Horizon Extraction Research Articles

Generating more constrained seeds for seismic horizon extraction

Enhancing accuracy remains the main challenge for the current horizon extraction algorithms. Adding more manually interpreted seeds helps to improve the accuracy of horizon extraction. We intend to generate more seeds prior to the 3D horizon extraction based on seeds that are manually interpreted on the coarse grid of the seismic survey. First, we compute four horizons for the same seismic sequence boundary using deep learning and constrained least-squares algorithms under the constraints of manually interpreted horizon seeds. Next, we generate more seeds by checking the consistency among those four horizons for each seismic trace, and this process is implemented in the same way as the tying loops of manual horizon interpretation. Finally, we extract horizons using constrained least-squares algorithms with the newly computed seeds. The applications of two real seismic surveys demonstrate that our method can generate horizons with higher accuracy when compared with the horizons computed with only manually interpreted seeds.

Stratal surfaces honoring seismic structures and interpreted geologic time surfaces

Seismic horizons play a significant role in reservoir model construction and sedimentary facies interpretation, providing crucial low-frequency constraints for seismic inversion. In basin and regional interpretations, the assumption that seismic reflections represent a stratigraphic surface with constant geologic time is significant for guiding seismic interpretation. This assumption may fail when applied to local reservoir scales due to common geologic time transgressions of a particular event in regular wavelet frequency. There will be inconsistencies between seismic events and stratigraphic surfaces. To address this issue and obtain relatively accurate stratal interpretations, we develop a hybrid horizon extraction method honoring seismic structures and time-stratigraphic frameworks, in which seismic reflection structures provide local details and interpreted geologic time surfaces offer critical constraints. First, we develop concepts and a workflow using a realistic outcrop model. We develop a new geology-guided structure tensor by fitting a gradient vector of seismic images and geologic time surfaces. We also consider existing geologic conditions, such as unconformities, and fuse them into our method to calculate accurate slopes and generate reliable relative geologic time images at a fine scale, followed by making slices. Further, we extend our method to 3D seismic data volumes. Our experiments, conducted using simulated and field data, show the superiority and accuracy of our hybrid method compared with the slope-based and stratal slicing methods. These results highlight the potential for applying our method to fine-scale subsurface modeling.

Horizon extraction using absolute seismic phase obtained by unwrapping techniques

Automatic detection of horizons and other geological structures of interest is often employed as a preliminary step in seismic interpretation. Although necessary, the identification and extraction of horizons can be a laborious and time-consuming step for the interpreter. Therefore, we present a semi-automatic horizon extraction algorithm based on the absolute phase of seismic data obtained through phase unwrapping techniques. We assess the algorithm accuracy and robustness in several examples, including noisy data sets. In addition, we present comparative results between two phase unwrapping methods, showing their robustness when associated with seismic horizon extraction. We compare the results obtained by our methods with the results obtained by seismic horizon extraction algorithms available in the literature.

ClinoformNet-1.0: stratigraphic forward modeling and deep learning for seismic clinoform delineation

Abstract. Deep learning has been widely used for various kinds of data-mining tasks but not much for seismic stratigraphic interpretation due to the lack of labeled training datasets. We present a workflow to automatically generate numerous synthetic training datasets and take the seismic clinoform delineation as an example to demonstrate the effectiveness of using the synthetic datasets for training. In this workflow, we first perform stochastic stratigraphic forward modeling to generate numerous stratigraphic models of clinoform layers and corresponding porosity properties by randomly but properly choosing initial topographies, sea level curves, and thermal subsidence curves. We then convert the simulated stratigraphic models into impedance models by using the velocity–porosity relationship. We further simulate synthetic seismic data by convolving reflectivity models (converted from impedance models) with Ricker wavelets (with various peak frequencies) and adding real noise extracted from field seismic data. In this way, we automatically generate a total of 3000 diverse synthetic seismic datasets and the corresponding stratigraphic labels such as relative geologic time models and facies of clinoforms, which are all made publicly available. We use these synthetic datasets to train a modified encoder–decoder deep neural network for clinoform delineation in seismic data. Within the network, we apply a preconditioning process of structure-oriented smoothing to the feature maps of the decoder neural layers, which is helpful to avoid generating holes or outliers in the final output of clinoform delineation. Multiple 2D and 3D synthetic and field examples demonstrate that the network, trained with only synthetic datasets, works well to delineate clinoforms in seismic data with high accuracy and efficiency. Our workflow can be easily extended for other seismic stratigraphic interpretation tasks such as sequence boundary identification, synchronous horizon extraction, and shoreline trajectory identification.

