Prestack depth migration requires an accurate kinematic velocity model to image the subsurface correctly. Wave equation migration velocity analysis techniques aim to update the background velocity model by minimizing image residuals to achieve the correct model. The most commonly used technique is differential semblance optimization (DSO), which depends on applying an image extension and penalizing the energy in the non-physical extension. However, studies show that the conventional DSO gradient is contaminated with artifact noise and unwanted oscillations which might lead to local minima. To deal with this issue and improve the stability of DSO, recent studies proposed to use an inversion formula rather than migration to obtain the image. Migration is defined as the adjoint of Born modeling. Since the inversion is complicated and expensive, a pseudo inverse is used instead. A pseudo inverse formula has been developed recently for the horizontal space shift extended Born. This formula preserves the true amplitude and reduces the artifact noise even when an incorrect velocity is used. Although the theory for such an inverse is well developed, it has only been derived and tested on laterally homogeneous models. This is because the formula contains a derivative of the image with respect to a vertical extension evaluated at zero offset. Implementing the vertical extension is computationally expensive, which means this derivative needs to be computed without applying the additional extension. For laterally invariant models, the inverse is simplified and this derivative is eliminated. I implement the full asymptotic inverse to the extended Born to account for laterally heterogeneity. I compute the derivative of the image with respect to a vertical extension without performing any additional shift. This is accomplished by applying the derivative to the imaging condition and utilizing the chain rule. The fact that this derivative is evaluated at zero offset vertical extension, makes it possible to compute the derivative without applying the extension. I also verify the newly proposed inversion formula on a laterally variant velocity model. In addition, I test the effect of the computed derivative and compare its contribution with the full formula. This additional term has overall limited influence on conventional images. Its largest impact is on vertical reflectors such as salt flanks, granted the velocity is varying laterally in the background as often is in this case. Otherwise, for most applications, we can obtain good quality extended images without this additional term.

The presence of gas hydrate and free gas within marine sediments, deposited along the South Shetland margin, offshore the Antarctic Peninsula, was confirmed by low and high resolution geophysical data, acquired during three research cruises. Seismic data analysis has revealed the presence of a bottom simulating reflector that is very strong and continuous in the eastern part of the margin. This area can be considered as a useful site to study the seismic characteristics of sediments containing gas hydrate, with a particular focus on the estimation of gas hydrate and free gas amounts in the pore space. Pre-stack depth migration and tomographic inversion were performed to produce a regional velocity field of gas-phase bearing sediments and to obtain information about the average thickness of gas hydrate and free gas layers. Using these data and theoretical models, the gas hydrate and free gas concentrations can be estimated. Moreover, the common image gather semblance analysis revealed the presence of detailed features, such as layers with small thickness characterised by low velocity alternating with high velocity layers, below and above the bottom simulating reflector. These layers are associated with free gas trapped within the hydrate stability zone and deeper sediments. Thus, the use of the detailed and the regional velocity field analysis is important to give a more reliable estimate of gas content in the marine sediments.

