This work describes a method for solving large, positive-defined, block-structured, dense linear systems with multiple right-hand sides that uses high-performance parallel computing. The system solution is obtained through a generalized Levinson recursion that uses the linear combination of smaller forward and backward solutions associated with lower order subsystems. The new implementation is described for parallel computing and is based on a partitioned matrix algorithm. The algorithm was separated into two subroutines, the first that computes the backward solution and the error energy matrix for smaller orders, and the second that computes the solution recursively. The algorithm was implemented for three types of systems: shared memory systems, distributed memory systems, and GPU systems. In each case, the lowest order systems were calculated using appropriate libraries. In the first, the OpenBLAS or MKL library was used; in the second, SCALAPACK; and finally, for systems with GPUs, we implemented an OUT-OF-CORE algorithm, in which the lowest order systems were calculated using MAGMA. In all three cases, the final solution is compared with the complete system solution using LAPACK, SCALAPACK, and MAGMA, respectively. In all three cases, the first part of the algorithm proved to be more computationally expensive compared to the Cholesky decomposition. However, the second part that computes the solution proved to be more efficient than the successive solution of two triangular systems when the right side of the system has a significant size, generally a few times the value of N. The error in the estimated model does not present significant variations compared to the reference solution. Finally, we present the use of the algorithm in frequency-domain seismic wave modeling, which involves the solution of large, sparse linear systems. These results show a disadvantage of the algorithm in sparse non-Toeplitz systems, as it increases the computational cost and memory consumption.

The present work analyzes and maps the occurrence of Internal Waves (OI) o the Amazon Continental Shelf (PCA). Using a 21-year database (2000 to 2020) of satellite images captured by the MODIS (Moderate Resolution Imaging Spectroradiometer) sensor onboard the AQUA and TERRA satellites. The study region had already been identified as an important site for OI in the equatorial Atlantic. This work deepens the knowledge about this oceanographic phenomenon in the region. Using images belonging to the visible spectrum and containing sunglint it was possible to identify and measure the OI. Collecting important information such as crest length (distance measured along the crest), number of waves per packet, distances betweenwaves (from the same pack or from different packs) and also making calculations to infer therelationship between the tidal harmonic components and the generation and propagation ofthese waves. The interaction between tidal waves and the relief of the ocean floor is the main source of OI in the study area. This project compiled 21 years of satellite images, resulting in 1483 mapped OI that were divided into 226 images. Places with higher formation rates, average size of InternalWaves, and areas with higher propagation density, are questions that can be clarified by the work developed here.

Data from satellite and in situ observations show significant variations in the pattern and intensity of sea surface salinity (SSS) from 2010 and 2012 in the western boreal tropical Atlantic Ocean. This SSS variability is associated with extreme hydrological precipitation events in the Amazon River basin that caused an abnormal reduction and increase in Amazon River discharge in 2010 and 2012, respectively. The objective of this work was to investigate the SSS seasonal variation and the physical mechanisms that rules the salinity balance of the mixed layer (MLS) in the western boreal tropical Atlantic Ocean between 2010 and 2012. Thus, SSS data from the SMOS satellite and 1/12◦ resolution HYCOM model results were used. The use of the model also allowed evaluating the impact of the tides on the MLS balance from two numerical simulations in a nested grid system, with and without the effect of the tides. The Amazon River plume was about 1.2 times less saline and covered a greater area during the boreal spring and summer of 2010 compared to 2012, even with the expressive increase in discharge from the Amazon River during the boreal spring of 2012 compared to 2010. This MLS variability is associated with the rise in the precipitation in the region and strengthening transport of less saline waters in a northwesterly direction during spring and boreal summer. The variability of these forcings is associated with the influence of the Atlantic meridional mode. The MLS balance of the model showed that surface freshwater flux, horizontal advection and vertical entrainment dominate the region’s MLS seasonal cycle, showing the same order of magnitude. In the region close to the mouth of the Amazon and Pará rivers, the results of the model show that the increase in the Amazon river discharge in 2012 compared to 2010 caused a greater contribution of the surface flow of fresh water, which implied a greater penetration of salinity in the region of the base of the mixed layer, which favors the increase of MLS. The MLS balance of the model showed strong compensation between the atmospheric component and the vertical oceanic processes, particularly in the North Equatorial countercurrent flow region and in the central area of the boreal tropical Atlantic. The impact of tidal oscillations on the SCM balance of the western boreal tropical Atlantic Ocean region occurs mainly in the area close to the mouth of the Amazon and Pará rivers, especially in the northwest portion of the mouth of the Amazon river up to approximately Cassiporé Cape. The tidal forced model shows an increase in MLS throughout the entire seasonal cycle (about 1.2), as well as a decrease in the zonal advection contribution to the MLS balance in the region, leading to a reduction in the northwestern component of the horizontal salt advection term.