A Multi-Task Learning Method for Relative Geologic Time, Horizons, and Faults With Prior Information and Transformer

Horizon extraction and fault detection are essential in seismic interpretation and closely related to each other. Most existing methods tend to deal with these two tasks independently, and may not work well in interpreting seismic images with complex geologic structures. We propose a multi-task learning (MTL) network with two branches to extract all horizons and detect faults simultaneously by estimating a relative geologic time (RGT) map as well as computing a fault map. These two branches share training datasets, feature maps, and network parameters during the training. The RGT estimation branch, constructed with a transformer architecture, is more lightweight compared to previous CNN methods but provides a larger and structure-oriented receptive field to adaptively capture global structural information for estimating a globally optimal RGT map. The fault detection branch is a simple convolutional neural network (CNN) which merges feature maps shared by the transformer and the derivatives of the estimated RGT to compute a fault map. The fault detection branch provides boundary control for the RGT estimation branch while the later provides global constraints for the former to improve its robustness to noise. Note that our RGT estimation by globally fitting all structures in a seismic image is a volumetric horizon interpretation method with which we are able to obtain a whole volume of horizons, all at once, by simply extracting contours of the RGT map. In our method, we further enable convenient human interactions by integrating manually interpreted horizons (or horizon segments) into the network, which imposes expert knowledge on the network to estimate reasonable RGT results from seismic images with complex fault systems, unconformities, and poor data quality. Moreover, when using 3D horizons as constraints, we are able to decompose the computationaly expensive 3D RGT estimation from a seismic volume into independently parallel 2D estimations slice by slice and combine them to obtain a laterally consistent 3D result.

Automatic Extraction of Seismic Data Horizon Across Faults

Most of the traditional horizon interpretation methods are based on the dip of the seismic data. Since the dip does not correctly reflect the discontinuous reflection characteristics of the events, these methods usually cannot extract the horizons containing faults well. Meanwhile, the field seismic data observed in complex areas are generally with low signal-to-noise ratio (SNR), which makes the extraction of horizons difficult. To address these problems, an automatic horizon extraction approach under fault control is investigated in this article. First, the horizon dip is obtained by using plane wave destruction (PWD), and the dip information can be utilized to control the overall trend of the horizon curve. Based on the dip information, we can also calculate the weight coefficients to characterize the faults. The fault characterization results can better guide the horizon extraction. Then, the dynamic time warping (DTW) is used for phase matching to further obtain the similar slope and describe the details of the horizontal curve. Finally, the horizon extraction equation is developed with the dip and phase information as the guide and solved iteratively to obtain the final horizon curve. The effectiveness of the presented approach is tested by both synthetic and field data. The results demonstrate that the presented approach can accurately pick the horizon information across faults and is more stable in the condition of low SNR compared with previous methods.

