Seismic Interpretation Workflow Research Articles

Comparative study of CNN-based and conventional fault interpretation methods: a study of the deep-water Orange Basin, South Africa

Image-based deep learning methods, especially convolutional neural networks (CNNs), are gaining traction in seismic interpretation, but their application still demands manual validation. This study compares a U-Net structured CNN, called Fault-Net, with conventional edge-enhancing seismic attributes of variance and chaos, which serve as a scientific baseline. We adopted two seismic interpretation workflows: (1) conventional attributes to enhance fault features, and (2) a deep learning-based workflow for fault segmentation using CNNs. Both workflows were applied to a high-resolution 3D seismic dataset from the structurally complex deep-water Orange Basin (South Africa). While deep learning-based software packages are commercially available, it is unclear whether they are suitable for the Orange Basin and in an academic setting due to their proprietary architectures and generally closed training data. This study provides public evidence of the feasibility of automated structural interpretation in complex seismic datasets using deep learning, revealing both key benefits and limitations. Where high-quality labelled data are available, the deep learning approach is faster and tends to produce a cleaner and more accurate depiction of larger faults compared to conventional methods. The open availability of Fault-Net makes deep learning-based interpretation particularly advantageous for academic settings, offering significant time and resource efficiency while enhancing the understanding of complex subsurface structures.

Machine learning for seismic exploration: Where are we and how far are we from the holy grail?

Machine-learning (ML) applications in seismic exploration are growing faster than applications in other industry fields, mainly due to the large amount of acquired data for the exploration industry. The ML algorithms are constantly being implemented for almost all the steps involved in seismic processing and interpretation workflow, mainly for automation, processing time reduction, efficiency, and in some cases for improving the results. We carry out a literature-based analysis of existing ML-based seismic processing and interpretation published in SEG and EAGE literature repositories and derive a detailed overview of the main ML thrusts in different seismic applications. For each publication, we extract various metadata about ML implementations and performances. The data indicate that current ML implementations in seismic exploration are focused on individual tasks rather than a disruptive change in processing and interpretation workflows. The metadata indicate that the main targets of ML applications for seismic processing are denoising, velocity model building, and first-break picking, whereas, for seismic interpretation, they are fault detection, lithofacies classification, and geobody identification. Through the metadata available in publications, we obtain indices related to computational power efficiency, data preparation simplicity, real data test rate of the ML model, diversity of ML methods, etc., and we use them to approximate the level of efficiency, effectivity, and applicability of the current ML-based seismic processing and interpretation tasks. The indices of ML-based processing tasks indicate that current ML-based denoising and frequency extrapolation have higher efficiency, whereas ML-based quality control is more effective and applicable compared with other processing tasks. Among the interpretation tasks, ML-based impedance inversion indicates high efficiency, whereas high effectivity is depicted for fault detection. ML-based lithofacies classification, stratigraphic sequence identification, and petro/rock properties inversion exhibit high applicability among other interpretation tasks.

Rock physics modeling and quantitative seismic interpretation workflow for organic-rich mudrocks

A quantitative seismic interpretation workflow based on rock physics was developed for an organic-rich mudrock formation in order to estimate porosity, and organic matter and clay contents from seismically-derived acoustic and shear impedances. This deterministic workflow consists of: developing a rock physics model that accurately estimates the elastic properties of the formation; finding inter-relationships between the formation's petrophysical properties; and validating the model after upscaling and applying it to seismically-derived acoustic and shear impedances. By analyzing wireline data from this organic-rich mudrock formation, we find that the P-wave velocity at fixed porosity may vary by as much as 1.0 km/s, while the respective variation in the S-wave velocity reaches 0.5 km/s. A robust discriminator for velocity appears to be, in addition to the total porosity, the volumetric fraction of the soft components of the solid matrix where the kerogen is included, namely the sum of the clay and kerogen. The ensuing velocity-porosity-mineralogy relations are accurately described by the soft-sand rock physics model. Thus, an attempt to interpret seismically-derived impedance volumes based on rock physics modeling is performed after simplifying the rock's petrophysical variables by building site-specific internal relations between mineral's volume fractions that comprise the rock's solid matrix. This workflow helps in hydrocarbon exploration by identifying favorable areas with relatively high porosities and organic matter content.

