Produção Científica
Artigo em Revista
A constrained version for the stereology inverse problem: Honoring power law and persistences of the fracture traces exposed on arbitrary surfaces We present a stereological study in a cave setting that is part of a karstic carbonate system located in the SÃ£o Francisco Craton, Brazil. Using a Lagrangian approach, a constrained version of the nonlinear inverse problem of stereology is solved. Besides the classical demand of fitting the histogram of fracture traces measured on arbitrary exposed surface, it is imposed that the solution honors also measures of surface intensity (p21) and power law exponent obtained from fracture traces on the same exposed surfaces. Estimates of volumetric intensities (p32) of conjugate fracture pairs might be also imposed to be close values. The resulting cost functional is minimized using the Particle Swarm Optimization (PSO) method. The implemented version of PSO furnishes the best solution and a set of suboptimal quasisolutions, from which the solution uncertainty is evaluated. A key aspect of the implemented approach is that all terms composing the cost functional are normalized, obtaining as a result, robustness for the weighting parameters to changes in the input data. The resulting stochastically simulated discrete fracture networks honor all statistical observations but, in general, do not reproduce the positions of the observed fracture traces. The methodology is applied to synthetic and field data examples. All obtained solutions are stable and geologically reliable. 

Artigo em Revista
Well log analysis for lithology and fluid contacts in Rovuma Basin â€“ Mozambique: Application of cluster and discriminant analyses This study applies the cluster and discriminant analyses in geophysical well log data from the Rovuma sedimentary Basin  Mozambique. The main objective was to determine the lithological profile and fluid contacts in reservoirs. Well log data from five wells drilled on the same basin 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, namely, the separation of shale/nonshale layers along the entire logged section, separation of water/hydrocarbon within reservoirs and the separation of oil/gas within hydrocarbon bearing zones. The two methods, cluster analysis and discriminant analysis, were applied in parallel and the results are compared in each classification phase. The quality of reservoirs was also assessed by applying cutoffs in relation to shale content and effective porosity, delineating net reservoirs. 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. 

Artigo em Revista
Comparison between oil spill images and lookalikes: an evaluation of SARderived observations of the 2019 oil spill incident along Brazilian waters Three SARderived observations of dark surface patches along the Northeastern Brazilian coastline by the end of 2019 were misreported in the Brazilian media as oil spillrelated. Unfortunately, these observations were misled by false positives or lookalikes. Therefore, this paper aims to technically evaluate these lookalike classes by analyzing image attributes found to be helpful to the identification of ocean targets, including oil spills, rain cells, biofilms, and low wind conditions. We use image augmentation to extend our dataset size and create the probability density function curves. The processing includes image segmentation, optimal attribute extraction, and classification with random forest classifiers. Our results contrast with the opensource oil spill detection system and patch classifier methodology called â€œRIOSS.â€ Analysis of the feature probability density functions based on optimal attributes is promising since we could capture most of the false positive targets in the three SARreported images in 2019. The only exception was the biofilm slick observed on October 28th, where the RIOSS mistakenly classified this organic patch as a low wind region with oil spots. This pitfall is acceptable at this project stage since we had only five biogenic film samples to train the algorithm. 

Artigo em Revista
Lowrank seismic data reconstruction and denoising by CUR matrix decompositions. Lowrank reconstruction methods assume that noiseless and complete seismic data can be represented as lowrank matrices or tensors. Therefore, denoising and recovery of missing traces require a reducedrank approximation of the data matrix/tensor. To calculate such approximation, we explore the CUR matrix decompositions, which use actual columns and rows of the data matrix, instead of the costly singular vectors derived from singular value decomposition. By allowing oversampling columns and rows, CUR decompositions obviate the need for the exact rank. We evaluate three different procedures for randomly selecting columns and rows to obtain the CUR. Once the lowrank approximation is estimated, data reconstruction is achieved by an iterative optimization scheme. To demonstrate the effectiveness of CUR matrix decompositions for multidimensional seismic data recovery, we present examples of 3D and 4D synthetic and field data. Results derived by CUR compare well to conventional eigenimagefamily methods. 

