The impact of shallow stratification on air-sea CO2 flux in the summer Arctic Ocean

Air-sea carbon dioxide (CO₂) flux is often indirectly estimated by the bulk method using the in-situ air-sea difference in CO₂ fugacity and a wind speed dependent parameterisation of the gas transfer velocity (K). In the summer, sea-ice melt in the Arctic Ocean generates strong shallow stratification with significant gradients in temperature, salinity, dissolved inorganic carbon (DIC) and alkalinity (TA), and thus a near-surface CO₂ fugacity  (fCO_2w) gradient. This gradient can cause an error in bulk air-sea CO₂ flux estimates when the fCO_2w is measured by the ship’s underway system at ~5 m depth. Direct air-sea CO₂ flux measurement by eddy covariance (EC) is free from the impact of shallow stratification because the EC CO₂ flux does not rely on a fCO_2w measurement. In this study, we use summertime EC flux measurements from the Arctic Ocean to back-calculate the sea surface fCO_2w and temperature and compare them with the underway measurements. We show that the EC air-sea CO₂ flux agrees well with the bulk flux in areas less likely to be influenced by ice melt (salinity > 32). However, in regions with salinity less than 32, the underway fCO_2w is higher than the EC estimate of surface fCO_2w and thus the bulk estimate of ocean CO₂ uptake is underestimated. The fCO_2w difference can be partly explained by the surface to sub-surface temperature difference. The EC estimate of surface temperature is lower than the sub-surface water temperature and this difference is wind speed-dependent. Upper-ocean salinity gradients from CTD profiles suggest likely difference in DIC and TA concentrations between the surface and sub-surface water. These DIC and TA gradients likely explain much of the near-surface fCO_2w gradient. Accelerating summertime loss of sea ice results in additional meltwater, which enhances near-surface stratification and increases the uncertainty of bulk air-sea CO₂ flux estimates in polar regions.