Seismic Horizon Identification Using Semi-Supervised Learning With Virtual Adversarial Training

Seismic horizon extraction is important for subsurface structure interpretation and reservoir modeling. Recently developed learning-based seismic interpretation methods show great success in the case of sufficient labeled data but may fail when the labels are limited. Thus, we propose a simple but effective network for seismic horizon identification using only a limited number of labels. To avoid overfitting in the training, we introduce the mechanism of semi-supervised learning (SSL) with virtual adversarial training (VAT). With several seed points, the method can provide a good prediction and suggest regions lacking control points. By adding several seed points in these suggested regions, the performance of the network can be further improved, which can be regarded as an interactive way. In addition, iteratively retraining the network by using the previous high-confidence prediction can further refine the horizon identification. We, finally, compute a full horizon surface without holes and outliers by optimally fitting the horizon points identified by our SSL and reflection slopes estimated from the seismic amplitude image. Applications to two field datasets show our method is superior to conventional methods in picking a seismic horizon with significant waveform variations or across complex discontinuities, such as faults.

Seismic horizon extraction with dynamic programming

Horizon picking is a fundamental and crucial step for seismic interpretation, but it remains a time-consuming task. Although various automatic methods have been developed to extract horizons in seismic images, most of them may fail to pick horizons across discontinuities such as faults and noise. To obtain more accurate horizons, we have developed a dynamic programming algorithm to efficiently refine manually or automatically extracted horizons so that they can more accurately track reflectors across discontinuities, follow consistent phases, and reveal more geologic details. In this method, we first compute an initial horizon using an automatic method, manual picking, or interpolation with several control points. The initial horizon may not be accurate, and it only needs to follow the general trend of the target horizon. Then, we extract a subvolume of amplitudes centered at the initial horizon and meanwhile flatten the subvolume according to the initial horizon. Finally, we use the dynamic programming to efficiently pick the globally optimal path that passes through the global maximum or minimum amplitudes in the subvolume. As a result, we are able to refine the initial horizon to a more accurate horizon that follows consistent amplitude peaks, troughs, or zero crossings. Because our method does not strictly depend on the initial horizon, we prefer to directly interpolate an initial horizon from a limited number of control points, which is computationally more efficient than automatically or manually picking an initial horizon. In addition, our method is convenient to interactively implement to update the horizon while editing or moving the control points. More importantly, these control points are not required to be exactly placed on the target horizon, which makes the human interaction highly convenient and efficient. We develop our method with multiple 2D and 3D field examples that are complicated by noise, faults, and salt bodies.

A High-resolution GPR horizon extraction method based on local reflection and global correlation

Horizon extraction is a very basic and important part of ground penetrating radar (GPR) interpretation. Most GPR horizon extraction methods are based on local reflection of the GPR event such as the trace correlation algorithm, and hidden Markov model algorithm. These methods use only local structural characteristics of the GPR data and often fail to pick the layer correctly across discontinuous caused by faults and noise. In order to overcome shortcomings of the above methods, a novel method based on gradient structure tensor algorithm (GST) and dynamic time warping algorithm (DTW) is used in this paper to extract the horizon accurately which has been proved to be effective in the complex seismic data. GPR electromagnetic wave and seismic wave have certain similarity in kinematics and dynamics, we apply this method to realistic GPR data and prove its effectiveness.

Roadbed defect detection from ground penetrating radar B-scan data using Faster RCNN

Ground penetrating radar (GPR) is the main technical method to detect roadbed subgrade defect. The recognition of subgrade defect is still mainly based on manual interpretation, which requires high professional knowledge of interpreters, resulting in the demand for automatic detection technology. In this paper, a solution for automatic detection of roadbed defect by implementing Faster RCNN with GPR system is presented. We simulated 30000 roadbed defect GPR B-scan data by simulation software gprMax, labeled them appropriately and automatically. Specifically, Faster RCNN was chosen, as a compromise between accuracy and ease of comparison. Preliminary detection results show that the AP (Average Precision) is 0.8067, proving that our simulation for defect is reasonable and reliable. And the Faster RCNN trained on the simulation dataset without any actual data also has excellent performance on the actual GPR data. Our method of detecting defects automatically with CNN can be easily generalized to other GPR tasks, e.g., detecting pipe, horizon extraction. It’s the biggest open-source GPR B-scan dataset as far as we know. Our simulation dataset and trained model will be made available.

