Salt Dome Detection Using Context-Aware Saliency

SaltFormer: A hybrid CNN-Transformer network for automatic salt dome detection

Delineating salt is an inherently difficult problem. Currently, there are several methods to detecting salt bodies in seismic. However, the result and volumes are often noisy, and are therefore hard to use for voxel based segmentation interpretation. We present a new method of using a 3D Sobel edge detector in combination with dip guiding to generate clearer images. We also introduce elements of weighting and normalization to achieve our results.

Limitations of the seismic method in mapping salt diapirs increase exploration risks while drilling in near-salt areas. An innovative method of 3D joint inverse problem solution for gravity, well-log, and seismic data makes it possible to verify and refine the shape of a salt dome and map hydrocarbon accumulations right beside the salt wall under the salt wing and over the salt dome. It is illustrated by the results for four salt domes in the central axial part of the Dniper-Donets Basin. In half of the cases, joint gravity inversion proved that seismic data provided a reliable salt shape that may be used as the biggest outline. In the other half of the cases, the salt stem outlined by 3D gravity inversion significantly differed from that by 2D and 3D seismic data interpretation. In one case, gravity inversion showed that the size of the salt stem was three times smaller. For all the cases under the salt wing in proximity to the salt dome, near the salt wall, and over the salt dome, localized low-density rock areas were mapped. These areas were associated with commercial hydrocarbon pools. For one, the salt dome location of the commercial pools was verified by drilling. As a result, a new oil field was discovered and named after Academician Schpak. Also, over the same salt dome, the size of the Runovshina field shallow gas pool outline was six times expanded. Four wells drilling outcomes ― three commercial and one dry ― fully confirm the correctness of the shape of the mapped commercial pool with a 100 % commercial probability of success.

Saliency mechanism has been considered crucial in the human visual system and helpful to object detection and recognition. This paper addresses an information theoretic model for visual saliency detection. It consists of two steps: first, using the Non-negative Matrix Factorization with sparseness constraints (NMFsc) algorithm to learn the basis functions from a set of randomly sampled natural image patches; and then, applying information theoretic principle to generate the saliency map by the Salient Information (SI) which is calculated from the coefficients represented by basis functions. We compare our model with the previous methods on natural images. Experimental results show that our model performs better than existing approaches.

Thin (<20 m), sand injectites, too small to resolve in conventional seismic reflection data, can constitute a large part of the net-to-gross volume and affect fluid flow in the reservoir. However, they may also cause challenges for well placement and reservoir development because they are too small to be reliably imaged in seismic reflection data. It is therefore important to better understand how small-scale injectites influence seismic images, and how they may be recognized and characterized above reservoirs. The Grane Field (North Sea) hosts numerous small-scale sand injectites above the main reservoir unit, challengin well placement, volume estimates and seismic interpretation. Here, we investigate how such small-scale sand injectites influence seismic images and may be characterized by (1) using well, 3D seismic and outcrop data to investigate geometries of small-scale sand injectites (0–15 m thick), (2) performing seismic convolution modelling to investigate how these would be imaged in seismic data, and (3) comparing these synthetic seismic images with actual 3D seismic data from the well-investigated Grane Field. Our results show that despite injectites being below seismic resolution, small-scale sand injectites can be detected in seismic data. They are more likely to be detected if they have high thickness (>5 m), steep dip (>30°), and when densely spaced. Furthermore, as the fraction of sand injectites increases the top reservoir amplitude will decrease. Moreover, comparison of the synthetic seismic images with real seismic data from the Grane Field indicates that the low-amplitude anomalies and irregularities observed above the reservoir may be a result of the overlying sand injectites. Additionally, the comparison strongly suggests that the Grane Field hosts sand injectites that are thicker and located further away from the top reservoir than is indicated by well observations. These results may be used to improve well planning and develop reservoirs with overlying sand injectites. Supplementary material: A PDF file containing all the seismic modelling results allowing the reader to flip back and forth between the models is available at https://www.doi.org/10.6084/m9.figshare.14333102 and well logs from well 25/11-18T2 are available at https://factpages.npd.no/pbl/wellbore_documents/2358_25_1_18_COMPLETION_REPORT_AND_LOG.pdf

