affect the calculation of ice or carbon content estimates. The work presented here suggests that state of the art methods may reveal surprising results requiring a re-assessment of carbon fluxes at the few permafrost sites where it is measured using eddy flux methods, and a re-assessment of permafrost organic carbon and ice content based on permafrost cores.