Mineral ballasting enhances carbon export from the surface to the deep ocean; however, little is known about the role of this process in the ice-covered Arctic Ocean. Here, we propose gypsum ballasting as a new mechanism that likely facilitated enhanced vertical carbon export from an under-ice phytoplankton bloom dominated by the haptophyte Phaeocystis. In the spring 2015 abundant gypsum crystals embedded in Phaeocystis aggregates were collected throughout the water column and on the sea floor at a depth below 2 km. Model predictions supported by isotopic signatures indicate that 2.7 g m−2 gypsum crystals were formed in sea ice at temperatures below −6.5 °C and released into the water column during sea ice melting. Our finding indicates that sea ice derived (cryogenic) gypsum is stable enough to survive export to the deep ocean and serves as an effective ballast mineral. Our findings also suggest a potentially important and previously unknown role of Phaeocystis in deep carbon export due to cryogenic gypsum ballasting. The rapidly changing Arctic sea ice regime might favour this gypsum gravity chute with potential consequences for carbon export and food partitioning between pelagic and benthic ecosystems.

Snow depth on sea ice is an essential state variable of the polar climate system and yet one of the least known and most difficult to characterize parameters of the Arctic and Antarctic sea ice systems. Here, we present a new type of autonomous platform to measure snow depth, air temperature, and barometric pressure on drifting Arctic and Antarctic sea ice. “Snow Buoys” are designed to withstand the harshest environmental conditions and to deliver high and consistent data quality with minimal impact on the surface. Our current dataset consists of 79 time series (47 Arctic, 32 Antarctic) since 2013, many of which cover entire seasonal cycles and with individual observation periods of up to 3 years. In addition to a detailed introduction of the platform itself, we describe the processing of the publicly available (near real time) data and discuss limitations. First scientific results reveal characteristic regional differences in the annual cycle of snow depth: in the Weddell Sea, annual net snow accumulation ranged from 0.2 to 0.9 m (mean 0.34 m) with some regions accumulating snow in all months. On Arctic sea ice, the seasonal cycle was more pronounced, showing accumulation from synoptic events mostly between August and April and maxima in autumn. Strongest ablation was observed in June and July, and consistently the entire snow cover melted during summer. Arctic air temperature measurements revealed several above-freezing temperature events in winter that likely impacted snow stratigraphy and thus preconditioned the subsequent spring snow cover. The ongoing Snow Buoy program will be the basis of many future studies and is expected to significantly advance our understanding of snow on sea ice, also providing invaluable in situ validation data for numerical simulations and remote sensing techniques.

Water vapor transport has been highlighted as a critical process in Arctic snowpacks, shaping the snow cover structure in terms of density, thermal conductivity, and temperature profile among others. Here, we present an attempt to describe the thermally-induced vertical diffusion of water vapor in the snow cover and its effects of the snowpack structure using the SNOWPACK model. Convection, that may also constitute a significant part of vapor transport, is not addressed. Assuming saturated conditions at the upper boundary of the snowpack and as initial condition, the vapor flux between snow layers is expressed by a 1-dimensional transient diffusion equation, which is solved with a finite difference routine. The implications on the snowpack of this vertical diffusive flux, are analyzed using metrics such as the cumulative density change due to diffusive vapor transport, the degree of over-or undersaturation, the instantaneous snow density change rate, and the percentage of snow density change. We present results for four different regions sampling the space of natural snow cover variability: Alpine, Subarctic, Arctic, and Antarctic sea ice. The largest impact of diffusive water vapor transport is observed in snow on sea ice in the Weddell Sea and the shallow Arctic snowpack. The simulations show significant density reductions upon inclusion of diffusive water vapor transport: cumulative density changes from diffusive vapor transport can reach −62 and −66 kg m −3 for the bottom layer in the shallow Arctic snowpack and snow on sea ice, respectively. For comparison, in deeper snow covers, they rarely exceed −40 kg m −3. This leads to changes in density for shallow snowpacks at the soil-snow interface in the range of −5 to −21%. Mirroring the density decease at depth is a thicker deposition layer above it with increase in density around 7.5%. Similarly, for the sea ice, the density decreased at the sea ice-snow interface by −20%. We acknowledge that vapor transport by diffusion may in some snow covers-such as in thin tundra snow-be small compared to convective transport, which will have to be addressed in future work.

Abstract. Sea ice is an important component of the global climate system. The presence of a snowpack covering sea ice can strongly modify the thermodynamic behavior of the sea ice, due to the low thermal conductivity and high albedo of snow. The snowpack can be stratified and change properties (density, water content, grain size and shape) throughout the seasons. Melting snow provides freshwater which can form melt ponds or cause flushing of salt out of the underlying sea ice, while flooding of the snow layer by saline ocean water can strongly impact both the ice mass balance and the freezing point of the snow. To capture the complex dynamics from the snowpack, we introduce modifications to the physics-based, multi-layer SNOWPACK model to simulate the snow–sea-ice system. Adaptations to the model thermodynamics and a description of water and salt transport through the snow–sea-ice system by coupling the transport equation to the Richards equation were added. These modifications allow the snow microstructure descriptions developed in the SNOWPACK model to be applied to sea ice conditions as well. Here, we drive the model with data from snow and ice mass-balance buoys installed in the Weddell Sea in Antarctica. The model is able to simulate the temporal evolution of snow density, grain size and shape, and snow wetness. The model simulations show abundant depth hoar layers and melt layers, as well as superimposed ice formation due to flooding and percolation. Gravity drainage of dense brine is underestimated as convective processes are so far neglected. Furthermore, with increasing model complexity, detailed forcing data for the simulations are required, which are difficult to acquire due to limited observations in polar regions.

Abstract. Sea ice is an important component of the global climate system. The presence of a snowpack covering sea ice can strongly modify the thermodynamic behaviour of the sea ice, due to the low thermal conductivity and high albedo of snow. The snowpack can be strongly stratified and change properties (density, water content, grain size and shape) throughout the seasons. Fresh water percolation during snow melt can decrease the salinity of the underlying ice, while flooding of the snow layer by saline ocean water can strongly impact both the ice mass balance and the freezing point of the snow. To capture the complex dynamics from the snowpack, we introduce modifications to the physics-based, multi-layer SNOWPACK model to simulate the snow-sea ice system. This involves modifications to the model thermodynamics and to describe water and salt transport through the snow – sea ice system by coupling the transport equation to the Richards equation. These modifications allow the snow microstructure descriptions developed in the SNOWPACK model to be applied to sea ice conditions as well. Here, we drive the model with data from Snow and Ice Mass-balance Buoys installed in the Weddell Sea in Antarctica. The model is able to simulate the temporal evolution of snow density, grain size and shape and snow wetness. The model simulations show abundant depth hoar layers and melt layers, as well as superimposed ice formation due to flooding and percolation. Gravity drainage of dense brine is underestimated as convective processes are so far neglected. Furthermore, with increasing model complexity, detailed forcing data for the simulations is required, which is difficult to acquire due to limited observations in polar regions.