|
Disertaciones |
|
1
|
-
ÁTILA SARAIVA QUINTELA SOARES
-
U-Net and Curvelet based debluring LSRTM filters
-
Líder : REYNAM DA CRUZ PESTANA
-
MIEMBROS DE LA BANCA :
-
DIEGO FERNANDO BARRERA PACHECO
-
OSCAR FABIAN MOJICA LADINO
-
REYNAM DA CRUZ PESTANA
-
Data: 18-ene-2022
-
-
Resumen Espectáculo
-
The concept of deblurring filters is a novel approach to the least-squares migration, trying to achieve similar results but with only one iteration, without resorting to computationally expensive iterative methods. There have been several studies proposing different solutions to this problem, such as using Fourier-based transforms and neural networks. In the end, all of them try to achieve some sort of reparametrization and deblurring of the image by using those new parameters, to approximate the inverse of the Hessian. This technique is not new, being employed in digital cameras to filter the blur inherent to the equipment by using a similar concept. This dissertation studies the use of the U-Net neural network topology, applied to the curvelet transform domain and to patches in the spatial domain. The U-Net is a specific type of convolutional neural network (CNN) that has encoding and decoding blocks, which enhances its ability to recognize features at different scales. By applying it to the curvelet domain, which is separated by scale, angle, and location, it has the opportunity to better grasp different aspects of the same features. The idea is to train the network to match the pair of remigrated and migrated images, and then apply it to the migrated image. Our research shows that training the U-Net deep learning model in the curvelet domain can improve resolution at the deep regions of the migrated seismic section.% However, is very susceptible to noise, and seems to work on marine data better. This is one of the very few recent studies attempting to use neural networks in the curvelet transform domain, and a lot of research is still needed to fully grasp what are the possibilities of using such a technique. The filter based in the U-Net network was tested on two synthetic data sets (Marmousi and Sigsbee), presenting encouraging results, on top of showing that the network application after the training process produced a seismic image with better resolution compared to the conventionally migrated section.
|
|
2
|
-
LUCAS SOUZA DA FONSECA
-
Upwelling and the estuarine gravitational circulation: a feedback system in a tropical estuary in the South Atlantic
-
Líder : GUILHERME CAMARGO LESSA
-
MIEMBROS DE LA BANCA :
-
CARLOS AUGUSTO FRANÇA SCHETTINI
-
GUILHERME CAMARGO LESSA
-
MAURO CIRANO
-
Data: 21-mar-2022
-
-
Resumen Espectáculo
-
The present work examines the effects of coastal upwelling on the longitudinal density gradient within Todos os Santos Bay (BTS) and, therefore, its effect on gravitational circulation at the entrance of the estuary, as well as the effect of gravitational circulation on the coastal upwelling. A one-year data set of observed water temperature, mean circulation, and river input, as well as 2 years of numerical simulation are the basis for this investigation. The results show that the upwelling has a considerable effect on the regulation of the thermohaline field in front of the BTS, decreasing water temperature and increasing density, having suficient intensity to accelerate or establish the estuarine gravitational circulation. It was also observed a frequent anticipation of increasing flow shear intensification in relation to temperature drops, suggesting that the estuarine gravitational circulation acts as a facilitator to the upwelling process. Furthermore, based on the results obtained with the numerical simulations, it appears that the coastal upwelling events was also able to reestablish an estuarine density gradient in a period when the estuarine circulation was interrupted.
|
|
3
|
-
GLEIDSON DINIZ FERREIRA
-
Prediction of discontinuities using spatial and temporal filtering - application in 3D seismic data from Alto de Cabo Frio.
