Le Venezuela est traversé par une zone de limite de plaque. Ce système tectonique accommode les mouvements relatifs de trois plaques majeures: la plaque Sud-Américaine, la plaque Caraïbe et la plaque de Nazca. Ce système est constitué de failles décrochantes actives qui ont généré au Venezuela de nombreux séismes de magnitude supérieure à 6-7. Parmi ces failles, cette thèse se focalise sur la faille de Boconó et la faille d’El Pilar.Le but de cette thèse est d'étudier l'activité de ces failles sur plusieurs échelles de temps en utilisant une approche multidisciplinaire qui combine analyses morphotectonique, paléosismologique et géodésique. Cette approche a permis de préciser le régime de déformation de ces failles indispensable à l'estimation de l'aléa sismique.Antérieurement à cette thèse, dans la région de Yaracuy, l’activité tectonique du segment nord de la faille de Boconó était mal contrainte faute de données géodésiques ou géochronologiques suffisantes. Or cette partie de la faille a provoqué en 1812 un séisme de Mwi 7.4 qui a détruit les villes de la région.Grâce à la datation par Béryllium-10 de la surface d’exposition de cônes alluviaux décalés par la cinématique dextre de la faille, cette thèse montre que la vitesse quaternaire de la faille est comprise entre 5.0 et 11.2 mm/an.En comparant cette vitesse estimée sur ~ 200 ka et le taux de glissement estimé en champ lointain par des mesures géodésiques (~ 12 mm/an), il peut être proposé que la faille de Boconó accommode une grande partie de l'extrusion du Bloc Nord Andin. La réalisation de carte de vitesses moyennes de déformation à partir d'images SAR a montré l'absence de glissement asismique le long de la faille de Boconó entre 2007 et 2011. En extrapolant ce comportement aux derniers 200 ans, il en résulte que, depuis le dernier séisme en 1812 il y a une accumulation de déficit de glissement de quelques mètres selon la vitesse de glissement considérée. Cette faille représente donc un aléa sismique important pour la région. Une tranchée réalisée pour préciser cet aléa a montré que trois évènements sismiques de Mw > 6-6.5 ont lieu depuis 1300 ap. J.-C., le dernier de ces événements étant probablement le séisme historique de 1812.Au Nord Est du Venezuela, la faille d’El Pilar accommode l’intégralité du mouvement relatif entre la plaque Sud-Américaine et la plaque Caraïbe (~ 20 mm/an). Après le séisme de Ms 6.8 en 1997, le segment émergé de cette faille a subit un important « afterslip ». Des mesures géodésique réalisées en 2003, 2005 et 2013 ont montré que ce segment glisse encore asismiquement (~12 – 13 mm/an). Cette thèse présente une carte des vitesses de déformation entre 2007 et 2011 calculée par interférométrie radar. Celle-ci a permis de montrer que cette faille glisse asismiquement de façon non uniforme dans l’espace et le temps. L’analyse en série temporelle des déplacements a révélé que le glissement asismique de certains tronçons de la faille subit une accélération en Juin 2009 avec des vitesses de glissement asismiques supérieures au déplacement relatif entre les plaques. Cette observation permet d’interpréter que le glissement asismique a un comportement transitoire, en effet, des périodes de blocage et des périodes de larges glissements se succèdent. Cette succession doit probablement se poursuivre tout le long de la période intersismique comme le suggèrent le faible nombre de séismes historiques et préhistoriques au regard de la vitesse de coulissage le long de la faille. Enfin ce glissement asismique présentant des variations spatiales et temporelles est probablement contrôlé par la présence de serpentinites et d’une activité hydrothermale le long de la zone de faille, contexte connu pour favoriser des comportements rhéologiques de ce type. / Venezuela is crosscut by a plate boundary zone, this tectonic setting accommodates the relative displacements of three plates: the South America, the Caribbean and the Nazca Plate. This tectonic system is constituted of active strike-slip faults which have provoked several events of Mw > 6-7. Among these faults, this thesis focuses on the Boconó Fault and on the El Pilar Fault.The aim of this dissertation is to study fault activities on several time scales using a multidisciplinary approach. This approach, which combines morphotectonic, paleoseismologic and geodetic analyses, leads to clarify the deformation pattern. This knowledge is essential to the seismic hazard assessment.Previously to this thesis, in the Yaracuy valley, the tectonic activity of the Boconó fault was poorly constrained due to the lack of geodetic and geochronological data; although a part of this fault triggered in 1812 an earthquake of Mwi 7.4. Through 10-Beryllium surface exposition dating of two alluvial fans shifted by the fault, this thesis shows that the Quaternary slip rate of the fault ranges from 5.0 to 11.2 mm/yr. By comparing this rate estimated on ~ 200 ka with the slip rate estimated in far field with geodesy (~ 12 mm/yr), it can be proposed that the Boconó fault accommodates a major part of the North Andean Block extrusion. Velocity map of ground displacements calculated using SAR images shows the lack of aseismic slip along the Boconó Fault during the 2007-2011 period. The extrapolation of this locked activity since the 1812 event, implies that there is a slip deficit of several meters. Therefore, the Boconó Fault have to be taken into account in the regional seismic hazard assessment. A paleoseismological trench across the studied segment is also presented in this thesis in order to constrain this assessment. Three events of Mw > 6 - 6.5 have been recorded in this trench since 1300 CE, the last of these events is probably the 1812 historical earthquake.In the north-western region of Venezuela, the El Pilar Fault accommodates the whole relative displacement between South-America and Caribbean Plates (~ 20 mm/yr). After the last event in 1997 (Ms 6.8) the on-shore segment of this fault undergoes an important afterslip. Geodetic campaign measurements performed in 2003, 2005 and 2013 showed that this segment was still creeping (~ 12 - 13 mm/yr). This thesis presents an InSAR analysis performed with 18 SAR images spanning the 2007-2011 period. The velocity map shows that the aseismic slip is not uniform along the El Pilar Fault. Time-series analysis reveals locally a creep acceleration. This transient is characterised by a rate exceeding the rate of surrounding plate motion. Therefore, the El Pilar fault seems to be partially locked during several years and then undergoes transient creep during several months. This succession should last during the whole interseismic period as suggested by the low seismic activity and paleoseismological trenches. This creep showing spatio-temporal variations is probably controlled by the existence of serpentinites lenses and the hydrothermal activity, which are known to promote this kind of rheological behaviour.

