Snow on Arctic sea ice: model representation and last decade changes

• Main Authors: Castro-Morales, K., Ricker, R., Gerdes, R.
• Format: Other/Unknown Material
• Language: English
• Published: 2018
• Subjects: Arctic; Sea ice
• Online Access: https://doi.org/10.5194/tcd-9-5681-2015; https://tc.copernicus.org/preprints/tc-2015-162/

Together with sea ice, Arctic snow has experienced vast changes during the last decade due to a warming climate. Thus, it is relevant to study the past and present changes of Arctic snow to understand the implications to the sea ice component, precipitation, heat and radiation budgets. In this study, we analyze the changes of snow depth between 2000 and 2013 at regional scale represented in an Arctic coupled sea ice-general circulation model. We evaluate the model performance by direct comparison of the modeled snow depths ( h s_mod ) to snow depths from radar measurements from the NASA Operation IceBridge ( h s_OIB ) during the flight campaigns completed from 2009 to 2013. Despite the description of the snow in our model is simple (i.e. single layer without explicit snow redistribution processes) as in many current sea-ice models; the latitudinal distribution of h s_mod in the western Arctic is in good agreement to observations. The h s_mod is on average 3 cm thicker than h s_OIB in latitudes > 76° N. According to the model results, the h s in 2013 decreased 21 % with respect to the multi-year mean between 2000 and 2013. This snow reduction occurred mainly in FYI dominated areas, and is in good agreement to the year-to-year loss of sea ice, also well reproduced by the model. In a simple snow mass budget, our results show that 65 % of the yearly accumulated snow is lost by sublimation and snowmelt due to the heat transfer between the snow/ice interface and the atmosphere. Although the snow layer accumulates again every year, the long-term reduction in the summer sea-ice extent ultimately affects the maximum spring accumulation of snow. The model results exhibit a last decade thinning of the snowpack that is however one order of magnitude lower than previous estimates based on radar measurements. We suggest that the later is partially due to the lack of explicit snow redistribution processes in the model, emphasizing the need to include these in current sea-ice models to improve the snow representations.