[1] In this paper, we present the results obtained with the eddy correlation method applied to data acquired during the Flux, Etat de mer et Télédetection en Condition de Fetch variable experiment onboard the R/V L'Atalante. We discuss the corrections made to account for platform motion and for the effects of mean airflow distortion. The data are compared to those obtained from a moored Air-Sea Interaction Spar (ASIS) buoy and from the L'Atalante using the inertial dissipation method (IDM). The main results from eddy correlation method on L'Atalante are that the momentum flux in-line with the mean wind, Àhu 0 w 0 i, is overestimated by 18%, likely due to turbulent flow distortion around the ship. In contrast, the results for heat flux do not appear to be contaminated by turbulent flow distortion. Indeed, heat fluxes obtained using the sonic temperature on the L'Atalante and the ASIS buoy are very similar. The eddy correlation latent heat fluxes obtained on the L'Atalante using a refractometer are significantly higher than those obtained from the same sensor using the IDM.

[1] The inertial dissipation method (IDM) is commonly used to measure turbulent fluxes over the ocean. It has the advantage over more direct methods in that it depends on the turbulent fluctuations only in the high frequencies of the so-called inertial subrange. These frequencies are above those of typical ship motions and are considered to be relatively unaffected by flow distortion. However, a drawback in applying the method is that the problem is underdetermined: estimation of the fluxes requires knowledge of the Obukhov length L, which is itself a function of the fluxes. The problem is typically solved by iteration, using an initial L estimated from bulk formulae. This introduces a possible dependency on the initial bulk estimate along with problems of convergence. Recently, several authors have proposed improvements to the basic algorithm. For instance, Dupuis et al. [1997] proposed a parameterization of the ''imbalance term'' in the budget of turbulent kinetic energy (TKE). We explore an alternative approach to the problem. In order to constrain the equations resulting from the IDM we use the vertical velocity variance, s w, measured from the research vessel L'Atalante and an ASIS buoy, both deployed during the 1998 FETCH experiment. These data are compared to several parameterizations of s w on stability derived in experiments. For unstable cases, the data are found to be well described by the Panofsky and Dutton [1984] parameterization, although the scatter of the data is higher for swell conditions than for pure wind sea, indicating a likely sea state effect. Using measured values of s w along with this parameterization, the inertial dissipation problem is fully specified. The convergence of the method is satisfactory, and it offers u * estimates independent of bulk formulae.