Ica Techniques Applied to Sar Interferometry

This paper deals with the validation of Blind Source Separation techniques for an automatic elimination of effects due to atmospheric fluctuations and microclimates or any other spurious artifact eventually present in SAR interferograms. Specifically, an Independent Component Analysis algorithm (the FastICA) is applied on arrays of ERS-1/ERS-2 interferograms and the results are shown for mountainous areas of Southern Italy. The characteristics of the extracted independent components are illustrated. Starting from observational recurrences, the deduction of the criteria for identifying the topographical and atmospheric sources is illustrated. Some validations for these criteria are provided by comparisons with standard DEM (Digital Elevation Model).