Analyse InSAR des déformations de volcans actifs

Development of Interferometric Synthetic Aperture

Radar Interferometric Synthetic Aperture Radar (InSAR), also referred to as SAR Interferometry, is a recently developed technique of geodesy, based on the phase difference between two SAR images. These images taken over a same region can be acquired at the same time by two CHAPTER 1: INTRODUCTION 24 antennas or acquired at two different times, making it possible to measure the topography or the ground deformation of an area, especially the inaccessible areas. The estimation of topography had been the main focus for the early applications (in the 1970s and 1980s) of InSAR, yielding elevation accuracies comparable with optical methods. One of the most valuable examples of application of InSAR to measure the topography is the construction of a global Digital Elevation Model (DEM) (Shuttle Radar Topography Mission, NASA; Farr & Kobrick, 2000), by treatment of SAR images simultaneously acquired with two different viewing angles. The obtained DEM is an estimate of the altitude at each point of the world, with a metric precision and a spatial resolution of 30 m on the territory of the United States of American and 90 m on the rest of the world. This mission has been ranked among the most successful missions in NASA. Acquiring almost simultaneously the SAR data can also be accomplished by the missions involving tandem satellites such as TerraSAR-X/TanDEM-X. With development of geodetic technique in the late 1980s, the geophysical applications of InSAR and Differential InSAR (hereafter using the term InSAR representing the both) enabled geodesists, since the early 1990s and the launch of the ERS-1 satellite, to monitor displacement of ground surface (Massonnet & Rebaute, 1985; Gabriel et al., 1989; Massonnet et al., 1993). The relative LOS movement of scatters with respect to a reference location in the image could be measured as a fraction of the wavelength, yielding cm to mm accuracies for L-band, C-band, and X-band radars (Gabriel et al., 1989). One of the first spectacular examples of InSAR measuring ground deformation was reported by Massonnet et al. (1993) concerning the displacement field of the Landers, Californie earthquake. They produced a displacement map in the area associated with the seismic rupture, based on two acquisitions of ERS-1 satellite acquired before and after the earthquake. Since then, numerous InSAR studies have been carried out focusing on ground deformation associated with various natural causes or human activities, such as volcano dynamics (e.g. Lu et al., 1997; Lanari et al., 1998; Pritchard & Simons, 2002, 2004; Froger et al., 2004; Brooks et al., 2008), glacier movement (e.g. Joughin et al., 1995; Kumar et al., 2011), earthquakes (e.g. Simons et al., 2002; Fialko et al., 2005; Bignami et al., 2012), urban subsidence (e.g. Amelung et al., 1999), mining activity (e.g. Gourmelen et al., 2007; Herrera et al., 2007), etc.. These works thus highlighted the main contribution of InSAR compared with other geodetic techniques (e.g. the Global Positioning System), which is, the CHAPTER 1: INTRODUCTION 25 excellent resolution and high spatial coverage making it possible to quantify the spatial variability of numerous phenomena. 

InSAR monitoring volcano deformation

Comparing with classical ground-based geodetic instruments that are widely used for volcano monitoring such as, theodolite, tiltmeters, electronic distance meter, seismic, and GNSS networks, the space-based technique, InSAR, is particularly valuable because it provides regional and spatially continuous measurements of ground surface displacement with unprecedented accuracy, and can be easily applied remotely for regions where there may be limitations for working on site with traditional instrumentations. At present, many active volcanoes in the world such as Etna (Italy), Stromboli (Italy), Kilauea (Hawaii), Piton de la Fournaise (La Réunion), and many andesitic volcano of the Andean Cordillera (the Llaima, Lastarria, Lascar, etc..), are regularly monitored by InSAR (e.g. Pritchard & Simons, 2002, 2004; Antonello et al., 2004; Palano et al., 2008; Neri et al., 2009; Fournier et al., 2010; Fukushima et al., 2010; Shirzaei et al., 2013; Di Traglia et al., 2014; Froger et al., 2015; Jo et al., 2015; Remy et al., 2015). The InSAR observed surface deformation alone is interesting, but more importantly, the interpretations are especially useful at volcanoes where surface displacement are associated with magma activity (e.g. Chadwick et al., 2006; Lundgren et al., 2013), movement of lava flows (e.g. Briole et al., 1997; Stevens et al., 2001a; Bato et al., 2016; Chaussard, 2016; this study), or flank instability (e.g. Solaro et al., 2010; Intrieri et al., 2013; Froger et al., 2015). Therefore, routinely monitoring surface deformation at volcanoes helps infer subsurface processes that cannot be directly observed and can improve our understanding of the state of volcano to further assess potential volcano hazard. In spite of all the advantages of InSAR for monitoring surface deformation at volcanoes, their capability remains limited in some cases. Many volcanoes are located in intertropical or temperate areas where the frequent precipitations and fertile soils provide optimal conditions for the development of dense vegetation that can cause temporal decorrelation of interferometric phases (Pinel et al., 2011). Snow cover, summit ice caps can also contribute to temporal decorrelation (Lu & Dzurisin, 2014). In addition, spatial and temporal variations of the troposphere can induce strong tropospheric (long- and short-wavelength) artifacts on InSAR products (Beauducel et al., 2000; Remy et al., 2015). All these factors have negative impacts on InSAR applicability to volcano deformation monitoring, by reducing or even preventing the CHAPTER 1: INTRODUCTION 27 ability to distinguish ground deformation signals from other radiometric contributions (Zebker et al., 1997; Remy et al., 2003; Doin et al., 2009; Pinel et al., 2011; Hooper et al., 2012). 

