Representative surface snow density on the East Antarctic Plateau

Surface mass balance estimates of polar ice sheets are essential to estimate the contribution of ice sheets to sea level rise, in response global warming. One of the largest uncertainties in the interior regions of the ice sheets, such as the East Antarctic Plateau (EAP), is the determination of a precise surface snow density. Wrong estimates of snow and firn density can lead to significant underestimations of the surface mass balance. We present density data from snow profiles taken along an overland traverse in austral summer 2016/17 covering over 2000 km on the Dronning Maud Land plateau. The sampling strategy included investigation on various spatial scales, from regional to local, with sampling locations 100 km apart as well as a high-resolution study in a trench at 30° E 79° S with thirty 3 m deep snow profiles. Density of the surface snow profiles has been measured volumetrically as well as using μ-computer tomography. With an error of less than 2 %, the volumetric liner density provides higher precision than other sampling devices of smaller volume. With four spatially independent snow profiles per location we derive a representative and precise 1 m mean snow density with an error of less than 1.5 %. The average liner density along the traverse across the EAP is 355 kg m−3, which we identify as representative surface snow density between Kohnen station and Dome Fuji. The highest horizontal variability in density can be seen in the upper 0.3 m. Therefore, we do not recommend vertical sampling in intervals of less than several decimeters, as this does neither adequately cover seasonal variations in high accumulation areas nor the annual accumulation in low accumulation areas. From statistical analysis of the liner density on regional scale we identify representative spatial distributions of density based on geographical and thus climatic conditions. Our representative density of 355 kg m−3 is considerably different from the density of 320 kg m−3 provided by a regional climate model. This difference of more than 10 % indicates the necessity for further calibration of density parameterizations. The difference in the total mass equivalent of measured and modelled density yields a 3 % underestimation by models, which translates into 5 cm sea level equivalent. We do not find a statistically significant temporal trend in density changes over the last two decades. Our data provide a solid baseline for tuning parameterizations of the surface snow density for regions with low accumulation and low temperatures like the EAP to improve surface mass balance estimates of polar ice sheets.