Ground-truth uncertainty-aware metrics for machine learning applications on seismic image interpretation: Application to faults and horizon extraction

We propose new metrics to measure the performance of a deep learning model applied to seismic interpretation tasks such as fault and horizon extraction. Faults and horizons are thin geologic boundaries (1 pixel thick on the image) for which a small prediction error could lead to inappropriately large variations in common metrics (precision, recall, and intersection over union). Through two examples, we show how classical metrics could fail to indicate the true quality of fault or horizon extraction. Measuring the accuracy of reconstruction of thin objects or boundaries requires introducing a tolerance distance between ground truth and prediction images to manage the uncertainties inherent in their delineation. We therefore adapt our metrics by introducing a tolerance function and illustrate their ability to manage uncertainties in seismic interpretation. We compare classical and new metrics through different examples and demonstrate the robustness of our metrics. Finally, we show on a 3D West African data set how our metrics are used to tune an optimal deep learning model.

Horizon extraction using ordered clustering on a directed and colored graph

Extracting horizons from a seismic image has been playing an important role in seismic interpretation. However, how to fully use global-level information contained in the seismic images such as the order of horizon sequences is not well-studied in existing works. To address this issue, we have developed a novel method based on a directed and colored graph, which encodes effective context information for horizon extraction. Following the commonly used framework, which generates horizon patches and then groups them into horizons, we first build a directed and colored graph by representing horizon patches as vertices. In addition, edges in the graph encode the relative spatial positions of horizon patches. This graph explicitly captures the geologic context, which guides the grouping of the horizon patches. Then, we conduct premerging to group horizon patches through matching some predefined subgraph patterns that are designed to capture some special spatial distributions of horizon patches. Finally, we have developed an ordered clustering method to group the rest of the horizon patches into horizons based on the pairwise similarities of horizon patches while preserving geologic reasonability. Experiments on real seismic data indicate that our method can outperform the autotracking approach solely based on the similarity of local waveforms and can correctly pick the horizons even across the fault without any crossing, which demonstrates the effectiveness of exploring the spatial information, i.e., the order of horizon sequences and special spatial distribution of horizon patches.

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.

Least-squares horizons with local slopes and multigrid correlations

Most seismic horizon extraction methods are based on seismic local reflection slopes that locally follow seismic structural features. However, these methods often fail to correctly track horizons across discontinuities such as faults and noise because the local slopes can only correctly follow laterally continuous reflections. In addition, seismic amplitude or phase information is not used in these methods to compute horizons that follow a consistent phase (e.g., peaks or troughs). To solve these problems, we have developed a novel method to compute horizons that globally fit the local slopes and multigrid correlations of seismic traces. In this method, we first estimate local reflection slopes by using structure tensors and compute laterally multigrid slopes by using dynamic time warping (DTW) to correlate seismic traces within multiple laterally coarse grids. These coarse-grid slopes can correctly correlate reflections that may be significantly dislocated by faults or other discontinuous structures. Then, we compute a horizon by fitting, in the least-squares sense, the slopes of the horizon with the local reflection slopes and multigrid slopes or correlations computed by DTW. In this least-squares system, the local slopes on the fine grid and the multiple coarse-grid slopes will fit a consistent horizon in areas without lateral discontinuities. Across laterally discontinuous areas where the local slopes fail to correctly correlate reflections and mislead the horizon extraction, the coarse-grid slopes will help to find the corresponding reflections and correct the horizon extraction. In addition, the multigrid correlations or slopes computed by dynamic warping can also assist in computing phase-consistent horizons. We apply the proposed horizon extraction method to multiple 2D and 3D examples and obtain accurate horizons that follow consistent phases and correctly track reflections across faults.

