Distributed surface mass balance of an avalanche-fed glacier

Abstract

Local snow redistribution processes such as avalanches can considerably impact the spatial variability of accumulation on glaciers. However, this spatial variability is difficult to quantify with traditional surface mass balance measurements or geodetic observations. Here, we leverage high-quality and high-resolution surface velocity and elevation change maps for the period 2012–2021 from Pléiades stereo images and ice thickness measurements of Argentière Glacier (France) to invert for its distributed surface mass balance. Three inversions are conducted using three different ice thickness modelling approaches, two of which are constrained by observations. The inversions all show very good agreement between inverted surface mass balance and in situ measurements (RMSE between 0.50 and 0.96 mw.e.yr-1 for the 11-year average). The detected spatial variability in surface mass balance is consistent between the modelling approaches and much higher than what is predicted from an enhanced-temperature-index model calibrated with measurements from a dense network of stakes. In particular, we find high accumulation rates at the base of steep headwalls on the left-hand side of the glacier, likely related to avalanche deposits at these locations. We calculate distributed precipitation correction factors to reconcile the outputs from the enhanced-temperature-index model with the inverted surface mass balance data. These correction factors agree with the outputs of a parametrisation of snow redistribution by avalanching, indicating an additional 60 % mass input relative to the accumulation from solid precipitation at these specific locations, which was equivalent to an additional 20 % mass accumulation at the scale of Argentière Glacier without its two smaller tributaries. Using these correction factors in a forward-modelling exercise, we show that explicitly accounting for avalanches leads to twice more ice being conserved in the Argentière catchment by 2100 in an RCP 4.5 climate scenario and to a considerably different ice thickness distribution. Our results highlight the need to better account for such spatially variable accumulation processes in glacio-hydrological models.