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.

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.

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. 

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.

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.

3D seismic processing, First break picking, Dynamic time warping, Static correction, Velocity analysis.

This study applies the spectral and fractal analyses in climate series and the cluster and discriminant analyses in geophysical well log data. In the first case, we intended to understand the climatic dynamics of the southern region of Mozambique by evaluating the periodicity, persistence, tendency and long-term variability of the time series of precipitation and extreme temperatures related to 1960-2018. The results of the analyzes are related to the current global climate observations and projections contained in the fifth assessment report of the Intergovernmental Panel on Climate Change (IPCC-AR5), regarding the Austral Africa which include Mozambique. In particular, the signal of climate change in Mozambique is visible, and the trends converge to those reported in the IPCC-AR5. In the second case, the main objective was to determine the lithological profile and fluid contacts in reservoirs. Well log data from five wells over the Rovuma sedimentary Basin - Mozambique were used. For the discrimination, a reference well was chosen for training, and the obtained functions from it were then applied to the remaining wells. The classification process comprehended three main phases. The two methods were applied in parallel and the results are compared in each classification phase. In general, both methods converged to the same lithological model and fluidtypes in reservoirs. Gas has been indicated as the most predominant hydrocarbon in the basin.

Data assimilation methods are fundamental pieces of operational oceanography and have strong impact on ocean models previsibility, since they produce model initial condition. The present work focus on the development of these methods and operational oceanography in Brazil in the context of the Oceanographic Modeling and Observation Network (REMO). On the first part of the work, emphasis is given to different strategies to assimilate vertical profiles of temperature (T) and salinity (S). On the second part, emphasis is given on the construction and application of an Ensemble Kalman Filter (EnKF). Hydrographic TS profiles are crucial for constraining thermohaline structure in ocean models. For the Hybrid Coordinate Ocean Model (HYCOM) these data can be assimilated on its original coordinate (Z level) or can be projected into the model vertical coordinate (isopycnal). These two strategies were compared assimilating TS data from ARGO profiles into HYCOM every three days for six years. Three experiments were realized: (i) ARGO_DP with innovation calculation on model vertical space, isopycnal coordinate, (ii) $ARGO_Z with innovation calculation in observational space, Z coordinate and (iii) ARGO_ZT same as last, but assimilating only T. Both schemes are suitable for assimilating hydrographic profiles, as they reduced model TS error with respect to ARGO data. However, ARGO_Z showed a more accurate analysis, smaller hindcast error, better representation of ocean heat content and model isopycnals depths. Consequently, achieved a better representation of the mean sea surface height (SSH), western boundary currents (WBCs) and the Brazil Current (BC) transport closer to references. It was also showed the importance of assimilating salinity in order to produce greater and more accurate increments. A new version of the Remo Ocean Data Assimilation System (RODAS) capable of realizing a joint assimilation of all available observations with the EnKF was developed. This was named RODAS_EnKF and compared to the older version with the Ensemble Optimal interpolation (EnOI), named RODAS_EnOI. RODAS_EnKF employs 11 members and perturbations on the wind forcing and observations. It was jointly assimilated hydrographic profiles, sea surface temperature (SST) and absolute dynamic topography (ADT) every ten days for seven months for both experiments. A good horizontal and vertical representation of the spread was observed and the error mean of each member was greater than the error considering the mean of all members, not occurring filter divergence. RODAS_EnKF presented the best SST and salinity, and RODAS_EnOI the best SSH and temperature. For the latter, RODAS_EnOI presented a tendency of cooling and RODAS_EnKF of warming. Both produced positive impacts on the eddy kinetic energy and on the WBCs, with RODAS_EnKF presenting slightly better results. Therefore, RODAS_EnKF was successfully implemented as no filter divergence was observed and the mean error considering each member was always greater than the error considering the mean of all members. However, it is necessary to increase ensemble size and implement a hybrid assimilation scheme, combining covariance matrix from each scheme, in order to improve its impacts.