Machine Learning Assisted Seismic Interpretation: a Case Study of the Loppa High Area, Barents Sea

The interpretation of faults and horizons from seismic data forms a critical part of the geoscience workflow, enhancing our understanding of the subsurface and ultimately the chance of success in extracting hydrocarbons. The depiction of these vital seismic interpretations has long been restricted to conventional manual and semi-automatic techniques, which require geoscientists to work line by line over ever-expanding volumes of seismic data. The advent of machine learning (ML) and cloud computing technology has revolutionised tasks across multiple industries, enabling the identification of patterns in large multi-factor datasets. In this case study, we predict fault locations using 2D U-Net architecture convolutional neural network (CNNs) and predict horizons by employing Radial Basis Functions (RBFs) and Neural Networks (NNs). We aim to demonstrate the gains in efficiency and geological insight found using ML technology as the bedrock of the seismic interpretation workflow, through the interpretation of a broadband seismic dataset from the Loppa High area, Barents Sea.

Multilevel Composition: A new method for revealing complex geological features in three-dimensional seismic reflection data

Advanced seismic data and multi-attribute visualization techniques, such as color blending of attributes, have considerably enhanced the capability of interpreters to characterize geological features in three-dimensional (3D) seismic reflection datasets. However, high resolution investigation of complex, vertically linked geological features such as channel systems and fluid conduits, remains challenging. These features may appear in the dataset as pronounced attribute anomalies, such as high-amplitude or spectrally or structurally enhanced seismic reflectivity bands, at several depth levels. Vertical linkages between these features, however, may not be readily established. We have developed an innovative method, Multilevel Composition, for an intuitive display of vertically connected features. Our method involves the composition of attribute maps from three different depth/time windows or slices onto a single map, in which inter-window/layer depth information is coded in colors. Multilevel Composition starts with the identification of suitable seismic attributes, such as high amplitudes in the examples displayed here, to map features of geological interest. At least one reference horizon is then identified and mapped in the vicinity of the target window of interest. Three sub-windows are then defined with respect to the reference horizon(s) based on the vertical and spatial distribution of the geological features. Relevant seismic attributes are computed for each of the sub-windows, and the resulting maps, one from each sub-window, are assigned basic color channels and are co-rendered to reveal multilevel linkages between these features. We demonstrate the efficacy of this method by applying it to two 3D seismic datasets, one illuminating deep-water depositional elements in the eastern Nile fan, eastern Mediterranean and the other targeting seafloor seeps and underlying gas migration systems beneath the Omakere Ridge, offshore New Zealand. The new method is simple and should be easy to implement to enhance seismic interpretation workflows.

A high-resolution fault assessment workflow for unconventional reservoirs exploration: A case study on the onshore Sergipe-Alagoas Basin, Brazil

Apart from Vaca Muerta’s case history in Argentina, the unconventional exploration in South America is still underdeveloped, and one of the causes is the scarcity of geological studies. The present paper enlightens the embryonic unconventional exploration in the Sergipe-Alagoas Basin by introducing a seismic interpretation workflow to map thin shale layers and the natural fracture system, a challenge in unconventional reservoir explorations. Fracture and faults are critical elements in exploring ultra-low permeability shale oil/gas reservoirs. Identifying and locating these structures influence modeling fluid flow, hydraulics, and simulation models to improve horizontal well-drilling plans. The workflow comprises four main steps: data pre-conditioning, well-to-seismic tie, seismic spectral blueing (for frequency enhancement), and, finally, spectral decomposition and advanced fault attributes. The proposed workflow was applied to a seismic dataset from the onshore Brejo Grande oilfield focusing on the source rocks of the Coqueiro Seco Formation, the main unconventional reservoir candidate in the basin. That approach allowed mapping the lateral continuity, truncation, faults, and fractures in these thin shale layers previously below seismic resolution. The mapped structures belong to the complex natural fracture system with faults distributed to all trend directions. These faults are distributed along two zones with distinct fault densities with a possible sweet spot outline.

