Assessing accurately the surface friction velocity is a key issue for predicting and quantifying aeolian soil erosion. This is usually done either indirectly from the law of the wall (LoW) of the mean wind velocity profile or directly from eddy covariance (EC) of the streamwise and vertical wind velocity fluctuations. However, several recent experiments have reported inconsistency between friction velocities deduced from both methods. Here we reinvestigate the determination of aerodynamic parameters (friction velocity and surface roughness length) over an eroding bare surface and look at the possible reasons for observing differences on these parameters following the method. For that purpose a novel field experiment was performed in South Tunisia under the research program WIND-O-V (WIND erOsion in presence of sparse Vegetation). We find no significant difference between friction velocities obtained from both law of the wall and EC approaches when the friction velocity deduced from the EC method was extrapolated to the surface. Surface roughness lengths show a clear increase with wind erosion, with more scattered values when deduced from the EC friction velocity. Our measurements further suggest an average value of the von Karman constant of 0.407±0.002, although individual wind events lead to different average values due probably to the definition of the ground level or to the stability correction. Importantly, the von Karman constant was found independent of the wind intensity and thus of the wind soil erosion intensity. Finally, our results lead to several recommendations for estimating the aerodynamic parameters over bare surface in order to evaluate saltation and dust fluxes.< Réduire