The adequate modelling of power systems under heavily penetration of Renewable Energy sources crucially depends on the accurate representation of the spatial and temporal features of intermittent power resources supplying the system. In the case of wind energy, two main ingredients are needed for a suitable assessment of its temporal features: a detailed knowledge of the technical parameters of the power generators and the proper description of the physical parameters leading to actual power generation. in this paper we compare the relative weight of uncertainties in the wind power assessment arising from the limited knowledge of the wind farms technical parameters (namely hub height, turbine type and wind farm positioning) with uncertainties originated by the limited description of wind speed fields at hub height provided by meteorological reanalyses. The quantitative comparison of error sources has been achieved by means of a sensitivity analysis taking into consideration the most crucial parameters impacting wind power estimation. The analysis has shown the overwhelming importance of coupling an accurate data base of operating wind farms with a proper representation of the input wind fields at the most possibly detailed level. To our knowledge, this is the first time that the sources of uncertainties for wind power generation estimates have been compared on a continental scale.