-
Líder : MILTON JOSE PORSANI
-
MIEMBROS DE LA BANCA :
-
ROGERIO DE ARAUJO SANTOS
-
MILTON JOSE PORSANI
-
SERGIO ADRIANO MOURA OLIVEIRA
-
Data: 05-ago-2022
-
-
Resumen Espectáculo
-
The detection of discontinuities has been highlighted in recent years due to the improvement in the quality of seismic data, thus contributing to the improvement in the interpretation of available data from oil and gas reservoirs. This work presents a methodology for detecting discontinuities in post-stack 3D seismic volumes that is divided into three steps: (i) spatial filtering is applied (inline x direction followed by crossline y direction) to attenuate horizontal and sub-horizontal events (dominantly associated with reflections in the layers), thus enhancing vertical and sub-vertical structures; (ii) then, the Hilbert filter is used, for each dimension of the volume, and the amplitude of the complex volume is obtained, and finally, (iii) a phase rotation is performed for each direction of the complex volume, placing highlighting the fractures and flaws of the original volume. We illustrate the application of the new methodology in a 2D synthetic seismic data and in the PSTM 3D seismic data of the Alto de Cabo Frio area located between the Santos and Campos Basins.The results obtained demonstrate the applicability of the new methodology in the mapping of structural discontinuities in 3D seismic volumes, thus being able to assist in the delimitation of hydrocarbon reservoirs.
|
|
4
|
-
NEI DAVI COSTA FIGUEIREDO
-
Migração, imageamento, RTM, inversão, conjugado gradiente
-
Líder : REYNAM DA CRUZ PESTANA
-
MIEMBROS DE LA BANCA :
-
EDVALDO SUZARTHE DE ARAUJO
-
MICHELANGELO GOMES DA SILVA
-
REYNAM DA CRUZ PESTANA
-
Data: 18-ago-2022
-
-
Resumen Espectáculo
-
The construction of seismic images through reverse time migration (RTM) on pre-stacked data and in the time domain requires the cross-correlation between the wavefield of the direct propagation of the source and the one coming from the reverse propagation of the receivers. In the present work, in addition to what is conventionally done, RTM will be executed by direct propagation of the source wavefield and reverse propagation of the field at the receivers performed simultaneously and at each extrapolation step, being transformed to the frequency domain. In this new domain, the imaging condition is applied, and the image is built without storing the direct propagation snapshots of the source.
Furthermore, such RTM results are input for least-squares RTM (LSRTM) in the image domain. There are several ways to do LSRTM and obtain an approximation of the inverse of the Hessian matrix to remove its effect on the final migrated image. This matrix is the defocusing operator responsible for the blurring effect in the migrated image from the conventional migration. In this paper, we seek to calculate non-stationary focusing filters on the image domain that approximate the Hessian and then remove it from the migrated image. Finally, it is shown here to be more appropriate, in the least-squares approach, to estimate the filter that approximates the Hessian and then assess the final image instead of the filter that matches the inverse of the Hessian from the application of this, obtaining the final image. The results obtained here were generated using the Marmousi and Pluto models, demonstrating that the approach of using matching filters in the image domain can produce results superior to those obtained with conventional RTM.
|
|
5
|
-
JORGE LUIS DOS SANTOS SANTANA
-
Multiscale Full Waveform Inversion Using Recurrent Neural Network
-
Líder : REYNAM DA CRUZ PESTANA
-
MIEMBROS DE LA BANCA :
-
ALEXSANDRO GUERRA CERQUEIRA
-
OSCAR FABIAN MOJICA LADINO
-
REYNAM DA CRUZ PESTANA
-
Data: 19-ago-2022
-
-
Resumen Espectáculo
-
Seen as a fundamental parameter for a reliable geological image of the subsurface, and consequent exploratory success, seismic velocity is one of the prerequisites in the seismic exploration chain. Such velocity relates directly to the construction quality of seismic images through robust algorithms such as reverse time migration (RTM) or other seismic imaging techniques. In this work, we use machine learning environments to obtain high resolution seismic velocity through the full waveform inversion (FWI) technique.
In summary, the FWI technique aims to compare data from real observations with calculated data obtained through seismic modeling from the solution of a specific wave equation. The residual of this comparison is minimized and the gradient is used to update, with an iterative optimization algorithm, the velocity model that at the end of the process will be able to correspond to the real data. Here we will use a Recurrent Neural Network (RNN), based on the governing physics (acoustic wave equation), to derive the real data and the calculated data regarding the direct seismic modeling step, since we will deal with purely synthetic data. In addition, learning environments, such as Pytorch, provide us with tools for calculating the gradient (automatic differentiation) and the mini-batch strategy important in terms of reduction memory and higher processing velocity.