Depuis son origine, la surface de la Terre est façonnée par de nombreux processus. Qu’ils soient géologiques ou, plus récemment, anthropiques, ces processus façonnent la topographie, à différentes échelles de volume et de temps. Devant l’augmentation des risques (combinaison des enjeux humains et économiques et des aléas de type subsidence ou glissements de terrains), la nécessité de développer des méthodes de mesures fiables et précises de ces processus apparaît évidente. C’est l’objectif des mesures dites géodésiques. Le but de cette thèse est de créer une méthodologie de combinaison des déplacements de surface mesurés par différentes techniques de mesures géodésiques (GNSS, InSAR, Nivellement) afin d’améliorer la connaissance de la déformation sur de nombreux cas d’études. La première étape de la combinaison est l’harmonisation des références spatiales de la déformation. Ensuite, les champs de vitesses issus des différentes techniques sont pondérés suivant leurs incertitudes natives. Enfin la combinaison de ces champs de vitesses est effectuée par la méthode du krigeage par régression pondérée. Le résultat obtenu est une grille de déformation qui, selon les méthodes et données utilisées, présentera une résolution spatiale en une à trois dimensions. Associée à celle-ci, une grille d’indices de fiabilité est créée, prenant en considération les incertitudes natives de chaque méthode d’une part, et les incertitudes liées à la méthode d’interpolation d’autre part. L’étude conjointe de ces deux grilles permet une discussion sur la déformation.Notre méthodologie a été confrontée à un cas d’étude : la subsidence de la saline de Vauvert (Gard, France). Ce site apparaît idéal car (1) sa situation géographique proche de Montpellier facilite son instrumentation et les mesures, (2) une subsidence y est déjà mesurée, dont l’origine est l’extraction de sel entre 2000 et 3000 m de profondeur, et (3) des jeux de données sont disponibles en InSAR, GPS et Nivellement. Deux périodes de mesures sont retenues (2002-2009 et 2015-2017) car permettant l’utilisation conjointe de plusieurs techniques de mesures. Les résultats de la combinaison des champs de vitesses montrent la présence d’un bol de subsidence de magnitude maximale de 2.2 cm/an sur la première période et 2.4 cm/an sur la deuxième période. Son extension atteint les quartiers sud-est de la ville voisine de Vauvert. Globalement cette subsidence (dimension du bol et magnitude) varie peu au cours du temps, néanmoins une zone localisée proche des sites historiques de la Galine, à l’est de la zone d’étude, présente une accélération de la subsidence d’environ 8 mm/an entre les deux périodes. Cette accélération semble montrer un effet en surface de la purge des puits à proximité, purges initiées en 2015.Enfin l’utilisation d’un algorithme/méthode d’inversion sur les données issues de la méthodologie de combinaison (vitesses et incertitudes) permet d’apporter de nouveaux éléments de discussion sur l’anisotropie locale de la déformation. Une accommodation des structures géologiques en profondeur (plis, failles, décollements) est montrée. / Since its origin, the surface of the Earth has been shaped by many processes. Whether they are geological or, more recently, anthropogenic, these processes shape the topography at different scales of volume and time. Given the increase in risks (a combination of human and economic issues with subsidence or landslide hazards), the necessity to develop reliable and accurate methods of measuring these processes is clear. This is the goal of the so-called geodesic measurements.The aim of this thesis is to create a new methodology for the combination of surface displacements measured by different geodetic techniques (GNSS, InSAR, Leveling) to improve the knowledge of the deformation on many cases of studies. The first step of the combination is the harmonization of the spatial references of the deformation. Then, the velocity fields resulting from the different techniques are weighted according to their native uncertainties. Finally, a combination of these velocity fields is performed by the weighted regression kriging method. The result is a deformation grid, which, according to the data used, will present a spatial resolution in one to three dimensions. Associated to the grid, a second one of reliability index is created, taking into account the native uncertainties of each method in one hand, and the uncertainties related to the interpolation method on the other hand. The joint use of these two grids improve the discussion on the deformation.Our methodology was confronted with a case study: the subsidence of the Vauvert saline (Gard, France). This site appears ideal because of (1) its geographical situation close to Montpellier facilitating its instrumentation and measurements, (2) a subsidence is already measured, whose origin is the salt extraction between 2000 and 3000 m in depth, and (3) datasets are available in InSAR, GPS and Leveling. Two measurement periods are selected (2002-2009 and 2015-2017) because they allow the joint use of several measurement techniques. The results of the combination of velocity fields show the presence of a subsidence bowl with a maximum magnitude of 2.2 cm / year in the first period and 2.4 cm / year in the second period. Its extension reaches the southeastern area of the neighboring town of Vauvert. On the whole this subsidence (size of the bowl and magnitude) varies a little, nevertheless a zone closed to the historical sites of the Galine (east of the zone of study) presents an acceleration of the subsidence of approximately 8 mm / year between the two periods. This acceleration seems to reflect the surface effect of the closest wells purging , initiated in 2015.Finally, the use of an inversion algorithm of the data from the combination methodology (velocities and uncertainties) bring some new informations on the deformation anisotropy. An accommodation of geological structures in depth (folds, faults, detachments) is shown.

