Modeling the Impact of Snow Drift on the Decameter-Scale Variability of Snow Properties on the Antarctic Plateau

Abstract:

On the Antarctic Plateau, the annual snow accumulation and the physical properties of snow close to the surface are characterized by a large spatial variability at the scale of a few metres. As a consequence, two snowpits measured a few metres apart from each other can show significant differences. This variability mainly results from the combination of low annual amounts of precipitation and drift events that redistribute snow. The latter physical process is not simulated by one-dimensional snow evolution models. Here we describe how the detailed snowpack model Crocus is adapted to Antarctic conditions and modified to account for this drift-induced variability using a stochastic snow redistribution scheme. For this, 50 simulations are run in parallel and the corresponding numerical snowpacks are allowed to exchange snow mass according to rules driven by wind speed and snow characteristics. These simple rules were developed and calibrated based on in situ pictures of the snow surface recorded at Dome C for two years, which show the occurrence of snow drift and its impact on snow height variations. At Dome C, the results of these parallel simulations show three substantial improvements with respect to standard Crocus simulations. First, significant and rapid variations of snow height observed in hourly measurements are well reproduced, highlighting the crucial role of snow drift in snow accumulation. Second, the statistics of annual accumulation is successfully simulated, including the years with negative net ablation which are as frequent as 15% in the observations and 11% in the simulation. Eventually, the simulated vertical profiles of snow density and specific surface area down to 50 cm depth are compared to 98 profiles measured at Dome C during the summer 2012–2013. The observed spatial variability is partly reproduced by the new model, especially close to the surface. The erosion/deposition processes also explain why layers with density lower than 250 kg m⁻³ or specific surface area larger than 30 m² kg⁻¹ can be found deeper than 10 cm. The proposed parameterization of snow drift may help with the interpretation of ice cores, which largely relies on assumptions about local snow accumulation.