L'imagerie sismique de la Terre permet de retrouver la structure de sous-sol. Cette opération est classiquement décomposée en deux étapes : une première phase a pour objectif de déterminer la cinématique de propagation des ondes (modèle de vitesse de référence) ; la seconde phase vise à retrouver la position des interfaces dans le sous-sol. Si cette seconde phase est maintenant classique, la détermination du modèle de vitesse de référence reste un sujet d’actualité en imagerie sismique. Une technique pour estimer le modèle de référence sépare les données en sous-ensembles, par exemple en points de tir. Une image partielle du sous-sol est obtenue pour chaque point de tir. Si ces images sont cohérentes les unes avec les autres, alors le modèle de vitesse qui a servi à obtenir ces images est dit correct. Cette technique s'est avérée très utile, en particulier pour l'imagerie des zones complexes, car elle ne requière pas a priori d'identifier des événements. Cependant, la technique est intrinsèquement liée aux données réfléchies. Elle ne prend pas en compte les arrivées transmises, les réflexions multiples, les ondes de surface, … L'objectif de la thèse de doctorat est de prendre en compte à la fois les arrivées réfléchies et les arrivées transmises (ondes directes et plongeantes) pour l'analyse de vitesse. Pour cela, le formalisme de l'analyse de vitesse doit être revu pour prendre en compte des développements récents (modèle « étendu » défini en 2008). Si une telle approche s’avère possible, alors il ne serait plus nécessaire d'extraire les ondes réfléchies des données. Par ailleurs, les ondes réfléchies ont des trajets essentiellement verticaux. Les ondes transmises ont des trajets plutôt horizontaux. La prise en compte des deux types d'ondes permettrait donc de mieux contraindre le modèle de vitesse et alors de déterminer des paramètres d'anisotropie. Au cours de la thèse, il s'agira de modifier le formalisme actuel d'analyse de vitesse pour prendre en compte les ondes réfléchies et transmises. Des applications seront développées sur les données synthétiques et réelles pour démontrer l'apport de l'approche combinée. Enfin, une analyse fine devra être menée pour voir si les ondes réfractées (en opposition aux ondes plongeantes) peuvent également être prises en compte. / Seismic imaging aims at retrieving the Earth's structures. It is classically split into two steps: firstly, the objective is to retrieve the background velocity model containing the large-scale structures of the velocity model. Secondly, one needs to determine the reflectivity part with the positions of the interfaces. The first part still remains a difficult challenge. Migration Velocity Analysis consists of migrating subsets of the total input data set (e.g. single shots). If all associated images are consistent, then the model used for imaging is said to be correct. This is a very useful method, in particular for the imaging of complex structures, as it does not a prioiri require any picking nor event identification. However, it is intrinsically restricted to reflected data: it does not take into account transmitted waves, multiple reflections, surface waves, ... The objective of the Ph.D work is to simultaneously consider reflected and transmitted (direct and diving) waves in the context of Migration Velocity Analysis. For that, one needs to reconsider the basis for velocity analysis. Recent developments around the « extended model » should be considered. If feasible, then it would not be anymore needed to extract reflected energy from a shot gather. More importantly, the combined inversion would better constrain the model: reflected waves are mainly associated to vertical propagation, whereas transmitted waves recorded at large offsets are more associated to horizontal propagation. This could offer the possibility to better estimate anisotropy parameters.

In this thesis, several aspects of GPR data processing for RC bridge decks are studied. First, autofocusing techniques are proposed to replace the previous expensive and unreliable human visual inspections during the iterative migration process for the estimation of the velocity/dielectric permittivity distribution from GPR data. Second, F-K filtering with dip relaxation is proposed for interference removal that is important for both imaging and the performance of post-processing techniques including autofocusing techniques and CS-based migration studied in this thesis. The targeted interferes here are direct waves and cross rebar reflections. The introduced dip relaxation is for accommodating surface roughness and medium inhomogeneity. Third, the newly developed CS-based migration is modified and evaluated on GPR data from RC bridge decks. A more accurate model by accounting for impulse waveform distortion that leads to less modeling errors is proposed. The impact of the selection of the regularization parameter on the comparative amplitude reservation and the imaging performance is also investigated, and an approach to preserve the comparative amplitude information while still maintaining a clear image is proposed. Moreover, the potential of initially sampling the time-spatial data with uniform sampling rates lower than that required by traditional migration methods is evaluated.

[pt] Este trabalho tem por objetivo investigar um método para auxiliar na quantificação de características sísmicas do subsolo. O modelo sísmico bidimensional de reflexão usa a equação Normal Move Out (NMO), para calcular os tempos de trânsito das ondas sísmicas, tipo P, refletidas em camadas isotrópicas e inclinadas. Essa equação usa a velocidade raiz quadrática média RMS como valor representativo das velocidades intervalares das camadas unidas. No processo de inversão para múlltiplas camadas, as velocidades RMS representam o problema principal para estimar as velocidades intervalares. Conseqüentemente, o método proposto estima sequencialmente os parâmetros do modelo sísmico, para resolver esse problema a partir dos tempos de trânsito com Algoritmos Genéticos Híbridos (algoritmo genético e algoritmo Nelder Mead Simplex). Os tempos de trânsito são sintéticos e a estimação de parâmetros é tratada como um problema de minimização. Com o método proposto foi obtido um alto grau de exatidão, além de reduzir o tempo de computação em 98,4 % em comparação com um método de estimação simultânea de parâmetros. Para aliviar a complexidade e a demora na geração de um modelo em três dimensões se contrói um modelo sísmico em três dimensões formado com modelos bidimensionais, sob cada unidade retangular da malha de receptores do levantamento sísmico, para camadas isotrópicas curvadas, com variações suaves das pendentes e sem descontinuidades. Os modelos bidimensionais formam polígonos que representam as superfícies de interface que são projetadas sob os retângulos da malha. Dois conjuntos de superfícies poligonais são gerados para auxiliar na localização das camadas. / [en] The objective of the present work is to investigate a method to help in the quantification of seismic characteristics underground. The two-dimensional seismic model of reflection employs the equation Normal Move Out (NMO) to calculate the travel times of P waves reflected on inclined and isotropic layers. This equation uses the root mean square velocity as a representative value of the joined layers velocities. At the inversion process, for multiple layers, the root mean square velocities are the main problem to estimate the layer velocities. Consequently, to solve that problem, the proposed method estimates sequentially the parameters of the seismic model using travel times and the Hybrid Genetic Algorithms (Genetic algorithm and the Nelder Mead Simplex algorithm). The travel times are synthetic and the estimation of parameters is treated as a minimization problem. With proposed method was obtained high grade of accurate, and the reduction of 98.4 % of computing time when it was compared to a simultaneous parameters estimation method. For decreasing the complexity and the delay to generate the models in three dimensions is proposed the construction of a three-dimensional seismic model formed with two-dimensional models, under every rectangular cell of the mesh of receptors of the seismic survey, for curved isotropic layers with soft variations in the gradient and without discontinuities. The two-dimensional models form polygons that represent the surfaces of interfaces that are designed under the rectangles of the surface or soil. Two sets of polygonal surfaces are generated to help at the geometric localization of layers.