Large-scale volcanic deformation recently detected by radar interferometry (InSAR) provides new information and thus new scientific challenges for understanding volcano-tectonic activity and magmatic systems. The destabilization of such a system at depth noticeably affects the surrounding environment through magma injection, ground displacement and volcanic eruptions. To determine the spatiotemporal evolution of the Lazufre volcanic area located in the central Andes, we combined short-term ground displacement acquired by InSAR with long-term geological observations. Ground displacement was first detected using InSAR in 1997. By 2008, this displacement affected 1800 km2 of the surface, an area comparable in size to the deformation observed at caldera systems. The original displacement was followed in 2000 by a second, small-scale, neighbouring deformation located on the Lastarria volcano. We performed a detailed analysis of the volcanic structures at Lazufre and found relationships with the volcano deformations observed with InSAR. We infer that these observations are both likely to be the surface expression of a long-lived magmatic system evolving at depth. It is not yet clear whether Lazufre may trigger larger unrest or volcanic eruptions; however, the second deformation detected at Lastarria and the clear increase of the large-scale deformation rate make this an area of particular interest for closer continuous monitoring. / Vulkanische Deformationen in großem Maßstab, die mittels InSAR gemessen wurden, liefern neue Informationen und dadurch einen neuen Blickwinkel auf vulkan-tektonische Aktivitäten und das Verständnis von langlebigen, magmatischen Systemen. Die Destabilisierung eines solchen Systems in der Tiefe beeinflusst dauerhaft die Oberfläche durch Versatz des Bodens, magmatische Einflüsse und vulkanische Unruhen. Mit der Kombination aus kleinräumigem Bodenversatz gemessen mittels InSAR, numerischer Modellierung und langfristigen geologischen Beobachtungen, analysieren wir die Gegend um den Vulkan Lazufre in den Zentralanden, um die raumzeitliche Entwicklung der Region zu bestimmen. Bodenversatz wurde hierbei im Jahr 1997 mittels Radar-Interferrometrie (InSAR) gemessen, was eine Fläche von 1800 km² ausmacht, vergleichbar mit der Größe der Deformation des Kraters. Im Jahr 2000 wurde zusätzlich eine kleinräumige Deformation am Nachbarvulkan Lastarria entdeckt. Wir sehen räumliche als auch zeitliche Verbindungen zwischen der Deformation des Vulkans und vulkanischen Strukturen innerhalb der betroffenen Gegend. Wir folgern daraus, dass diese Beobachtungen der Ausdruck eines langlebigen, magmatischen Systems in der Tiefe an der Oberfläche sind. Es ist noch nicht klar, ob Lazufre größere vulkanische Unruhen, wie zum Beispiel Eruptionen auslösen könnte, aber die Deformation am Vulkan Lastarria und ein Anstieg der großräumigen Deformationsrate, machen diese Region interessant für eine zukünftige, kontinuierliche Überwachung.

Crustal deformation can be the result of volcanic and tectonic activity such as fault dislocation and magma intrusion. The crustal deformation may precede and/or succeed the earthquake occurrence and eruption. Mitigating the associated hazard, continuous monitoring of the crustal deformation accordingly has become an important task for geo-observatories and fast response systems. Due to highly non-linear behavior of the crustal deformation fields in time and space, which are not always measurable using conventional geodetic methods (e.g., Leveling), innovative techniques of monitoring and analysis are required. In this thesis I describe novel methods to improve the ability for precise and accurate mapping the spatiotemporal surface deformation field using multi acquisitions of satellite radar data. Furthermore, to better understand the source of such spatiotemporal deformation fields, I present novel static and time dependent model inversion approaches. Almost any interferograms include areas where the signal decorrelates and is distorted by atmospheric delay. In this thesis I detail new analysis methods to reduce the limitations of conventional InSAR, by combining the benefits of advanced InSAR methods such as the permanent scatterer InSAR (PSI) and the small baseline subsets (SBAS) with a wavelet based data filtering scheme. This novel InSAR time series methodology is applied, for instance, to monitor the non-linear deformation processes at Hawaii Island. The radar phase change at Hawaii is found to be due to intrusions, eruptions, earthquakes and flank movement processes and superimposed by significant environmental artifacts (e.g., atmospheric). The deformation field, I obtained using the new InSAR analysis method, is in good agreement with continuous GPS data. This provides an accurate spatiotemporal deformation field at Hawaii, which allows time dependent source modeling. Conventional source modeling methods usually deal with static deformation field, while retrieving the dynamics of the source requires more sophisticated time dependent optimization approaches. This problem I address by combining Monte Carlo based optimization approaches with a Kalman Filter, which provides the model parameters of the deformation source consistent