Ocean winds retrieved from X-band radar-image sequences

A new method for retrieving wind speeds and directions using nautical radar-image sequences is presented. The method consists of two parts, one for wind direction and another for wind speed retrieval. Wind directions are locally extracted from wind induced streaks, which are approximately in line with the mean wind direction. The algorithm assumes wind direction as normal to the local gradients of the amplitude image. Wind speeds are derived from the radar cross section, by parameterization of its dependency on the wind vector, which is performed by training of a neural network. For verification of the method the wind direction and speed from nearly 1400 radar-image sequences are compared to in situ data from a wind sensor. The accuracy and limitations of the method are discussed. A second new method is introduced, which enables to retrieve spatial and temporal wind fields from radar-image sequences. Thereby the wind streaks are available in space and time. The local velocity and direction of the wind pattern of each point in the investigated area is determined using tensor-based techniques. This method has the advantage that no calibration of the radar images or training of a neural network is necessary.