SAR interferometry and statistical topography

The paper introduces a parametric model for the power spectrum density of synthetic aperture radar (SAR) interferograms. This model is derived by assuming the interferogram as a stationary frequency-modulated process and by exploiting the statistical description of earth topography provided by multifractal fields. Despite the small number of parameters involved (three to six in most cases), the model has been proven robust and accurate in most of the cases tested. In particular, the parametric probability distribution assumed for ground terrain slopes fitted the actual histogram (derived from existing digital elevation models) in a range of three to five decades. Applications to blind baseline estimation and phase unwrapping are then briefly discussed.