Three-dimensional InSAR displacement profiles exploiting multi-platform SAR acquisitions: Application to the slow-varying landslide of Gorgoglione (Italy)

Ground displacement time-series are standard outputs of consolidated multi-temporal Synthetic Aperture Radar Interferometric (InSAR) algorithms. These products make it possible to remotely detect the spatial and temporal evolution of the deformation field relative to investigated stable targets on the ground with centimeter or even millimeter precision. The occurred deformation is perceived as a change in the round-trip Line-Of-Sight (LOS) path from the sensor to the target; therefore, unambiguous projection in three-dimensional (3-D) space is an undetermined problem. Instead, discerning the vertical [up-down (UD)] and horizontal [east-west (EW)] profiles can be achieved using complementary ascending and descending satellite orbits, assuming however the north-south (NS) profile is negligible. Furthermore, the addition of other independent ascending and descending observations is valuable in order to get bidimensional estimates with greater precisions, but at the same time it is not sufficient to recover the NS profiles due to the lack of sensitivity of the polar orbits to that component. In this context, multi-platform SAR acquisitions from complementary views can be used with the polar-orbiting satellite counterpart to resolve full 3-D displacement profiles.

In this work, we propose a multi-platform procedure to integrate satellite SAR data collected from ascending and descending orbits with data collected from ground-based SAR (GB-SAR) systems to compute 3-D InSAR ground displacement maps and time-series. At this aim, the multi-platform data are first processed independently in order to obtain single-look georeferenced InSAR LOS products, using advanced or canonical multi-temporal InSAR processors and, then, the data are integrated by solving a determined system of linear equations where the unknowns are the temporal samples common to all multiplatform datasets. Note that, as not all data are acquired in the same temporal instant, a quasi-synchronous temporal matching procedure is applied. Also, a theoretical variance-covariance-based framework is proposed to assess the precision of the 3-D estimates.

The developed algorithm was applied to the slow-varying landslide of Gorgoglione in the south-western part of Matera Province (Basilicata Region, southern Italy) on a hilly area at about 800 m a.s.l. by processing ascending and descending Copernicus Sentinel-1 A/B C-Band acquisitions, and a set of GB-SAR IBIS-L Ku-Band images; the results shown that during the analyzed period (September 2016 - July 2017) the landslide area was subject to a deformational trend along the southern profile and to a vertical subsidence trend in accordance with the morphology of the landslide itself.