Innovative methods for the spatial and temporal analysis of interferometric data at national scale

The present PhD thesis represented the culmination of a three-years of research conducted within the collaboration between the Department of Earth Sciences of the University of Florence and the Institute for Electromagnetic Sensing of the Environment, which is affiliated with the National Research Council, based in Naples. The PhD project aimed to develop innovative, reproducible and scalable methodologies to enhance the quality and usability of wide-scale interferometric SAR satellite data. The advent of freely and openly available Sentinel-1 constellation of the European Space Agency, coupled with the expansion of its archives, has rendered the application of DInSAR analysis feasible over country and even continent. The interferometric measurements pose a significant challenge due to the presence of overlapping signals, which can be originated by a variety of factors, including anthropogenic activities, local natural phenomena, encompassing subsidence and landslides, volcanic and tectonic activities, and large-scale geodynamic processes. In order to obtain reliable interferometric measurements, it is essential to discern the low- and high-frequency signals. In the absence of proper identification and removal, the low-frequency signals have the potential to mask or distort the interpretation of high-frequency signals component. In this scenario, the purpose of the present PhD thesis was twofold: firstly, to enhance the quality of the low-frequency signals component; and secondly, to improve the usability of the DInSAR product with the design, implementation, testing, and validation of two post-processing tools. The low-frequency signals component, which is commonly associated with large-scale tectonic displacement at the European scale, was investigated with the development of a procedure involving the acquisition of the freely available GNSS stations data from the Nevada Geodetic Laboratory database, which comprises a total of 5,980 GNSS stations. The GNSS measurements provide three-dimensional time series displacement with millimetric accuracy, encompassing the East-West, North-South, and vertical components. The time series from each GNSS station was then subjected to a screening process to remove from jumps or artefacts, in order to retrieve an accurate estimation of the mean deformation rate. The implementation of the data was conducted using a Kriging interpolation technique, excluding the incorporation of fault constraints. This resulted in the generation of a continuous European tectonic model. The most relevant signals described were referred to the Eastern Mediterranean region, as well as the Fennoscandian countries and Iceland. The outcome of DInSAR technique is referred to the high-frequency signals component. A pivotal challenge that emerges in the context of wide-scale applications is the management and interpretation of the massive volume of interferometric data currently available. In contrast to local-scale analysis, where manual inspection and interpretation remain feasible options, wide-scale analysis requires the implementation of scalable, automated, and statistically robust methodologies. Consequently, the thesis was directed towards the development of two DInSAR post-processing tools. The TS-DInSAR tool introduced an automatic, scalable, and unsupervised spatial and temporal analysis procedure of DInSAR datasets at the Italian national scale. This methodology was successfully applied to the horizontal and vertical components retrieved from Sentinel-1 imagery, processed using the P-SBAS technique. The application of the TS-DInSAR tool resulted in the identification of the deformation triggering factors to approximately 600,000 measurement points, including landslide (33.4%), subsidence (18.4%), and soil erosion (14.6%). Still concerning the increasing usability of DInSAR products, a methodological procedure was developed for the assessment of the vulnerability of buildings to slow-moving landslides, via the construction of fragility and vulnerability curves. The potential economic consequences of landslides could be considerable, resulting in damage to exposed structures and infrastructures. In order to determine the most efficacious risk mitigation strategies, the scientific community was engaged in the analyses aimed at assessing the expected consequences of landslide cause. In particular, in the quantitative risk assessment procedure the vulnerability parameter was typically the most challenging to assess. In this context, the Norther Apennines were selected as the area of study to test the developed methodology. A total of approximately 3,000 buildings were investigated with the objective of cataloguing the damage severity caused by landslides. Subsequently, the application of the retrieved vulnerability curve was applied within the QRA procedure, in which were interested over than 700.000 buildings in the entire Northern Apennines. The result of the application assessed a total risk of approximately 1.8 billion euros. In summary, the central contribution of the presented thesis lied in the methodological efforts, derived from a critical analysis of the existing literature, to enhance the quality and usability of the satellite DInSAR data. The procedures represented a methodological advancement towards enhancing the interpretability of DInSAR products, with the potential to be further refined through the integration of additional parameters and the extension of its application to both C- and L-bands.