Surface current maps derived from High Frequency (HF) Doppler radar observations provide valuable information for numerical ocean models' validation, as well as for improvement of model nowcasts and forecasts. Assimilation of HF radar (CODAR)-derived surface currents into the fine-resolution numerical ocean model of the Monterey Bay Area is discussed in this paper. Effective data assimilation techniques for surface information rely on methods for projecting this information into the interior of the ocean for improving dynamical forecasts of the currents and hydrographic structure of the ocean's interior. For this reason, special attention is devoted in this paper to the sub-surface projection of CODAR-derived surface information in three-dimensional data assimilation of surface currents. Different approaches based on the modification of the ICON model surface forcing (wind forcing) as well as direct instantaneous projection of surface information into the subsurface of the model are presented and compared. The significant improvement of correlation between the model and ADCP subsurface currents was achieved when the subsurface projection of the corrections to the model surface currents was based on the Ekman theory.