Seismic interpretation workflows for deep-water systems: A practical guide for the subsurface

Effective exploration, appraisal and development of deep-water reservoirs requires systematic mapping, description and characterization of the depositional systems that host them. Success in these areas requires that there is a firm understanding of hydrocarbon volumes and uncertainties, facies distributions to locate exploration and appraisal wells, and a framework to build reservoir models. We present a practical seismic interpretation workflow methodology which will be applicable in most cases and with most seismic data sets. The workflows are for channels, sheets and mass transport deposits (MTDs). Each step of the workflow is illustrated with subsurface and/or outcrop examples. The channels workflow guides the interpreter through identification of three hierarchical levels of channel architecture and the recognition of six key seismic facies: channel axis; channel margin; basal lags; slumps; thin-bedded facies (internal levees and terraces); and external levees. The sheet sandstone workflow emphasizes a series of steps including: understanding the regional and local context; understanding the evolution of sea-floor topography at basinal and depositional scales; and generating detailed amplitude maps. Mass transport complexes are rarely reservoirs but can have a major impact on reservoir distribution. The workflow encourages enquiries into these impacts which include: potential erosion of underlying reservoirs; ponding on and around the MTD; diversion and deflection of subsequent reservoir-hosting depositional systems; and their potential as sealing facies. The workflows are designed to be efficient and practical, and will provide an appropriate characterization encompassing exploration and appraisal to early development phases.

Rock Physics Based Interpretation of Seismically Derived Elastic Volumes

A rock physics based seismic interpretation workflow has been developed to extract volumetric rock properties from seismically derived P- and S-wave impedances, Ip and Is. This workflow was first tested on a classic rock physics velocity-porosity model. Next, it was applied to two case studies: a carbonate and a clastic oil field. In each case study, we established rock physics models that accurately relate elastic properties to the rock’s volumetric properties, mainly the total porosity, clay content, and pore fluid. To resolve all three volumetric properties from only two inputs, Ip and Is, a site-specific geology driven relation between the pore fluid and porosity was derived as a hydrocarbon identifier. In order to apply this method at the seismic spatial scale, we created a coarse-scale elastic and volumetric variables by using mathematical upscaling at the wells. By using Ip and Is thus upscaled, we arrived at the accurate interpretation of the upscaled porosity, mineralogy, and water saturation both at the wells and in a simulated vertical impedance section generated by interpolation between the wells.

Lithofacies-dependent rock physics templates of an unconventional shale reservoir on the North Slope, Alaska

Over the past 20 years, oil and gas companies have turned their attention to producing petroleum directly from organic-rich shale. Successful exploration, appraisal, and production strategies for source rocks critically depend on reliable identification of their organic components (kerogen, in particular) and generation potential. There is mounting demand to evaluate organic richness in terms of quantity (i.e. total organic carbon) and quality (i.e. hydrogen index) from seismic data, which is usually the only source of information in the early development period of emerging shale plays. We delineated major seismic lithofacies on the Alaska North Slope using elastic, seismic, and petrophysical properties. We performed a well-established quantitative seismic interpretation workflow to integrate geochemical data in the lithofacies definition. Rock physics templates of seismic parameters, Acoustic Impedance, (AI), versus P-wave to S-wave velocity ratio, (VP/VS), are constructed for each lithofacies to assess variations in pore fluid and lithology. We proposed correlations between source rock properties (hydrogen index, total organic carbon) and petrophysical properties (bulk density, porosity, sonic velocity ratio) of the major lithofacies. These correlations, together with facies-specific rock physics templates, can be utilized to predict organic richness and source rock properties away from drilled wells. The models are validated by training data from 2 regional wells to observe their applicability on the Alaska North Slope.

