No description available.

Subduction zone fault processes range from tsunami-generating megathrust events to aseismic creep along the deeper portions of the fault. Episodic tremor and slow slip (ETS) represents the transition between these two regimes, where slip occurs at semi-regular recurrence intervals of months-to-years. These events are also accompanied by low frequency earthquakes, referred to as tremor. The study of ETS in Cascadia has been made possible by the enhancement of large-scale seismic and geodetic networks. In this dissertation, I use a range of geodetic and seismic observations at sub-daily to decadal time scales to investigate the kinematic behavior of individual ETS events, as well as the long-term behavior of the ETS zone and its relationship with the updip seismogenic zone. In Cascadia, current seismic hazard maps use the ETS zone as the downdip limit of rupture during future megathrust events. In Chapter II, I utilize uplift rates derived from 80 years of leveling measurements to explore the possibility that long-term strain accumulation exists near the ETS zone. The uplift rates are consistent with a region of 10-20% locking on the updip side of the ETS zone. The lack of associated topography indicates that the accumulated strain must be released during the megathrust cycle. The correlation of tremor and slip in Cascadia suggests there is an inherent relationship between the two. In Chapter III, I develop a method for using tremor as a proxy for slip to assess the spatial relationship of tremor and slip. I compare predictions of tremor-derived slip models to results from static inversions of GPS offsets by modeling slip based on the density of tremor. These comparisons suggest that the correlation of tremor and slip is variable along strike and along dip. In Chapter IV, I explore how borehole strainmeters can improve our resolution of slip on the plate interface. I incorporate strainmeters into joint, time-dependent kinematic inversions with GPS data. The temporal resolution of strainmeters provides improved constraints when deriving time-dependent slip estimates during slow slip events, allowing us to better image the kinematics of slow slip. This dissertation includes previously published and unpublished material.

Three-dimensional mechanical simulations of the San Andreas fault (SAF) within the Coachella Valley in California produce deformation that match geologic observations and demonstrate the impact of fault geometry on uplift patterns. Most models that include the Coachella Valley segment of the SAF have assumed a vertical orientation, but recent studies suggest that this segment dips 60-70° northeast. We compare models with varied fault geometry and evaluate how well they reproduce observed uplift patterns. Our model with a dipping SAF matches geologic observations, while models containing a vertical fault do not. This suggests that the active Coachella Valley segment of the SAF dips 60-70° northeast. Since ~1.5 Ma, the SAF in this region has undergone a major reorganization that entailed initiation of the San Jacinto fault and termination of slip on the West Salton detachment fault. The trace of the SAF itself has also evolved, with several shifts in activity through the San Gorgonio Pass. Despite a rich geologic record of these changes, the mechanisms that controlled abandonment of faults, initiation of new strands, and shifting loci of uplift are poorly understood. We model snapshots in time through the evolution of the fault system, and assess the mechanical viability of our snapshots by comparison with uplift patterns inferred from the stratigraphic record. Model results are compared with vertical axis rotation. We examine incipient faulting using maps of strain energy density, and explore changes to the mechanical efficiency of the system to better understand the evolution of this fault system.

Induced seismicity (earthquakes caused by injection or extraction of fluids in Earth's subsurface) is a major, new hazard in the United States, the Netherlands, and other countries, with vast economic consequences if not properly managed. Addressing this problem requires development of predictive simulations of how fluid-saturated solids containing frictional faults respond to fluid injection/extraction. Here we present a numerical method for linear poroelasticity with rate-and-state friction faults. A numerical method for approximating the fully coupled linear poroelastic equations is derived using the summation-by-parts-simultaneous-approximation-term (SBP-SAT) framework. Well-posedness is shown for a set of physical boundary conditions in 1D and in 2D. The SBP-SAT technique is used to discretize the governing equations and show semi-discrete stability and the correctness of the implementation is verified by rigorous convergence tests using the method of manufactured solutions, which shows that the expected convergence rates are obtained for a problem with spatially variable material parameters. Mandel's problem and a line source problem are studied, where simulation results and convergence studies show satisfactory numerical properties. Furthermore, two problem setups involving fault dynamics and slip on faults triggered by fluid injection are studied, where the simulation results show that fluid injection can trigger earthquakes, having implications for induced seismicity. In addition, the results show that the scheme used for solving the fully coupled problem, captures dynamics that would not be seen in an uncoupled model. Future improvements involve imposing Dirichlet boundary conditions using a different technique, extending the scheme to handle curvilinear coordinates and three spatial dimensions, as well as improving the high-performance code and extending the study of the fault dynamics.

