Interpreting Satellite Radar for Bridges: Integrating MT-InSAR with Physics-Based and Data-Driven Models for Scalable Structural Health Monitoring

Abstract

Multi-Temporal Interferometric Synthetic Aperture Radar (MT-InSAR) has matured into a credible, non-contact technique for monitoring bridge deformation from individual structures to regional portfolios. The main challenge for routine engineering use is no longer measuring millimetre-scale line-of-sight (LOS) displacement, but interpreting those measurements into defensible statements about structural condition. This review addresses that interpretation problem by synthesising the physical and algorithmic fundamentals of MT-InSAR, bridge-typology-dependent applicability, LOS projection ambiguity, current sensor capabilities, and emerging open data services. Particular emphasis is placed on signal decoupling because measured bridge displacement combines reversible thermal response, transient live-load effects, progressive deformation associated with damage or settlement, and measurement noise. Two complementary interpretation strategies are examined: physics-based integration using finite-element analysis, model updating and digital twins, and data-driven integration using clustering, anomaly detection, forecasting and deep learning. Their convergence through physics-informed methods is also discussed. The review proposes a hybrid framework in which data-driven methods screen bridge portfolios and identify anomalous deformation, while physics-based modelling provides structural interpretation and supports validation using inspection records, in-situ measurements and uncertainty quantification. The synthesis highlights the strengths and limitations of current practice and identifies research needs in multi-geometry analysis, dense spatial sampling, physics-informed learning, portfolio-scale digital twins and prospective validation. The resulting framework is intended to support scalable, physically interpretable and evidence-based use of MT-InSAR in bridge structural health monitoring.