[pt] Este trabalho quantifica o erro que se comete durante o processamento sísmico quando uma abordagem isotrópica é utilizada na análise de velocidade e correção de sobretempo normal (NMO) em dados anisotrópicos. Esta quantificação é realizada através da repetição do experimento de análise de velocidade em dados sísmicos sintéticos construídos a partir de um modelo geológico simples (porém realista) representando uma seqüência sedimentar típica de águas profundas com camadas horizontais onde uma delas é anisotrópica (VTI). A variação da anisotropia é conhecida e a diferença entre as velocidades NMO encontradas para cada modelo é quantificada. O resultado também é quantificado através da diferença na profundidade de uma camada resultante da conversão tempo- profundidade a partir da velocidade obtida no procedimento e mostra que, dependendo da anisotropia, o erro na velocidade (e/ou profundidade) por não se considerar anisotropia pode chegar a 10-15 porcento para valores de anisotropia que são encontrados habitualmente na natureza. A análise desta quantificação também mostra que o parâmetro de anisotropia delta é muito mais influente sobre o erro da velocidade que o parâmetro e e que a espessura da camada utilizada no experimento. Para complementar o trabalho, é verificada a melhor abordagem anisotrópica de correção de sobretempo normal para dados reais a partir de equações existentes na literatura, tanto para a correção de NMO quanto para a estimativa de parâmetros de anisotropia. A utilização de equações que incluem anisotropia permitem uma melhor correção do sobretempo normal e uma boa estimativa dos parâmetros de anisotropia épsilon e delta. Os parâmetros estimados foram utilizados em um algoritmo de migração para meio anisotrópico pré-empilhamento em profundidade e foi verificado que quando existe uma boa estimativa de delta, a profundidade migrada do refletor sofre uma distorção, desprezível em comparação à sua posição original. Em conjunto com este estudo, também é proposta uma técnica alternativa para análise de velocidade utilizando uma função de discriminantes lineares chamada gradiente descendente. Esta metodologia permite adaptar várias equações de NMO e obter todos os parâmetros da equação de uma única vez (t0 ,VNMO e eta), diferentemente do método de coerência geralmente utilizado para realizar análise de velocidade e que só permite a obtenção de dois parâmetros ao mesmo tempo (t0 e VNMO), tornando necessária uma segunda análise de velocidade. O método do gradiente descendente proposto neste trabalho foi testado para quatro funções de sobretempo normal com dados sintéticos e um dado real e foi considerado rápido, robusto e eficiente. / [en] This work quantifies the error created during seismic processing when isotropic approach is used to normal moveout correction and velocity analysis and the seismic data is anisotropic. This quantification is made by performing velocity analysis in several synthetic seismograms built from a simple (but realistic) geological model with some horizontal layers, one of them being anisotropic (VTI), representing a common deep water sedimentary sequence. The anisotropy in the model is known and the difference between the NMO velocities found by seismic processing (semblance analysis) is quantified for each model. The result is also shown through the difference in depth obtained from time-depth conversion with NMO velocity. Depending on the anisotropic degree, the velocity (and/or depth) error produced when the anisotropy is not considered can be up to 10-15 percent for anisotropic values commonly found in nature. This quantification analysis also concludes that delta parameter influences more on velocity error than e parameter and layer depth. To complement this work, the best anisotropic approach for normal moveout correction to be used in real data is investigated, by comparing some NMO functions found in technical papers, analyzing the data correction and the parameter estimation. The usage of NMO equation with anisotropic approach in anisotropic seismic data allows a better normal moveout correction and the anisotropic parameters ( épsilon and delta) can be estimated from velocity analysis. The estimated anisotropic parameters were applied in a pre-stack anisotropic depth migration algorithm and it was verified that when delta is well estimated, the migrated position of a seismic reflector is not very distorted from its real position. It is also proposed in this Thesis one alternate technique for velocity analysis using one linear discriminant function called gradient descendent. This methodology allows adopting several normal moveout functions and obtaining all the equation parameters (t0 , VNMO and eta) at once, differently from semblance method used in conventional velocity analysis that only allows obtaining two parameters at the same time (t0 and VNMO ), which requires a second velocity analysis to obtain all equation parameters. The proposed gradient descendent method was tested with four NMO equations and it was shown to be fast, robust and efficient.

