Air-sea Exchange of Dimethylsulfide (DMS) - Separation of the Transfer Velocity to Buoyancy, Turbulence, and Wave Driven Components

In the past several years, we have measured the sea-to-air flux of DMS directly with eddy covariance on five cruises in distinct oceanic environments, including the equatorial Pacific (TAO 2003), Sargasso Sea (Biocomplexity 2004), Northern Atlantic (DOGEE 2007), Southern Ocean (SO-GasEX 2008), and Peruvian/Chilean upwelling region (VOCALS-REx 2008). Normalizing DMS flux by its concurrent air-sea concentration difference gave us the transfer velocity of DMS (kDMS). Our wealth of kDMS measurements (~2000 hourly values) in very different oceans and across a wide range of wind speeds (0.5~20.5 m/s) provides an opportunity to evaluate existing parameterizations of k and quantify the importance of various controlling factors on gas exchange. Gas exchange in different wind speed regimes is driven by distinct physical mechanisms. In low winds (<4 m/s), buoyancy-driven convection results in a finite and positive kDMS. In moderate winds (4~10 m/s), turbulence from wind-stress prevails, as we found a near linear dependence of kDMS on wind speed and on friction velocity (u*). In high winds (>10 m/s), there is additional bubble-mediated exchange from wave-breaking, which depends on gas solubility (a function of temperature and to a lesser degree, salinity). When normalizing kDMS to a reference temperature of 20°C, we found the oft-used Schmidt number correction (for diffusivity) to be inadequate because it does not account for the temperature dependence in solubility. To quantify the solubility effect, we subtract the small buoyancy-driven term computed by the NOAA-COARE model 3.0a from k660 (kDMS corrected to a Schmidt number of 660). A linear fit to the residual k660 in the moderate wind regime allows us to further separate the turbulence-driven and wave-breaking components. A solubility correction is applied to the latter, which is then added back to the buoyancy and turbulence-driven terms to give k660,C. Compared to k660, k660,C shows a significant reduction in scatter from moderate to high winds, as ~20% more variance can be explained by wind speed after the solubility correction. For the less soluble CO2 and SF6 that have greater bubble-mediated components, this solubility correction is potentially even more important. Transfer velocity of DMS before and after solubility correction. The amount of scatter is significantly reduced after the correction.