The advent of the Copernicus Sentinel-1 mission has substantially broadened access to Synthetic Aperture Radar (SAR) data, enabling continuous, moderate-resolution monitoring of the Earth's surface at an unprecedented scale. While the theoretical framework for Persistent Scatterer Interferometry (PSI) is well-established, and Free and Open Source Software (FOSS) tools like ESA SNAP and StaMPS exist, the operational integration of these resources remains a significant challenge. The complexity of existing workflows, often fragmented across different environments and requiring high-level expertise, creates a bottleneck that limits the transition from raw satellite data to actionable risk management information. This thesis presents PHASE, a comprehensive MATLAB-based software suite designed to bridge this gap. PHASE integrates the complete processing pipeline into a unified, user-friendly environment, optimizing the workflow from raw data download to the final generation of displacement models. The software is structured into two core modules. Module 1 is as an automated wrapper for the SNAP-StaMPS ecosystem, streamlining DInSAR preprocessing and PSI analysis. By automating critical steps, such as master image selection, interferogram generation, and parameter tuning, it significantly reduces operator workload and processing time while ensuring scientific rigor. Module 2 addresses the post-processing challenge: transforming sparse, noisy persistent scatterers into continuous deformation fields. The thesis introduces two distinct methodological approaches for spatio-temporal reconstruction: a stochastic approach based on Least Squares Collocation (LSC), and a deterministic approach utilizing multi-dimensional splines. These methods are adapted for both 1D linear-like infrastructures (e.g., highways, railways) and 2D planar surfaces (e.g., dams, landslides), allowing for the estimation of displacement and uncertainty across space and time. Importantly, PHASE goes beyond numerical estimation by focusing on the interpretation and dissemination of results. A fully automated reporting engine aggregates the statistical outputs into structured tabular reports (Excel format) and GIS-compatible outputs. The methodology was validated through two critical case studies: the earth-filled Nocelle dam in Italy and the H4 highway in Slovenia. Comparisons with in-situ GNSS networks confirmed the statistical consistency of the estimated deformations, demonstrating that the modeled SAR displacements and their associated uncertainty bounds successfully overlap with the ground truth measurements. By combining robust statistical modeling with operational accessibility, PHASE represents a significant step forward in the FOSS landscape, enabling researchers and practitioners to effectively exploit satellite interferometry for the continuous monitoring of critical infrastructure and territory.
L'avvento della missione Copernicus Sentinel-1 ha ampliato considerevolmente l'accesso ai dati SAR (Synthetic Aperture Radar), consentendo un monitoraggio continuo e a risoluzione moderata della superficie terrestre senza precedenti. Sebbene il contesto teorico della Persistent Scatterer Interferometry (PSI) sia ben consolidato e siano disponibili strumenti Free and Open Source (FOSS) come ESA SNAP e StaMPS, l'integrazione operativa di tali risorse rimane una sfida significativa. La complessità dei flussi di lavoro esistenti, spesso frammentati tra ambienti diversi e richiedenti un elevato livello di competenza, costituisce una limitazione al passaggio dai dati satellitari grezzi a informazioni concrete per la gestione del rischio. Questa tesi presenta PHASE, una suite software basata su MATLAB, progettata per colmare questa lacuna. PHASE integra l'intero flusso di lavoro in un ambiente unificato e di facile utilizzo, ottimizzando dal download dei dati grezzi alla generazione finale dei modelli di deformazione. Il software è strutturato in due moduli principali. Il Modulo 1 funge da wrapper automatizzato per l'ecosistema SNAP-StaMPS, semplificando il preprocessing DInSAR e l'analisi PSI. Automatizzando le fasi più critiche, come la selezione dell'immagine master, la generazione degli interferogrammi e la regolazione dei parametri, riduce significativamente il carico di lavoro dell'operatore e i tempi di elaborazione, garantendo al contempo il rigore scientifico. Il Modulo 2 affronta la sfida del post-processing: trasformare i persistent scatterer, sparsi e rumorosi, in campi di deformazione continui. La tesi introduce due distinti approcci metodologici per la ricostruzione spazio-temporale: un approccio stocastico