Multiscale Surface Roughness and Backscattering

Surface roughness is usually described with the two classical constant parameters, rms height and correlation length, which is determined using the autocorrelation function. Unfortunately, the rms height and correlation length of natural surfaces, such as sea ice, depend on the measured distance [1,2]. This is understandable since the shape of natural solid surfaces is often a result of a number of random changes superposed on an initial surface. Spectral analysis would permit multiscale roughness, but the spectrum of a sparse data set does not represent the real target spectrum well. A description for a surface with multiscale roughness was developed using autocorrelation functions and is presented in this paper.