Determination of the Geophysical Model Function of NSCAT and itscorresponding variance by the use of Neural Networks

We have computed two Geophysical Model Functions (one for the vertical and one for the horizontal polarization) for the NSCAT scatterometer by using neural networks. These Neural Network Geophysical Model Functions (NN-GMF) were estimated with NSCAT scatterometer sigma-0 measurements collocated with ECMWF analyzed wind vectors during the period 15 January 1997 to 15 April 1997. We performed a Student t-test showing that the NN-GMFs estimate the NSCAT sigma0 with a confidence level of 95%. Analysis of the results shows that the mean NSCAT signal depends on the incidence angle, on the wind speed and presents the classical bi-harmonic modulation with respect to the wind azimuth. The NSCAT sigma-0 increases with respect to the wind speed and presents a well marked change at around 7 m/s. The upwind-downwind amplitude is higher for horizontal polarization signal than for vertical polarization indicating that the use of horizontal polarization can give additional information for wind retrieval. Comparison of the sigma-0 computed by the NN-GMFs against the NSCAT measured sigma0 show a quite low RMS except at low wind speeds. We also computed two specific neural networks for estimating the variance associated to these GMFs. The variances are analyzed with respect to geophysical parameters. This lead us to compute the geophysical signal to noise ratio, i.e. Kp. The Kp values are quite high at low wind speed and decreases at high wind speed. At constant wind speed, the highest Kp are at cross-wind directions showing that the cross wind values are the most difficult to estimate. These neural networks can be expressed as analytical functions and Fortran subroutines can be provided.