in time. I found there are numerous deformation sources at Hawaii Island which are spatiotemporally interacting, such as volcano inflation is associated to changes in the rifting behavior, and temporally linked to silent earthquakes. I applied these new methods to other tectonic and volcanic terrains, most of which revealing the importance of associated or coupled deformation sources. The findings are 1) the relation between deep and shallow hydrothermal and magmatic sources underneath the Campi Flegrei volcano, 2) gravity-driven deformation at Damavand volcano, 3) fault interaction associated with the 2010 Haiti earthquake, 4) independent block wise flank motion at the Hilina Fault system, Kilauea, and 5) interaction between salt diapir and the 2005 Qeshm earthquake in southern Iran. This thesis, written in cumulative form including 9 manuscripts published or under review in peer reviewed journals, improves the techniques for InSAR time series analysis and source modeling and shows the mutual dependence between adjacent deformation sources. These findings allow more realistic estimation of the hazard associated with complex volcanic and tectonic systems. / Oberflächendeformationen können eine Folge von vulkanischen und tektonischen Aktivitäten sein, wie etwa Plattenverschiebungen oder Magmaintrusion. Die Deformation der Erdkruste kann einem Erdbeben oder einem Vulkanausbruch vorausgehen und/oder folgen. Um damit drohende Gefahren für den Menschen zu verringern, ist die kontinuierliche Beobachtung von Krustendeformationen eine wichtige Aufgabe für Erdobservatorien und Fast-Responce-Systems geworden. Auf Grund des starken nicht-linearen Verhaltens von Oberflächendeformationsgebiet in Zeit und Raum, die mit konventionellen Methoden nicht immer erfasst werden (z.B., Nivellements), sind innovative Beobachtungs- und Analysetechniken erforderlich. In dieser Dissertation beschreibe ich Methoden, welche durch Mehrfachbeobachtungen der Erdoberfläche nit satellitengestützem Radar eine präzise und akkurate Abbildung der raumzeitlichen Oberflächendeformationen ermöglichen. Um die Bildung und Entwicklung von solchen raumzeitlichen Deformationsgebieten besser zu verstehen, zeige ich weiterhin neuartige Ansätze zur statischen und zeitabhängigen Modellinversion. Radar-Interferogramme weisen häufig Gebiete auf, in denen das Phasensignal dekorreliert und durch atmosphärische Laufzeitverzögerung verzerrt ist. In dieser Arbeit beschreibe ich wie Probleme des konventionellen InSAR überwunden werden können, indem fortgeschrittene InSAR-Methoden, wie das Permanent Scatterer InSAR (PSI) und Small Baseline Subsets (SBAS), mit einer Wavelet-basierten Datenfilterung verknüpft werden. Diese neuartige Analyse von InSAR Zeitreihen wird angewendet, um zum Beispiel nicht-lineare Deformationsprozesse auf Hawaii zu überwachen. Radar-Phasenänderungen, gemessen auf der Pazifikinsel, beruhen auf Magmaintrusion, Vulkaneruption, Erdbeben und Flankenbewegungsprozessen, welche durch signifikante Artefakte (z.B. atmosphärische) überlagert werden. Mit Hilfe der neuen InSAR-Analyse wurde ein Deformationsgebiet ermittelt, welches eine gute Übereinstimmung mit kontinuierlich gemessenen GPS-Daten aufweist. Auf der Grundlage eines solchen, mit hoher Genauigkeit gemessenen, raumzeitlichen Deformationsgebiets wird für Hawaii eine zeitabhängige Modellierung der Deformationsquelle ermöglicht. Konventionelle Methoden zur Modellierung von Deformationsquellen arbeiten normalerweise mit statischen Daten der Deformationsgebiete. Doch um die Dynamik einer Deformationsquelle zu untersuchen, sind hoch entwickelte zeitabhängige Optimierungsansätze notwendig. Dieses Problem bin ich durch eine Kombination von Monte-Carlo-basierten Optimierungsansätzen mit Kalman-Filtern angegangen, womit zeitlich konsistente Modellparameter der Deformationquelle gefunden werden. Ich fand auf der Insel Hawaii mehrere, raumzeitlich interagierende Deformationsquellen, etwa Vulkaninflation verknüpft mit Kluftbildungen und Veränderungen in bestehenden Klüften sowie zeitliche Korrelationen mit stillen Erdbeben. Ich wendete die neuen Methoden auf weitere tektonisch und vulkanisch aktive Gebiete an, wo häufig die eine Interaktion der Deformationsquellen nachgewiesen werden konnte und ihrer bedeutung untersucht wurde. Die untersuchten Gebiete und Deformationsquellen sind 1) tiefe und oberflächliche hydrothermale und magmatische Quellen unterhalb des Campi Flegrei Vulkans, 2) gravitationsbedingte Deformationen am Damawand Vulkan, 3) Störungsdynamik in Verbindung mit dem Haiti Beben im Jahr 2010, 4) unabhängige blockweise Flankenbewegung an der Hilina Störungszone, und 5) der Einfluss eines Salzdiapirs auf das Qeshm Erdbeben im Süd-Iran im Jahr 2005. Diese Dissertation, geschrieben als kumulative Arbeit von neun Manuskripten, welche entweder veröffentlicht oder derzeit in Begutachtung bei ‘peer-review’ Zeitschriften sind, technische Verbesserungen zur Analyse von InSAR Zeitreihen vor sowie zur Modellierung von Deformationsquellen. Sie zeigt die gegenseitige Beeinflussung von benachbarten Deformationsquellen, und sie ermöglicht, realistischere Einschätzungen von Naturgefahren, die von komplexen vulkanischen und tektonischen Systemen ausgehen.

