Development of Satellite Interferometry Methods for Volcanic Surveillance
Resumen Abstract Índice Conclusiones
Przeor Przeor, Monika
2025-A
Descargar PDF 
Resumen
Esta tesis se centra en métodos de análisis de datos de radar satelital aplicados a la vigilancia volcánica. La importancia de este estudio radica en la aplicación del Differential Interferometric Synthetic Aperture Radar with a Small Baseline Subset (DInSAR- SBAS) para mejorar la predicción de posibles eventos volcánicos en el futuro y comprender los procesos que tienen lugar en las zonas volcánicas de interés. Un método geodésico de gran éxito, como la interferometría por radar satelital, permite la generación de interferogramas y series temporales que muestran el comportamiento de cada píxel de datos analizado a lo largo del tiempo. En esta tesis, aplicamos una herramienta estadística a los resultados de las series temporales, denominada Análisis de Componentes Independientes (ICA), para eliminar posibles fuentes de ruido y descubrir patrones de deformación ocultos en señales complejas. El uso de la herramienta ICA en el conjunto de datos DInSAR‑SBAS es el aspecto más innovador de este trabajo. Finalmente, los datos de deformación se someten a una inversión matemática para modelar los parámetros de la fuente que generan las anomalías observadas.
Aplicamos el método descrito al estudio de tres áreas volcánicamente activas en las islas de Hawái, La Palma y Tenerife. Los resultados demuestran que llevar a cabo estos estudios puede ayudar a comprender los procesos volcánicos que podrían ocurrir en el futuro, proporcionando una ventaja a las poblaciones que viven y comparten espacio con los volcanes. Las zonas seleccionadas se eligieron para monitorizar volcanes en áreas con alto potencial de actividad volcánica. En el caso de Hawái, la aplicación del ICA al conjunto DInSAR‑SBAS permitió reconocer los patrones de deformación de dos volcanes en la isla que interactúan de manera opuesta. Este estudio contribuyó a comprender procesos volcánicos que habían sido objeto de debate científico durante más de un siglo. En el caso de La Palma, el estudio se centró en visualizar la vía magmática en la corteza seguida durante la fase pre-eruptiva de la erupción del volcán de Tajogaite y durante sus primeros días. Finalmente, la isla de Tenerife pone de manifiesto la elevada actividad de fondo del Teide. El objetivo de estudiar esta zona fue comprender la fuente y los procesos que tuvieron lugar durante la crisis sísmica de 2004‑2005. Los resultados de esos estudios ponen de relieve la eficacia de los métodos aplicados a volcanes activos, permitiendo entender los procesos geodésicos y la predicción de futuras actividades volcánicas.
This thesis focuses on satellite radar data analysis methods applied to volcanic surveillance. The
importance of this study lies in the application of Differential Interferometric Synthetic Aperture
Radar with a Small Baseline Subset (DInSAR- SBAS) to improve the prediction of potential volcanic
events in the future and to understand the processes taking place in the volcanic areas of interest.
A highly successful geodetic method, such as satellite radar interferometry, allows the creation of
interferograms and time series showing the behaviours of each analysed data pixel over time. In
this thesis, we applied the statistical tool to the time series results, named Independent Component
Analysis (ICA), to eliminate potential noise sources and uncover hidden deformation patterns in
the complex signals. Employing the ICA statistical tool to the DInSAR SBAS dataset is the most
innovative aspect of this thesis. Finally, the deformation data undergo mathematical inversion to
modelise the source parameters that create the observed anomalies. We applied the aforementioned
method to study three volcanically active areas on Hawaii, La Palma, and Tenerife islands. The results
show that conducting such studies can help understand the volcanic processes that could occur in the
future, providing an advantage to society living and sharing space with volcanoes. The selected
areas were chosen to monitor volcanoes in areas with high potential for volcanic events. In the case
of Hawaii, the importance of applying the ICA to the DInSAR SBAS dataset allowed the recognition
of the deformation patterns of two volcanoes on the island that interact in opposite ways. This study
helped to understand volcanic processes that had been the objective of scientific debate for more than
100 years. In the case of La Palma, the study focused on imaging the magmatic path in the crust that
was followed in the pre-eruptive phase of the eruption of Tajogaite and during its first days. Finally,
Tenerife island showcases the high background activity of the Teide volcano. The aim of studying
this area was to understand the source and the processes that took place during the seismic crisis of
2004-2005. The outcomes of those studies shed light on the effectiveness of the applied methods
to active volcanoes, enabling the understanding of geodetic processes and the prediction of future
volcanic activity.
