Local mixed layer response to wind and buoyancy forcing in the western equatorial Pacific

The behavior of three types of widely used one-dimensional mixed layer mode ls (bulk, turbulence closure, and shear instability) is evaluated under ide alized conditions representative of atmospheric forcing over the western eq uatorial Pacific Ocean. Results show that under heavy precipitation (freque ntly observed in this region), models with a shear instability mixing compo nent capture the fundamental behavior of the mixed layer better than do tho se without. On the basis of this comparison, a version of the Price-Weller- Pinkel (PWP) [Price et al., 1986; Schudlich and Price, 1992] mixed layer mo del is selected for the purpose of simulating nearsurface temperature and s alinity measurements in this region. Heat and salt budgets calculated from observations taken during the Tropical Ocean Global Atmosphere-Coupled Ocea n-Atmosphere Response Experiment (TOGA-COARE) pilot study suggest that abov e the top of the thermocline, Local atmospheric forcing is indeed of first- order importance in the evolution of the temperature field. Horizontal adve ction of salt, however, is observed to be as important as local forcing (ev aporation minus precipitation) in the evolution of the salinity field. Driv en by observed fluxes, numerical simulation using the PWP model confirm the primary importance of local forcing for the temperature field and the impo rtance of salinity advection for the salinity field. During the TOGA-COARE intensive observing period, simulated and observed temperatures agree well at 9 m depth under weak winds, but the model/data discrepancy gradually inc reases during periods of stronger westerly winds. The addition of a simplif ied meridional temperature advection term, based on the mean observed tempe rature gradient, reduces but does not eliminate the model data discrepancy. The magnitude of model/data salinity discrepancies is consistent with conc urrent measurements of instantaneous horizontal salinity gradients and velo cities, but not with their time-mean values.