Les séismes le long de grandes failles crustales représentent un danger énorme pour de nombreuses populations. Le mécanique de ces failles est influencé par des zones endommagées qui entourent le coeur de faille. La fracturation dans ces zones contrôle chaque étape du cycle sismique. En effet, cette zone contrôle la mécanique de la rupture sismique, elle est un conduit pour les fluides, réagit chimiquement sous l'effet de fluides réactifs, et facilite la déformation pendant les périodes post- et inter-sismiques. Dans cette thèse de doctorat, des expériences de laboratoire ont été réalisées pour mieux comprendre 1) la façon dont l'endommagement est généré pendant le chargement transitoire co-sismique, 2) comment l'endommagement permet de mieux contraindre le chargement co-sismique le long de grandes failles, et iii) comment les fractures peuvent se cicatriser au fil du temps et contrôler l'évolution de la perméabilité et de la résistance mécanique de la faille.L'introduction de la thèse propose une revue critique de la littérature sur la génération de dommages co-sismiques et en particulier sur la formation des roches pulvérisées. Le potentiel de ces roches comme marqueur des déformations co-sismiques est discuté. Bien que ces roches pulvérisées soient prometteuses pour ces aspects, plusieurs questions restent ouvertes.L'une de ces questions concerne les conditions de chargement transitoire nécessaires pour atteindre la pulvérisation. Le seuil de taux de deformation pour atteindre la pulvérisation peut être réduit par des endommagemments progressifs, au cours de ruptures sismiques successives. Des barres de Hopkinson ont été utilisées pour effectuer des chargements dynamique successifs d'une roche cristalline (monzonite). Les résultats montrent que le seuil pour atteindre la pulvérisation est réduit d'au moins 50% lorsque des chargements successives sont imposés. Cette thèse discute aussi pourquoi les roches pulvérisées sont presque toujours observées dans des roches cristallines et peu dans des roches sédimentaires poreuses. Pour comprendre cette observation, des expériences à haute vitesse de déformation ont été effectuées sur des grès de Rothbach. Les résultats montrent que la pulvérisation des grains eux mêmes ne se produit pas dans les grès. L'endommagement reste se produit principalement à une échelle supérieure à celle grains, et des bandes de compaction sont observées. La compétition entre l'endommagement inter- et intra-granulaire est expliquée par les paramètres microstructuraux en combinant deux modèles micromécaniques classiques. Les microstructures observées dans les grès peuvent se former dans le régime quasi-statiques et aussi dans le régime dynamique. Par conséquent, il est recommandée d'être prudent lors de l'interprétation du mécanisme de deformation dans les roches sédimentaires proches de la surface. La dernière question abordée durant la thèse est la cicatrisation post-sismique de fractures co-sismiques. Des expériences ont été réalisées pour cicatriser des fissures par précipitation de calcite. Le but est l'étude du couplage entre l'augmentation de résistance mécanique de la roche fissurée et l'évolution de la perméabilité. Les échantillons fracturées ont été soumis à des conditions de pression et températures similaires de la croûte supérieure et à une percolation d'un fluide sursaturé en calcite pendant plusieurs mois. Ce couplage non-existe dans les premières étapes de la cicatrisation. Il est révélé par l'imagerie par tomographie aux rayons X que le scellement naissant des fractures se produit dans les porosités situées en aval de barrières d'écoulement, et donc dans des régions qui ne touchent pas les principales voies d'écoulement du fluide. Le découplage entre l'augmentation de résistance de la roche et la perméabilité suggère que les zones d'endommagement peu profondes dans les failles actives peuvent rester des conduits actifs pour les fluides plusieurs années après un séisme. / Earthquakes along large crustal scale faults are a huge hazard threatening large populations. The behavior of such faults is influenced by the fault damage zone that surrounds the fault core. Fracture damage in such fault damage zones influences each stage of the seismic cycle. The damage zone influences rupture mechanics, behaves as a fluid conduit to release pressurized fluids at depth or to give access to reactive fluids to alter the fault core, and facilitates strain during post- and interseismic periods. Also, it acts as an energy sink for earthquake energy. Here, laboratory experiments were performed to come to a better understanding of how this fracture damage is formed during coseismic transient loading, what this fracture damage can tell us about the earthquake rupture conditions along large faults, and how fracture damage is annihilated over time.First, coseismic damage generation, and specifically the formation of pulverized fault damage zone rock, is reviewed. The potential of these pulverized rocks as a coseismic marker for rupture mechanisms is discussed. Although these rocks are promising in that