Landslides are one of the biggest natural hazards in Georgia, a mountainous country in the Caucasus. So far, no systematic monitoring and analysis of the dynamics of landslides in Georgia has been made. Especially as landslides are triggered by extrinsic processes, the analysis of landslides together with precipitation and earthquakes is challenging. In this thesis I describe the advantages and limits of remote sensing to detect and better understand the nature of landslide in Georgia. The thesis is written in a cumulative form, composing a general introduction, three manuscripts and a summary and outlook chapter. In the present work, I measure the surface displacement due to active landslides with different interferometric synthetic aperture radar (InSAR) methods. The slow landslides (several cm per year) are well detectable with two-pass interferometry. In same time, the extremely slow landslides (several mm per year) could be detected only with time series InSAR techniques. I exemplify the success of InSAR techniques by showing hitherto unknown landslides, located in the central part of Georgia. Both, the landslide extent and displacement rate is quantified. Further, to determine a possible depth and position of potential sliding planes, inverse models were developed. Inverse modeling searches for parameters of source which can create observed displacement distribution. I also empirically estimate the volume of the investigated landslide using displacement distributions as derived from InSAR combined with morphology from an aerial photography. I adapted a volume formula for our case, and also combined available seismicity and precipitation data to analyze potential triggering factors. A governing question was: What causes landslide acceleration as observed in the InSAR data? The investigated area (central Georgia) is seismically highly active. As an additional product of the InSAR data analysis, a deformation area associated with the 7th September Mw=6.0 earthquake was found. Evidences of surface ruptures directly associated with the earthquake could not be found in the field, however, during and after the earthquake new landslides were observed. The thesis highlights that deformation from InSAR may help to map area prone landslides triggering by earthquake, potentially providing a technique that is of relevance for country wide landslide monitoring, especially as new satellite sensors will emerge in the coming years. / Erdrutsche zählen zu den größten Naturgefahren in Georgien, ein gebirgiges Land im Kaukasus. Eine systematische Überwachung und Analyse der Dynamik von Erdrutschen in Georgien ist bisher nicht vorhanden. Da Erdrutsche durch extrinsische Prozesse ausgelöst werden, wird ihre Analyse zusammen mit Niederschlag und Erdbeben zu einer besonderen Herausforderung. In dieser Dissertation beschreibe ich die Potenziale und Limitierungen der Fernerkundung für die Detektion und das Verständnis von Erdrutschen in Georgien. Die Arbeit ist in einer kumulativen Form geschrieben, und besteht aus einer allgemeinen Einführung, drei Manuskripten sowie einer Zusammenfassung und einem Ausblick. In der vorliegenden Arbeit, Gestimme ich die Oberflächenverschiebung von aktiven Erdrutschen mit Methoden der Radarinterferometrie (InSAR). Die langsamen Erdrutsche (cm pro Jahr) konnten im einfachen Vergleich zeitlich unterschiedlicher Radaraufnahmen (two-pass InSAR), gut nachgewiesen werden. Die extrem langsamen Erdrutsche (mm pro Jahr) konnten hingegen nur mit InSAR Zeitreihentechniken nachgewiesen werden. Der Erfolg der angewandten InSAR Techniken wird durch die erfolgreiche Identifikation von bisher unbekannten Erdrutschen in Zentral Georgien veranschaulicht. Sowohl das Ausmaß als auch die Verschiebungsrate der Erdrutsche wurden quantifiziert. Ferner, um die mögliche Tiefe und Lage von potentiellen Gleitflächen zu bestimmen, wurden inverse Modelle entwickelt. Inverse Modellierung sucht nach Parametern der Quelle, welche die beobachtete Verschiebungsverteilung reproduzieren können. Ferner habe ich anhand der ermittelten Verschiebungsverteilung aus InSAR in Verbindung mit der Morphologie aus Luftaufnahmen das Volumen der untersuchten Erdrutsche empirisch abgeleitet. Ich habe eine Volumenformel für unseren Fall angepasst, und die verfügbaren Datensätze bezüglich Seismizität und Niederschlag kombiniert, um potenzielle auslösende Faktoren zu analysieren. Eine leitende Frage hierbei war: Was sind die Ursachen für die Beschleunigung von Erdrutschen, wie sie in den InSAR Daten beobachtet werden konnte? Das Untersuchungsgebiet in Zentral Georgien ist seismisch sehr aktiv. Als zusätzlichen Produkt der InSAR Datenanalyse wurde ein Deformationsgebiet gefunden, welches im Zusammenhang mit dem Mw=6.0 Erdbeben vom 7. September 2009 zusammenhängt. Beweise für Oberflächenbrüche, die direkt mit dem Erdbeben zusammenhängen, konnten in dem Gebiet nicht gefunden werden, jedoch konnten während und nach dem Erdbeben neue Erdrutsche beobachtet werden. Die Dissertation unterstreicht, dass Verformungsinformationen aus InSAR Analysen helfen können ein Gebiet, welches von Erdbebeninduzierten Erdrutschen gefährdet ist, zu kartieren. Potenziell stellt InSAR eine Technik dar, die von Bedeutung für die landesweite Überwachung von Erdrutschen sein kann, insbesondere im Hinblick auf die neuen Satellitensensoren, die in den kommenden Jahren verfügbar sein werden.

