Air–sea exchanges in the equatorial area from the EQUALANT99 dataset: Bulk parametrizations of turbulent fluxes corrected for airflow distortion

Turbulent fluxes of momentum, sensible heat and water vapour were calculated using both the eddy covariance (EC) and the inertial dissipation (ID) methods applied to data collected on board the research vessel La Thalassa during 40 days of the EQUALANT99 oceanographic campaign. The aim of this experiment was to establish accurate parametrizations of air–sea fluxes for the equatorial Atlantic area from a large dataset. However, the accuracy of turbulent fluxes measured aboard ships is strongly affected by the distortion of airflow patterns generated by obstacles such as the ship and mast. For the EQUALANT99 experiment, the effects of airflow distortion were estimated using physical simulations in a water channel. To reproduce the conditions of the campaign, a neutral boundary layer was simulated in the water channel and a detailed model of the ship La Thalassa was built. Correction coefficients for the mean wind speed were evaluated from these physical simulations. They show a dependence on both the azimuth angle of the flow (i.e. the horizontal direction of the flow with respect to the ship's longitudinal axis) and the angle of incidence of the wind. The correction for airflow distortion was applied to the measured wind speed and also included in the flux computation using the ID method. Compared with earlier studies which applied a single correction per flux sample, it appears that our results for the corrected transfer coefficients present greater dependence on neutral wind speed than the previous parametrizations; the method also shows encouraging results, with a decrease in the scatter of the transfer coefficients parametrization. However, the distortion could not be corrected for in the fluxes calculated using the EC method, because this technique integrates a wide range of turbulence scales for which the airflow distortion cannot be simulated in a water channel. Fluxes computed using the ID and EC methods are presented and compared in order to determine which method, in the configuration of the EQUALANT99 experiment, provides the best resulting transfer coefficients. According to the results, fluxes of momentum and latent heat computed by ID were better for deriving the drag and humidity coefficients. The EC method seemed better adapted to calculate sensible-heat fluxes than the ID method, although a high scatter remained on the Stanton neutral number. Copyright © 2005 Royal Meteorological Society