An Intercomparison of Snow Mass Budget over Arctic Sea Ice Simulated by CMIP6 Models

Abstract

Abstract The Arctic has experienced rapid changes in recent decades. For the first time, we intercompare five snow mass budget processes over Arctic sea ice simulated by 22 models from the Coupled Model Intercomparison Project phase 6 (CMIP6) using new diagnostics that have not been available for previous CMIPs. Our analysis suggests that snowfall accumulation (melt) is the dominant process contributing to nearly 100% (70.4% ± 10.1%) of the annual snow growth (loss). Snow mass change through sea ice dynamics, snow–ice conversion, and sublimation contribute 10.9% ± 4.9%, 9.7% ± 5.9%, and 9.0% ± 7.7% to the total snow mass loss. The seasonal cycle of various snow processes simulated by most of the CMIP6 models generally follows similar variations. There is reduced snowfall accumulation, melt, and sea ice dynamics during 1993–2014. However, substantial temporal and spatial discrepancies are noteworthy between the CMIP6 models. There is a large spread of snowfall accumulation and snowmelt in summer and fall, snow–ice conversion from autumn to spring, sublimation in late spring and summer, and snow mass change due to sea ice dynamics from winter to midspring. About half the models show decreasing trends of snowfall accumulation during 1993–2014, with no trends in others. Divergent trends in snow–ice conversion and sublimation occur in the Greenland and Barents Seas. The discrepancies are attributed equally to internal variability and model structural differences. Future projections that remove the identified outlier models suggest a significant reduction in snowfall accumulation, snowmelt, and snow mass change due to sea ice dynamics in the Arctic Ocean from 2015 to 2099. Snow–ice conversion and sublimation are also projected to be reduced but with less confidence.