An accurate numerical prediction of the oceanic upper layer velocity is a demanding requirement for many applications at sea and is a function of several near-surface processes that need to be incorporated in a numerical model. Among them, we assess the effects of vertical resolution, different vertical mixing parameterization (the so-called Generic Length Scale -GLS- set of k-??, k-??, gen, and the Mellor-Yamada), and surface roughness values on turbulent kinetic energy (k) injection from breaking waves. First, we modified the GLS turbulence closure formulation in the Regional Ocean Modeling System (ROMS) to incorporate the surface flux of turbulent kinetic energy due to wave breaking. Then, we applied the model to idealized test cases, exploring the sensitivity to the above mentioned factors. Last, the model was applied to a realistic situation in the Adriatic Sea driven by numerical meteorological forcings and river discharges. In this case, numerical drifters were released during an intense episode of Bora winds that occurred in mid-February 2003, and their trajectories compared to the displacement of satellite-tracked drifters deployed during the ADRIA02-03 sea-truth campaign. Results indicted that the inclusion of the wave breaking process helps improve the accuracy of the numerical simulations, subject to an increase in the typical value of the surface roughness z₀. Specifically, the best performance was obtained using ??_CH = 56,000 in the Charnok formula, the wave breaking parameterization activated, k-?? as the turbulence closure model. With these options, the relative error with respect to the average distance of the drifter was about 25% (5.5 km/day). The most sensitive factors in the model were found to be the value of ??_CH enhanced with respect to a standard value, followed by the adoption of wave breaking parameterization and the particular turbulence closure model selected. ?? 2009 Elsevier Ltd.