Micro-seismology is attracting more and more attention in the exploration seismology community. The main goal in micro-seismic imaging is to find the source location and the ignition time in order to track the fracture expansion, which will help engineers monitor the reservoirs. Conventional imaging methods work fine in this field but there are many limitations such as manual picking, incorrect migration velocity and low signal to noise ratio (S/N). In traditional surface survey imaging, full waveform inversion (FWI) is widely used. The FWI method updates the velocity model by minimizing the misfit between the observed data and the predicted data. Using FWI to locate and image microseismic events allows for an automatic process (free of picking) that utilizes the full wavefield. Use the FWI technique, and overcomes the difficulties of manual pickings and incorrect velocity model for migration. However, the technique of waveform inversion of micro-seismic events faces its own problems. There is significant nonlinearity due to the unknown source location (space) and function (time). We have developed a source independent FWI of micro-seismic events to simultaneously invert for the source image, source function and velocity model. It is based on convolving reference traces with the observed and modeled data to mitigate the effect of an unknown source ignition time. The adjoint-state method is used to derive the gradient for the source image, source function and velocity updates. To examine the accuracy of the inverted source image and velocity model the extended image for source wavelet in z-axis is extracted. Also the angle gather is calculated to check the applicability of the migration velocity. By inverting for the source image, source wavelet and the velocity model simultaneously, the proposed method produces good estimates of the source location, ignition time and the background velocity in the synthetic experiments with both parts of the Marmousi and the SEG Overthrust model. On the other hand, a new imaging condition of natural Green’s function has been implemented to mitigate the effect of the unknown velocity model. It is based on putting receivers in a horizontal well close to the micro-seismic events so that only a small part of the velocity model is required for the imaging. In order to focus the multi scattering energy to the source location, as well as to suppress the influence of the noise in the data, we introduced a new method to compensate the energy in the receiver wavefield. It is based on reflection waveform inversion (RWI) theory. We simply migrate for the scatters (reflectors) in the medium, and set the image as a secondary source to compensate for the multi scattering energy in the receiver wavefield. By applying the same imaging condition, the energy of those scattering events can be traced to the source location. Thus the source point has higher energy in the source image. A simple two-layer medium test demonstrates the features.

Le cycle sismique résulte de la compétition entre des mécanismes de renforcement et d'endommagement. Le temps de récurrence entre les grands séismes fait qu'il est difficile d'observer des cycles complets. L'étude expérimentale des mécanismes de déformation et de nucléation des fractures a permis de mieux contraindre les processus à l'origine des séismes. Le rôle de la cicatrisation sur l'évolution de la résistance d'une faille soumise à une déformation stationnaire à été étudié expérimentalement par Weiss et al (2016). Dans cette expérience, une faille est créée dans une plaque de glace par cisaillement. Les mécanismes de cicatrisation sont obtenus par le regel de l'eau présente dans la zone de déformation. Dans le cadre de cette thèse, ce dispositif expérimental a été étendu pour permettre le suivi micro-sismique de la déformation imposée. Les mécanismes de déformation fragile émettent des ondes élastiques détectables qui se propagent dans le milieu, nous permettant de les caractériser. En raison de la géométrie en plaque du milieu, on observe la propagation d'ondes guidées similaire aux modes de Lamb symétrique et antisymétrique.Les fractures de grandes tailles se distribuent selon une loi de puissance en $10^{-bm}$ similaire à ce qui est observé en sismologie. Cependant, lors des expériences de déformation stationnaires, la valeur de $b$ est large ($b=3$), et bien supérieure à ce qui est observée dans la croûte terrestre ($b=1$). Une valeur de $b$ aussi élevée traduit le fait que la déformation est principalement accommodée de façon asismique ou part des