Abstract. The Greenland ice sheet (GrIS) has experienced significant changes in recent decades. Data confirming those changes are derived from remote sensing, regional climate models (RCMs), firn cores and automatic weather stations (AWSs) on the ice sheet. Data sources comprise different extents in area coverage. While remote sensing and RCMs cover at least regional scales with an extent ranging from 1–10 km, AWS data and firn cores are point observations. To link such regional scales with point measurements, we investigate the spatial variability of snow accumulation within areas of approximately 1–4 km2 and its temporal changes. At three different sites of the southwestern GrIS (Swiss Camp, KAN-U, Dye-2), we performed extensive ground-penetrating radar (GPR) transects and numerous snow pits. In dry snow conditions, radar-measured two-way travel time can be converted to snow depth and snow accumulation if the density is known. Density variations per site for snow pits within distances of up to 1 km are found to be consistently within ±5 %. GPR transects were further filtered to remove small scale surface-related noise. The combined uncertainty of density variations and spatial filtering of radar transects is at 7–8 % per regional scale. To link point observations with regional scales, we analyze for spatial representativeness of snow pits. It occurs that with a probability of p = 0.8 (KAN-U) to p > 0.95 (Swiss Camp and Dye-2), randomly selected snow pits are representative in snow accumulation for entire regions with an offset of ±10 % from arithmetic means. However, to achieve such high representativeness of snow pits, it is required to average snow depth for an area of at least 20 m x 20 m. Interannual accumulation pattern at Dye-2 are very persistent for two subsequent accumulation seasons with similarity probabilities of p > 0.95, if again an error of ±10 % is included. Using target reflectors placed at respective end-of-summer-melt horizons, we additionally analyzed for occurrences of lateral redistribution within one melt season. In this study, we show that at Dye-2 lateral flow of meltwater cannot be evidenced in the current climate. Such studies of spatial representativeness and temporal changes in accumulation are inevitable to assess reliability of the linkage between point measurements and regional scale data and predictions, which are used for validation and calibration of remote sensing data and RCM outputs.;
