Integrating InSAR monitoring into infrastructure vulnerability assessment

: (Alternative Format Thesis)

Abstract

Bridge infrastructure worldwide faces growing challenges from ageing structures, increasing traffic demands, and climate change effects, requiring effective monitoring and maintenance strategies to ensure public safety. Traditional approaches rely on periodic visual inspections and limited deployment of Structural Health Monitoring (SHM) systems, but these methods suffer from subjectivity, high costs, and limited coverage that particularly disadvantage developing regions. Current prioritisation frameworks often overlook social equity considerations, potentially perpetuating infrastructure disparities.

Multi-Temporal Interferometric Synthetic Aperture Radar (MT-InSAR) offers significant advantages for bridge monitoring, providing millimetre-level displacement measurements across large geographic areas. Studies have demonstrated the effectiveness of MT-InSAR in detecting structural changes and geo-hazard effects on bridges. Satellite constellations like Sentinel-1 enable global coverage and frequent monitoring updates, offering more equitable access to infrastructure monitoring capabilities.

However, widespread adoption of MT-InSAR for bridge management faces several critical barriers. The technique relies on radar targets called Persistent Scatterers (PSs), which may not be available for all structures, making it difficult to predict monitoring feasibility before expensive data processing. Integration into comprehensive risk assessment frameworks requires accounting for monitoring system availability. Equitable resource allocation demands systematic incorporation of social vulnerability considerations alongside technical performance metrics. Therefore, the objective of this thesis is to advance the integration of spaceborne remote sensing technologies with infrastructure vulnerability assessment frameworks to enable more effective monitoring strategies and socially equitable resource allocation decisions on regional to global scales.
This involved: (1) the development of predictive methodologies for MT-InSAR measurement availability using machine learning techniques; (2) the integration of spaceborne monitoring capabilities into comprehensive geo-hazard risk assessment frameworks; (3) the incorporation of social vulnerability indicators into bridge performance assessment for equitable prioritisation at the administrative level.

First, a machine learning method using convolutional neural networks to predict MT-InSAR measurement availability for infrastructure monitoring was developed. The approach utilises interferometric coherence decay characteristics and infrastructure location data to estimate PS density without requiring time-consuming analysis of entire SAR data stacks. A U-Net architecture was implemented and trained using regional-scale data from the Netherlands, achieving low prediction errors with over 80\% of estimates $\pm 1$ of actual values. The methodology demonstrated strong generalisation performance when applied to previously unseen datasets, providing stakeholders with a practical tool for assessing monitoring feasibility before committing resources to full interferometric analysis.

Second, a global geo-hazard risk assessment framework that integrates monitoring system availability into structural vulnerability calculations was developed. The methodology incorporates both traditional SHM sensors and MT-InSAR capabilities to account for how these monitoring systems reduce uncertainties in structural health assessment. Applied to a database of 744 long-span bridges worldwide, the framework revealed significant monitoring gaps and demonstrated how spaceborne monitoring could substantially reduce the number of structures classified as high-risk. The approach provides actionable guidance for optimising monitoring deployment and creating more dynamic risk assessments that account for the temporal dimension of structural vulnerability.

Third, an integrated socio-structural vulnerability framework for equitable bridge prioritisation at the administrative unit level was established. The methodology combines bridge performance metrics with social vulnerability indicators using data-driven weighting approaches to eliminate subjective bias in prioritisation decisions. Additionally, the approach incorporates adaptive capacity dimensions, including economic resilience, inspection burden, and monitoring availability, into comprehensive vulnerability assessments. Applied to Californian bridge infrastructure, the framework identified counties where high structural vulnerability coincides with elevated social vulnerability, enabling prioritisation of areas where bridge failures would trigger the most severe societal consequences.

The proposed approach has the potential to enhance bridge management practice by enabling more cost-effective monitoring deployment, comprehensive risk assessment, and socially equitable resource allocation. The research provides practical tools for stakeholders to assess MT-InSAR feasibility, integrate monitoring capabilities into risk frameworks, and incorporate social equity considerations into infrastructure prioritisation decisions, ultimately advancing both operational effectiveness and social equity in bridge infrastructure management.

Date of Award	25 Mar 2026
Original language	English
Awarding Institution	University of Bath
Supervisor	Cormac Reale (Supervisor), Giorgia Giardina (Supervisor), Chris Blenkinsopp (Supervisor), Kevin Briggs (Supervisor) & Pietro Milillo (Supervisor)