Energy-balance modeling of glacier mass balance in Iceland

The detailed, physically-based, one-dimensional snowpack model SURFEX/ISBA-Crocus was used to simulate the seasonal and annual surface mass balance (SMB) of the five largest Icelandic glaciers (Vatnajökull, Langjökull, Hofsjökull, Mýrdalsjökull and Drangajökull) for the period of 1990–2024, for which annual mass-balance measurements are available. The model is forced using the high-resolution Copernicus Arctic Regional Reanalysis (CARRA) dataset. Near-surface air temperature from CARRA was downscaled using a daily lapse rate (multi-year average for each day of the year) using Digital Elevation Models (DEM) with 100-m or 250-m resolution. In addition, a precipitation/elevation gradient of 10% per 100 m was applied using the elevation difference between the higher-resolution DEM and the CARRA DEM following a reclassification of rainfall and snowfall to take into account the downscaled temperature. Daily albedo from the Moderate Resolution Imaging Spectroradiometer (MODIS), data for 2000–2024, and observations from the Icelandic Glacier Automatic Weather Station (ICE-GAWS) network, data for 2001–2024, as well as the glacier mass-balance data, are used to calibrate the model. An intermediate complexity modeling strategy was considered in this work, which takes into account the effect of light absorbing particles (LAP) on albedo using an LAP-informed and spatially variable snow-darkening coefficient to control the time evolution of snow albedo in the visible range. Roughness length of pure snow (Z_0snow) in the range of 0.1–10 mm was also used as a calibration parameter. Multi-year mean albedo values from MODIS were calculated for the period of 2000–2017 to represent the end-of-summer ice albedo, used as inputs to the model at each of the SMB measuring stakes. The modeled seasonal and annual SMB results were compared to measurements at more than 150 stakes on the five glaciers. Summer and winter mass balances were well predicted by the model (the model explains 70–80% of the variance, RMSE = 0.6–0.9 m w.e.). The modeled albedo was compared with the observed albedo values from MODIS and ICE-GAWS data. The model captures the temporal evolution of albedo relatively well, but generally underestimates high albedo values and overestimates low albedo values. The model was also not able to capture variability in ice albedo during the ablation season due to the use of a constant multi-year average ice albedo, which results in an underestimation of the highest melt values. Further improvements in the model are under development, including a correction of the multi-year late-summer ice albedo from MODIS to improve summer melt estimation.