The effect of wind speed products and wind speed–gas exchange relationships on interannual variability of the air–sea CO2 gas transfer velocity

The lack of a firm relationship between wind speed (U10) and gas transfer velocity (k) is considered to be one of the factors that hinders accurate quantification of interannual variations of ocean–atmosphere CO2 fluxes. In this paper the interannual variations of k of using four different k–U10 parametrizations are examined using wind speed data from the NCEP/NCAR reanalysis project. The extent to which interannual variations are faithfully reproduced in the NCEP/NCAR data is also investigated. This is carried out through comparison with QuikSCAT data. Compared with 4 yr of QuikSCAT data, NCEP/NCAR data reproduce interannual k variations, although the absolute magnitude of k is underestimated. Interannual k variation shows great sensitivity to selection of k–U10 parametrization, and in the Westerlies it changes by a factor of three depending on k–U10 parametrization. Use of monthly mean winds speeds leads to overestimation of interannual k variations compared with k variations computed using 6-hourly wind speeds and the appropriate k–U10 parametrization. Even though the effect of changing k–U10 parametrization is large enough to be an issue that needs to be considered when computing interannual air–sea CO2 flux variations through combining estimates of k with data for the air–sea CO2 gradient, it is not sufficient to bridge the gap between such estimates and estimates based on analyses of atmospheric oxygen, CO2 and δ13C data. Finally it is shown that the ambiguity in the relationship between wind speed and k introduces an uncertainty in interannual flux variations comparable to a bias of interannual ΔpCO2 variations of at most ±5 μatm.