Quantitative Seismic Interpretation Identifies Sweet Spots in Organic-Rich Mudrocks

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 197290, “Quantitative Seismic Interpretation Work Flow for Sweet-Spot Identification in Organic-Rich Mudrocks,” by Mustafa A. Al Ibrahim, Tapan Mukerji, SPE, and Allegra Hosford Scheirer, Stanford University, prepared for the 2019 SPE Abu Dhabi International Petroleum Exhibition and Conference, Abu Dhabi, 11–14 November. The paper has not been peer reviewed. Estimating the lateral heterogeneity of geochemical properties of organic-rich mudrocks is important for unconventional resource plays. Well-penetration data from emerging plays are sparse and so traditional geostatistical methods will not yield good results. In this study, the authors use a quantitative seismic interpretation work flow (QSI) based on a rock-physics template in estimating the uncertainty of the geochemical properties of organic mudrocks of the Shublik formation of the Alaskan North Slope. By combining the rock-physics template and the results of seismic inversion, multiple realizations of total organic content (TOC), matrix porosity, and brittleness index are generated. These parameters can be used for sweet-spot detection. Introduction Data for physical rock parameters are generally restricted to well locations. To extend data laterally, two general directions can be taken: a data-oriented approach and a physics-oriented approach. The data-oriented approach is powerful but requires a large amount of data commonly not available in emerging areas where well penetration is scarce. Physics-oriented approaches can be used to obtain the desired parameters used in decision-making; however, the results are highly dependent on the inputs, such as TOC and hydrogen index, and these typically are not defined regionally. The QSI is a powerful tool that can be used to estimate the lateral and vertical heterogeneity of rocks beyond wellbores. QSI combines both of the previously mentioned approaches by using well data to calibrate rock-physics models and then using statistical methods to derive relationships between rock-physics models and seismic data. In the complete paper, the authors demonstrate a work flow to apply the QSI to organic-rich mudrocks by incorporating a constructed rock-physics template for mudrocks. Joint estimation of rock properties is used to estimate the geochemical and mechanical properties important in determining sweet spots. Multiple possible realizations of Earth models are generated to capture the nonuniqueness associated with the inversion problem. Study Area The Triassic Shublik source rock is the main petroleum source in the North Slope of Alaska. The organic-rich calcareous interval is interpreted to be deposited during a marine upwelling period. Lateral and vertical heterogeneity in the Shublik formation has been documented with multiple facies identified.

Automatic extraction of dislocated horizons from 3D seismic data using nonlocal trace matching

Extracting key horizons from seismic images is an important element of the seismic interpretation workflow. Although numerous computer-assisted horizon extraction methods exist, they are typically sensitive to structural and stratigraphic discontinuities. As a result, these computer-assisted methods have difficulties in extracting noncoherent dislocated horizons. We have developed a new data-driven method to correlate, track, and extract horizons from seismic volumes with complex geologic structures. Our method correlates seismic horizons across discontinuities and does not require user input in the form of seed points or prior identification of faults. Furthermore, the method is robust toward amplitude changes along a seismic horizon and does not jump from peak to trough or vice versa. We use a large sliding window and match full-length seismic traces using nonlocal dynamic time warping to extract grids of correlated points for our target horizons. Through computed accuracy measurements, we discard nonaccurate correlations before interpolating complete seismic horizons. Because our method does not require manually picked seed points or prior structural restoration, it does not rely on interpretive experience or geologic knowledge. The proposed method is applied on different real and complex seismic images, with two case examples from the southwestern Barents Sea, and one on the open source Netherlands F3 seismic data.

A practical workflow using quantitative interpretation of seismic amplitudes to derisk infill wells in deepwater Gulf of Mexico: The Auger example