As the FWI is based on the iterative minimization of a cost function between observed and calculated data, in order to avoid convergence to local minima, we use the multiscale approach of frequency. In addition, we tested the FWI response when subjected to noisy observation data and less accurate initial models, and compared it with the inversion response added to the multiscale approach of frequency technique, to demonstrate the performance of this approach when it comes to mitigating these limitations.
The results obtained in three sets of data demonstrate the efficiency and applicability of the technique used in the attempt to obtain high resolution seismic velocity fields.
|
|
6
|
-
VINICIUS ARIMATEIA RIBEIRO OLIVEIRA
-
Gravimetric and magnetic interpretation of the crustal structure between the Province of Borborema Meridional Oeste and the North of the San Francisco Cráton.
-
Líder : ALANNA COSTA DUTRA
-
MIEMBROS DE LA BANCA :
-
DAVID LOPES DE CASTRO
-
ALANNA COSTA DUTRA
-
ALEXANDRE BARRETO COSTA
-
Data: 01-sep-2022
-
-
Resumen Espectáculo
-
The Southern Borborema Province and the Northern São Francisco Craton experienced major tectonic events in the Neoproterozoic Era and their tectonic history is still not completely clarified due to insufficient geotectonic and geophysical studies. This work aims investigating the main boundaries of the crustal blocks between the Southern Borborema Province and the Northern São Francisco Craton for the reconstruction of the tectonic history of this area. Gravimetric, magnetic, geoid anomaly and elevation data were used to accomplish the objectives. The lateral extensions of the sources were determined through magnetic field transformations and the depths were estimated by integrating geoid anomaly and topography data as well as by An-Euler method. The obtained source depths of less than 1 km and more than 5 km with structural index 1 to 3 are present in the São Francisco Craton as in the Borborema cross section and west of the Sergipano belt. The obtained crustal thickness shows a crustal thinning in the Riacho do Pontal belt with 37.25 km in relation to the north of the São Francisco Craton, with values ranging from 41.6 km, these results are in accordance with seismic data. These results were used as a priori information, in the direct crustal 3D modeling to obtain the density contrast distribution in the subsurface. The lithosphere-asthenosphere interface (LAB) results revealed culminating values in the north of the craton reaching 216.0 km. In relation to the Pernambuco-Alagoas domains and the Sergipe belt, LAB values reach 203.0 to 207.0 km thick. The results of the interpreted geophysical maps reveal geological structures marked by lineaments with preferential direction NE-SW, consistent with the features originating from the Brasiliano orogenic event, as well as the contact boundary between the craton and the Borborema Meridional Province, and the boundary between the Pernambuco-Alagoas domains and the belt: Sergipano and Riacho do Pontal. The suture zone was identified from high magnetic signatures.
|
|
7
|
-
LUCAS SILVA BITENCOURT
-
Reverse time migration in TTI media using pseudo-acoustic wave equation for pure P-wave
-
Líder : REYNAM DA CRUZ PESTANA
-
MIEMBROS DE LA BANCA :
-
REYNAM DA CRUZ PESTANA
-
OSCAR FABIAN MOJICA LADINO
-
DIEGO FERNANDO BARRERA PACHECO
-
Data: 15-dic-2022
-
-
Resumen Espectáculo
-
Although it is possible to image important geological structures by assuming that the Earth's subsurface is a homogeneous and isotropic medium, there is inevitably a loss of information, especially in more complex geological media. Therefore, it is needed to include anisotropy in seismic imaging, particularly the most common in geophysics: the transverse isotropy. However, this also means a considerable increase in the computational cost of reverse time migration (RTM). Thus, this work proposes new pseudo-acoustic wave equations for pure P wave in tilted transversely isotropic (TTI) media, which can also be implemented with finite difference method, aiming to both reduce the computational cost and increase the quality of the RTM image in relation to other equations found in the literature. Accordingly, these new equations are tested on three synthetic data, and their efficacy is demonstrated by modeling and migrating TTI data found in the literature. Hence, it is shown that these new equations are efficacious for RTM and the new equation repositions the reflectors and makes them better defined, in addition to providing more details of some geological structures.
|
|
|
Tesis |
|
1
|
-
FERNANDA FIGUEIREDO FARIAS
-
Approaches for efficient imaging in linear and non-linear acoustic inversion.
