Neural networks based algorithms for sea surface wind speed retrieval using SSM/I data and their validation

The algorithms for sea surface wind speed retrieval from Special Sensor Microwave/Imager data based on a neural networks approach and some physical limitations are considered. Different physical limitations, based on the atmospheric absorption and applied to training and testing data sets, are used. It is shown that the best results (root mean square difference=1.05 m/s) were obtained for the limitations of horizontal uniformity of the atmosphere with the absence of large absorption.