The current interest in exploring unconventional reservoirs in the Recôncavo Basin, due to their high potential in shale gas, seeks to overcome obstacles due to the low signal-noise ratio of seismic data. For this, an optimal seismic processing should be performed for shale gas reservoirs, which is the purpose of this work. To optimize data resolution, some filtering methods are introduced during the seismic processing steps. Among the filtering techniques presented is the Singular Spectrum Analysis which, based on SVD decomposition, aims to attenuate the presence of low frequency noise, known as ground roll, and the deconvolution method, which aims to show high frequencies, resulting in higher resolution especially in the shallower region of the data. After seismic processing we seek to characterize regions containing possible shale gas reservoirs. For this you can use information from wells by performing seismic-well correlations. However, in some regions, the identification of shale gas related seismic patterns is unclear, so the use of seismic attributes can be used to reduce uncertainties and assumptions, allowing a better interpretation of the regions of interest. This work will use Seisspace, Seismic Unix processing software, Fortran programming language, and DUG Insight software for the seismic-well correlation step.

Brazil has 7 documented impact structures in its territory up to date. The Santa Marta impact structure is located in the southwestern state of Piauí, in Corrente County. It is a complex impact structure, about 10 km in diameter with a central uplift 3,5 km-wide. It has an estimated age between 93 and 100 Ma and was formed in a complex geological context, where the Mesozoic rocks of the Sanfranciscana Basin are overlying the sequences of the Paleozoic Basin of Parnaíba. In this sense, understanding its stratigraphy and its formation process is a big challenge. This thesis aims to characterize the units of the Santa Marta using magnetotelluric data. Broadband magnetotellurics data were collected along two profiles, which were analyzed using 1D and 3D inversions. The results show that the western anular basin is characterized by Sanfranciscan Basin rocks (_ > 50 :m) and breccias (_ > 800:m) resulting from the impact event. Under this layer, slightly smaller resistivity units are observed in the range of 100 - 800 :m probably related to the basement of the Parnaíba Basin. Its central region is characterized by the discontinuity of the basement and a peak of Bouguer anomaly. These characteristics are possibly related to faults resulting from the central uplift formation, and may indicate the location and diameter of the central elevation, estimated at about 4 km. Finally, the eastern annular basin is characterized by low resistivity values in the range of 5 to 100 :m, which are identified as rocks of the Parnaíba Basin and by the presence of a conductor body of unknown origin, but possibly not related to the impact event. The results demonstrate the great complexity of the Santa Marta structure due to the proposed mapping, and the distribution of anomalies shows that, the geological units are not distributed as expected, although it has a circular shape. On the contrary, the pattern of the anomalies shows that Santa Marta is an asymmetric structure, which is roughly related to the impact angle.

The present work proposes an interpretation of the magnetotelluric data acquired in the region of Ilha de Itaparica, Bahia, Brazil, the limit region between the Recôncavo and Camamu Basins. In a phase prior to the processing of this data, this research proposes its filtering based on the premises of the singular spectral analysis and directly on its time series. This filtering was carried out in two complementary stages: decomposition of the initial time series and reconstruction of the original decomposed series. A two-dimensional synthetic model (2-D) was generated to test the efficiency of this filter, which proved to be satisfactory both in synthetic applications and in real data. From the filtering, processing and 2-D inversion of the apparent resistivity and phase data in two orthogonal directions (TE and TM modes), it was possible to perform a qualitative analysis of the inverted sections allowing us to identify the structural and sedimentological distribution of the studied region. In addition to the identification of the Barra Fault, which marks the boundary between the two mentioned basins, the final section interpreted suggests a structural regime characterized by a fault system that is believed to be linked to the transfer zone reflection with polarity inversion between the direction dip of the Salvador and Barra Fault.