Two case studies: Piton de la Fournaise (La Réunion, France) and Llaima (Chile) 

The main objective of this dissertation is to monitor and interpret the ground deformation at both Piton de la Fournaise and Llaima volcanoes by taking full advantage of InSAR data acquired by different satellites. We are aiming to provide accurate results and valuable insights for understanding the state of volcano and the subsurface processes. Piton de la Fournaise is one of the most active basaltic volcanoes in the world with 44 occurrences of eruption between 1998 and 2015 (Roult et al., 2012; observation of Observatoire Volcanologique du Piton de la Fournaise). Most of the recent activity occurred either in the summit crater or directly on the flank of the Central Cone. Exceptionally, fissures opened at low elevation far from the Central Cone in the Plaine des Osmondes or in the Grand Brulé (Peltier et al., 2009a). Among all the recent eruptions, the March-April 2007 eruption is considered as the largest eruption ever observed at this volcano during the 20th and 21th centuries, characterized by a distal effusion of 210 ×106 m3 volume of lava at low elevation in the Grand Brulé and a 340 m consequent collapse of the Dolomieu crater (e.g. Staudacher et al., 2009; Bachèlery et al., 2010). A notable widespread downslope motion (up to 1.4 m to the east and up to 0.37 m of uplift) affecting the Eastern Flank during co-eruptive period was observed by previous studies (Augier, 2011; Froger et al., 2015). Moreover, this downslope movement of the Eastern Flank had lasted from the end of the eruption to at least July 2008, with overall up to 0.18 m to the east and up to 0.35 m of subsidence during this one-year post-eruptive period (Augier, 2011; Froger et al., 2015). Therefore, it is essential and urgent to characterize and interpret the ground deformation field at Piton de la Fournaise, in particular on the Eastern Flank, in order to reduce the possibility of flank instabilities, which can result in large catastrophic flank destabilizations and tsunami (McGuire, 1996). The main challenge in using InSAR data to study ground deformation at Piton de la Fournaise is the topography-correlated displacement. Further, the external data-based atmospheric artifact correction methods are inapplicable due to the CHAPTER 1: INTRODUCTION 28 unavailability of external data corresponding to the InSAR data over Piton de la Fournaise. In order to overcome these difficulties, we adopt an enhanced Small Baselines (SB) -InSAR method based on Stanford Method for Persistent Scatterers (StaMPS; Hooper et al., 2012) to perform a detailed time series analysis on SAR data acquired over Piton de la Fournaise. We propose a new correction approach grounded on a principal component decomposition implemented in Principal Component Analysis-based Inversion Method package (PCAIM; Kositsky and Avouac, 2010) for discriminating the displacement signal from artifact in interferograms. This approach is proved to be efficient for the case of this study, allowing us to finally obtain high-precision InSAR time series maps of ground displacement at Piton de la Fournaise. Llaima is not only one of the biggest volcanoes in Chile but also one of the most historically active volcanoes in South America with a record of around 50 documented eruptions between 1640 and 2009 (Naranjo & Moreno, 1991; Naranjo & Moreno, 2005; Moreno et al., 2009). Its recent activity has been characterized by relatively small eruptions between 1979 and 2003 (11 eruptions with VEI ≤ 2) and by a slight increase in explosivity for the last eruptive cycle, between March 2007 and June 2009 (with 2 ≤ VEI ≤ 3) (Smithsonian Institution Global volcanism report available at http://www.volcano.si.edu). Differing from basaltic volcano Piton de la Fournaise, the andesitic volcano Llaima has steep slopes and large scale near-summit glaciers. Thus, relatively minor eruptions can create hazards (as the eruptions in January 2008 and April 2009) due to the presence of large scale near-summit glaciers, which are highly susceptible to melting and lahar generation during energetic fountaining (Bouvet de Maisonneuve et al., 2012). Therefore, detecting ground deformation at Llaima volcano can contribute to preventing potential damage to local population and property. For Llaima, we first perform a careful analysis of the water vapor variations using Medium-Resolution Imaging Spectrometer (MERIS) and Moderate Resolution Imaging Spectroradiometer (MODIS) near-infrared water vapor products, and then we invert wrapped interferograms for both topographic correlated phase delays and a simple model source strength. Rigorous analysis allows us to reach a revised interpretation with respect to that of Bathke et al. (2011).