Seismic direct hydrocarbon indicators (DHIs) are routinely used in the identification of hydrocarbon reservoirs and in the positioning of drilling targets. Understanding seismic amplitude reliability and character, including amplitude variation with offset (AVO), is key to correct interpretation of the DHI and to enable confident assessment of the commercial viability of the reservoir targets. In many cases, our interpretation is impeded by limited availability of data that are often less than perfect. Here, we present a seismic quantitative interpretation (QI) workflow that made the best out of imperfect data and managed to successfully derisk a multiwell drilling campaign in the Auger and Andros basins in the deepwater Gulf of Mexico. Data challenges included azimuthal illumination effects caused by the presence of the Auger salt dome, sand thickness below tuning, and long-term production effects that are hard to quantify without dedicated time-lapse seismic. In addition, seismic vintages with varying acquisition geometries led to different QI predictions that further complicated the interpretation story. Given these challenges, we implemented an amplitude derisking workflow that combined ray-based illumination assessments and prestack data observations to guide selection of the optimal seismic data set(s) for QI analysis. This was followed by forward modeling to quantify the fluid saturation and sand thickness effects on seismic amplitude. Combined with structural geology analysis of the well targets, this workflow succeeded in significantly reducing the risk of the proposed opportunities. The work also highlighted potential pitfalls in AVO interpretation, including AVO inversion for the characterization of reservoirs near salt, while providing a workflow for prestack amplitude quality control prior to inversion. The workflow is adaptable to specific target conditions and can be executed in a time-efficient manner. It has been applied to multiple infill well opportunities, but for simplicity reasons here, we demonstrate the application on a single well target.

Full-volume 3D seismic interpretation methods: A new step towards high-resolution seismic stratigraphy

Following decades of technological innovation, geologists now have access to extensive 3D seismic surveys across sedimentary basins. Using these voluminous data sets to better understand subsurface complexity relies on developing seismic stratigraphic workflows that allow very high-resolution interpretation within a cost-effective timeframe. We have developed an innovative 3D seismic interpretation workflow that combines full-volume and semi-automated horizon tracking with high-resolution 3D seismic stratigraphic analysis. The workflow consists of converting data from seismic (two-way traveltime) to a relative geological time (RGT) volume, in which a relative geological age is assigned to each point of the volume. The generation of a horizon stack is used to extract an unlimited number of chronostratigraphic surfaces (i.e., seismic horizons). Integrated stratigraphic tools may be used to navigate throughout the 3D seismic data to pick seismic unconformities using standard seismic stratigraphic principles in combination with geometric attributes. Here, we applied this workflow to a high-quality 3D seismic data set located in the Northern Carnarvon Basin (North West Shelf, Australia) and provided an example of high-resolution seismic stratigraphic interpretation from an Early Cretaceous shelf-margin system (Lower Barrow Group). This approach is used to identify 73 seismic sequences (i.e., clinothems) bounded by 74 seismic unconformities. Each clinothem presents an average duration of approximately 63,000 years (fifth stratigraphic order), which represents an unprecedented scale of observation for a Cretaceous depositional system on seismic data. This level of interpretation has a variety of applications, including high-resolution paleogeographical reconstructions and quantitative analysis of subsurface data. This innovative workflow constitutes a new step in seismic stratigraphy because it enables interpreters to map seismic sequences in a true 3D environment by taking into account the full variability of depositional systems at high frequency through time and space.

Scale effects of velocity dispersion and attenuation (Q−1) in layered viscoelastic medium

We have used numerical modeling of normal incidence P-waves in periodic and nonperiodic viscoelastic layered media to help improve understanding of the scale effects of heterogeneity on velocity dispersion and attenuation. The improved understanding of these effects facilitates better interpretation and integration of data acquired at different scales, such as seismic data, well-log data, and laboratory measurements. We developed a direct method for estimating velocity and attenuation for viscoelastic media, the viscoelastic Kennett-Frazer (vKF) method, which is an invariant imbedding (reflectivity) method that uses reflection and transmission transfer functions. The vKF method is used to estimate rigorous dispersion and attenuation curves for periodic and nonperiodic cases. The results from our studies validate and quantify the intuitive qualitative understanding that dispersion and attenuation for a layered viscoelastic medium depend on the ratio of wavelength of the waves ([Formula: see text]) and the spatial period of the medium ([Formula: see text]), similar to the elastic case. We also decoupled the total effective attenuation obtained from the viscoelastic case into scattering attenuation that can be modeled from the elastic case and the intrinsic effective attenuation that is present in addition to the scattering attenuation in the viscoelastic case. The calculated intrinsic effective attenuation curves matched the theoretical values at the ray theory and effective medium theory limits. We derived analytical expressions for the long-wavelength limit of velocity and attenuation in a viscoelastic medium. Our expressions can be used directly for the upscaling of well logs to seismic scale considering viscoelastic effects. The expressions indicate the coupling between effective velocity and effective intrinsic attenuation in the long-wavelength limit. Finally, we use the derived expressions to determine the difference between elastic versus viscoelastic upscaling and to highlight the impact on traveltime and amplitude by properly considering viscoelastic information, especially for quantitative seismic interpretation workflows.

