Influence of convection on mixed-layer evolution: comparison of two mixing parameterizations with buoys data in the Bay of Biscay

In this study, we compare two 1-D mixing parameterizations developed by Gaspar et al. [J. Geophys. Res. C95 (C9) (1990) 16179] (G90) and Large et al. [Rev. Geophys. 32 (1994) 363] (L94), respectively. Both models are tested against drifting Marisonde bouys deployed in the Bay of Biscay during PRECOCE experiment (1997–1998) [Mariette, V., Ratsivalaka, C., Verbéque V., Leborgne, E., 1999. CAMPAGNE PREOCOCE (PREdiction du comportement des Couches superficielles de lʹOcéan le long des Côtes Européennes, Tomes 1, 2 and 3, Rapport EPSHOM/CMO/RE/NP 11 du 31 mai 1999]. Periods of stabilizing and destabilizing conditions are successively examined by using both realistic and schematic dynamical and thermodynamical air–sea fluxes. Schematic conditions applied over one diurnal cycle evidence the relative performance of G90 and L94 parameterizations as a function of surface inputs and stratification. The results obtained from these schematic cases are used to compare the results obtained by G90 and L94 over periods of 2 to 10 weeks along three Marisonde buoy trajectories. The ability of both models to simulate the seasonal thermocline formation in Spring as well as its destruction in Fall is discussed. If the nonlocal parameterization used by L94 is taken in its complete form (including the diapycnal mixing), it allows the mixed-layer deepening in Fall in a more satisfactory way than the local parameterization used by G90. The results obtained in Spring by both models are debatable.