aspect, several open questions remain.One of these open questions is if the transient loading conditions needed for pulverization can be reduced by progressively damaging during many seismic events. The successive high strain rate loadings performed on quartz monzonites using a split Hopkinson pressure bar reveal that indeed the pulverization strain rate threshold is reduced by at least 50%.Another open question is why pulverized rocks are almost always observed in crystalline lithologies and not in more porous rock, even when crystalline and porous rocks are juxtaposed by a fault. To study this observation, high strain rate experiments were performed on porous Rothbach sandstone. The results show that pervasive pulverization below the grain scale, such as observed in crystalline rock, does not occur in the sandstone samples for the explored strain rate range (60-150 s-1). Damage is mainly occurs at a scale superior to that of the scale of the grains, with intragranular deformation occurring only in weaker regions where compaction bands are formed. The competition between inter- and intragranular damage during dynamic loading is explained with the geometric parameters of the rock in combination with two classic micromechanical models: the Hertzian contact model and the pore-emanated crack model. In conclusion, the observed microstructures can form in both quasi-static and dynamic loading regimes. Therefore caution is advised when interpreting the mechanism responsible for near-fault damage in sedimentary rock near the surface. Moreover, the results suggest that different responses of different lithologies to transient loading are responsible for sub-surface damage zone asymmetry.Finally, post-seismic annihilation of coseismic damage by calcite assisted fracture sealing has been studied in experiments, so that the coupling between strengthening and permeability of the fracture network could be studied. A sample-scale fracture network was introduced in quartz monzonite samples, followed exposure to upper crustal conditions and percolation of a fluid saturated with calcite for several months. A large recovery of up to 50% of the initial P-wave velocity drop has been observed after the sealing experiment. In contrast, the permeability remained more or less constant for the duration of the experiment. This lack of coupling between strengthening and permeability in the first stages of sealing is explained by X-ray computed micro tomography. Incipient sealing in the fracture spaces occurs downstream of flow barriers, thus in regions that do not affect the main fluid flow pathways. The decoupling of strength recovery and permeability suggests that shallow fault damage zones can remain fluid conduits for years after a seismic event, leading to significant transformations of the core and the damage zone of faults with time.

Spot core from the San Andreas Fault Observatory at Depth (SAFOD) borehole provides the opportunity to characterize and quantify damage and mineral alteration of siliciclastics within an active, large-displacement plate-boundary fault zone. Deformed arkosic, coarse-grained, pebbly sandstone, and fine-grained sandstone and siltstone retrieved from 2.55 km depth represent the western damaged zone of the San Andreas Fault, approximately 130 m west of the Southwest Deforming Zone (SDZ). The sandstone is cut by numerous subsidiary faults that display extensive evidence of repeating episodes of compaction, shear, dilation, and cementation. The subsidiary faults are grouped into three size classes: 1) small faults, 1 to 2 mm thick, that record an early stage of fault development, 2) intermediate-size faults, 2 to 3 mm thick, that show cataclastic grain size reduction and flow, extensive cementation, and alteration of host particles, and 3) large subsidiary faults that have cemented cataclastic zones up to 10 mm thick. The cataclasites contain fractured host-rock particles of quartz, oligoclase, and orthoclase, in addition to albite and laumontite produced by syn-deformation alteration reactions. Five structural units are distinguished in the subsidiary fault zones: fractured sandstones, brecciated sandstones, microbreccias, microbreccias within distinct shear zones, and principal slip surfaces. We have quantified the particle size distributions and the particle shape of the host rock mineral phases and the volume fraction of the alteration products for these representative structural units. Shape characteristics vary as a function of shear strain and grain size, with smooth, more circular particles evolving as a result of increasing shear strain. Overall, the particle sizes are consistent with a power law distribution over the particle size range investigated (0.3 µm < d < 400 µm). The exponent (fractal dimension, D) is found to increase with shear strain and volume fraction of laumontite. This overall increase in D and evolution of shape with increasing shear strain reflects a general transition from constrained comminution, active at low shear strains to abrasion processes that dominate at high shear strains.

