The temperature distribution in permafrost soils isaffected by a wide variety of parameters, which canvary over small distances and on short timescale. Anadequate representation of these small-scale heterogeneitiesin permafrost models remains a challengingtask. Energy balance models calculate the surfacetemperature based on the partitioning of energyat the surface. The surface temperature is then projectedinto deeper soil layers. In principle, such permafrostmodels can account for small-scale spatialheterogeneity, if only a sufficiently resolved set ofall input parameters is provided. In practice, suchdata sets rarely exist, so it is necessary to identify thecrucial parameters and the spatial and temporalscales, over which they must be accounted for toachieve a satisfactory accuracy of the model. Forthis purpose, a detailed understanding of the surfaceenergy budget is indispensable.The study site is located between the glacier Brøggerbreenand the Kongsfjorden at 78° 55N, 11°50E, approximately 2km SW of the village of Ny-Alesund on Svalbard. It is situated in hilly tundra atthe foot of two major glaciers at elevations of 15mto 25m above sea level and is characterized bysparse vegetation alternating with exposed soil androck fields.We present continuous measurements of all componentsof the surface energy budget at a high-arcticpermafrost site on Svalbard over the course of oneyear (Westermann et al. 2009). An eddy covariancesystem is used to determine the turbulent landatmosphereexchange processes. Our results highlightthe importance of sensible and latent heatfluxes for the formation of the surface temperature.During the snow-free period, the surface temperaturesof an area of about 100 x 100 m² have beenmonitored at spatial resolutions below one meter usinga thermal camera system. Strong temperaturedifferences between wet and dry areas are found onshort timescales of a few hours. Using an energybalance approach, this can be explained by differentevaporation rates and hence a different energy partitioningbetween the sensible and the latent heat flux.However, on timescales of one week to one month,the differences between wet and dry areas widelyaverage out, so that they are negligible for the formationof spatial differences in subsurface temperature.During winter, an average temperature difference ofmore than 3K is found between the air temperatureat 10m height and the surface temperature. Thisstrong near-surface temperature inversion is a strikingfeature, which clearly limits the use of air temperaturesas surrogate for the temperature of thesnow surface.Furthermore, the temperature at the snow-soil interfaceand the temperature profile to a depth of 1.5 mhave been monitored at 14 different locations withinan area of half a square kilometre. In contrast tosummer, sustained average temperature differencesof up to 6 K between different locations are found atthe snow-soil interface, although energy balance calculationsand direct measurements suggest little spatialvariation of the temperature of the snow surface.The temperature differences is directly related to thethickness of the snow cover and possibly also its historyof formation. They result in strong site-to-sitevariations of the soil temperatures at 1.5 m depth,which range from -6°C to -0.3°C in March. Thesnow cover is therefore found to be the prime sourceof spatial variability of the permafrost temperaturesat the study site.Westermann, S., Lüers, J., Langer, M., Piel, K., and Boike, J. The annualsurface energy budget of a high-arctic permafrost site onSvalbard, Norway, The Cryosphere, 3, 245263, http: //www.thecryosphere.net/3/245/2009/, 2009.