This work makes an attempt to explain the origin, features and potential applications of the elevation bias of the synthetic aperture radar interferometry (InSAR) datasets over areas covered by vegetation. The rapid development of radar-based remote sensing methods, such as synthetic aperture radar (SAR) and InSAR, has provided an alternative to the photogrammetry and LiDAR for determining the third dimension of topographic surfaces. The InSAR method has proved to be so effective and productive that it allowed, within eleven days of the space shuttle mission, for acquisition of data to develop a three-dimensional model of almost the entire land surface of our planet. This mission is known as the Shuttle Radar Topography Mission (SRTM). Scientists across the geosciences were able to access the great benefits of uniformity, high resolution and the most precise digital elevation model (DEM) of the Earth like never before for their a wide variety of scientific and practical inquiries. Unfortunately, InSAR elevations misrepresent the surface of the Earth in places where there is substantial vegetation cover. This is a systematic error of unknown, yet limited (by the vertical extension of vegetation) magnitude. Up to now, only a limited number of attempts to model this error source have been made. However, none offer a robust remedy, but rather partial or case-based solutions. More work in this area of research is needed as the number of airborne and space-based InSAR elevation models has been steadily increasing over the last few years, despite strong competition from LiDAR and optical methods. From another perspective, however, this elevation bias, termed here as the “biomass impenetrability”, creates a great opportunity to learn about the biomass. This may be achieved due to the fact that the impenetrability can be considered a collective response to a few factors originating in 3D space that encompass the outermost boundaries of vegetation. The biomass, presence in InSAR datasets or simply the biomass impenetrability, is the focus of this research. The report, presented in a sequence of sections, gradually introduces terminology, physical and mathematical fundamentals commonly used in describing the propagation of electromagnetic waves, including the Maxwell equations. The synthetic aperture radar (SAR) and InSAR as active remote sensing methods are summarised. In subsequent steps, the major InSAR data sources and data acquisition systems, past and present, are outlined. Various examples of the InSAR datasets, including the SRTM C- and X-band elevation products and INTERMAP Inc. IFSAR digital terrain/surface models (DTM/DSM), representing diverse test sites in the world are used to demonstrate the presence and/or magnitude of the biomass impenetrability in the context of different types of vegetation – usually forest. Also, results of investigations carried out by selected researchers on the elevation bias in InSAR datasets and their attempts at mathematical modelling are reviewed. In recent years, a few researchers have suggested that the magnitude of the biomass impenetrability is linked to gaps in the vegetation cover. Based on these hints, a mathematical model of the tree and the forest has been developed. Three types of gaps were identified; gaps in the landscape-scale forest areas (Type 1), e.g. forest fire scares and logging areas; a gap between three trees forming a triangle (Type 2), e.g. depending on the shape of tree crowns; and gaps within a tree itself (Type 3). Experiments have demonstrated that Type 1 gaps follow the power-law density distribution function. One of the most useful features of the power-law distributed phenomena is their scale-independent property. This property was also used to model Type 3 gaps (within the tree crown) by assuming that these gaps follow the same distribution as the Type 1 gaps. A hypothesis was formulated regarding the penetration depth of the radar waves within the canopy. It claims that the depth of penetration is simply related to the quantisation level of the radar backscattered signal. A higher level of bits per pixels allows for capturing weaker signals arriving from the lower levels of the tree crown. Assuming certain generic and simplified shapes of tree crowns including cone, paraboloid, sphere and spherical cap, it was possible to model analytically Type 2 gaps. The Monte Carlo simulation method was used to investigate relationships between the impenetrability and various configurations of a modelled forest. One of the most important findings is that impenetrability is largely explainable by the gaps between trees. A much less important role is played by the penetrability into the crown cover. Another important finding is that the impenetrability strongly correlates with the vegetation density. Using this feature, a method for vegetation density mapping called the mean maximum impenetrability (MMI) method is proposed. Unlike the traditional methods of forest inventories, the MMI method allows for a much more realistic inventory of vegetation cover, because it is able to capture an in situ or current situation on the ground, but not for areas that are nominally classified as a “forest-to-be”. The MMI method also allows for the mapping of landscape variation in the forest or vegetation density, which is a novel and exciting feature of the new 3D remote sensing (3DRS) technique. Besides the inventory-type applications, the MMI method can be used as a forest change detection method. For maximum effectiveness of the MMI method, an object-based change detection approach is preferred. A minimum requirement for the MMI method is a time-lapsed reference dataset in the form, for example, of an existing forest map of the area of interest, or a vegetation density map prepared using InSAR datasets. Preliminary tests aimed at finding a degree of correlation between the impenetrability and other types of passive and active remote sensing data sources, including TerraSAR-X, NDVI and PALSAR, proved that the method most sensitive to vegetation density was the Japanese PALSAR - L-band SAR system. Unfortunately, PALSAR backscattered signals become very noisy for impenetrability below 15 m. This means that PALSAR has severe limitations for low loadings of the biomass per unit area. The proposed applications of the InSAR data will remain indispensable wherever cloud cover obscures the sky in a persistent manner, which makes suitable optical data acquisition extremely time-consuming or nearly impossible. A limitation of the MMI method is due to the fact that the impenetrability is calculated using a reference DTM, which must be available beforehand. In many countries around the world, appropriate quality DTMs are still unavailable. A possible solution to this obstacle is to use a DEM that was derived using P-band InSAR elevations or LiDAR. It must be noted, however, that in many cases, two InSAR datasets separated by time of the same area are sufficient for forest change detection or similar applications.

Evaluation of influence of interpolation methods on coregistration of radar images Abstract SAR interferogram processing requires subpixel coregistration of SAR image pair for accurate phase differencing. Errors in alignment introduce phase noise in SAR interferogram. Last step in coregistration is resampling one of SAR images. Also this step introduces errors in SAR interferogram. The resampling algorithms Nearest Neighbor, Bilinear interpolation, Cubic Convolution and advanced methods such as Raised Cosine kernel, Knab interpolation kernel and Truncated Sinc were tested on ERS tandem data and compared. The results were compared with the theory and simulations of earlier investigations (Hanssen, Bamler, 1999), (Migliaccio, Bruno, 2003) and (Cho ... [et al.], 2005). The main experiment in this work was to examine and compare resampling methods on real data to evaluate their effect on the interferometric phase quality and DEM generation. The coregistration performance was evaluated by the coherence (Touzi ... [et al.], 1999) and the sum of phase differences (Li ... [et al.], 2004). No evidence showed that computationally intensive algorithms produced better quality of interferogram than Cubic Convolution. The possibilities of evaluating by means of the accuracy of the final InSAR DEM (Li, Bethel, 2008) were...