fractures trop petites pour être détectées par notre méthode. Lorsque le rôle de la cicatrisation est renforcée par rapport à l'endommagement, on observe une diminution de la valeur de $b$. Ce changement de distribution est probablement dû à la diminution des hétérogénéités de structure dans la faille et à une augmentation de sa capacité à accumuler une contrainte plus élevée avant la rupture, permettant aux fractures de se propager sur de plus longues distances. Une partie importante de la sismicité correspond à des multiplets qui semblent être des produits passifs de la déformation. Ce comportement est similaire à ce qui est observé pour les essaims de séismes déclenchés par des transitoires de déformation : valeur de $b$ grande, absence de choc principal et peu de déclenchement de répliques. Pour des taux de déformation faibles, on observe une augmentation des chutes de couple avec la magnitude de la forme $Delta Gamma sim M_0 sim 10^{1.2m}$, similaire à ce qui est observé dans la croûte terrestre, $M_0 sim 10^{1.5m}$. Il est donc possible que la relation observée en sismologie s'étende aux petites magnitudes observées ici. Une diminution du couplage sismique est observé avec l'augmentation du taux de glissement $Omega$. Pour finir, pour une fracture de magnitude donnée, on observe une diminution de la chute de couple avec l'augmentation de $Omega$. Ce comportement peut être expliqué par la diminution du couplage sismique et/ou une dépendance du taux de cicatrisation. / The deformation observed along a seismic fault can be described as the succession of phases for which the fault accumulate stress imposed by the steady deformation of the surrounding regions, and phases of sudden sliding during which the stress is relaxed: the earthquakes. After the rupture, strengthening mechanisms are required to make possible the new accumulation of elastic stress. Therefore, the seismic cycle results in the steady competition between strengthening and damage. The aim of this study is to explore the role of cohesion-healing on the fault deformation dynamic, as well as to characterize the effect of slip rate on the seismicity. The experimental set-up designed by Weiss et al (2016) has been extended in this study to carry out a micro-seismic monitoring of the deformation. This experiment consists in the shear deformation of a fault created in a thin ice plate overlying a water column. Cohesion-healing mechanisms are achieved through freezing of the water along the fault. The damage mechanisms and the spatial and temporal distribution of the deformation can be characterized thanks to the detectable elastic waves emitted by the fracturing. Because of the plate geometry and underlying water column, we observed guided waves similar to the Lambs symmetric and antisymmetric modes.The largest fractures distribute according to a power law of the form $10^{-bm}$ that is similar to the one observed in seismology. At a constant sliding rate, we observe a large $b$ value, $simeq 3$, which is much larger than the value observed in the Earth's crust ($b=1$). This large $b$ value indicates that the deformation is mainly accommodated aseismically or by small, undetected, fractures. During Slide-Hold-Slide experiment that corresponds to a case for which the cohesion-healing is enhanced compared to the damage, we observe a decrease in the $b$-value likely due to a decrease in fault heterogeneity and an increase of the fault ability to store more elastic stress before the rupture, allowing the fractures to grow larger. An important part of the fractures are multiplets, swarms of fractures, which seem to be passive by-products of the imposed deformation. This behaviour is similar to the one observed for swarm seismicity triggered by slip transient: high $b$-value, no identified mainshock, and very little triggering. For small driving rate $Omega$, we observe an increase in torque drop amplitude with magnitude, $Delta Gamma sim M_0 sim 10^{1.2m}$, similar to the relation observed in seismology, $M_0 sim 10^{1.5m}$. Thus, the latter could be extended to small magnitudes observed in this study. A decrease of the seismic coupling is observed through the decrease in the number of fractures per unit of slip, and because in average a fracture behaves similarly at the different $Omega$ tested. Finally, for a given magnitude interval, we observe a decrease in torque drop amplitude with the increase in $Omega$. This could be explained by the observed decrease in seismic coupling or by a decrease in strengthening rate with $Omega$ that is not observed.