A comprehensive observational data set was used to examine the shoaling and breaking characteristics of shoreward propagating nonlinear internal tides on the edge of the Australian North West Shelf (NWS) during the late winter of 2013. The internal wave events supported the grouping of field observations into two distinct categories: (1) prior-; and, (2) post- wave breaking. It was found that the transition from (1) to (2) was marked by the rise of nonlinear steepening () and reduction in dispersion (), both coefficients that parameterize nonlinear wave effects on the Korteweg-de Vries (KdV) equation. We developed a criterion for wave breaking from the dimensionless parameter that relates these two terms: where wave breaking occurs when In the first group, the rear of the internal tide was initially composed of a series of internal solitary waves (ISWs) of depression, these high-frequency waveforms were apparently merged into the rear face of a longer wave, which steepened and formed a shoaling bore while decreases. In the second, as dropped below the unit, the wave broke and formed multiple turbulent boluses at the rear. The kinematics of the wave field, taken together with a number of empirical parameters adapted from numerical solutions for ISWs of depression, suggested that these turbulent features have resulted from collapsing-plunging breakers and that the boluses have developed trapped cores. Therefore, explaining strong events of diapycnal mixing and onshore transport of water and material observed during late winter. Nothing similar has ever been described during summer months for the region, thus opposing preconceived suggestions that the nonlinear internal wave field is of secondary importance on the NWS in winter. Additionally, we have proposed the adjustment of the breaking location criterion to long internal tides, which produced a prognostic that was confirmed empirically through glider’s profiles.

Reverse time migration (RTM) is based on the solution of the full wave-equation and can be defined as the modeling adjoint operator. The image is generated from a cross-correlation between the forward-propagated source wavefield and the backward-propagated receiver wavefield. RTM has the power to image complicated structures without dip limita-tion. Is commonly used for complex imaging areas, like subsalt, because of the significant contrast of velocity and steep dips. Artefatos de migração e iluminação irregular motivaram o surgimento do conceito de migração por mínimos quadrados (Least Squares Migration - LSM ). O objetivo é aprimorar ainda mais o resultado do modelo de refletividade Migration artifacts and irregular illumination motivated the start of the concept of Least squares migration (LSM).The method was proposed to further improve the reflectivity model by minimizing the residual between predicted data and observed data via an iterative lin-earized inversion. But the LSM application is expensive, for each iteration is needed one forward modeling and two migrations. We propose a method of preconditioning in data and image domain that works as a least squares solving the problem in one step. First, was used a Wiener filter with a stable division algorithm via Taylor expansion that solves the problem with division by zero. The main goal is to approximate the calculated data with the observed one. Using this result, we applied an iterative filter in F-k domain that correlates the raw migrated image and the reflectivity, to achieve an image that corresponds to least-squares migration image. The combination of these two methods was tested using Marmousi velocity model, with complex structures and high velocity contrast, to see if the combination of these two pre-conditioning could achieve an image with less artifacts and better continuity. The iterative F-k filter can be also used as a tecnique to improve the signal/noise ratio using only the raw migration image. Both methods were tested in Marmousi velocity model. The filter was also tested using Pluto model, that also has complex salt structures.