Artigo em Revista
Probabilistic Estimation of Seismically Thinâ€‘Layer Thicknesses with Application to Evaporite Formations The identifcation of potassium (K) and magnesium (Mg) salts prior to the well drilling is a key factor to avoid washouts, closing pipes, fuid loss damage, and borehole collapse. The Bayesian classifcation combines the outcomes from statistical rock physics and seismic inversion, providing the spatial occurrence of the mostprobable salt types. It serves as a facies identifer of Mgâ€“Krich salts (bittern salts) before drilling. Nevertheless, the mostprobable classifcation is limited to the seismic resolution which may underestimate seismically thinlayer thicknesses. Along with the mostprobable facies, the Bayesian classifcation renders the facies probability volume. We demonstrate that the facies probability and faciesspecifc total thickness highly correlate to each other even under the threshold of seismic resolution. Thus, we employ the bitternsalts probability volume to predict thinbed bitternsalts thickness in undrilled locations. To capture the variability of the seismic estimation, we resort to Monte Carloassisted simulations of wells that emulate the layering patterns of a sitespecifc deposition environment. These simulations are crucial to assist the estimation of the joint probability density function between the facies volume and the total thickness. Therefore, given the facies probability, the joint probability density function enables us to derive the conditional expectation and percentiles of thinbed thicknesses. Furthermore, this paper proposes a method to quantify the negative infuence of seismic noise in the estimation of thinbed thicknesses. The blind well confrms the consistency of this technique to unfold the uncertainty in the seismic predictability of thin layers. We argue that this procedure is extendable to other facies. 

Artigo em Revista
Timemigration velocity estimation using FrÃ©chet derivatives based on nonlinear kinematic migration/demigration solvers Advanced seismic imaging and inversion are dependent on a velocity model that is sufficiently accurate to render reliable and meaningful results. For that reason, methods for extracting such velocity models from seismic data are always in high demand and are topics of active investigation. Velocity models can be obtained from both the time and depth domains. Relying on the former, time migration is an inexpensive, quick and robust process. In spite of its limitations, especially in the case of complex geologies, time migration can, in many instances (e.g. simple to moderate geological structures), produce image results compatible to the those required for the project at hand. An accurate timevelocity model can be of great use in the construction of an initial depthvelocity model, from which a highquality depth image can be produced. Based on available explicit and analytical expressions that relate the kinematic attributes (namely, traveltimes and local slopes) of local events in the recording (demigration) and migrated domains, we revisit tomographic methodologies for velocitymodel building, with a specific focus on the time domain, and on those that makes use of local slopes, as well as traveltimes, as key attributes for imaging. We also adopt the strategy of estimating local inclinations in the timemigrated domain (where we have less noise and better focus) and use demigration to estimate those inclinations in the recording domain. On the theoretical side, the main contributions of this work are twofold: 1) we base the velocity model estimation on kinematic migration/demigration techniques that are nonlinear (and therefore more accurate than simplistic linear approaches) and 2) the corresponding FrÃ©chet derivatives take into account that the velocity model is laterally heterogeneous. In addition to providing the comprehensive mathematical algorithms involved, three proofofconcept numerical examples are demonstrated, which confirm the potential of our methodology. 

Artigo em Revista
EXTENDING THE USAGE OF GRAPHICS PROCESSING UNITS ON THE CLOUD FOR COST SAVINGS ON SEISMIC DATA REGULARIZATION The usage of graphics processing units is already known as an alternative to traditional multicore CPU processing, offering faster performance in the order of dozens of times in parallel tasks. Another new computing paradigm is cloud computing usage as a replacement to traditional inhouse clusters, enabling seemingly unlimited computation power, no maintenance costs, and cuttingedge technology, dynamically on user demand. Previously those two tools were used to accelerate the estimation of Common Reflection Surface (CRS) traveltime parameters, both in zerooffset and finiteoffset domain, delivering very satisfactory results with large time savings from GPU devices alongside cost savings on the cloud. This work extends those results by using GPUs on the cloud to accelerate the Offset Continuation Trajectory (OCT) traveltime parameter estimation. The results have shown that the time and cost savings from GPU devicesâ€™ usage are even larger than those seen in the CRS results, being up to fifty times faster and sixty times cheaper. This analysis reaffirms that it is possible to save both time and money when using GPU devices on the cloud and concludes that the larger the data sets are and the more computationally intensive the traveltime operators are, we can see larger improvements. 