Automatically interpreting all faults, unconformities, and horizons from 3D seismic images

Extracting fault, unconformity, and horizon surfaces from a seismic image is useful for interpretation of geologic structures and stratigraphic features. Although others automate the extraction of each type of these surfaces to some extent, it is difficult to automatically interpret a seismic image with all three types of surfaces because they could intersect with each other. For example, horizons can be especially difficult to extract from a seismic image complicated by faults and unconformities because a horizon surface can be dislocated at faults and terminated at unconformities. We have proposed a processing procedure to automatically extract all the faults, unconformities, and horizon surfaces from a 3D seismic image. In our processing, we first extracted fault surfaces, estimated fault slips, and undid the faulting in the seismic image. Then, we extracted unconformities from the unfaulted image with continuous reflectors across faults. Finally, we used the unconformities as constraints for image flattening and horizon extraction. Most of the processing was image processing or array processing and was achieved by efficiently solving partial differential equations. We used a 3D real example with faults and unconformities to demonstrate the entire image processing.

Moving faults while unfaulting 3D seismic images

Unfaulting seismic images to correlate seismic reflectors across faults is helpful in seismic interpretation and is useful for seismic horizon extraction. Methods for unfaulting typically assume that fault geometries need not change during unfaulting. However, for seismic images containing multiple faults and, especially, intersecting faults, this assumption often results in unnecessary distortions in unfaulted images. We have developed two methods to compute vector shifts that simultaneously move fault blocks and the faults themselves to obtain an unfaulted image with minimal distortions. For both methods, we have used estimated fault positions and slip vectors to construct unfaulting equations for image samples alongside faults, and we have constructed simple partial differential equations for samples away from faults. We have solved these two different kinds of equations simultaneously to compute unfaulting vector shifts that are continuous everywhere except at faults. We have tested both methods on a synthetic seismic image containing normal, reverse, and intersecting faults. We also have applied one of the methods to a real 3D seismic image complicated by numerous intersecting faults.

Vision-Based Horizon Extraction Method under Kalman Filter Framework

As the demands of UAV’s visual navigation technology, we bring out a new horizon extraction method in this paper. Firstly, we propose a horizon extraction algorithm for single image. We employ dark channel in single image to avoid the interferences from clouds and fogs, and use Sobel operator extract edges, among which we can extract the true horizon through an algorithm mentioned in Paragraph II. Secondly, we propose a horizon extraction algorithm for video streaming under Kalman Filter (KF) framework based on the horizon extraction algorism for single image. The position of horizon in each frame will be estimated by using the priori horizon positions under KF framework at first, and a search neighborhood will be determined around the estimated position, in which we can get the true position of the horizon through a certain search algorithm. Simulations and analyses are carried out with aerial video streaming, the results show that such algorithms work well on those videos with noise, clouds and fogs, while the time overhead decrease by about 50% than traditional algorithms.

SeiVis: An Interactive Visual Subsurface Modeling Application

The most important resources to fulfill today's energy demands are fossil fuels, such as oil and natural gas. When exploiting hydrocarbon reservoirs, a detailed and credible model of the subsurface structures is crucial in order to minimize economic and ecological risks. Creating such a model is an inverse problem: reconstructing structures from measured reflection seismics. The major challenge here is twofold: First, the structures in highly ambiguous seismic data are interpreted in the time domain. Second, a velocity model has to be built from this interpretation to match the model to depth measurements from wells. If it is not possible to obtain a match at all positions, the interpretation has to be updated, going back to the first step. This results in a lengthy back and forth between the different steps, or in an unphysical velocity model in many cases. This paper presents a novel, integrated approach to interactively creating subsurface models from reflection seismics. It integrates the interpretation of the seismic data using an interactive horizon extraction technique based on piecewise global optimization with velocity modeling. Computing and visualizing the effects of changes to the interpretation and velocity model on the depth-converted model on the fly enables an integrated feedback loop that enables a completely new connection of the seismic data in time domain and well data in depth domain. Using a novel joint time/depth visualization, depicting side-by-side views of the original and the resulting depth-converted data, domain experts can directly fit their interpretation in time domain to spatial ground truth data. We have conducted a domain expert evaluation, which illustrates that the presented workflow enables the creation of exact subsurface models much more rapidly than previous approaches.