This work presents evidence of regional climatic changes in the vicinities of Todos os Santos Bay (NE Brazil) and investigates the impact of these changes upon the salinity field in the bay. A large data set of meteorological and oceanographic data was used, which included more than 50 years of rainfall, river discharge and air temperature data,7 years of hydrodynamic numerical simulations of the bay and adjacent shelf, and 3 year of observations of the salinity field. Temporal trends were identified and quantified with the aid of Mann-Kendall e Sen's Slope, and the results show a clear aridification trend (less humid and hotter) in the last 5-6 decades. The discharge of the three main rivers debouching into the bay, Paraguaçu, Jaguaripe and Subaé rivers, were depleted by 62% , 72% and 24% , respectively, in relation to their climatological (1988-2017) means. The total annual rainfall was reduced by 24% in Salvador, 29% in Cruz das Almas and 50% in Feira de Santana, again in relation to their climatological means. Less river discharge and rainfall caused an overall decrease of the mean water influx to the bay of 100 m 3 /s, which correspond to 57 % of the climatological mean. The mean air temperature increased by 0.75 the mean maximum air temperature in Salvador increased by 1.25 ◦C . ◦C regionally, but The reduction of freshwater inflow to the bay caused an increase of the mean surface salinity of 0.6 g/kg for the whole bay of 1.0 g/kg in the most internal sections. Hypersalinity events appear to have come around in the early 90's. The mean seasonal salinity amplitude is 1.5 g/kg at the bay entrance and 4.0 g/kg in the NW section of the bay. The salinity field throughout the bay is sensitive to the summer discharges of the Paraguaçu river. The results suggest that the regional aridification trend is weakening the gravitational circulation and compromising the exchange capabilities of the bay, thus increasing both the residence and flushing times.

The variability of the Atlantic Meridional Overturning Circulation (AMOC) and the Meridional Freshwater Transport (Mov) at 24 N, 11 S and 30 S, simulated with the GFDL-CM3 model under the historical (1860-2005) and RCP 4.5 and 8.5 scenarios (2006-2100) are examined. The results are compared with the climatology and observations in the search for salinity biases that would compromise the accuracy of the state of the future climate predicted by this model. The AMOC index shows a predominant decadal variability as well as at 24 N. On the other hand, the seasonal signal is more prominent in the South Atlantic. The magnitude of the meridional freshwater transport varies considerably among the latitudes, with Mov increasing at 24 N, rather stable 11 S and decreasing at 30 S. Such pattern is followed by a weakening of the AMOC regardless of the latitude until the end of the twenty-first century, indicating that a reduction in Mov in the southern South Atlantic and increase in the North Atlantic are potential drivers of the AMOC destabilization. However, the model and observations disagree on the stability of the AMOC with the first featuring a rather monostable regime and the second a stable bistable regime. scenario the weakening by 2100 is 47% Under the RCP 4.5' of the individual model s historical mean, whereas under the RCP 8.5 scenario over the same period, the weakening is 53% of the historical mean. The stream function plots revealed not only the AMOC weakening but also an overall shoaling of the circulation suggesting a freshening of the Antarctic Bottom Water (AABW). Salinity biases were found mostly in the South Atlantic in the upper 1000 m of the water column, with the model results significantly deviating from the observations and climatology in the mid-Atlantic. Possible causes for the Mov results, AMOC weakening and variability, including potential drivers for salinity biases are discussed.

Total Organic Carbon (TOC) content is a measure of the amount of carbon available in an organic compound and is an essential parameter in the evaluation of potentially hydrocarbon source rocks. The use of well logs in geochemical evaluation is an important technique, not just for its usefulness as a quick scan for identification of such rocks, but also by their ability to identify the TOC content of these rocks. Geophysical log data can be used to identify source rocks formations in the early stages of well drilling. Consequently, the records used for source rock assessments and TOC calculation commonly includes density, sonic, gamma rays, neutron and resistivity, through several widely spread techniques. In this work, a suite of well logs along with laboratory-measured TOC data of core samples from 10 boreholes in Santos Basin, were used. to obtain TOC profiles continuously over the entire profiled range so that it is possible to identify potentially generating rocks where there is no measured TOC data. Given the abundance (45% of the data available) of TOC data measured in sandy formations (sandstones and calcarenites), it was possible to make predictions in these regions as well, with purely scientific interest, since the TOC content is not a parameter of industry interest in conventional reservoirs (sandstones and calcarenites, for example). Three algorithms were applied and compared, among which include the classic multiple linear regression (MLR), in addition to the Support Vector Machine (SVM) and Random Forest (RF). Finally, the mean of the results given by the three algorithms was used as a meta-regressor. Although multiple linear regression is an often underrated algorithm compared to others more complex, it returned interesting results in this research, with the smallest errors in clayey rocks and the second smallest error in sandy rocks (after SVM). However, a closer assessment shows that the MLR displays extrapolated values in an especific formation in oppose to SVM and RF.