This paper is the second one in a series of tutorials on the usage of modern migration methods for seismic imaging in areas of intense salt tectonics. In the paper, it is shown the importance of building an adequate velocity model for effective seismic imaging and subsequent mapping of hydrocarbon traps associated with salt domes. To solve this problem, ray-based tomography, delayed imaging time scans and full waveform inversion are successfully used. Moreover, the paper shows the importance of evaluating properly and taking into account velocity anisotropy when making seismic images. This is demonstrated with real seismic data obtained in areas of heavy halokinesis.

Defining the root zone of mud volcanoes (MVs), structural interpretation, and geologic modeling of their body is a problematic task when only seismic data are available. We have developed a strategy for integration of gravity and seismic data for better structural interpretation. Our strategy uses the concept of the normalized full gradient (NFG) for integration of gravity and seismic data to define geometry and the root zone of MVs in the southeast onshore of the Caspian Sea. Our strategy will increase the resolution of the seismic envelope compared with the conventional Hilbert transform. Prior to interpretation, we applied the NFG method on field gravity data. First, we perform a forward-modeling step for accurate NFG parameter definition. Second, we estimate the depth of the target, which is the root zone of the MV here. Interpretation of field gravity data by optimized NFG parameters indicates an accurate depth of the root zone. Subsequently, we apply the NFG method with optimized parameters on a 2D seismic data. Application of our strategy on seismic data will enhance resolution of the seismic image. The depth of the root zone and the geometry of the MV and mud flows was interpreted better on the enhanced image. It also illustrates the complex structure of a giant buried MV, which was not well-interpreted on conventional seismic image. Interpretation of the processed data reveals that the giant MV had lost its connection to its reservoir, whereas the other MV is still connected to the mud reservoir. The giant MV is composed of complex bodies due to pulses in the mud flows. Another MV in the section indicates narrow neck with anticline and listric normal faults on its top. Thus, application of the NFG concept on seismic image could be considered as an alternative to obtain enhanced seismic image for geologic interpretation.

Our research project is dedicated to the development of a comprehensive model for analysing the distribution, dimensions, and evolution of Pleistocene tunnel valleys and their deposits in northern Germany and adjacent areas. The primary objective is to leverage these findings to assess the likelihood of future tunnel-valley formation, with potential implications for the long-term (over the next 1 million years) safety of a radioactive waste repository.To achieve our goal, we are relying on a 3D seismic dataset. Previously, the mapping of tunnel valleys on land is primarily based on 2D seismic and boreholes, which unfortunately do not provide the required accuracy. Therefore, we have opted to utilize a marine seismic dataset. This 3D seismic dataset 'GeoBasis3D' was acquired by the BGR in 2021.The 3D seismic dataset is situated within the German Exclusive Economic Zone (EEZ) in the 'Entenschnabel' area. In this region, two intersecting tunnel valleys exist, with one located above the 'Belinda' salt dome. The interpretation of the tunnel-valley base based on the seismic data, and we will observe the influence of the crestal faults above the salt dome on the genesis and filling of the tunnel valley. The filling of the tunnel valleys will be described in terms of seismic facies. Different sedimentary processes can be interpreted from the seismic data. The deepest parts of the tunnel valley are directly filled, and the valley widens above. Some slumping can be detected along the steep slopes of the tunnel valley. Different phases of sedimentation can be observed within the tunnel valley, including both glacifluvial and glacilacustrine phases with parallel and homogenous reflectors. Since there are no available geological cores for the Quaternary in the area of the seismic surveys, we will have to rely on cores from Danish North Sea for the lithostratigraphic description of the sediments and for their chronological classification.Our aim is to analyse sediment facies to draw conclusions about the backfilling process and repeated erosion phases. This will enable us to compare the findings with the development of onshore tunnel valleys in the next step. The tunnel valleys are a type of glacial erosion that can reach depths of up to 600 meters above sea level in northern Germany. They can have an impact on the long-term safety of a repository, which is required by law to be located at a minimum depth of 300 meters below ground level.

