Abstract. In this study, two types of vertical turbulence closure models are tested in the Mediterranean Sea in a one-dimensional configuration. The numerical experiments are performed at different locations in the Mediterranean for which the year 2004 is simulated. The model results are then compared and validated with in-situ temperature observations. For the model simulations, initial profiles of temperature and salinity come from the ARGO (Array for Real-time Geostrophic Oceanography) profiles. The surface forcing (momentum, heat) is calculated from bulk formulae using 6-hourly atmospheric data from the European Center for Medium Range Weather Forecast (ECMWF). The vertical mixing schemes tested in this study are a second-order statistical model (k-ε) and the non-local K-profile parameterization (KPP). Both schemes yield similar results in terms of reproducing the water column dynamics. A major source of discrepancy between model and observations comes from the uncertainties in the atmospheric forcing parameterization. At this point, net shortwave radiation data from NCEP atmospheric reanalysis has been used obtaining a more realistic Sea Surface Temperature (SST) compared with satellite observations for the summer months.