PSI - Issue 84

Alice Vitaletti et al. / Procedia Structural Integrity 84 (2026) 191–198

195

3.2. InSAR data Satellite-based data measure deformation along the sensor’s line of sight, providing a one-dimensional projection of the actual three-dimensional displacement. This limitation is particularly relevant for landslides, where movement is mainly horizontal, potentially leading to underestimated, or undetected, displacements. A reliable interpretation of slope kinematics thus requires converting LOS measurements into actual displacement components. Since ascending and descending observations are rarely spatially and temporally coincident, the datasets must be resampled and temporally aligned through post-processing. A grid-based resampling procedure, following Proietti et al. (2009), was applied by dividing the study area into 60 x 60 m cells. For each cell, the centroid (synthetic PS) was assigned the average LOS velocity of all PS within the cell, for both geometries and all acquisition dates. Temporal interpolation was also carried out over the common monitoring interval (06/01/2018-22/12/2022). This approach ensured a uniform spatial distribution of measurement points and enabled a consistent representation of deformation across the landslide (Fig. 3a).

a

b

Fig. 3. (a) Grid-based resampling of ascending (left) and descending (right) datasets. (b) Local reference system (L, T, V).

Estimating the 3D displacement vector requires at least three independent observations from different geometries. Since only two geometries (ascending and descending) are available, the system is underdetermined. To overcome this limitation, a local reference system (L, T, V) fixed to the deformation phenomenon was adopted, following the methodology proposed by Farneti et al. (2023) (Fig. 3b). In this reference system, the longitudinal axis (L) is parallel to the iso-elevations lines of the slope, whereas the transverse axis (T) corresponds to the downslope direction. Thus, LOS displacement can be expressed as a linear combination of the actual components: = (1) where: ={ } , = [ ( + ) ( + ) ( + ) ( + ) ] , = { }. where θ is the LOS incidence angle and φ is the angle defining the transverse axis of the actual displacement in the horizontal plane with respect to the global reference system (E, N, V). With two LOS measurements and three unknowns, an assumption regarding the landslide displacement direction is required. Therefore, the direction derived from the FEM model was used to define the orientation of the local reference system (L, T, V). The numerical simulations revealed predominantly downslope movement (i.e., along the transverse axis, T), supporting the assumption of negligible longitudinal displacement ( =0 ) and allowing the projection of LOS measurements onto the (T, V) plane (Brouwer et al. 2021). Under this assumption, the vector ̂ ,