-
Líder : REYNAM DA CRUZ PESTANA
-
MIEMBROS DE LA BANCA :
-
BRUNO PEREIRA DIAS
-
EDVALDO SUZARTHE DE ARAUJO
-
JESSE CARVALHO COSTA
-
JOERG DIETRICH WILHELM SCHLEICHER
-
REYNAM DA CRUZ PESTANA
-
Data: 05-abr-2022
-
-
Resumen Espectáculo
-
Attempts to obtain information from the Earth's subsurface usually require accurate numerical simulations of the wave equation, essential in the inversion processes which has as final goal the construction of the velocity model or the seismic imaging. In this thesis, topics are addressed that aim to make both the linear inversion, which uses the Born approximation as well as the nonlinear inversion, which uses the complete wave equation more efficiently or can produce better resolved images when compared to conventional methods. In the nonlinear inversion case, the question addressed here is based on its application to a reduced portion of the velocity model, a common situation where the geoscientist is interested in the variations of physical properties in the vicinity of the reservoir, and how these variations occur in subsequent surveys, such as in 4D seismic analyses, for example. Comparing two different strategies to calculate the wavefield locally, it was noted that calculating it in a numerically exact way did not bring significant improvements, when compared to inversion that neglects iterations of the local wavefield with the external model, being more sutable to stop at the simplest implementation of the finite differences injection method. On issues addressed to improve performance of the least-squares reverse time migration, two ways of improving the spatial resolution were addresed, the first focused on using the directional wavefield in conjunction with the commonly used pressure field. Since the adjoint of the Born modeling operator when properly handling this directional data, uses the energy of the receiver ghost during imaging, attenuating the noise commonly seen in data with low receiver coverage. This technique has been applied before for conventional reverse time migration and for the full waveform inversion, this thesis extends this approach to the least-squares reverse time migration. The second proposed approach to accelerate convergence also in the least-squares reverse time migration, consisted of applying a stochastic optimizer that takes into account an approximation of the Hessian, and was originally proposed for machine learning problems. In this case, the Devito framework was used, as it is capable of generating as efficient code involving the discretization of differential equations, including the use of GPUs. Being Devito written in Python language makes it convenient to use it along with techniques involving the quite popular machine learning.
|
|
2
|
-
DIEGO MENEZES NOVAIS
-
Spherical and Ellipsoidal Harmonic Representations and Bouguer Correction for Elipsoidal Cap Applied to the Analysis of Satellite Gravimetric Data of the State of Bahia.
-
Líder : EDSON EMANOEL STARTERI SAMPAIO
-
MIEMBROS DE LA BANCA :
-
LUIZ GALLISA GUIMARAES
-
ALANNA COSTA DUTRA
-
EDSON EMANOEL STARTERI SAMPAIO
-
JAILMA SANTOS DE SOUZA DE OLIVEIRA
-
MILTON JOSE PORSANI
-
Data: 11-abr-2022
-
-
Resumen Espectáculo
-
Harmonic modeling is used for gravimetric field representation of massive bodies, such as the Earth. The solution of Laplace's equation based on spherical coordinates, the spherical harmonic series, is widely used in bibliographies on physical geodesy. Taking into account the Earth's ellipsoidal shape, it would be more adequate to use the solution of the Laplace equation via ellipsoidal coordinates, reaching the ellipsoidal harmonic series. In the present study, the harmonic series - spherical and ellipsoidal - are applied to gravimetric satellite data over the state of Bahia. For the calculation of harmonic coefficients of spherical and ellipsoidal harmonic modeling, the ordinary and weighted least squares methods are used. For the weighting matrices of the weighted least square method, diagonal matrices with principal diagonal values are used based on the functions: related to the inverse of the residue minus the variance and the second related to the inverse of the geometric means of the residue minus the variance. Bar charts provide evaluative metrics for comparing the different models. Theoretical gravity at the level of the Earth's reference ellipsoid is used for gravimetric corrections and its formula can be conceived through the derivative of the gravitational potential added to the centrifugal potential of rotation of the Earth. The present work presents two theoretical gravity maps calculated through the spherical and ellipsoidal harmonics. After harmonic analysis, it will present the gravimetric corrections, mainly the Bouguer correction. Bouguer's simple correction, commonly used, is based on approximating the subsurface to an infinite horizontal plate of thickness equal to the elevation of the observation point, but this conception lacks realism to the terrestrial format. Based on the Bullard B correction for a spherical cap, the thesis proposes a new Bouguer correction for an ellipsoidal cap establishing an appropriate approximation for the terrestrial shape. In this way, the Bouguer corrections for the infinite plate, spherical cap and ellipsoidal cap are computed and arranged in maps for the state of Bahia. To support the geological interpretation, a gravimetric inversion was performed to estimate the depth of the Moho through the data processed with the ellipsoidal harmonics. The obtained results allowed to sketch three east-west geological profiles for the state of Bahia with interpretations of the geological blocks together with a model of the crust and the mantle.