A SPECTRAL METHOD OF LITHOFACIES DIFFERENTIATION WITHIN A HYDROCARBON RESERVIOR UNIT USING SEISMIC AND WELL DATA FROM TOMBOY FIELD NIGER DELTA

Seismic facies analysis is a key component of Seismic interpretation workflow, as much information on depositional process, environment, reservoir architecture and potential can be determined from the seismic data. This enhances the seismic visibility of subtle stratigraphic features, and prediction of new prospects for drilling due to lateral continuity of sand bodies (the hydrocarbon reservoir) deposited within the sedimentary column. This study presents the results of the application of seismic amplitude inversion and spectral decomposition to differentiate lithofacies within a thin-sand reservoir and show their lateral and vertical continuity using amplitude-derived attributes (rock properties) in the interpretation of 3D seismic and well log data from “TOMBOY” field Niger Delta. A thin-sand reservoir was delineated from lithologic log signatures within the interval 11,135.61-11,210.35ft which corresponds to the time window 2.752 - 2.768secs on the seismic data along an arbitrary line drawn to connect the six wells (01-06) in the field. Crossplot of acoustic impedance against density across the reservoir bed showed three compartments interpreted as gas filled sand, oil filled sand and shale. The interpretation was enhanced via spectral analysis by computing frequency maps of acoustic impedance, density and velocity using discrete Fourier transform (DFT) technique at top (2.752sec) of the reservoir in terms of magnitude, phase and frequency. Phase response for the three attributes (acoustic impedance, density and velocity) showed lateral continuity and discontinuity of sand lithofacies at the various well locations which were not evident on their original amplitude time slice, while frequency response showed bed thickness, with low frequency indicating presence of hydrocarbon or sand rich beddings, while high frequency indicates thin shale beddings. It was observed that the top (2.752sec) of the reservoir for wells 02, 05, 01 and 04 showed low frequency values for the three attributes (acoustic impedance, density and velocity) which indicates potential zones for hydrocarbon production and development in the field, while wells 03 and 06 within the same time slice (2.752sec) showed a higher value in acoustic impedance frequency, density frequency and velocity frequency and suggests the presence of thin shale facies at the top (2.752sec) of the thin-sand reservoir. By defining the lateral and vertical continuity of sand facies within the reservoir bed, hydrocarbon fairways were better defined and knowledge of this can be incorporated into drilling decisions for field development. The results gave enhanced structural disposition of the reservoir bed and more insights into the variation of sand lithofacies with depth. Keywords: Lithofacies, Amplitude inversion, Geologic transform, Seismic attributes, Cross plots and Spectral decomposition. DOI : 10.7176/JEES/9-3-14 Publication date :March 31 st 2019

Fault deformation, seismic amplitude and unsupervised fault facies analysis: Snøhvit Field, Barents Sea

We present an integrated seismic imaging and fault interpretation workflow to characterize the seismic expression in and around an E-W trending central fault system imaged in near-angle stack seismic data of the Snøhvit Field, Barents Sea. Three E-W normal fault systems offset five Triassic-Lower Cretaceous seismic horizons across the field. Fault throw is largest at depth and decreases with shallowing. Dip distortion (DD) decreases in magnitude and extent with shallowing. Fault enhancement (FE), a filter used to detect edges, was applied on a blend of tensor, semblance and dip attributes, and allowed us to classify fault zones into four unsupervised seismic fault facies (mid-high FE). High FE facies occur at the center of the fault zones and are abundant in the highest thrown eastern part of the field. The FE facies decrease radially outwards field wide. Facies correlate with throw and dip separation gradient, which are in turn related to mechanical stratigraphy controlling fault propagation. We observe systematic seismic amplitude variations: a major amplitude drop on the fault plane, and a brightening and dimming linked to fault-related synclines and anticlines, respectively. Our workflow establishes a methodology for fault interpretation, linking fault throw, DD, seismic attributes and fault facies classification.