basato sulla Least Squares Collocation (LSC) e un approccio deterministico che utilizza spline multidimensionali. Tali metodi sono adattati sia a infrastrutture lineari 1D (ad es. autostrade, ferrovie) sia a superfici piane 2D (ad es. dighe, frane), consentendo la stima dello spostamento e dell'incertezza nello spazio e nel tempo. PHASE non si limita alla sola stima numerica, ma punta all'interpretazione e alla diffusione dei risultati. Un motore di reportistica completamente automatizzato aggrega gli output statistici in report tabulari strutturati (formato Excel) e in output compatibili con i principali software GIS. La metodologia è stata validata attraverso due casi studio: la diga in terra di Nocelle, in Italia, e l'autostrada H4, in Slovenia. Il confronto con reti GNSS in-situ ha confermato la consistenza statistica delle deformazioni stimate, dimostrando che gli spostamenti SAR modellati e i relativi intervalli di incertezza si sovrappongono con successo alle misure di riferimento. Combinando una robusta modellazione statistica con l'accessibilità operativa, PHASE rappresenta un significativo passo avanti nel panorama FOSS, mettendo ricercatori e professionisti in condizione di sfruttare efficacemente l'interferometria satellitare per il monitoraggio continuo delle infrastrutture critiche e del territorio.
The PHASE framework: spatio-temporal modeling of SAR persistent scatterer interferometry for deformation analysis
MONTI, ROBERTO
2026
Abstract
The advent of the Copernicus Sentinel-1 mission has substantially broadened access to Synthetic Aperture Radar (SAR) data, enabling continuous, moderate-resolution monitoring of the Earth's surface at an unprecedented scale. While the theoretical framework for Persistent Scatterer Interferometry (PSI) is well-established, and Free and Open Source Software (FOSS) tools like ESA SNAP and StaMPS exist, the operational integration of these resources remains a significant challenge. The complexity of existing workflows, often fragmented across different environments and requiring high-level expertise, creates a bottleneck that limits the transition from raw satellite data to actionable risk management information. This thesis presents PHASE, a comprehensive MATLAB-based software suite designed to bridge this gap. PHASE integrates the complete processing pipeline into a unified, user-friendly environment, optimizing the workflow from raw data download to the final generation of displacement models. The software is structured into two core modules. Module 1 is as an automated wrapper for the SNAP-StaMPS ecosystem, streamlining DInSAR preprocessing and PSI analysis. By automating critical steps, such as master image selection, interferogram generation, and parameter tuning, it significantly reduces operator workload and processing time while ensuring scientific rigor. Module 2 addresses the post-processing challenge: transforming sparse, noisy persistent scatterers into continuous deformation fields. The thesis introduces two distinct methodological approaches for spatio-temporal reconstruction: a stochastic approach based on Least Squares Collocation (LSC), and a deterministic approach utilizing multi-dimensional splines. These methods are adapted for both 1D linear-like infrastructures (e.g., highways, railways) and 2D planar surfaces (e.g., dams, landslides), allowing for the estimation of displacement and uncertainty across space and time. Importantly, PHASE goes beyond numerical estimation by focusing on the interpretation and dissemination of results. A fully automated reporting engine aggregates the statistical outputs into structured tabular reports (Excel format) and GIS-compatible outputs. The methodology was validated through two critical case studies: the earth-filled Nocelle dam in Italy and the H4 highway in Slovenia. Comparisons with in-situ GNSS networks confirmed the statistical consistency of the estimated deformations, demonstrating that the modeled SAR displacements and their associated uncertainty bounds successfully overlap with the ground truth measurements. By combining robust statistical modeling with operational accessibility, PHASE represents a significant step forward in the FOSS landscape, enabling researchers and practitioners to effectively exploit satellite interferometry for the continuous monitoring of critical infrastructure and territory.| File | Dimensione | Formato | |
|---|---|---|---|
|
PhD_Thesis_RM_final.pdf
accesso aperto
Licenza:
Tutti i diritti riservati
Dimensione
188.59 MB
Formato
Adobe PDF
|
188.59 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/384574
URN:NBN:IT:POLIMI-384574