Abstract An accurate offshore wind resource assessment is a key tool for planning marine wind renewable exploitation. To achieve this goal, without resort to an extensive and costly network of anemometric stations or buoys, it becomes necessary to use the so-called atmospheric mesoscale models. This work presents a high spatial resolution (1x1 km) offshore wind resource Atlas for Portugal and the model calibration steps. During the calibration steps, the most adequate: i) atmospheric parameterizations - physics options, ii) initial and boundary conditions (IBC) meteorological datasets, and iii) data assimilation scheme were achieved through sensitivity tests using the common statistical metrics and hourly simulated/observational data. Results show that the most significant improvements are associated with the IBC dataset and the data assimilation scheme used. Thus, the results show that the assimilation procedures coupled with the new ERA-5 reanalysis dataset reduce significantly the errors of the wind speed and direction, especially the normalized mean square error. This reduction, depending on the different calibration setup, can be above 50%. The new Atlas confirms the previous indicators, Portugal presents a high wind power potential, especially for deep offshore regions.