Evaluation of ERA5, ERA5-Land, CERRA and NEWA datasets in reproducing observed near-surface wind speeds across Spain

Near-surface wind speed (NSWS) is a key variable for climate and energy applications, yet it remains difficult to model because of its sensitivity to local conditions that are often inadequately represented in reanalysis and regional datasets. This study evaluates daily NSWS from four widely used gridded products, ERA5, ERA5-Land, CERRA, and NEWA, against homogenized observations from 748 stations of the Spanish Meteorological Agency for 1/Jan/1990–31/Dec/2018. Performance was quantified using correlation, bias, and error metrics, with results stratified by synoptic circulation types (Jenkinson–Collison classification) and station setting (inland, coastal, mountain). Among the datasets, CERRA showed the best agreement with observations (r = 0.77, cRMSE = 0.95 m/s), largely by reducing unrealistic elevation biases present in ERA5 and ERA5-Land. NEWA systematically overestimated NSWS (MBE = 1.56 m/s) but reproduced the observed elevation dependence more effectively than the other products. Across circulation regimes, all datasets performed best under westerly and south-westerly flows, and worst under weak or easterly conditions, especially in summer and in complex terrain. These findings identify CERRA as the most reliable product for NSWS studies in Spain and regions with similar geographical diversity, while highlighting important limitations of NEWA for applications requiring accurate NSWS.