Les matériaux soumis à de grandes déformations présentes pour la plupart l’apparition de déformations inélastiques. Ce phénomène est souvent accompagné d’une localisation des déformations dans une zone étroite, précurseur de la rupture. Un cas particulier, mais très fréquent, est les bandes de cisaillement qui apparaissent pour beaucoup de géomatériaux. Ces bandes peuvent être rencontrées à des échelles allant de l’échelle kilométrique pour les zones de subduction à l’échelle micrométrique à l’intérieur des zones de faille. Etudier et modéliser la création de ces zones d’instabilité est fondamental pour décrire la rupture des géomatériaux et des phénomènes associés comme les glissements sismiques dans les zones de faille mature de la lithosphère. Les conditions de pression, de température, l’interaction de l’eau interstitielle avec un matériau finement fracturé conduisent à l’apparition de multiples processus physiques impliqués dans les glissements sismiques. Dans ce travail, nous nous attachons à modéliser la création de bandes de cisaillement à l’intérieur des gouges de faille en prenant en compte l’effet de la microstructure par l’intermédiaire des milieux continus de Cosserat, ainsi que les couplages thermo-hydro-mécanique. L’utilisation de la théorie de Cosserat permet non seulement de régulariser le problème de localisation des déformations par l’introduction d’une longueur interne dans les lois constitutives, mais en même temps de prendre en compte l’effet de la microstructure. Deux approches sont employées pour étudier le système d’équations couplées aux dérivées partielles non linéaires : L’analyse de stabilité linéaire et la méthode des éléments finis. L’analyse de stabilité linéaire permet d’examiner les conditions d’apparitions d’instabilités pour un système mécanique avec des couplages multi-physiques. Par ailleurs, des considérations sur les perturbations appliquées au système permettent aussi de déterminer l’épaisseur de la zone de cisaillement, un paramètre clé pour la compréhension du mécanisme mécanique des failles. Ces estimations sont confirmées par l’intégration numérique pour des déformations restant dans une gamme donnée. Elles sont confrontées aux observations expérimentales et in situ et présentent une bonne corrélation. D’autre part, les simulations numériques permettent d’obtenir la réponse mécanique de la gouge de faille et de donner des informations sur l’influence des différents couplages dans le budget énergétique d’un tremblement de terre / When materials are subjected to large deformations, most of them experience inelastic deformations. It is often accompanied by a localization of these deformations into a narrow zone leading to failure. One particular case of strain localization is the formation of shear bands which are the most common patterns observed in geomaterials. In geological structures, they appear at very different scales, from kilometer scale for subduction zones, to micrometric scale inside fault cores. Studying their occurrence and evolution is of key importance to describe the failure of geomaterials and model seismic slip for mature crustal faults. The pressure and temperature conditions in these faults and the interaction with the pore water inside a highly fractured materials highlight the importance of different physical processes involved in the nucleation of earthquakes. In this thesis, we study the occurrence and evolution of shear bands inside fault gouges taking into account the material microstructure by resorting to elastoplastic Cosserat continua and also the effect of thermo-hydro mechanical couplings. The use of Cosserat theory introduces information about the gouge microstructure, namely the grain size, and permits to regularize the mathematical problem of in the post-localization regime by introducing an internal length into the constitutive equations. Two approaches are used to study the coupled non-linear partial differential set of equations: linear stability analysis and finite element simulations. Linear stability analysis allows to study the occurrence of localized deformation in a mechanical system with multi-physical couplings. Considerations on the dominant wave length of the perturbations permit also to determine the width of the localized zone. This shear band thickness is confirmed by numerical integration in the post-localization regime for a certain range of deformation. The obtained widths of the localized zone are key parameters for understanding fault behavior, are in agreement with experimental and field observations. Moreover, numerical finite element computations enable to model the mechanical response of a fault gouge during seismic slip and give insights into the influence of various physical couplings on the energy budget