[pt] A presente tese tem como objetivo avaliar a interação trovão-solo no que toca a ocorrência de trovões e suas características microssísmicas, apontando a influência de vibrações induzidas para a redução do fator de segurança em uma análise (pseudo-estática) de estabilidade de encostas. Considerando a abordagem inédita, foi efetuado um levantamento teórico com o intuito de apresentar as principais características dos relâmpagos e suas correlações com o solo. Como material, foi utilizado um solo coluvionar, composto principalmente de quartzo, feldspato e biotita. O referido foi caracterizado através de métodos padrão (complementados com o uso da microtomografia 3D), e a retenção e disponibilidade de água foram reveladas. Foi efetuado um monitoramento de campo de longo prazo para avaliar a correlação entre os dados climáticos (incluindo incidência de raios) e o potencial hídrico dos solos. Por fim, foi utilizado uma estação para monitoramento sismográfico para captar as vibrações induzidas por trovões nos solos. Os resultados mostram 39 ocorrências de raios próximos ao local de monitoramento de campo. O monitoramento sísmico mostrou que os trovões causam sinais microssísmicos compostos por acelerações de pico do solo até 0,02 m/s ao quadrado. Em conclusão, para fins geotécnicos, o trovão é um assunto que pode ser avaliado à luz de um carregamento sísmico. / [en] The present thesis aims to evaluate thunder-soil interaction verifying the influence of its induced vibrations to the reduction of the factor of safety in a pseudo-static slope stability analysis. In order to carry out this research, considering the unpublished approach, a theoretical survey was made in order to present the main characteristics of the lightning and its correlations with the soil. As a material, a colluvial soil, mainly composed of quartz, feldspar and biotite was characterized by standard methods (supplemented with the use of 3D microtomography) in order to reveal its mineral composition, structural arrangement and water retention. After field and laboratory calibration of the water potential and volumetric moisture sensors, a long-term field monitoring was performed to evaluate the correlation between climatic data (including lightning incidence) and soil water potential. Finally, a seismographic monitoring station was used to capture the vibrations induced by thunder in the soils. The results depicted 39 lightning events near the field monitoring site. However, no rapid variation of water potential was revealed during thunderstorm days. Seismic monitoring showed that thunder caused micro-seismic signals composed of ground peak accelerations up to 0.02 m/s squared. In conclusion, for geotechnical purposes, thunder is a subject that can be evaluated in the light of pseudostatic loads. However, further researches are required to verify the vibrations of larger magnitudes, induced by rays that occur at smaller distances of the seismic monitoring point.

Cette thèse porte sur l’application de la technique de corrélation de bruit sismologique ambiant pour l'imagerie et le suivi des réservoirs géothermiques de Rittershoffen (ECOGI) et de Soultz-sous-Forêts (GEIE-EMC). La forte variabilité spatio-temporelle des sources du bruit de fond sismologique dans la gamme de période 0.2-7s limite la reconstruction des fonctions de Green. Cela induit des erreurs dans la construction des modèles de vitesse. Deux approches sont proposées pour s’affranchir des effets de la non-uniformité spatiale du bruit. Par ailleurs, la variabilité temporelle des sources de bruit est un facteur limitant pour le suivi du réservoir. On estime que les perturbations de vitesse doivent être de l’ordre de 0.1% à 1% pour pouvoir être détectées par les réseaux disponibles. Ce seuil n’a pas été franchi lors de la construction du site Rittershoffen mais une modification probable des propriétés diffractantes du milieu a été observée à la suite d’une stimulation. / This work focuses on the application of the ambient seismic noise correlation technique for the imaging and monitoring of deep geothermal reservoirs near Rittershoffen (ECOGI) and Soultz-sous-Forêts (GEIE-EMC). The strong spatial and temporal variability of the noise sources in the period range 0.2-7s limits the reconstruction of the Green’s functions. This results in significant errors in the velocity models. Two approaches are proposed to overcome the spatial non-uniformity of the noise and to improve the quality of the velocity models. Besides that, the temporal variability of the noise sources is a limiting factor for monitoring purposes. We estimate that the speed variations should be larger than 0.1% to 1% to be detected by the available networks. This threshold was not reached at Rittershoffen during the drillings or the stimulations. However, a probable change of the diffracting properties of the medium was observed following a hydraulic stimulation.