Seismic reflection data are used by geologists to identify the best site for oil and gas explorations. The raw seismic image cannot be directly used for explorations because the noises may hinder the primary reflections and result in misinterpretations. In the proposed research work, the denoising method with adaptive dictionary learning is attempted to attenuate the coherent and random noises which affect the seismic reflections. In this method, Karhunen Loeve Transform (KLT) is combined with K-means Singular Value Decomposition (KSVD) to improve the Signal to Noise Ratio (SNR) of the seismic data. The combination of fixed transform using KLT and learning-based dictionary using KSVD remove the redundant data while retaining the necessary data for further interpretations. KLT is applied on the whole seismic image to decorrelate the coefficients and retain the primary reflections. The horizontal events in the seismic image are preserved as they represent the eigen images with large energy. The KSVD is applied to the KLT resultant data, which denoises the patch of data while simultaneously updating the dictionary. The noise is eliminated using local sparsity of the image where mutually overlapping small image blocks are learned to yield the self-adaptive redundant dictionary. It is then used to obtain the sparse representation of the image blocks by eliminating noise. The algorithm is tested for both synthetic and field seismic images, and the results indicate a better reduction of random and coherent noise with the acceptable execution time compared to other denoising algorithms used. The KLT combined with KSVD method decorrelate the seismic data from noise by preserving the primary reflections and discarding the redundant noise.

Interpretation of seismic structural traps for accurate hydrocarbon reservoirs characterization is a challenging task. Seismic interpreters learn to accurately delineate subsurface structures after going through a lengthy process of training and expertise-acquiring that is challenging and time-consuming. In this paper, we propose a novel semantic segmentation model for salt domes and faults identification in a real concurrent scenario using an improved encoder-decoder deep neural network that achieves high detection accuracy for both salt domes and faults. We also introduce transfer learning to alleviate the everlasting scarcity issue of labeled seismic data and develop a robust model whose performance is not affected by event similarities among various discontinuities in seismic data. In addition, we use residual blocks in our deep neural network to make it even more robust. To demonstrate the effectiveness of our model, extensive experiments were conducted through validation and testing on real-world seismic data from the publicly available Netherlands offshore F3 block, the LANDMASS, and the TGS datasets. Both qualitative and quantitative evaluations are provided to confirm the superior performance achieved by our deep learning based workflow under the challenging scenario of multiple events detection in subsurface surveys.

Seismic reflection amplitude data are increasingly used in reservoir modeling to provide information on changes in earth properties away from well locations. In geostatistical reservoir modeling, the most common application is to use seismic data as background data in some form of comodeling. Seismic data image reflectors and not earth layer properties. Therefore, prior to use in comodeling, seismic data must first be transformed into an earth layer property. Typically, the transform is to acoustic impedance using an appropriate seismic inversion method. Seismic inversion methods generate results that are generally band limited in nature, resulting in limits to vertical resolution. The vertical resolution achieved can be an order of magnitude below the vertical model resolution required from geostatistical reservoir modeling, which is in the order of well-log resolution. Hence, in using seismic data, geostatistical modelers encounter a problem of downscaling, not the more commonly encountered upscaling problem. This difference in scale introduces scatter between the primary data with well-log order resolution and the secondary seismically derived rock property data used in the comodeling. As a result, to preserve vertical heterogeneity, only limited use of the secondary data can be made in comodeling procedures. This results in models that only partially fit the seismic data, i.e., only limited use is made of the seismic information. If the secondary data are more strongly imposed, the fit to the seismic data improves, but the required vertical heterogeneity is not preserved. The inability to overcome this difference in scale issue, therefore, limits the value of the application of comodeling methods to integrate seismic data into reservoir models. One class of geostatistical methods that overcomes this limitation relies on iterative geostatistical modeling. In these methods, referred to as geostatistical seismic inversion, the iterative modeling process is conditioned such that the final models generated closely match the seismic data while maintaining the required vertical heterogeneity. The application of these methods is computationally expensive relative to comodeling methods but is now practical for large models on today's desktop hardware. Relative to comodeling, geostatistical seismic inversion methods make full use of the information carried in the seismic data, resulting in a significant reduction in model uncertainty away from well control.

