A NEURAL-NETWORK APPROACH FOR MODELING NONLINEAR TRANSFER-FUNCTIONS -APPLICATION FOR WIND RETRIEVAL FROM SPACEBORNE SCATTEROMETER DATA

The present paper shows that a wide class of complex transfer function s encountered in geophysics can be efficiently modeled using neural ne tworks. Neural networks can approximate numerical and nonnumerical tra nsfer functions. They provide an optimum basis of nonlinear functions allowing a uniform approximation of any continuous function. Neural ne tworks can also realize classification tasks. It is shown that the cla ssifier mode is related to Bayes discriminant functions, which give th e minimum error risk classification. This mode is useful for extractin g information from an unknown process. These properties are applied to the ERS1 simulated scatterometer data. Compared to other methods, neu ral network solutions are the most skillful.