Assessment of MT-InSAR processing techniques for slow-moving landslides monitoring in Cuenca (Ecuador) through double-band SAR satellite

Abstract

Landslides are among the most intense geological disasters worldwide. Several remote sensing techniques have been used to study these phenomena over the last few decades, including Global Navigation Satellite Systems (GNSS) and Multi-Temporal Interferometric Synthetic Aperture Radars (MT-InSAR), in particular, Small Baseline Subset (SBAS) and Permanent Scatterer Interferometry (PSI). The University of Azuay and highway infrastructure (Cuenca - Ecuador) were affected by slow-moving landslides which in urban areas had more noticeable deformation effects due to their crowns. Furthermore, evidence indicates that the landslide toe was developed in a rural area. To study landslide boundary and deformation kinematics comprehensively, we need to observe two distinct scattering surfaces using SAR imagery (high coherence areas in the urban area and low coherence areas in the rural area). We can also observe different distributions in X- and C-bands due to the rural area covered by vegetation and trees at the toe of the landslide. Two sets of radar images were analyzed and compared for monitoring the aforementioned slow-moving landslide kinematics, including Sentinel1-A (S1-A) and COSMO-SkyMed (CSK). PSI and SBAS techniques have been used to measure the rate of surface deformation and displacement time series at CSK over the period (2017-2019) and S1-A units over the period (2015-2018), respectively. Furthermore, a GNSS station in the stable area was used as a reference station to integrate the PSI and SBAS results over space. Then, an unsupervised machine learning algorithm (K-medians) was used to determine the most appropriate intervals for preparing mean deformation rate maps. Finally, a comparison was made between the outputs obtained from this study and displacement recorded at two additional GNSS stations located in the area affected by deformation. As a result, by comparing and monitoring double-band SAR satellites and different SAR image processing techniques, it became possible to analyze the whole landslide of our case study appropriately.