In Synthetic Aperture Radar Interferometry (InSAR), phase unwrapping holds the key to accurate inversion of digital elevation data. Two new techniques are introduced in this paper that can perform automatic phase unwrapping. The first one is an "optimal" branch-cut algorithm and the second one a hybrid branch-cut/least-square technique, in which pole locations form the weighting basis for the weighted least-square approach. Application of both techniques to ERS-1 data indicates that the height inversion errors are comparable and offer over fifty percent reduction in root mean square (rms) height error compared to the straight least squares method and over thirty-five percent reduction in rms height error compared to the weighted least squares method based on coherence data weighting schemes. The hybrid technique is especially appealing due to its computational efficiency and robustness when compared to traditional branch-cut algorithms.

We study the electromagnetic wave propagation in three-dimensional (3-D) dense random discrete media containing dielectric spheroidal scatterers. We employ a Monte Carlo method in conjunction with the Method of Moments to solve the volume integral equation for the electric field. We calculate the effective permittivity of the random medium through a coherent-field approach and compare our results with a classical mixing formula. A parametric study on the dependence of the effective permittivity on particle elongation and fractional volume is included.