The rapid rate of environmental change in the Arctic alters the exchange of water, carbon, and energy fluxes between the land surface and the atmosphere with global impacts on ecosystems and climate. This thesis investigates the effect of mixed satellite signals on land cover mapping and on the estimation of latent heat fluxes, QE, and land surface temperature (LST) in three Arctic tundra environments in the Lena Delta (Siberia, Russia), on Bathurst Island (Canadian High Arctic), and the Barrow Peninsula (Alaska, USA). Land cover maps were derived from optical and radar remote sensing data with resolutions of 4m or better to decompose satellite mixed pixels with resolutions of 17m (CHRIS/PROBA) and 30m (Landsat5-TM). Downscaling land/water cover via Landsat surface albedo increased the total water surface area of the Lena Delta from 13% to 20%. Ponds, i. e., water bodies with a surface area smaller than 10^4 m, made over 95% of the total number of water bodies at all sites. Water body size-distributions deviated from a power law function for ponds and very large lakes which could only be detected with high-resolution water body mapping. Maximum spatial differences of up to 22 W/m^2 for QE and 10°C for LST were associated with fair weather perios dominated by high net radiation and little precipitation. Uncertainties of �35% would arise in Landsat-based QE mapping, and of �30% in MODIS-based LST mapping when subpixel land cover heterogeneities are not considered. Results of this thesis highlight the importance of integrating detailed field studies with multi-scale remote sensing data to determine fine-scale spatial differences in energy fluxes over larger areas in Arctic tundra landscapes. Land cover maps with spatial resolutions of 2m or better are necessary to ensure the quality and representativeness of land cover statistics. This thesis proposes to compile improved subpixel land cover statistics in different Arctic ecosytems to facilitate upscaling of the surface energy balance as well as carbon fluxes to larger-scale grids. This is a crucial task regarding the great uncertainty associated with the global estimation of feedbacks between the Arctic surface and the atmosphere under a changing climate.