Fast seismic horizon reconstruction based on local dip transformation

Zinck, Guillaume; Donias, Marc; Daniel, Jacques; Guillon, Sébastien; Lavialle, Olivier

doi:10.1016/j.jappgeo.2013.06.010

Cited by 22 publications

(8 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…RGT estimation from a seismic image is a more challenging task than is fault detection. Several types of methods including phase unwrapping (Stark, 2003;Wu and Zhong, 2012) and slope-based methods (Lomask et al, 2006;Fomel, 2010;Parks, 2010;Wu andHale, 2013, 2015;Zinck et al, 2013) have been proposed to perform automatic RGT estimation. However, estimating geologically consistent RGT values across faults, especially the complicated crossing faults, remains a highly challenging problem for all of the methods.…”

Section: Train a Cnn For Rgt And Seismic Horizonsmentioning

confidence: 99%

“…To assist in seismic horizon interpretation, various methods such as phase-unwrapping (Stark, 2003;Wu and Zhong, 2012), waveform classification (Figueiredo et al, 2007(Figueiredo et al, , 2014(Figueiredo et al, , 2015, slope-based methods (Lomask et al, 2006;Fomel, 2010;Parks, 2010;Wu andHale, 2013, 2015;Zinck et al, 2013;Monniron et al, 2016), and multigrid correlation (Wu and Fomel, 2018b) have been proposed to automate the horizon mapping from a seismic image. These methods, however, face a common challenge of dealing with complex structures such as intensive crossing faults and complicated folding, which requires a globally optimal correlation of all the structure elements in a seismic image.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Building realistic structure models to train convolutional neural networks for seismic structural interpretation

et al. 2020

View full text Add to dashboard Cite

Seismic structural interpretation involves highlighting and extracting faults and horizons that are apparent as geometric features in a seismic image. Although seismic image processing methods have been proposed to automate fault and horizon interpretation, each of which today still requires significant human effort. We improve automatic structural interpretation in seismic images by using convolutional neural networks (CNNs) that recently have shown excellent performances in detecting and extracting useful image features and objects. The main limitation of applying CNNs in seismic interpretation is the preparation of many training data sets and especially the corresponding geologic labels. Manually labeling geologic features in a seismic image is highly time-consuming and subjective, which often results in incompletely or inaccurately labeled training images. To solve this problem, we have developed a workflow to automatically build diverse structure models with realistic folding and faulting features. In this workflow, with some assumptions about typical folding and faulting patterns, we simulate structural features in a 3D model by using a set of parameters. By randomly choosing the parameters from some predefined ranges, we are able to automatically generate numerous structure models with realistic and diverse structural features. Based on these structure models with known structural information, we further automatically create numerous synthetic seismic images and the corresponding ground truth of structural labels to train CNNs for structural interpretation in field seismic images. Accurate results of structural interpretation in multiple field seismic images indicate that our workflow simulates realistic and generalized structure models from which the CNNs effectively learn to recognize real structures in field images.

show abstract

Section: Train a Cnn For Rgt And Seismic Horizonsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Building realistic structure models to train convolutional neural networks for seismic structural interpretation

et al. 2020

View full text Add to dashboard Cite

show abstract

“…A more stable way of slope-based horizon picking is to compute a weighted least-squares fit of the horizon slopes with the preesti- (Lomask et al, 2006;Wu andHale, 2013, 2015;Zinck et al, 2013). In cases in which the estimated reflection slopes are highly noisy, we can further impose smooth regularizations on the horizon and build the following equations for the 2D slope-based horizon extraction:…”

Section: Least-squares Horizons With Local Slopesmentioning

confidence: 99%

“…A straightforward slope-based horizon extraction method is to pick horizons by starting from seed points and then recursively following the local slopes (Fomel, 2010). Most slope-based methods (e.g., Lomask et al, 2006;Parks, 2010;Wu andHale, 2013, 2015;Zinck et al, 2013;Monniron et al, 2016) compute horizons by fitting the slopes of horizons with the local reflection slopes in the least-squares sense. By recursively following local slopes or fitting local slopes in the least-squares sense, the computed horizons can correctly follow laterally continuous reflections but they often fail to track corresponding reflections across faults because the local slopes cannot correctly correlate reflections across faults.…”

Section: Introductionmentioning

confidence: 99%

Least-squares horizons with local slopes and multigrid correlations

Fomel

2018

GEOPHYSICS

View full text Add to dashboard Cite

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.

show abstract

“…Dip information can be computed everywhere in the seismic volume [3] and used for flattening the seismic image [4], for stratal slicing or for guiding automated interpretations by correlating neighboring seismic traces [5] [6]. Local correlation between traces can also be exploited for simultaneously tracking multiple horizon patches and identifying those that should be merged to form large scale horizon surfaces [7].…”

Section: Automated Seismic Interpretationmentioning

confidence: 99%

Construction of Structurally and Stratigraphically Consistent Structural Models Using the Volume-Based Modelling Technology: Applications to an Australian Dataset

Souche¹,

Iskenova²,

Lepage³

et al. 2014

International Petroleum Technology Conference

View full text Add to dashboard Cite

This paper describes new techniques and algorithms for building consistent and accurate structural models of the subsurface in structurally and stratigraphically complex areas, and proposes an innovative workflow for iteratively building large-scale models in presence of sparse or noisy data.Contrary to traditional 3D modeling approaches, which consist in modeling first geological interfaces independently from each other before gluing them together to form a watertight representation of geological layers, the proposed technique creates first a 3D unstructured mesh representing the volume of interest, which is then sliced and diced by geological horizons and faults. It is based on a global interpolation method in which a scalar property representing the relative stratigraphic age of the formations is interpolated everywhere in the volume of interest. In the process, the geometry of each horizon is effectively constrained by interpretation data of all other horizons belonging to the same conformable sequence.The interpolation is practically insensitive to the complexity of the fault network (X faults, Y faults, thrusts) and yields a high resolution, 3D representation of geological layers, allowing the extraction of main horizons based on seismic input and of closely spaced intermediate horizon surfaces defined by well data.The proposed technique ensures the preservation of four key quality features:1. Geometrical accuracy: surfaces are as smooth as possible while fitting all input data geometry. 2. Topological correctness: all model contacts are perfect and watertight. 3. Stratigraphic consistency: horizon surfaces cannot cross each other; layering and truncation patterns dictated by the input stratigraphic column are honored everywhere in the model. 4. Structural coherency: variations of fault displacement should be consistent with their size; layer thickness changes should be as regular as possible.These benefits are illustrated through the step-by-step construction and refinement of a large scale geological model located offshore Western Australia. It is also demonstrated that this technique and methodology can successfully be applied to a large variety of structural and stratigraphic contexts.

show abstract

Fast seismic horizon reconstruction based on local dip transformation

Cited by 22 publications

References 12 publications

Building realistic structure models to train convolutional neural networks for seismic structural interpretation

Building realistic structure models to train convolutional neural networks for seismic structural interpretation

Least-squares horizons with local slopes and multigrid correlations

Construction of Structurally and Stratigraphically Consistent Structural Models Using the Volume-Based Modelling Technology: Applications to an Australian Dataset

Contact Info

Product

Resources

About