PSI - Issue 84

Luciano Pavesi et al. / Procedia Structural Integrity 84 (2026) 583–590

586

enhance the representation of topography and anthropogenic structures, starting from the generation of updated, high resolution Digital Elevation Models (DEMs). In particular, the Synthetic Aperture Radar (SAR) imagery acquired by the ASI’s COSMO-SkyMed constellation is processed using interferometric techniques to derive high-resolution elevation models, reaching meter-scale resolution. The availability of multi-geometry acquisitions (ascending and descending orbits) improves DEM quality in complex or poorly accessible areas and supports periodic updates of topographic information, enabled by the different viewing angles associated with specific satellite orbits. The Differential Interferometric Synthetic Aperture Radar (DInSAR) processing chain is used to derive a Digital Elevation Model (DEM) from a pair of SAR images. More generally, Interferometric Synthetic Aperture Radar (InSAR) exploits the phase difference between two complex SAR acquisitions obtained from slightly different positions of either the same satellite sensor or two satellites to extract information about the Earth’s surface. Interferometry involves combining the phase information of two SAR images (the reference and the secondary image).

Fig. 1. The RESCUE_SAT project.

Without including additional technical details, the interferometric phase is obtained from the combination of two accurately co-registered SAR images. After removing the flat Earth phase, it can be expressed as the sum of four components (Ferretti et al., 2001): topographic contribution ( topo ), ground deformation ( def ), atmospheric effects ( atm ), and noise ( noise ). int = topo + def + atm + noise (1) This assumption is generally valid for short temporal baselines. Atmospheric disturbances and noise, including thermal effects and decorrelation, may introduce errors and are mitigated through appropriate filtering and compensation techniques. For DEM generation, it is typically assumed that the interferometric phase is dominated by topographic information, implying that no significant surface displacement occurred between the two acquisitions. Under this assumption, def can generally be neglected when using an SAR image pair with a short temporal baseline of a few days. In addition, DEM extraction using the DInSAR approach requires several acquisition constraints to ensure high precision while minimizing artifacts and errors. These include the use of identical sensors, consistent viewing geometry (Ascending/Ascending or Descending/Descending), the same incidence angle, identical data types, and co-polarized images (VV or HH). The processing workflow involves interferogram generation, adaptive filtering and coherence estimation, phase unwrapping, orbit refinement, and finally conversion of phase values into elevation followed by geocoding. At the end