List of Figures xv
1 Introduction 1
1.1 Remote Sensing on active volcanoes . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Volcano Geodesy and Remote Sensing 5
2.1 Volcano geodesy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Active sensors for Remote Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Global Navigation Satellite System GNSS . . . . . . . . . . . . . . . . . . . . . . . 7
2.4 Interferometric Synthetic Aperture Radar InSAR . . . . . . . . . . . . . . . . . . . 8
3 Interferogram phase processing 15
3.1 Interferogram generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Adaptive filter and coherence generation . . . . . . . . . . . . . . . . . . . . . . . . 18
3.3 Phase unwrapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.4 Refinement and re-flattening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3.5 Phase to height conversion and geocoding . . . . . . . . . . . . . . . . . . . . . . . 21
3.6 Phase to displacement conversion and geocoding . . . . . . . . . . . . . . . . . . . 22
4 Interferometric Stacking SBAS 25
4.1 Connection graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.2 Interferometric Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.3 Inversion: First Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.4 Inversion: Second Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.5 Geocoding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5 Independent Component Analysis (ICA) 33
5.1 Independent Component Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.2 ICA for SBAS DInSAR dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
6 Volcano Deformation Source Models 37
6.1 Analytical volcano deformation source models . . . . . . . . . . . . . . . . . . . . 37
6.1.1 Mogi point pressure source . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.1.2 McTigue finite spherical model . . . . . . . . . . . . . . . . . . . . . . . . 39
6.1.3 Okada dike-like model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6.1.4 Sill-like model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
6.2 Numerical volcano deformation source models . . . . . . . . . . . . . . . . . . . . 43
7 Inverse Modelling 45
7.1 Linear inverse modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
7.2 Nonlinear inverse modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
8 Elastic interaction between Mauna Loa and Kīlauea evidenced by independent component
analysis 49
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
8.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
8.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
8.4 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
8.4.1 DInSAR SBAS time series . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
8.4.2 Independent component analysis (ICA) of DInSAR SBAS time series . . . . 63
8.4.3 Nonlinear inverse modelling . . . . . . . . . . . . . . . . . . . . . . . . . . 64
8.4.4 GPS data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
8.4.5 Numerical modelling of the stress and strain fields . . . . . . . . . . . . . . 65
9 Geodetic imaging of magma ascent through a bent and twisted dike during the Tajogaite
eruption of 2021 (La Palma, Canary Islands) 69
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
9.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
9.2.1 Preliminary non-linear inversion and dike geometry . . . . . . . . . . . . . . 71
9.2.2 Geodetic imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
9.3 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
9.4 Data and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
9.4.1 Data and processing of GNSS time series . . . . . . . . . . . . . . . . . . . 82
9.4.2 DInSAR Sentinel-1 data and processing . . . . . . . . . . . . . . . . . . . . 83
9.4.3 Non-linear inversion for the shallow dike geometry . . . . . . . . . . . . . . 84
9.4.4 Geodetic imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
10 Independent Component Analysis and Finite element modelling of the 2004-2005 ground
deformation in Tenerife (Canary Islands) 89
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
10.2 Data and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
10.2.1 SBAS DInSAR time series . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
10.2.2 Independent Component Analysis (ICA) of SBAS DInSAR time series . . . 93
10.2.3 Non-linear source modeling through the Finite Element Modeling . . . . . . 93
10.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
10.3.1 SBAS DInSAR time series . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
10.3.2 Application of the ICA to the DInSAR dataset . . . . . . . . . . . . . . . . 94
10.3.3 Non-linear optimization in Finite Element Modeling . . . . . . . . . . . . . 96
10.4 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
11 Conclusions 103
11.1 Conclusions for Elastic interaction between Mauna Loa and Kīlauea evidenced by ICA103
11.2 Conclusions for Geodetic Imaging of magma ascent of the Tajogaite eruption of 2021 104
11.3 Independent component analysis and finite element modelling of the 2004–2005
ground deformation in Tenerife (Canary islands) . . . . . . . . . . . . . . . . . . . . 104
11.4 General conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
References 107
Appendix A Supplementary material for: Elastic interaction between Mauna Loa and
Kīlauea evidenced by ICA 115
Appendix B Supplementary material for: Geodetic imaging of magma ascent through a
bent and twisted dike during the Tajogaite eruption of 2021 (La Palma, Canary Islands131
Appendix C Thesis Publications 147
Appendix D Other Publications 187
Applying the ICA to the DInSAR SBAS dataset in different volcanic areas revealed the presence of
hidden ground deformation patterns operating within the studied regions. In the case of Hawaii island,
it was possible to understand and model the characteristics of magmatic sources in these volcanoes
and comprehend complex volcanic interaction within the crust involving two volcanoes. We gained
insight into how the volcanoes of Mauna Loa and Kilauea exhibit contrasting behaviours due to stress
transfer mechanisms.
La Palma island and its most recent Tajogaite eruption in 2021 provided an excellent opportunity
to employ Geodetic Imaging techniques, allowing us to visualize the kinematic processes occurring
before and during the eruption. Modelling the sources responsible for ground deformation and
subsequently simulating their ascent facilitated understanding and imaging of the internal crustal
structure of La Palma, along with the specific ascent paths utilized by magma as it rose toward the
surface.
Finally, applying the ICA to the DInSAR SBAS dataset of Tenerife enabled us to comprehend
the mechanism of the degassing of the magma batch responsible for the seismic crisis of 2004-2005.
The ground deformation source modelling also revealed hydrothermal activity within the Teide-Pico
Viejo volcanic complex in the crust. This study underscored the importance of future research aimedat understanding volcanic processes on the island, which can aid in identifying magmatic or hydrothermal
sources beneath the islands and, in the future, modelling potential volcanic scenarios.
This study focuses on comprehending processes occurring in volcanic areas by utilising the DIn-
SAR SBAS dataset and subsequently analysing its hidden ground deformation patterns. Subsequent
modelling of the observed ground deformation data can enhance our understanding of processes occurring
in the studied areas, thereby enabling anticipation of potential future volcanic scenarios on the
islands. These findings are of great importance in volcanic regions as they allow society to prepare
for future volcanic phenomena.
The results of this work show detailed, high-resolution ground deformation models that provide
significant insight into volcano dynamics. Future studies will focus on analyzing more complex
datasets and integrating gravity and other geophysical datasets. An interesting addition to this thesis
is modelling the kinematic ascent of magma, which could be applied to other volcanic areas, such
as Iceland’s current volcanic activity (which commenced in 2023). These studies could potentially
provide insights into imaging and understanding the paths of magmatic branches and their magnitude,
aiding in the prediction of forthcoming volcanic scenarios.