Le système Géocube est un réseau de capteurs GPS conçu et développé par le Laboratoire d'OptoÉléctronique de Métrologie et d'Instrumentation (LOEMI) de l'Institut National de l'Information Géographique et Forestière (IGN) et maintenu par le même laboratoire et l'entreprise Ophelia- Sensors qui s'occupe de son industrialisation. Il a comme objectif de mesurer les déformations du sol avec une précision millimétrique. Ce réseau de capteurs a la particularité d'être à la fois très peu énergivore, d’un faible coût de revient, simple d’installation et d’utilisation. Il est donc bien adapté à l’usage dans un environnement difficile, comme les volcans. Ce système a déjà été testé avec succès lors d’une précédente thèse sur le glacier d’Argentière et sur un glissement de terrain proche de Super-Sauze en France. La première partie de cette thèse porte sur l’optimisation du système de calcul du Géocube pour l'adapter à des réseaux de tailles plus importantes horizontalement et verticalement en vue de son utilisation dans un contexte volcanique. Cela passe, d’abord, par l’intégration d’une stratégie pour l’estimation du biais troposphérique dans le filtre de Kalman qui constitue le coeur du logiciel de calcul du Géocube. Cette amélioration est ensuite validée en utilisant les données de quelques réseaux GNSS permanents nationaux et internationaux. La deuxième partie consiste à étudier l’apport d’un réseau dense de Géocubes à l’étude du volcanisme à travers une expérience conduite sur le flanc sud-est de l’Etna, où cinq Géocubes ont été déployés entre le 12 Juillet 2016 et le 10 Juillet 2017. Les résultats obtenus et les enseignements tirés de cette expérimentation sont discutés et analysés. Enfin, nous validons les résultats obtenus avec les Géocubes en appliquant une technique PSI (Persistent Scatterer InSAR) sur des interférogrammes RADAR calculés à partir des données des satellites Sentinel-1A/B et qui couvrent la période de déploiement des Géocubes sur l’Etna. Ces deux méthodes (GPS et RADAR) se sont avérées complémentaires puisque le RADAR apporte la densité spatiale des mesures et le système Géocube la précision et la continuité temporelle. / The Geocube system is a network of wireless GPS sensors designed and developed by the Laboratory of Opto-Electronics, Metrology and Instrumentation (LOEMI) of the National Institute of Geographical and Forest Information (IGN) and maintained by the same laboratory and Ophelia-Sensors, the company responsible for its industrialization. Its purpose is to measure ground deformations with millimetre accuracy. This sensor network has the particularity of being very low in energy consumption, low cost, easy to install and easy to use. It is suited for use in harsh environments, such as volcanoes. This system has already been successfully tested in a previous works on the Argentière glacier and a Super-Sauze landslide in France. The first part of this thesis deals with the optimization of the Geocube system for larger networks, horizontally and vertically, in order to use it in a volcanic context. First, a new strategy to estimate the tropospheric bias has been implemented into the Kalman filter (the heart of the Geocube processing software) in real time and in post-processing. This improvement is then validated using data from some national and international permanent GNSS networks. The second part consists in studying the contribution of a dense Geocubes network to the study of volcanism through an experiment conducted on the southeastern flank of Etna, where five Geocubes were deployed between July, 12th 2016 and July, 10th 2017. The results obtained from this experiment are discussed and analysed. Finally, the results obtained with Geocubes are validated by applying a PSI (Persistent Scatterer InSAR) technique on RADAR interferograms calculated from Sentinel-1A/B satellite data covering the period of deployment of the Geocubes on Etna. These two methods (GPS and RADAR) turned out to be complementary since RADAR provides the spatial density of measurements and the Geocube system provides accuracy and temporal continuity.

Les études des déformations de surface en relation avec l'activité volcanique permettent de quantifier les phénomènes de transfert de magma qui s'opèrent dans les structures superficielles et profondes d'un édifice volcanique. Ces études s'appuient essentiellement sur l'utilisation de séries temporelles acquises par des réseaux de récepteurs GNSS installés sur les flancs de l'édifice volcanique et sur l'utilisation d'images acquises par des satellites équipés de capteurs à ouverture de synthèse. Les objectifs de ce travail ont été de mettre en œuvre sur deux des volcans les plus actifs du monde des méthodes numériques pour détecter, analyser et interpréter les déformations du sol associées à l'activité. Sur le Piton de la Fournaise, nous avons analysé l'évolution spatiale et temporelle du champ de déplacement entre l'éruption historique d'avril 2007 et octobre 2014 à partir de l'analyse de mesures continues acquises par les stations GNSS et de longues séries temporelles d'images radar Cosmos-Skymed et TerraSAR acquises en bande X. Pour le traitement des données radars, nous avons adopté une approche classique qui exploite la redondance d'information dans les interférogrammes et nous avons mis en œuvre une méthode originale de correction des effets troposphériques reposant sur la décomposition des signaux radars en valeurs singulières. La complexité spatiale et temporelle du champ de déplacement obtenu indique qu'une partie importante de l'édifice volcanique est affectée par des déformations d'origines diverses qui se superposent spatialement et temporellement. Ainsi, on observe des processus de subsidence qui ne s'accompagnent pas de déplacements horizontaux sur les coulées de lave récentes. Nous montrons qu'il existe une relation linéaire entre cette subsidence et l'épaisseur de la coulée et que son amplitude décroit avec le temps. Ces relations nous permettent de construire une loi empirique pour estimer la contribution de ce processus dans le champ de déformation. Nous observons également que le cône central subside de manière persistante durant la période étudiée. Nous interprétons cette subsidence comme l'expression d'une relaxation des contraintes provoquée par l'effondrement de plus de 350 m du Dolomieu survenu lors de l'éruption d'avril 2007. Enfin, nous montrons qu'une large partie du flanc est de l'édifice volcanique est