Air-ice carbon pathways inferred from a sea ice tank experiment

Air-ice CO2 fluxes were measured continuously using automated chambers from the initial freezing of a sea ice cover until its decay. Cooling seawater prior to sea ice formation acted as a sink for atmospheric CO2, but as soon as the first ice crystals started to form, sea ice turned to a source of CO2, which lasted throughout the whole ice growth phase. Once ice decay was initiated by warming the atmosphere, the sea ice shifted back again to a sink of CO2. Direct measurements of outward ice-atmosphere CO2 fluxes were consistent with the depletion of dissolved inorganic carbon in the upper half of sea ice. Combining measured air-ice CO2 fluxes with the partial pressure of CO2 in sea ice, we determined strongly different gas transfer coefficients of CO2 at the air-ice interface between the growth and the decay phases (from 2.5 to 0.4 mol m−2 d−1 atm−1). A 1D sea ice carbon cycle model including gas physics and carbon biogeochemistry was used in various configurations in order to interpret the observations. All model simulations correctly predicted the sign of the air-ice flux. By contrast, the amplitude of the flux was much more variable between the different simulations. In none of the simulations was the dissolved gas pathway strong enough to explain the large fluxes during ice growth. This pathway weakness is due to an intrinsic limitation of ice-air fluxes of dissolved CO2 by the slow transport of dissolved inorganic carbon in the ice. The best means we found to explain the high air-ice carbon fluxes during ice growth is an intense yet uncertain gas bubble efflux, requiring sufficient bubble nucleation and upwards rise. We therefore call for further investigation of gas bubble nucleation and transport in sea ice.