Artigo em Revista
A statisticsbased descriptor for automatic classification of scatterers in seismic sections Discontin uities and small structures induce diffractions on seismic or groundpenetrating radar (GPR) acquisitions. Therefore, diffraction images can be used as a tool to access valuableinformation concerning subsurface scattering features, such as pinch outs, fractures, and edges. Usually, diffractionimaging methods operate on diffraction events previously detected. Patternrecognition methods are efficient to detect, image, and characterize diffractions. The use of this kind of approach, though, requires a numerical description of image points on a seismic section or radargram. We have investigated a new descriptor for seismic/GPR data that distinguishes diffractions from reflections. The descriptor consists of a set of statistical measures from diffraction operators sensitive to kinematic and dynamic aspects of an event. We develop experiments in which the proposed descriptor was incorporated into a patternrecognition routine for diffraction imaging. The obtained method is useful for performing the automatic classification of image points using supervised and unsupervised algorithms, as a complementary step to Kirchhoff imaging. We also develop a new type of filtering, designed to address anomalies on the diffraction operators caused by interfering events. We evaluate the method using synthetic seismic data and real GPR data. Our results indicate that the descriptor correctly discriminates diffractions and shows promising results for low signaltonoiseratio situations. 

Artigo em Revista
Numerical modeling of flexural stresses in the upper crust of NE Brazil â€“ Implications for stress regimes in intraplate settings We model the contribution of the stresses generated by the topography and density heterogeneities in the stress field of NE Brazil and the interplay between these and regional stresses. Our model consists of a horizontally layered elastic slab (the lithosphere) that floats above an inviscid fluid (the asthenosphere), where the surface load is generated by the topography and bathymetry, and the internal loads are calculated from the Bouguer anomalies. The deviatoric flexural stresses are calculated in the Fourier domain. Our results showed local flexural stresses with magnitudes comparable to those of the platewide stresses, and mainly controlled by the longwavelength Bouguer anomalies. Scenarios for the total stress field were calculated by superposing different values for an EWoriented regional compression. We predict the stress axes orientations and stress regimes for the equatorial margin and the Pernambuco Shear Zone (PESZ) region. However, we highlight the importance of spreading stresses, not considered here, in rotating the maximum horizontal compression (SHmax) predictions. The prediction of stress regime and stress axes orientation in PESZ using low to zero regional compression suggest that a decoupling of the shallow stresses from the platewide stresses might occur in this region. Another possible explanation found was that the spreading stresses might counterbalance the regional stresses. We conclude that the superposition of principally flexural and platewide stresses explains reasonably the observed stress regimes and SHmax orientations in a great part of NE Brazil. Furthermore, flexural stresses are playing an important role in the reactivation of the PESZ, possibly controlling the stress field. 

Artigo em Revista
Geophysical evidence for doming during the PanAfrican/Brasiliano orogeny in the SeridÃ³ belt, Borborema Province, Brazil The Neoproterozoic SeridÃ³ belt located in the Borborema Province, NE Brazil, was deformed and metamorphosed in the PanAfrican/Brasiliano orogeny. The basement rocks of the belt were submitted to intense partial melting to form domelike structures as documented in the Santa Luzia anatetic dome. The aeromagnetic map of the SeridÃ³ belt shows a longwavelength anomaly with elongated sigmoidal shape, whose limits coincide with shear zones. Comparison with gravity data shows that both anomalies share common sources, which are felsicrocks more magnetic and less dense than the surrounding crust. Correlation with geology shows that the geophysical anomalies are associated with magnetiterich migmatites and granites. We modelled the geophysical datasets using magnetic susceptibility and density values from measurements in cores of a stratigraphic borehole. The results are consistent with a migmatitedome system formed in the southcentral part of the belt. This largescale crustal structure is composed of elongated internal domes, which were geophysically modelled as four anomalous bodies, that resulted from partial melting of rocks mostly of the basement. The anomalies define an elongated migmatitedome system with the main axis in the NNESSW direction that progressively bends to EW approaching to the Patos shear zone, thus indicating that this shear zone had a key role in shaping the domic structure. The migmatitic domes of the SeridÃ³ belt compose a faultrelated dome system, formed in association with the broad strikeslip shear zone array of the PatosSeridÃ³ system, which evolved through the crustal extrusion resulting from the continental collisions that amalgamated West Gondwana in the Ediacara. 