affectée d'un mouvement lent le long de la pente entre 2007 et 2014. L'absence d'évidences sur la structure et sur la rhéologie de l'édifice nous amène à explorer différentes hypothèses pour expliquer l'origine de ce glissement qui pourrait être contrôlé par les propriétés frictionnelles des roches le long d'un ou de plusieurs plans de faille, ou bien être l'expression d'une déformation ductile dépendante de la viscosité du milieu. Le Llaima est un large strato-volcans andin dont les processus de déformations sont mal compris à cause principalement de la complexité de son mode de fonctionnement mais, également, aussi par l'absence de réseaux d'observation au sol. Dans ce contexte, les potentialités des données radar pour le suivi des déformations de surface de ces volcans constituent un intérêt scientifique majeur. / We address in this dissertation the use of Interferometric Synthetic Aperture Radar (InSAR) to measure and characterize the ground surface deformation at two volcanoes - Piton de la Fournaise (La Réunion Island, France) and Llaima (Chile). For Piton de la Fournaise, we analyzed the spatial pattern and temporal evolution of the ground displacement between the historical March-April 2007 eruption and October 2014, based on continuous measurements recorded by GNSS stations and X band COSMO-SkyMed and TerraSAR-X/TanDEM-X time series analysis. For the processing of radar data, we adopted a classical InSAR time series approach that exploits the information redundancy in the interferograms and we implemented an original method for correcting artifacts based on the principal component decomposition. The spatial and temporal complexity of the obtained deformation field indicates that an important part of the volcanic edifice is affected by deformations of various origins that overlap spatially and temporally. We observe also subsidence processes that are not accompanied by horizontal displacements in recent lava fields. We show that there exists a linear relationship between the subsidence and the thickness of lava and that the amplitude of subsidence decreases with time. These relationships allow us to construct an empirical law to estimate the contribution of post-lava emplacement process in the deformation field. We also observe that the Central Cone subsides persistently during the study period. We interpret this subsidence as the expression of a relaxation of the stresses caused by the Dolomieu collapse during the March-April 2007 eruption. Finally, we show that a widespread time-dependent moving sector on the Eastern Flank is affected by downslope motion during the 2007-2014 period. The uncertainties on both the structure and rheology parameters of the edifice leads us to explore different hypotheses to explain the origin of this flank motion which could be controlled by the frictional properties of the rocks along one or more fault planes, or be the expression of a dependent ductile deformation of the viscosity of the medium. Llaima is a large Andean stratospheric volcano whose deformation processes are poorly understood not only because of the complexity of its functioning mode but also because of the absence of observation networks on the ground. In this context, the potential of radar data for monitoring the ground deformations of these volcanoes is a main scientific interest. However, the complex environment conditions (steep slopes, snow- or ice-capped summit, dense vegetation cover, and strong tropospheric artifacts) and limited amount of available radar data make it challenging to accurately measure ground displacement with InSAR.

Atmospheric artifacts superimposed on interferometric synthetic aperture radar (InSAR) measurements have the potential to greatly impede the accurate estimation of deformation signals. The research presented in this dissertation demonstrates a novel InSAR time series algorithm, called HiRAPS algorithm, for effectively estimating high resolution atmospheric phase screens (APS) from differential InSAR measurements. In summary, the HiRAPS algorithm utilizes short time span differential interferograms and rearranges components of existing advanced InSAR techniques to identify a higher density of scatterers used to create the APS. The improved scatterer density allows one to estimate high spatial frequency atmospheric signals not recovered from existing InSAR time series techniques. The HiRAPS algorithm was tested with simulated and actual data, which contain phase contributions from linear and nonlinear deformation, topographic height errors, and atmospheric artifacts. Simulated differential interferograms were generated to have the same spatial and temporal baselines as the actual differential interferograms formed from RADARSAT-1 data over Phoenix, Arizona. The APS superimposed on simulated differential interferograms were then estimated and compared to simulated APS. The root mean square error (RMSE) between the estimated and simulated APS was calculated to qualitatively assess the different values obtained. The RMSE was 0.26 radians when utilizing the HiRAPS algorithm, compared to an RMSE value of 0.39 radians using an implementation of the permanent scatterer (PS) algorithm. The HiRAPS algorithm also showed its applicability for estimating high spatial frequency atmospheric signals for actual data. Sixty-six SAR images, starting from October 5, 2002 and spanning 5 years, were processed for this research. The APS pixel density obtained using the HiRAPS algorithm was 253 pixels per square kilometer, compared to 14 pixels per square kilometer utilizing the PS algorithm. The APS superimposed on the differential interferograms were estimated with both the proposed and PS algorithms. High resolution APS were estimated with the HiRAPS algorithm, whereas only low resolution APS were obtained with the PS algorithm. After estimating and removing estimated APS, the phase stability of APS-free differential interferograms was examined by identifying the permanent scatterers (PS). The final density of identified PS obtained with the HiRAPS algorithm was 453 PS per square kilometer, whereas the density of detected PS using the generic PS algorithm was 381 PS per square kilometer. The maximum difference in the deformation time series between the HiRAPS algorithm and the PS algorithm was less than 6 mm. However, the HiRAPS algorithm resulted in less apparent noise in the time series than the PS algorithm due to the precise estimation of APS. / text