Rethinking Craig and Gordon's approach to modeling isotopic compositions of marine boundary layer vapor

Abstract

We develop a one-dimensional (1-D) steady-state isotope marine boundary layer (MBL) model that includes meteorologically important features missing in models of the Craig and Gordon type, namely height-dependent diffusion and mixing, lifting to deliver air to the free troposphere, and convergence of subsiding air. Kinetic isotopic fractionation results from this height-dependent diffusion that starts as pure molecular diffusion at the air-water interface and increases with height due to turbulent eddies. Convergence causes mixing of dry, isotopically depleted air with ambient air. Model results fill aquadrilateral in D-18O space, of which three boundaries are defined by (1) vapor in equilibrium with various sea surface temperatures (SSTs), (2) mixing of vapor in equilibrium with seawater and vapor in subsiding air, and (3) vapor that has experienced maximum possible kinetic fractionation. Model processes also cause variations in d-excess of MBL vapor. In particular, mixing of relatively high d-excess descending and converging air into the MBL increases d-excess, even without kinetic isotope fractionation. The model is tested by comparison with seven data sets of marine vapor isotopic ratios, with excellent correspondence. About 95% of observational data fall within the quadrilateral predicted by the model. The distribution of observations also highlights the significant influence of vapor from nearby converging descending air on isotopic variations within the MBL. At least three factors may explain the 5% of observations that fall slightly outside of the predicted regions in D-18O and d-excess18O space: (1) variations in seawater isotopic ratios, (2) variations in isotopic composition of subsiding air, and (3) influence of sea spray.