The most robust evidence indicating that petroleum migrates in the Gulf of Mexico (GoM) is the presence of oil slicks at the surface of the sea coming from naturally sourced oil at the sea bottom. The mechanisms for oil migration in the GoM involve the interplay of several variables (including faults, pressure, salt diapirism, etc.) These are not fully understood locally, let alone regionally, and are not well constrained to basin evolution. The use of numerical models helps to represent and visualize oil migration from the source to the traps and or seeps but it is limited by the lack of a complete geological picture. However, despite these limitations, numerical models are an important means of estimating the timing of oil migration. Salt movement through time has a huge influence on the patterns of oil migration and re-migration. The underlying geometries of Allochthonous Salt can help focus oil to an up dip trap or in turn result in the breaching of a trap. Subsequent salt withdrawal may result in a weld which may or may not longer seal the hydrocarbons in place. The characterization of successful plays in the GoM involves understanding seeps, oil fluxes, surface anomalies or hard crusts, along with plausible migration pathways of leakage and remigration of oil as related to salt tectonics. These elements are some of the building blocks to high-grade prospective areas for oil exploration. Introduction The first and most obvious observation is the presence of natural seepage occurring at the surface like sea-bottom pock marks and oil slicks (Fig. 1) The MacDonald Image Laboratory of Florida State University posted a video on the web (http://www.sarsea.org/natural_seapage/gallery.html) where it is possible to fully appreciate the most important characteristics of oil seepage: spatial variability of the point of origin, presence of carbonate hard crust, and domes and pock-marks. Secondly, the identification of possible migration pathways relies on seismic data. By looking at deep seismic reflection data we can interpret mechanisms likely influencing petroleum movement from the source rock to the intermediate traps, and eventually into seeps. Interpreted faults and fractures seem to be the main conduits for oil migration. The examples in this work show the dominant impact of salt movement on the migration. This salt was originally deposited over 160 mya. and has had a complex history of movement in resulting diapirs and salt canopies. Therefore, improving the combined knowledge of halokinesis, oil geochemistry and structural evolution of the basin it is possible to have a better picture of the migrating oil into traps, intermediate reservoirs and ultimately seeps.

The extensive coverage of 3-D seismic data on the Santos Basin deep water offered an integrated vision of the local physiography and medium- to small-scale geomorphological features, which enhances the understanding of its constructive/deconstructive natural processes and their interrelations. The Santos Basin is a divergent marginal basin composed of three main physiographic provinces: its continental shelf, continental slope, and the São Paulo Plateau, whose dimensions, orientation, gradients, and relief interact with oceanographic, biological, and geological processes. This study geologically contextualized environmental data related to the Regional Environmental Characterization Project of the Santos Basin (hereinafter referred to as PCR-BS), especially regarding the regional physiographical architecture and deep-water geomorphological provinces as a background for the benthic communities and the distribution of several of its geochemical parameters. The integrated analysis of 3-D seismic and multibeam bathymetric data and its derivatives (gradient and edge maps) — calibrated by 108 piston cores — aimed to evaluate its seafloor physiographic features and geological processes. Edge detection seismic attributes enhance gradient contrast, which, in turn, can map innumerous medium- to small-scale geomorphic features (features solved in maps at a 1:1,000,000 scale or larger), such as canyons, channels, ravines, pockmark units, pockmark fields, lineaments, carbonate and coral mounds, salt-related features (crests, minibasins, and crestal grabens), scars and rugous relief associated with mass-transport deposits, and bottom current-related features (depressions and furrows). The slope and the gently dipping plateau (São Paulo Plateau) that follow it eastward show four geomorphological domains based on their direction, profile, average gradient, shelf break isobath, distribution of its medium-small scale geomorphic features, and the presence or absence of evaporites at subsurface. The central-northern São Paulo Plateau contains salt-related topographic elevations (salt diapirs and walls) and depressions (minibasins) that give a rough aspect to the seafloor relief.

