Wind Class Sampling of Satellite SAR Imagery for Offshore Wind Resource Mapping

Abstract

Abstract High-resolution wind fields retrieved from satellite synthetic aperture radar (SAR) imagery are combined for mapping of wind resources offshore where site measurements are costly and sparse. A new sampling strategy for the SAR scenes is introduced, based on a method for statistical–dynamical downscaling of large-scale wind conditions using a set of wind classes that describe representative wind situations. One or more SAR scenes are then selected to represent each wind class and the classes are weighted according to their frequency of occurrence. The wind class methodology was originally developed for mesoscale modeling of wind resources. Its performance in connection with sampling of SAR scenes is tested against two sets of random SAR samples and meteorological observations at three sites in the North Sea during 2005–08. Predictions of the mean wind speed and the Weibull scale parameter are within 5% from the mast observations whereas the deviation on power density and the Weibull shape parameter is up to 7%. These results are promising and may be improved further through a better population of the wind classes. Advantages of the wind class sampling method over random sampling include, in principle, selection of the most representative SAR scenes such that wind resources can be predicted from a lower number of SAR samples. Furthermore, the wind class weightings can be adjusted to represent any time period.