Evaluation of ERA5, COSMO-REA6 and CERRA in simulating wind speed along the French coastline for wind energy applications

Abstract. The quality of wind speed from different reanalyses (ERA5, COSMO-REA6 and CERRA) is assessed along the different coasts of mainland France. SYNOP (surface synoptic observations) wind measurements from Météo-France at 10 m height and floating LiDAR (Light Detection and Ranging) measurements from DGEC (Direction Générale de l'Énergie et du Climat) at 100 m height are used as reference. The inter-annual variability, distribution of wind, seasonal cycle, diurnal cycle and extremes are evaluated using several metrics (bias, correlation, normalized root mean square error). Results show that the shape of the 10 m wind seasonal cycle is well represented by all reanalyses. However, along the Mediterranean coast, wind speed is underestimated by ERA5, and overestimated by COSMO-REA6 during winter. COSMO-REA6 does not reproduce well the diurnal cycle along the Mediterranean coast, nor does ERA5 for the Atlantic coast. Overall, CERRA has better skills in representing surface wind speed on the three French seafronts, as well as for offshore wind speed at 100 m. The present study provides insights on the use of CERRA as a reference for offshore wind studies over the French maritime zone.