This work aims to evaluate the capability of satellite techniques for dams monitoring. In particular the attention is focused on earth dams, because they are the most common type and the existing models in literature are able to describe their displacements if mainly the structural and geotechnical parameters characterizing the dam composition are deeply evaluated. During the last years, it has been demonstrated that remote sensing satellite technologies are able to detect the structural behaviour of strategic structures (including the dam response), evaluating both displacements and deformation with high accuracy. In this work, the monitoring of an embankment dam (the Castello dam), located in southern Italy, has been carried out using different satellite techniques. In particular, Global Navigation Satellite System (GNSS) and Synthetic Aperture Radar Interferometry (InSAR) have been used for the detection of dam displacements, while classification technique and Object-Based Image Analysis (OBIA)have been involved for the estimation of the reservoir surface and levels. GNSS data from a Continuously Operating Reference Station (CORS) far away from the monitored site and remote sensing images with different spatial and radiometric resolution (optical and Synthetic Aperture Radar, SAR, images) and temporal coverage (from 2011 to 2016) have been used. The displacements by using the Differential GNSS (DGNSS) technique have been retrieved employing an innovative approach, based on the use of a CORS far away from the monitored site. The reversible and irreversible displacements of the dam have been analysed via Permanent Scatterers InSAR (PS-InSAR), using different Multi-Baseline Construction Methods (MBC), in particular, the star-graph and the full-graph. From the analysis of the GNSS displacements over ⁓1 year (April 2011 – March 2012), only the planimetric component reaches a suitable accuracy for monitoring the dam displacements (≈ 1–5 x 10-3 m). The non-linear trend shows a variability range of ≈ 2 mm y-1, while the linear trend highlights an estimated velocity orthogonal to the dam of ≈ -1 mm y-1. The comparison between the GNSS displacements and the measured water levels shows a non-linear relation between the variables. In particular, the dam response is related to the water levels during the emptying and filling periods. Also, the air temperature influences the dam displacements. Indeed, two maximum displacements have been recorded, the first corresponding to the maximum daily averaged air temperature, the other to the minimum water level; also, the minimum displacement corresponds to the minimum daily averaged air temperature, while the maximum water level was constant in the springtime. The use of different strategies for PS-InSAR analysis, highlights higher accuracy when more redundant connection are used (full-graph). The irreversible displacements are evaluated with a linear trend estimation and the results are comparable to those obtained via GNSS (⁓ -1 mm y-1). Analysing the reversible component of displacements, the best fitting model has been obtained superimposing a polynomial interpolating curve (r2 ≈ 0.80) on the temporal moving average of displacements. The classification techniques applied on both optical and SAR images allows estimating the water surface and levels with high accuracy (r2 > 0.95). The implanted OBIA techniques (edge and distance similarity indices) allowed estimating the water levels with accuracy of ≈ 2 m, (r2 > 0.90, with Sentinel-1A data). Although the application of PS-InSAR needs more deep analyses, the use of the same dataset (Sentinel-1A) for both dam displacements and water levels evaluation, represents an innovative approach of this work. Further studies will include the analysis of PS on both ascending and descending orbits over a wider time-span.
Il presente lavoro mira a valutare le potenzialità delle tecniche satellitari per il monitoraggio delle dighe. In particolare, l’attenzione è stata rivolta alle dighe di terra, poichè queste rappresentano la tipologia costruttiva più diffusa ed i modelli esistenti in letteratura sono in grado di descriverne gli spostamenti solo nell’ipotesi di conoscere accuratamente tutti parametri geotecnici e strutturali che caratterizzano la composizione della diga stessa. Negli ultimi anni, è stato ampiamente dimostrato che le tecniche satellitari da remoto sono in grado di individuare con elevata accuratezza, il comportamento di grandi strutture (incluse le dighe), valutandone spostamenti e deformazioni. In questo lavoro, è stato condotto il monitoraggio di una diga di terra (la diga Castello), ubicata nel sud Italia, utilizzando varie tecniche satellitari. In particolare, per la stima degli spostamenti della diga, sono stati utilizzate tecniche Global Navigation Satellite System (GNSS) ed interferometriche satellitari, mentre per la stima delle superfici e dei livelli di invaso, sono state messe a punto tecniche di classificazione di dati telerilevati e tecniche di Object-Based Image Analysis (OBIA). Sono stati utilizzati dati GNSS acqusiti da una Continously Operating Reference Station (CORS) distante dal sito monitorato ed immagini telerilevate aventi differente risoluzione spaziale e radiometrica (sia ottiche sia immagini Sinthetic Aperture Radar, SAR) acquisite in un ampio intervallo temporale (dal 2011 al 2016). Gli spostamenti sono stati determinati utilizzando la tecnica differenziale GNSS (DGNSS), attraverso un approccio sperimentale basato sull’utilizzo di una CORS posta distante dal sito monitorato. Gli spostamenti reversibili ed irreversibili del coronamento della diga sono stati analizzati tramite tecnica Permanent Scatterers Interferometrica SAR (PS-InSAR) utilizzando differenti Multi-Baseline Construction Methods (MBC), in particolare lo star-graph ed il full-graph. Dall’analisi degli spostamenti GNSS su un periodo di circa un anno (Aprile 2011 – Marzo 2012) si evince che solo la componente planimetrica possiede un’accuratezza idonea per il monitoraggio degli spostamenti della diga (≈ 1–5 x 10-3 m). L’analisi del trend non lineare degli spostamenti evidenzia un range di variabilità pari a ≈ 2 mm/anno, mentre l’analisi del trend lineare evidenzia una velocità