Bridge monitoring with satellite InSAR technology and clustering algorithm

• Novel clustering algorithm integrates spatial and temporal satellite data. • Automated Persistent Scatterers classification for bridge deformation. • Low-input approach for civil Structural Health Monitoring. • Validated on two Italian bridges, demonstrating scalability and adaptability. • Supports integration into bridge management systems for proactive decision-making. Satellite Interferometric Synthetic Aperture Radar (InSAR) technology offers a cost-effective, large-scale approach to Structural Health Monitoring (SHM). The Multi-Temporal InSAR (MT-InSAR) technique allows to detect millimetric displacements over time, providing a powerful tool for assessing the evolution of infrastructure deformation and deterioration. However, in order to extract meaningful structural information from the large amount of data generated by MT-InSAR, time-consuming analyses and careful interpretation by civil engineers are required. The present work fills this gap by presenting a clustering algorithm for the automated classification of Persistent Scatterers (PS) derived from Multi-Temporal InSAR (MT-InSAR) data, applied to the Structural Health Monitoring of bridges. Crucially, this low-input, speedy and accessible approach permits a rapid interpretation of MT-InSAR data and advances the integration of satellite information into Bridge Management System (BMS) and Decision Support System (DSS) frameworks. The proposed method accounts for both temporal and spatial domain of MT-InSAR data with a one-step analysis, enabling the identification of predominant PS time-series patterns and their association with physical entities within the study area. The algorithm was tested on two strategic bridges in Italy: the Po River Bridge and the Colle Isarco Viaduct, using COSMO-SkyMed (CSK) satellite data processed with the SarProZ® software. Results demonstrated: (i) the algorithm’s ability to identify anomalies and distinct behaviours across various portions of the structures; (ii) the robustness and adaptability of the algorithm across different case studies, proving its potential for large-scale applications; and (iii) the algorithm’s effectiveness in identifying structurally relevant deformation patterns with minimal user input.