Seismic oceanography can provide a two- or three-dimensional view of the water column thermocline structure at a vertical and horizontal resolution from the multi-channel seismic dataset. Several seismic imaging methods and techniques for seismic oceanography have been presented in previous research. In this study, we suggest a new formulation of the frequency-domain reverse-time migration method for seismic oceanography based on the analytic Green's function. For imaging thermocline structures in the water column from the seismic data, our proposed seismic reverse-time migration method uses the analytic Green's function for numerically calculating the forward- and backward-modeled wavefield rather than the wave propagation modeling in the conventional algorithm. The frequency-domain reverse-time migration with analytic Green's function does not require significant computational memory, resources, or a multifrontal direct solver to calculate the migration seismic images as like conventional reverse-time migration. The analytic Green's function in our reverse-time method makes it possible to provide a high-resolution seismic water column image with a meter-scale grid size, consisting of full-band frequency components for a modest cost and in a low-memory environment for computation. Our method was applied to multi-channel seismic data acquired in the Arctic Ocean and successfully constructed water column seismic images containing the oceanographic reflections caused by thermocline structures of the water mass. From the numerical test, we note that the oceanographic reflections of the migrated seismic images reflected the distribution of Arctic waters in a shallow depth and showed good correspondence with the anomalies of measured temperatures and calculated reflection coefficients from each XCDT profile. Our proposed method has been verified for field data application and accuracy of imaging performance.

This article, written by Special Publications Editor Adam Wilson, contains highlights of paper IPTC 18293, “Near-Surface Velocity Model of Dukhan Field From Multiphysics Survey To Enhance PSDM Seismic Imaging,” by K. Setiyono, Qatar Petroleum; S. Gallo, C. Boulanger, F. Bruere, F. Moreau, and B. Rondeleux, CGG; and J. Snow, Qatar Petroleum, prepared for the 2015 International Petroleum Technology Conference, Doha, Qatar, 7-9 December. The paper has not been peer reviewed. Dukhan field affords an opportunity to assess alternative near-surface velocity-modeling work flows that specifically integrate microgravity (MGR) and vertical electrical sounding (VES). A pilot study of the Dukhan seismic data demonstrates that integration of MGR/VES into prestack depth migration (PSDM) improves the fidelity of seismic images and results in an improved depth match. The resultant MGR/VES PSDM pilot model improved the velocity model between the surface and top Simsima formation on the basis of visible impact on the deeper structure obtained from imaging and better fit between the seismic depth horizons and well measured depths. Introduction The Dukhan field (Fig. 1) along the southwest coast of Qatar is an elongated anticline that has significant hydrocarbon reservoirs in the Jurassic Arab and deeper formations. In 2007, Qatar Petroleum invested in a new full-field, land and shallow-water 3D-seismic survey of the area. Despite considerable and continuous progress in seismic technology, obtaining an accurate, quantitative, well-focused seismic depth image at reservoir level remains challenging. In land seismic, a poor characterization of the near-surface geology is detrimental to the depth imaging of underlying structures. The proper knowledge of the near surface to correct for its perturbations onto the seismic wavefield remains one of the main challenges to be addressed. The near surface influences the kinematics as well as the dynamics of the source-generated seismic waves. Knowledge of the near surface to correct for its induced distortions is critical for a reliable use of seismic images, structures, and amplitude at reservoir level. The near-surface lateral and vertical velocity variations often are not accessible with seismic methods, even to the seismic refraction survey, because their depth of investigation is limited to the top of the shallowest hard layer. In addition, rapid lateral changes and large velocity contrasts are unfavorable to seismic refraction.