Offset vector tile domain wide-azimuth prestack seismic interpretation methods and applications

The offset vector tile (OVT) is a special prestack seismic gather type, and OVT technology is a seismic processing technology suitable for wide-azimuth seismic (WAZ) exploration. Because current WAZ interpretation is not yet mature and because of the lack of key techniques and tools for OVT-domain seismic interpretation, abundant offset and azimuth information possessed in the OVT gathers has not been fully used. To take full advantage of the OVT gathers to achieve more accurate geologic, reservoir, and fluid information, we have developed an OVT-domain wide-azimuth prestack seismic interpretation technology and a typical workflow. The OVT-domain wide-azimuth prestack seismic interpretation technology is composed of five key prestack analysis and interpretation techniques focused on OVT gathers: (1) sensitive offset and azimuth analysis for choosing their sensitive offset and azimuth ranges, (2) template-based OVT gather optimization for improving their signal-to-noise ratio and quality, (3) multiscale and multiazimuth attribute analysis, (4) OVT domain amplitude versus incidence angle analysis, and (5) azimuthal anisotropy analysis. These technologies are used to extract seismic attributes that vary with offset and/or azimuth from OVT gathers facing on reservoir characterization, fluid detection, and fracture prediction. These technologies can form an OVT-domain prestack seismic interpretation workflow.

Deep-learning tomography

Velocity-model building is a key step in hydrocarbon exploration. The main product of velocity-model building is an initial model of the subsurface that is subsequently used in seismic imaging and interpretation workflows. Reflection or refraction tomography and full-waveform inversion (FWI) are the most commonly used techniques in velocity-model building. On one hand, tomography is a time-consuming activity that relies on successive updates of highly human-curated analysis of gathers. On the other hand, FWI is very computationally demanding with no guarantees of global convergence. We propose and implement a novel concept that bypasses these demanding steps, directly producing an accurate gridding or layered velocity model from shot gathers. Our approach relies on training deep neural networks. The resulting predictive model maps relationships between the data space and the final output (particularly the presence of high-velocity segments that might indicate salt formations). The training task takes a few hours for 2D data, but the inference step (predicting a model from previously unseen data) takes only seconds. The promising results shown here for synthetic 2D data demonstrate a new way of using seismic data and suggest fast turnaround of workflows that now make use of machine-learning approaches to identify key structures in the subsurface.

Introduction to special section: Seismic inversion — Conventional seismic impedance inversion and advanced seismic inversion techniques: Developments, workflow, and case studies

Seismic inversion is a key component of the quantitative seismic interpretation workflows for hydrocarbons exploration and production. In exploration with limited well control, seismic inversion for acoustic impedance helps to understand the spatial variations and identify potential “pay zones

High resolution diffraction imaging for reliable interpretation of fracture systems

Small-scale subsurface features, such as natural fractures, act as scattering sources for seismic waves propagating through the subsurface. The wavefield generated by those source points is identified as diffraction energy. The amplitude of this type of energy is much smaller than the recorded events reflected from actual interfaces between different geological layers. Moreover, diffraction energy is normally suppressed by conventional processing and standard imaging algorithms, where summations and averaging processes are applied. The common objective in such processing workflows is to focus on the high specular amplitudes in order to enhance the continuity of seismic reflection events for improving the structural mapping of the subsurface. Our goal is to complement the traditional seismic interpretation workflow by integrating information relative to diffraction energy as another seismic attribute to be interpreted. The technique applied in this paper is based on a depth imaging algorithm that maps and bins the recorded surface information into multi-dimensional, local angle domain (LAD) common image gathers. The advantage of this system is its unique ability to decompose the wavefield into reflection and diffraction energy directly at the image locations. This paper provides a brief overview of the technology and illustrates its benefits when applied to the Eagle Ford and Barnett shale reservoirs, where seismic data can be of moderate quality, leading to accurate, high-resolution, and highcertainty seismic interpretation for risk-managed field development.