Development of an Empirical Model for Bidirectional Ocean Surface Slope Statistics for use in Reflected GPS Remote Sensing

The cross-correlation waveform generated from GPS signals bistatically scattered from the ocean surface has been shown to be dependent upon the slope statistics of that surface. This is usually characterized by a probability density function (PDF) of small facets which approximate the surface. Existing empirical models, such as those from Cox and Munk (1954), Shaw and Churnside (1997) or Liu, et al (1997) have been adapted to this problem. Recently, however, Bertuccelli and Garrison (2001) presented an empirical model derived directly from the GPS waveform observations themselves. That model approximated the random distribution of surface facets with an isotropic Gaussian PDF. An improved, bidirectional, model has been derived from a combination of GPS measurements, and surface truth (buoy) data. In this model, wind direction information from buoys or wind sondes is used to rotate the principal axes of the two-dimensional Gaussian PDF. Next, a nonlinear least squares technique is applied to small segments of reflected GPS data to estimate the upwind and cross-wind slope variances which describe that PDF. The dependence of slope variances on surface wind speed is then obtained through a second parameter estimation using wind speed data from the buoys or sondes. Application of this new model to bistatic GPS sensing of surface wind speed and direction will be discussed.