Presentation Description: European synthetic-aperture radars (SAR) provide a 20-year database of high-resolution sea surface roughness observations over the globe. These observations are a direct reflection of governing atmospheric phenomena, especially air-sea and land-sea interactions that are not well captured by mesoscale modelling. Dedicated processing is needed to remove artefacts nonrelated to wind stress (e.g. ships, pollution, etc.) to retrieve geophysical quantities such as the surface wind speed. An advanced extrapolation algorithm based on machine learning techniques has been developed and trained with 12 lidars in the North Sea in order to provide wind speed at altitudes relevant for offshore wind applications. The extrapolation is validated over 28 lidars in Germany, the Netherlands, Belgium, France, US (East and West coasts), China. In general, the mean absolute bias is 2% for the SAR-extrapolated wind speed compared to 4% for advanced mesoscale models. Here, we compare the performance against surface winds from NDBC met buoys and floating Lidars deployed in the US East Coast. These unique coverage, precision and resolution from SAR measurements bring great benefits such as 500-m resolution wind atlases with spatial heterogeneities for the characterization of wind conditions in coastal/offshore regions, hence helping in early screening of development zones and designing lidar campaigns. While the extrapolation methodology does not require any on-site measurements, on-site Lidar data can be integrated to improve the accuracy of the SAR-based wind resource, therefore taking advantage of the accuracy of Lidar measurements and of the spatial distribution assessed by SAR.
.png)
.png)