No description available.

Les expériences de sismique active sont couramment utilisées pour estimer la valeur d'un modèle de vitesse de propagation desondes P dans le sous-sol. Les méthodes dites d'« analyse de vitesse par migration » ont pour but la détermination d'un macro-modèle de vitesse, lisse, et responsable de la cinématique de propagation des ondes. Dans une première étape de « migration », une image de réflectivité est obtenue à partir des données enregistrées en utilisant une première estimation du macro-modèle. Cette image dépend d’un paramètre additionnel permettant dans un second temps d’estimer la qualité du macro-modèle puis de l'améliorer. Les images de réflectivité obtenues par les techniques de migration classiques sont cependant contaminées par des artefacts, altérant la qualité de la remise à jour du macro-modèle. En particulier, elles ne prennent pas en compte les réflexions multiples, habituellement retirées des données avant traitement. Cette étape reste cependant délicate et on se prive alors de l'information supplémentaire contenue dans les multiples.Nous proposons dans cette étude une stratégie d’optimisation imbriquée en itérant l'étape de migration avant de remettre à jour le macro-modèle. La migration itérative produit des images de réflectivité satisfaisantes pour l'analyse de vitesse et s’étend naturellement aux réflexions multiples. Un désavantage de la méthode est son coût de calcul. Un pseudo-inverse de l'opérateur de modélisation est alors utilisé comme préconditionneur pour limiter le nombre d’itérations dans la boucle interne. Une autre difficulté est l'instabilité de la remise à jour du modèle de vitesse calculée pour des modèles de réflectivité successifs proches les uns des autres. Une nouvelle approche plus robustesse est proposée, valide aussi dans le cas de multiples. Son efficacité est testée sur des jeux de données synthétiques 2D. / Active seismic experiments are commonly used to recover a model of the P-wave propagation velocity in the subsurface. “Migration Velocity Analysis” techniques aim at deriving a smooth background velocity model controlling the kinematics of wave propagation. First, a reflectivity image is obtained by “migration” of observed data using a first estimate of the background velocity. This image depends on an additional “subsurface-offset” parameter allowing to assess the quality of the background velocity model with a focusing criterion and to correct it. However classical migration techniques do not provide a sufficiently accurate reflectivity image, leading to inconsistent velocity updates. In particular they do not take into account multiple reflections, usually regarded as noise and removed from the data before processing. Multiple removal is however a difficult step, and additional information contained in multiples is discarded.In this thesis, we propose to determine the reflectivity model by iterative migration before subsequent velocity analysis, leading to a nested optimisation procedure. Iterative migration yields accurate reflectivity image and extends naturally to the case of multiples. One of its disadvantages is the associated increased computational cost. To limit the number of iterations in the innerloop, a preconditioner based on a pseudo-inverse of the modelling operator is introduced. Another difficulty is the instability of the velocity update obtained with very close successive reflectivity models. We propose a modified approach, valid in the presence of multiples, and discussed through applications on 2D synthetic data sets.