di spostamento ortogonale al coronamento pari a ≈ -1 mm/anno. Oltremodo, il confronto tra gli spostamenti GNSS e i livelli di invaso misurati evidenzia una relazione non lineare tra le variabili. In particolare la risposta della diga è legata alla variazione del livello idrico durante il periodo di svuotamento e riempimento dell’invaso. Parimenti, anche la temperatura dell’aria influenza gli spostamenti della diga. Si osservano, infatti, due massimi spostamenti, uno in corrispondenza della massima temperatura media giornaliera, l’altro al verificarsi del minimo livello di invaso; viceversa, lo spostamento minimo si ha in corrispondenza del valore minimo della temperatura media giornaliera, mentre il massimo livello di invaso veniva mantenuto costante per tutto il periodo primaverile. L’utilizzo di differenti strategie per la stima degli spostamenti PS InSAR, evidenzia un’accuratezza migliore nel caso di connessioni ridondanti (full-graph). La componente irreversibile degli spostamenti è ottenuta attraverso la stima del trend lineare e mostra valori comparabili con quelli ottenuti tramite GNSS (~ -1 mm/anno). La componente reversibile degli spostamenti è interpretabile attraverso l’impiego di un modello polinomiale (r2 ≈ 0.80) sulla media mobile temporale. Le tecniche di classificazione di immagini ottiche e radar consentono di stimare accuratamente (r2 > 0.95) sia le superfici sia i livelli dell’invaso. Le tecniche OBIA implementate (edge e distance similarity indices), hanno consentito di stimare i livelli di invaso con una accuratezza pari a ≈ 2 m (r2 > 0.9, dataset Sentinel-1A). Sebbene l’applicazione della tecnica PS-InSAR necessiti ulteriori approfondimenti, l’utilizzo di un unico dataset (Sentinel-1A) consente sia la stima degli spostamenti del coronamento, sia dei livelli di invaso, e pertanto costituisce un elemento di innovazione di questo lavoro. I suddetti approfondimenti includono l’integrazione delle analisi PS effettuate su scene descending ed ascending e su un periodo di indagine più ampio.
Dam displacements monitoring by using GNSS and remote sensing techniques.
Dam displacements monitoring by using GNSS and remote sensing techniques
Pipitone, Claudia
Abstract
This work aims to evaluate the capability of satellite techniques for dams monitoring. In particular the attention is focused on earth dams, because they are the most common type and the existing models in literature are able to describe their displacements if mainly the structural and geotechnical parameters characterizing the dam composition are deeply evaluated. During the last years, it has been demonstrated that remote sensing satellite technologies are able to detect the structural behaviour of strategic structures (including the dam response), evaluating both displacements and deformation with high accuracy. In this work, the monitoring of an embankment dam (the Castello dam), located in southern Italy, has been carried out using different satellite techniques. In particular, Global Navigation Satellite System (GNSS) and Synthetic Aperture Radar Interferometry (InSAR) have been used for the detection of dam displacements, while classification technique and Object-Based Image Analysis (OBIA)have been involved for the estimation of the reservoir surface and levels. GNSS data from a Continuously Operating Reference Station (CORS) far away from the monitored site and remote sensing images with different spatial and radiometric resolution (optical and Synthetic Aperture Radar, SAR, images) and temporal coverage (from 2011 to 2016) have been used. The displacements by using the Differential GNSS (DGNSS) technique have been retrieved employing an innovative approach, based on the use of a CORS far away from the monitored site. The reversible and irreversible displacements of the dam have been analysed via Permanent Scatterers InSAR (PS-InSAR), using different Multi-Baseline Construction Methods (MBC), in particular, the star-graph and the full-graph. From the analysis of the GNSS displacements over ⁓1 year (April 2011 – March 2012), only the planimetric component reaches a suitable accuracy for monitoring the dam displacements (≈ 1–5 x 10-3 m). The non-linear trend shows a variability range of ≈ 2 mm y-1, while the linear trend highlights an estimated velocity orthogonal to the dam of ≈ -1 mm y-1. The comparison between the GNSS displacements and the measured water levels shows a non-linear relation between the variables. In particular, the dam response is related to the water levels during the emptying and filling periods. Also, the air temperature influences the dam displacements. Indeed, two maximum displacements have been recorded, the first corresponding to the maximum daily averaged air temperature, the other to the minimum water level; also, the minimum displacement corresponds to the minimum daily averaged air temperature, while the maximum water level was constant in the springtime. The use of different strategies for PS-InSAR analysis, highlights higher accuracy when more redundant connection are used (full-graph). The irreversible displacements are evaluated with a linear trend estimation and the results are comparable to those obtained via GNSS (⁓ -1 mm y-1). Analysing the reversible component of displacements, the best fitting model has been obtained superimposing a polynomial interpolating curve (r2 ≈ 0.80) on the temporal moving average of displacements. The classification techniques applied on both optical and SAR images allows estimating the water surface and levels with high accuracy (r2 > 0.95). The implanted OBIA techniques (edge and distance similarity indices) allowed estimating the water levels with accuracy of ≈ 2 m, (r2 > 0.90, with Sentinel-1A data). Although the application of PS-InSAR needs more deep analyses, the use of the same dataset (Sentinel-1A) for both dam displacements and water levels evaluation, represents an innovative approach of this work. Further studies will include the analysis of PS on both ascending and descending orbits over a wider time-span.File | Dimensione | Formato | |
---|---|---|---|
PhD_thesis_Pipitone_2019_.pdf
accesso aperto
Dimensione
4.33 MB
Formato
Adobe PDF
|
4.33 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.