Retrieval and assessment of sub-mesoscalewind velocity vectors with Synthetic Aperture Radar

Wind vector fields are currently available from different sources at mesoscale resolutions (2 km to 200 km). The commonly provided wind resolution on 25 km imposes a limit in coastal areas and in the study of small-scale phenomena occurring in the ocean surface. Since wind is the driving force of the ocean, the study of its behavior in sub-mesoscale (< 10 km), improves the understanding of the ocean state. This paper presents a description and assessment of a wind velocity algorithm implemented for C-band (frequency 5.3 GHz) Synthetic Aperture Radar (SAR) data to detect sub-mesoscale wind fields. Results are obtained with the algorithm applied on 14 different SAR images. The quality assessment is performed by the comparison with wind velocity estimates from a numerical weather model and a scatterometer sensor. The statistics of the wind speed retrievals show a bias of about 0.5 m/s, a root mean square (rms) error of 2.6 m/s, and correlation of 79%. For the wind direction, the bias is lower than 8°, with a rms error of about 40° and a correlation of about 85%. The large magnitude of the rms error in the direction is attributed to the differences in variability and resolutions of the data