Improved Small Baseline processing by means of CAESAR eigen-interferograms decomposition

The Component extrAction and sElection SAR (CAESAR) is a method for the selection and filtering of scattering mechanisms recently proposed in the multibaseline interferometric SAR framework. Its strength is related to the possibility to select and extract multiple dominant scattering mechanisms, even interfering in the same pixel, since the stage of the interferograms generation, and to carry out a decorrelation noise phase filtering. Up to now, the validation of CAESAR has been addressed in the framework of SAR Tomography for the model-based detection of Persistent Scatterers (PSs). In this paper we investigate the effectiveness related to the use of CAESAR eigen-interferograms in classical multi-baseline DInSAR processing, based on the Small BAseline Subset (SBAS) strategy, typically adopted to extract large scale distributed deformation and atmospheric phase screen. Such components are also exploited for the calibration of the full resolution data for PS or tomographic analysis. By using COSMO-SKyMed (CSK) SAR data, it is demonstrated that dominant scattering component filtering effectively improves the monitoring of distributed spatially decorrelated areas (f.i. bare soil, rocks, etc.) and allows bringing to light man-made structures with dominant backscattering characteristics embedded in highly temporally decorrelated scenario, as isolated asphalt roads and block of buildings in non-urban areas. Moreover it is shown that, thanks to the CAESAR multiple scattering components separation, the layover mitigation in low-topography eigen-interferograms relieves Phase Unwrapping (PhU) errors in urban areas due to abrupt height variations.