Within the frame of the EUMELI program-component of FRANCE-JGOFS-in the Northeast tropical Atlantic ocean, we investigate the potential of a one-dimensional eddy-kinetic-energy model (Gaspar et al., 1990, GGL) to characterize the vertical dynamics of the oceanic mixed layer (ML) at the EUMELI oligotrophic site (21 degrees N, 31 degrees W) north of the Cape Verde Frontal Zone, The atmospheric forcings used are derived from two different sources: the operational Atmospheric General Circulation Model of ECMWF (over two 12-month periods: August 1985-July 1986 and the full year 1990) and climatologies (Esbensen and Kushnir, EK, 1981; Hsiung, H, 1986; Oort, 1983). At the site, depending on the data base, the annual mean of the total energy flux at the ocean-atmosphere interface differs in sign and intensity and its monthly evolution presents significant variation both in amplitude and timing of the maximum. The monthly wind stress evolution due to the regular north-east trade winds prevailing in this region is quite consistent as derived by the different data sources. In our area, a net evaporation rate occurs throughout the year. The simulated ML depth, based on GGL's ML depth definition, is always shallower than climatological observations of ML depth, whatever the surface atmospherical forcing used, the exception being the simulation performed with the atypical ECMWF85-86 forcing. The simulated SST's using H forcing compare rather well (within 1 degrees C) with the observed SST's of the climatologies of Lamb and EK. Sampling experiments on the surface boundary conditions showed that simulated evolutions of the ML depth and SST differ quite significantly due to differences in data bases rather than differences in forcing frequencies. An error analysis on the ocean surface energy fluxes and the prescription of evaporation and precipitation rates under various forms demonstrate the crucial need for heat, momentum and freshwater fluxes estimates as accurate as possible. From the distributions of the turbulent kinetic energy (TKE) budget on different time scales, it is found that most often shear production and viscous dissipation dominate in the ML. Gravitational production or destruction, turbulent diffusion and storage of TKE are of second-order. The use of a daily instead of a 3-hour forcing creates an underestimation of 19% of the total annual energy produced by the shear. When taking into account freshwater fluxes, gravitational production becomes of first order during fall and winter and intervenes in the balance between shear production and viscous dissipation.