Factors influencing fractures networks within Permian shale intervals in the Cooper Basin, South Australia

The future success of both enhanced (engineered) geothermal systems and shale gas production is reliant on the development of reservoir stimulation strategies that suit the local geo-mechanical conditions of the prospects. The orientation and nature of the in-situ stress field and pre-existing natural fracture networks in the reservoir are among the critical parameters that will control the quality of the stimulation program. This study provides a detailed investigation into the nature and origin of natural fractures in the area covered by the Moomba–Big Lake 3D seismic survey, in the southwest termination of the Nappamerri Trough of the Cooper Basin. These fractures are imaged by both borehole image logs and complex multi-traces seismic attributes (e.g. dip-steered most positive curvature and dip-steered similarity), are pervasive throughout the cube, and exhibit a relatively consistent northwest–southeast orientation. Horizon extraction of the seismic attributes reveal a strong variation in the spatial distribution of the fractures. In the acreage of interest, fracture density is at its highest in the vicinity of faults and on top of tight antiforms. This study also suggests a good correlation between high fracture density and high gamma ray values. The correlation between high fracture density and shale content is somewhat counterintuitive, as shale is expected to have a higher tensile and compressive strengths at shallow depths and typically contain fewer fractures (Lin, 1983). At large depths, however—and due to sandstone diagenesis and cementation—shale has lower tensile and compressive strength than sandstone and is expected to be more fractured (Lin, 1983). A similar correlation has been noted in other Australian Basins (e.g. Northern Perth Basin). Diagenetic effects, pore pressure, stiffness, variations in tensile versus compressive strength of the shale and the sandstone may explain this disparity.

Effects of elevated atmospheric carbon dioxide on amino acid and NH4+‐N cycling in a temperate pine ecosystem

AbstractRising atmospheric carbon dioxide (CO2) is expected to increase forest productivity, resulting in greater carbon (C) storage in forest ecosystems. Because elevated atmospheric CO2 does not increase nitrogen (N) use efficiency in many forest tree species, additional N inputs will be required to sustain increased net primary productivity (NPP) under elevated atmospheric CO2. We investigated the importance of free amino acids (AAs) as a source for forest N uptake at the Duke Forest Free Air CO2 Enrichment (FACE) site, comparing its importance with that of better‐studied inorganic N sources. Potential proteolytic enzyme activity was monitored seasonally, and individual AA concentrations were measured in organic horizon extracts. Potential free AA production in soils ranged from 190 to 690 nmol N g−1 h−1 and was greater than potential rates of soil NH4+ production. Because of this high potential rate of organic N production, we determined (1) whether intact AA uptake occurs by Pinus taeda L., the dominant tree species at the FACE site, (2) if the rate of cycling of AAs is comparable with that of ammonium (NH4+), and (3) if atmospheric CO2 concentration alters the aforementioned N cycling processes. A field experiment using universally labeled ammonium (15NH4+) and alanine (13C3H715NO2) demonstrated that 15N is more readily taken up by plants and heterotrophic microorganisms as NH4+. Pine roots and microbes take up on average 2.4 and two times as much NH4+ 15N compared with alanine 15N 1 week after tracer application. N cycling through soil pools was similar for alanine and NH4+, with the greatest 15N tracer recovery in soil organic matter, followed by microbial biomass, dissolved organic N, extractable NH4+, and fine roots. Stoichiometric analyses of 13C and 15N uptake demonstrated that both plants and soil microorganisms take up alanine directly, with a 13C : 15N ratio of 3.3 : 1 in fine roots and 1.5 : 1 in microbial biomass. Our results suggest that intact AA (alanine) uptake contributes substantially to plant N uptake in loblolly pine forests. However, we found no evidence supporting increased recovery of free AAs in fine roots under elevated CO2, suggesting plants will need to acquire additional N via other mechanisms, such as increased root exploration or increased N use efficiency.