Nowadays unconventional reservoirs have become very attractive for the industry, among those reservoirs there is shale gas. The objective of this work is, through the study of seismic attributes in a processed cube, to obtain a region of interest, the Sweet Spot. For this, a seismic data processing was performed composed of a simple processing flowchart and an analysis of the average frequency, TOC and TecVA attributes. The TOC analysis was obtained through a linear relationship between acoustic impedance and TOC, calculated using the textit Total Least Square method to minimize the error in both dimensions. With the study it was possible to identify a region of interest of exploration, however with great reservations to the quality of the results obtained compared to what is reported in the literature.

The Recôncavo Basin consists of an aulacogen developed from the opening of the Gondwana Supercontinent, a phenomenon that culminated in the formation of the South Atlantic Ocean. Located in Bahia, the basin was the first to produce hydrocarbons in national territory. The work consisted in the processing of a seismic line located in the northeast region of the Recôncavo Basin, in the tying of two wells and in the seismostratigraphic interpretation under the perspective of sequence stratigraphy applied to rift basins. Being a noisy seismic data, several filtering methods were used to generate the stacked section, which was later migrated by Kirchhoff post-stack. Among the main processing steps, we highlight SVD method in F-X domain to attenuate the direct wave, single-trace SVD for ground-roll attenuation and multichannel predictive deconvolution, which contributed significantly to a greater continuity and individualization of the reflectors in the seismic section. The seismostratigraphic interpretation was based on the model of Prosser (1993) – an adaptation of sequence stratigraphy for rift basins - which considers that tectonics, not eustasy, is the main factor controlling the relationship between accommodation rate and sedimentation. That is, in rift basins tectonics controls the movement of the shoreline and stacking patterns. Thus, rift sequence is divided into tectonic systems tract of rift initiation, rift climax and rift final, each associated with a distinct stage of rifting. The data from two wells close to the line, which were tied to seismic section and whose gamma ray profiles were the basis for identifying the stratigraphic surfaces separating the tectonic systems tracts, contributed to the seismostratigraphic interpretation.

Seismic tomography has been used in reservoir geophysics as an important method to image

in high resolution interest region. In this context, diffraction tomography stands out, which

uses the waveform rather than the traveltime. The classic Born approximation, used in

monochromatic diffraction tomography in the condition of weak scattering, presents as a li-

mitation the need for prior knowledge of the background velocity. In this thesis, tomographic

inversion approaches are proposed, with matrix formalism using the Born approximation.

Such approaches are iterative, where the background velocity field is updated in each itera-

tion, however knowledge of the true background is not necessary. In addition to the iterative

process, one can choose to use a single frequency or several frequencies at each inversion. In

the first approach, a single frequency is used, which is kept constant. In the second, several

frequencies are used simultaneously in each iteration which are also maintained constant.

In the third, in addition to the background velocity, the working frequency is also updated.

Finally, in the last approach, the multiple frequencies used simultaneously are updated ite-

ration by iteration. The role of the frequency in the different approaches was summarized

in a comparative way for each model used, evidencing the approach that presented smaller

RMS errors of the estimated model when compared to the true model. The inverse gene-

ralized was implemented through singular value decomposition, and the poor conditioning

of the tomographic matrix was attenuated using regularization by derivative matrices. On

the other hand, regularization demands the search for an optimal parameter, which was ob-

tained through generalized cross-validation and Reginska techniques. The Barbieri method

was also used for a qualitative appraisal of the inversions, where one can identify the regions

of the model where the inversion was less satisfactory. The proposed approaches were tested

in two synthetic models: the first containing a dipping layer and a paleochannel, with well

to well acquisition geometry, and the second presenting greater complexity, with two salt

bodies between different superimposed layers with reflection geometry, the data being con-

taminated with noise in both cases. The proposed iterative process generated results with

lower model parameter, velocity, and data errors. In addition, a Born approximation using

finite difference modeling was performed in test models and later this technique was used in an integrated inversion using diffraction tomography followed by a full waveform inversion,

resulting in a satisfactory recovery of the true model without prior knowledge of background

velocity.