Relative performance of empirical and physical models in assessing the seasonal and annual glacier surface mass balance of Saint-Sorlin Glacier (French Alps)

• Main Authors: M. Réveillet, D. Six, C. Vincent, A. Rabatel, M. Dumont, M. Lafaysse, S. Morin, V. Vionnet, M. Litt
• Format: Article
• Language: English
• Published: Copernicus Publications 2018-04-01
• Series: The Cryosphere
• Online Access: https://www.the-cryosphere.net/12/1367/2018/tc-12-1367-2018.pdf

This study focuses on simulations of the seasonal and annual surface mass balance (SMB) of Saint-Sorlin Glacier (French Alps) for the period 1996–2015 using the detailed SURFEX/ISBA-Crocus snowpack model. The model is forced by SAFRAN meteorological reanalysis data, adjusted with automatic weather station (AWS) measurements to ensure that simulations of all the energy balance components, in particular turbulent fluxes, are accurately represented with respect to the measured energy balance. Results indicate good model performance for the simulation of summer SMB when using meteorological forcing adjusted with in situ measurements. Model performance however strongly decreases without in situ meteorological measurements. The sensitivity of the model to meteorological forcing indicates a strong sensitivity to wind speed, higher than the sensitivity to ice albedo. Compared to an empirical approach, the model exhibited better performance for simulations of snow and firn melting in the accumulation area and similar performance in the ablation area when forced with meteorological data adjusted with nearby AWS measurements. When such measurements were not available close to the glacier, the empirical model performed better. Our results suggest that simulations of the evolution of future mass balance using an energy balance model require very accurate meteorological data. Given the uncertainties in the temporal evolution of the relevant meteorological variables and glacier surface properties in the future, empirical approaches based on temperature and precipitation could be more appropriate for simulations of glaciers in the future.