[pt] Esta dissertação investiga o uso de Algoritmos Genéticos aplicados em dados sísmicos com o objetivo de obter parâmetros físicos e atributos sísmicos que auxiliem na caracterização das rochas de um subsolo terrestre. Os dados sísmicos têm sido extensamente empregados no setor de exploração de petróleo. As aplicações envolvendo sísmica não se restringem na busca por novas reservas de petróleo, mas também são usadas para projetar novos poços e melhorar a produção dos reservatórios de petróleo. O levantamento de dados sísmicos permite analisar extensas áreas da subsuperfície com custo praticável em relação a outras técnicas. Entretanto, a interpretação desses dados com o objetivo de obter informações relevantes e acuradas não é uma tarefa simples. Para isto, várias técnicas de inversão sísmica vêm sendo desenvolvidas. Este trabalho consistiu em avaliar uma alternativa que emprega Algoritmos Genéticos para inverter parâmetros a partir de dados sísmicos. Existem 3 etapas principais neste trabalho. Primeiramente, foram estudados o tema da exploração sísmica e a técnica de Algoritmos Genéticos. Na segunda etapa foi definido um modelo, usando Algoritmos Genéticos, que busca, neste caso, minimizar uma medida de erro, para obtenção dos parâmetros objetivos. Finalmente, foi implementado um sistema a partir do modelo proposto e realizados os estudos de casos com dados sísmicos sintéticos para avaliar o seu desempenho. O modelo baseado em Algoritmos Genéticos foi avaliado submetendo-se seus resultados a um especialista e comparando-os com os da busca aleatória. Os resultados obtidos se mostraram consistentemente satisfatórios e sempre superiores aos da busca exaustiva. / [en] This dissertation investigates the use of Genetic Algorithms applied to seismic data with the objective of obtaining physical parameters and seismic attributes that would facilitate the characterization of rocks in terrestrial subsoil. The seismic data has been extensively utilized in the field of petroleum exploration. The applications involving seismic are not restrained to the search for new petroleum reserves, but are also used to project new wells and to improve the production of existing petroleum reservoirs. The survey of seismic data allows the analysis of extended areas of the subsurface at an affordable price relative to other techniques. However, the interpretation of the data with the objective of obtaining relevant and accurate information is not an easy task. For that, several seismic inversion techniques are being developed. This work consists in evaluating an alternative that uses Genetic Algorithms to invert parameters from seismic data. There are 3 main stages in this work. Initially, the theme of seismic exploration and the technique of Genetic Algorithms have been studied. On the second stage a model has been defined, using Genetic Algorithms, which aims, in this case, to minimize an error measurement, obtaining objective parameters. Finally, a system from the proposed model has been implanted and the study of cases with synthetic seismic data has been executed to evaluate its performance. The process of optimizing has been compared to the process of random search and the results obtained by the model have always been superior.

Les expériences sismiques actives sont largement utilisées pour caractériser la structure de la subsurface. Les méthodes dites d’analyse de vitesse par migration ont pour but la détermination d’un macro-modèle de vitesse, lisse, et contrôlant la cinématique de propagation des ondes. Le modèle est estimé par des critères de cohérence d’image ou de focalisation d’image. Les images de réflectivité obtenues par les techniques de migration classiques sont cependant contaminées par des artefacts, altérant la qualité de la remise à jour du macro-modèle. Des résultats récents proposent de coupler l’inversion asymptotique, qui donne des images beaucoup plus propres en pratique, avec l’analyse de vitesse pour la version offset en profondeur. Cette approche cependant demande des capacités de calcul et de mémoire importantes et ne peut actuellement être étendue en 3D.Dans ce travail, je propose de développer le couplage entre l’analyse de vitesse et la migration plus conventionnelle par point de tir. La nouvelle approche permet de prendre en compte des modèles de vitesse complexes, comme par exemple en présence d’anomalies de vitesses plus lentes ou de réflectivités discontinues. C’est une alternative avantageuse en termes d’implémentation et de coût numérique par rapport à la version profondeur. Je propose aussi d’étendre l’analyse de vitesse par inversion au domaine des données pour les cas par point de tir. J’établis un lien entre les méthodes formulées dans les domaines données et images. Les méthodologies sont développées et analysées sur des données synthétiques 2D. / Active seismic experiments are widely used to characterize the structure of the subsurface. Migration Velocity Analysis techniques aim at recovering the background velocity model controlling the kinematics of wave propagation. The first step consists of obtaining the reflectivity images by migrating observed data in a given macro velocity model. The estimated model is then updated, assessing the quality of the background velocity model through the image coherency or focusing criteria. Classical migration techniques, however, do not provide a sufficiently accurate reflectivity image, leading to incorrect velocity updates. Recent investigations propose to couple the asymptotic inversion, which can remove migration artifacts in practice, to velocity analysis in the subsurface-offset domain for better robustness. This approach requires large memory and cannot be currently extended to 3D. In this thesis, I propose to transpose the strategy to the more conventional common-shot migration based velocity analysis. I analyze how the approach can deal with complex models, in particular with the presence of low velocity anomaly zones or discontinuous reflectivities. Additionally, it requires less memory than its counterpart in the subsurface-offset domain. I also propose to extend Inversion Velocity Analysis to the data-domain, leading to a more linearized inverse problem than classic waveform inversion. I establish formal links between data-fitting principle and image coherency criteria by comparing the new approach to other reflection-based waveform inversion techniques. The methodologies are developed and analyzed on 2D synthetic data sets.