On physical mechanisms controlling air–sea CO 2 exchange

. Reducing uncertainties in the air–sea CO 2 flux calculations is one of the major challenges when addressing the oceanic contribution in the global carbon balance. In traditional models, the air–sea CO 2 flux is estimated using expressions of the gas transfer velocity as a function of wind speed. However, other mechanisms affecting the variability in the flux at local and regional scales are still poorly understood. The uncertainties associated with the flux estimates become particularly large in 5 heterogeneous environments such as coastal and marginal seas. Here, we investigated the air–sea CO 2 exchange at a coastal site in the central Baltic Sea using nine years of eddy covariance measurements. To the best of our knowledge, this is the longest record of direct observations of CO 2 fluxes and the corresponding gas transfer velocities ( k ) in a marine environment. Based on these observations we were able to capture the temporal variability of the air–sea CO 2 flux and other parameters relevant for the gas exchange. The analysis of water-side and atmospheric control mechanisms showed that during wind speeds above 8 m s − 1 10 the conditions on both sides of the air–water interface were relevant for the gas exchange. Our findings further suggest that at such relatively high wind speeds, sea spray was an efficient mechanisms for air–sea CO 2 exchange. In contrast to high wind-speed conditions, during low wind speeds (<6 m s − 1 ) only water-side processes were found to be relevant control mechanisms, in particular, water-side convective processes. Altogether, our results show that currently existing wind-based parametrizations of k might be good approximations as long-term averages for environments with coastal characteristics. However, in order to 15 reduce the uncertainty associated to these averages and produce reliable short-term k estimates, additional physical processes must be considered. 2008). A more detailed description of the Östergarnsholm site can be found in Rutgersson et al. (2020). et al. (2018) based on the skin temperature, we calculated a rough estimate of the effect of surfactants 355 on our long-term k 660 . The results suggest an overall effect of merely -0.1 %, suggesting a very small reduction of k 660 when surfactants are taken into account. A detail analysis of the effect of surface films is beyond the scope of this study. Nevertheless, we recognize that this process might be particularly relevant in coastal seas and other shallow bodies of water. The analysis of the effect of sea ice, precipitation, Langmuir circulation, and other processes in the Baltic Sea is still pending.