Sea-roughness spectrum retrieval from radar and radiometric measurements

Numerical simulations of radar backscattering cross sections from sea roughness have previously been performed for K/sub u/- and C-bands for various wind speeds and incident angles using the small-slope approximation. The Elfouhaily et al. surface-height spectrum with Gaussian sea-roughness statistics was used for numerical calculations of scattering, and the results were compared with empirical scattering models CMOD2-I3 and SASS-II. The simulations showed good agreement for the zeroth-order azimuthal harmonic (/spl sim/1.5 dB mismatch) of the backscattering cross section that dominates the expansion of the total backscattering cross section. However, the satellite scatterometry data for small incidence angles indicate that the scattering is nearly isotropic in this region, especially for low winds. This is incompatible with the Elfouhaily et al. model, which is significantly anisotropic for sufficiently long waves. To achieve better agreement with the scatterometric data for the second-order azimuthal harmonic of the normalized radar cross section (NRCS), a modification of the directional part of the Elfouhaily et al. spectrum was considered. The modification is wind-dependent and is most significant for low winds. Then the brightness temperature of the sea surface at 37 GHz was calculated based on the small-slope expansion both for original and for modified roughness spectra. Somewhat better agreement with measurements was also observed for the modified directional part of a spectrum. In contrast to NRCS, brightness temperature appeared to be sensitive to the spectral model modifications at large incidence angles. This work can be considered as an attempt to solve the inverse problem of inferring sea-roughness spectrum based on remote sensing data.