|
|
3
|
-
PETERSON NOGUEIRA SANTOS
-
A wavefield domain dynamic approach and multi-parameter seismic inversion in attenuating media.
-
Líder : MILTON JOSE PORSANI
-
MIEMBROS DE LA BANCA :
-
ITALO AUGUSTO SOUZA DE ASSIS
-
AMIN BASSREI
-
GENARO FERNANDES DE CARVALHO COSTA
-
JESSE CARVALHO COSTA
-
MILTON JOSE PORSANI
-
Data: 03-jun-2022
-
-
Resumen Espectáculo
-
Modern seismic imaging techniques are computationally expensive, demanding a lot of memory space for data storage, in addition to having a high execution time. As a rule, methods involving wave propagation in the time domain, such as RTM, LSRTM, and FWI, require that the wavefields of forward propagation be stored or recalculated. Reconstruction methods that are based on storing temporally spaced wavefields or storing their boundaries have been widely used in an attempt to circumvent the problem of memory unavailability.
Understanding the computational domain of these wavefields becomes essential to be able to manipulate it efficiently and reduce the amount of data to be stored. In the first part of this thesis, we propose an approach capable of reducing memory consumption and consequently the execution time. By considering the sparsity of the wavefield in the initial propagation times, we build a dynamic approach that, from an empirical filter, maps the wavefront limiting it in a rectangular region, and thus we store only the data of this region. That is, the dynamic approach tries to approximate the computational domain to the propagation domain. With the dynamic approach, we obtained significant gains in terms of memory consumption and computational time when compared to the effective boundary reconstruction technique and conventional RTM. We apply the dynamic approach to realistic 2D and 3D synthetic data in order to show how well the technique works, and its robustness when applied to noisy data. In experiments on 2D Pluto model data, we achieved an average memory savings of 28\% per shot, and a 34\% decrease in total execution time. In the numerical examples with 3D data from the SEG/EAGE salt model, the dynamic approach, considering all shots, achieved a memory saving of 23.73\% which corresponds to 19TB. On the other hand, it was only 8 minutes faster than the conventional approach, given the inherent limitations of the 3D empirical filter.
The wave propagation in attenuating media has also been studied in the field of seismic imaging due to the degradation that the effects of attenuation can cause in the subsurface image, if these are not treated. Understanding the attenuating phenomena becomes of fundamental importance to be able to derive wave equations capable of effectively simulating them in the forward propagation and compensating them in the reverse propagation.
In the second part of this thesis, we analyze different viscoacoustic equations based on different mechanical models of attenuation. We derived and evaluated the forward and adjoint equations for each rheological model presented. From the seismograms and wavefields, we explored the main differences in the forward modeling viscoacoustic equations. As for the adjoint equations, we analyzed the result of the migrated images. Thus, we conclude that the forward and adjoint modeling equations based on the standard solid linear model (SLS) are the ones that best simulate the attenuating phenomena and, consequently, produce a better migrated seismic image.
Therefore, we propose a multi-parameter approach in the context of Born approximation to the SLS equation, perturbing the inverse of the bulk modulus and the magnitude of the quality factor. The least-square migration methods in visco-acoustic media (Q-LSRTM) have been widely used to improve the resolution of seismic images. However, these methods consider that only the parameters related to velocity are those that characterize the subsurface reflectivity, obtaining only single information from the physical media. Our multi-parameter approach allows us to jointly estimate reflectivity models related to velocity and quality factor, producing two model parameters about the geological medium. For this, we derive a multi-parameter Born modeling equation, the respective adjoint equation, and the respective gradient formulations. Thus, we built a multi-parameter Q-LSRTM (M-QLSRTM) scheme capable of producing high-resolution reflectivity images concerning the inverse of the bulk modulus and the magnitude of the quality factor.
|
|
4
|
-
JORGE LUIS ABRIL BENJUMEA
-
1,5D and 2D inversion of electromagnetic and electric data using global optimization methods and parallel processing.
-
Líder : MARCOS ALBERTO RODRIGUES VASCONCELOS
-
MIEMBROS DE LA BANCA :
-
RENATO RAMOS DA SILVA DANTAS
-
AMIN BASSREI
-
MARCOS ALBERTO RODRIGUES VASCONCELOS
-
MILTON JOSE PORSANI
-
WALTER EUGENIO DE MEDEIROS
-
Data: 01-jul-2022
-
-
Resumen Espectáculo
-
The use of optimization global techniques in geophysics to solve inverse problems, allowing to incorporate with easiness several kinds of constraints to create an stable mathematically solution. The 1D forward modeling of time domain electromagnetic data (TDEM) and 2D-DC resistivity is computationally inexpensive, allowing to use global optimization methods (GOMs) to solve 1.5D and 2D inverse problems, incorporating several stabilizers. Nonetheless, these meta-heuristic techniques are time-consuming, specially when a high number of parameters is involved. In this sense, the use of paralleling processing tools have become paramount to alleviate the computing time in inverse problems.
In this work was developed a newly particle swarm optimization technique with elitism and mutation (EMPSO) using a hybrid parallel MPI/OpenMP version named (PEMPSO). This algorithm, also supported with the use of Total Variation (TV) and Global Smoothness (GS) constraints allowed studying the resistivity distribution in subsurface, with both synthetic and real data. The inversion outcomes reflect good results for both constraints, having better sensitivity with the Total Variation (TV) constraint in detecting high and low resistivity contrasts. The inversion times, also demonstrated a speedup of 17x in the assessed models in relation with the sequential processing times.
Apart from this 2D electrical resistivity method, we also presented an inversion approach for 1,5D TDEM data, that similarly with the 2D case aforementioned, presented good outcomes in the inverted models using different global optimization techniques such as Whale Optimization Algorithm (WOA), Particle Swarm Optimization (PSO) and Grey Wolf Optimizer (GWO). These algorithms allowed mapping and localizing high and low resistivity contrasts for both synthetic and real data. Similarly with 2D case, an approach of parallel processing using pure MPI allowed to speedup the computational time that involve the use of these meta-heuristic algorithms. The major outcomes of 1,5D electromagnetic case show that: (i) both constraints offer good results to map the resistivity distribution, (ii) the WOA and PSO algorithms offer better computational performance, converging first than the GWO, (iii) WOA presented the better performance in the cost function value attained than PSO and GWO and (iv) a pure MPI parallelized version provided a 17x speedup in the time processing for synthetic models and up to 50x in the time computing of real data studied.
|
|
5
|
-
Wilker Eduardo Souza
-
Processamento de dados sísmicos terrestres 3D: novas abordagens e soluções 3D land seismic data processing: new approaches and solutions.
-
Líder : MILTON JOSE PORSANI
-
MIEMBROS DE LA BANCA :
-
MICHELANGELO GOMES DA SILVA
-
MILTON JOSE PORSANI
-
RAFAEL RODRIGUES MANENTI
-
SERGIO ADRIANO MOURA OLIVEIRA
-
SERGIO LUCIANO MOURA FREIRE
-
Data: 18-nov-2022
-
-
Resumen Espectáculo
-
3D seismic processing, First break picking, Dynamic time warping, Static correction, Velocity analysis.
|
|