scholarly journals 3D seismic interpretation with deep learning: A brief introduction

2021 ◽  
Vol 40 (7) ◽  
pp. 524-532
Author(s):  
Thilo Wrona ◽  
Indranil Pan ◽  
Rebecca E. Bell ◽  
Robert L. Gawthorpe ◽  
Haakon Fossen ◽  
...  
Keyword(s):  
Internal Structure ◽  
Seismic Interpretation ◽  
Data Sets ◽  
3D Seismic ◽  
Deep Convolutional Neural Networks ◽  
Seismic Reflection Data ◽  
Reflection Data ◽  
Fundamental Goal ◽  
Seismic Volumes ◽  
Or Team

Understanding the internal structure of our planet is a fundamental goal of the earth sciences. As direct observations are restricted to surface outcrops and borehole cores, we rely on geophysical data to study the earth's interior. In particular, seismic reflection data showing acoustic images of the subsurface provide us with critical insights into sedimentary, tectonic, and magmatic systems. However, interpretations of these large 2D grids or 3D seismic volumes are time-consuming, even for a well-trained person or team. Here, we demonstrate how to automate and accelerate the analysis of these increasingly large seismic data sets with machine learning. We are able to perform typical seismic interpretation tasks such as mapping tectonic faults, salt bodies, and sedimentary horizons at high accuracy using deep convolutional neural networks. We share our workflows and scripts, encouraging users to apply our methods to similar problems. Our methodology is generic and flexible, allowing an easy adaptation without major changes. Once trained, these models can analyze large volumes of data within seconds, opening a new pathway to study the processes shaping the internal structure of our planet.

Hectometer-scale, shallow buried honeycomb-like structures on the continental shelf of the Otway Basin, southeastern Australia

2020 ◽  
Vol 8 (4) ◽  
pp. SR65-SR81 ◽  
Author(s):  
Yakufu Niyazi ◽  
Mark Warne ◽  
Daniel Ierodiaconou
Keyword(s):  
Seismic Reflection ◽  
Geospatial Analysis ◽  
Data Sets ◽  
Soft Sediment ◽  
3D Seismic ◽  
Burial Diagenesis ◽  
Seismic Reflection Data ◽  
Reflection Data ◽  
Southeastern Australia ◽  
2D And 3D

The Plio-Pleistocene Whalers Bluff Formation (WBF) of the offshore Otway Basin is composed of mixed siliciclastic-carbonate sediments. In seismic cross sections, this formation includes an interval that consists of higher amplitude seismic reflections that display alternating depressional ponds and raised ridges. This interval is shallowly buried and lies between 40 and 150 ms two-way traveltime below the present-day seafloor. In this study, we have used 2D and 3D seismic data sets in combination with the available shallow subsurface well logs to characterize the geomorphology and investigate the origin of these enigmatic features. The ponds are expressed as densely packed, circular to polygonal, and in some cases, hexagonal-shaped features in time-slice maps, and they closely resemble previously documented honeycomb structures. In our study area, the honeycomb-like structures (HS) are comprised of large (200–800 m diameter range) depressed ponds that are separated by narrow (approximately 20 m at the top) reticulate ridges. In total, these HS cover an area of 760 km2. Geospatial analysis shows that the ponds of HS, especially those in the northeast of the study area, are aligned along the northwest–southeast trend lines. There are several possible origins for the HS. The most probable mechanism is that the HS resulted from the bulk contraction of soft sediment, associated with shallow-burial diagenesis processes such as subaqueous dewatering of the fine-grained successions within the WBF. Interestingly, irregular furrows of various lengths on the seafloor correspond to the ridges of the HS, and we hypothesize that these furrows may have formed due to differential compaction of the underlying alternating ponds and ridges. Our results demonstrate the benefits of using seismic reflection data sets in combination with geospatial analysis to investigate the buried paleogeomorphologic features and their impact on the present-day seafloor physiography. Geological feature: Honeycomb-like, soft sediment deformation associated with shallow-burial diagenesis, Otway Basin, southeastern Australia Cross-section appearance: Alternating depressional ponds and raised ridges Map view appearance: Densely packed, oval to polygonal-shaped features Features with a similar appearance: Acquisition footprints, carbonate mounds/dissolution features, polygonal faults, pockmarks, opal-A to opal-CT transition Formation: Whalers Bluff Formation, offshore Otway Basin Age: Pliocene to recent Location: Continental shelf of the Otway Basin, southeastern Australia Data sets: 2D and 3D seismic reflection data, borehole data, from Geological Survey of Victoria, Australia Analysis tools: Interpretation and visualization (Petrel 2019 and DUG Insight, v.4.7, 2020), Geospatial analysis (ESRI‘s ArcMap 10.5)

Statistical approach to neural network imaging of karst systems in 3D seismic reflection data

2018 ◽  
Vol 6 (3) ◽  
pp. B15-B35
Author(s):  
Daniel R. Ebuna ◽  
Jared W. Kluesner ◽  
Kevin J. Cunningham ◽  
Joel H. Edwards
Keyword(s):  
Object Detection ◽  
Seismic Reflection ◽  
Carbonate Platform ◽  
Seismic Attributes ◽  
Seismic Interpretation ◽  
3D Seismic ◽  
Probability Models ◽  
Seismic Reflection Data ◽  
Reflection Data ◽  
Karst Systems

The current lack of a robust standardized technique for geophysical mapping of karst systems can be attributed to the complexity of the environment and prior technological limitations. Abrupt lateral variations in physical properties that are inherent to karst systems generate significant geophysical noise, challenging conventional seismic signal processing and interpretation. The application of neural networks (NNs) to multiattribute seismic interpretation can provide a semiautomated method for identifying and leveraging the nonlinear relationships exhibited among seismic attributes. The ambiguity generally associated with designing NNs for seismic object detection can be reduced via statistical analysis of the extracted attribute data. A data-driven approach to selecting the appropriate set of input seismic attributes, as well as the locations and suggested number of training examples, provides a more objective and computationally efficient method for identifying karst systems using reflection seismology. This statistically optimized NN technique is demonstrated using 3D seismic reflection data collected from the southeastern portion of the Florida carbonate platform. Several dimensionality reduction methods are applied, and the resulting karst probability models are evaluated relative to one another based on quantitative and qualitative criteria. Comparing the preferred model, using quadratic discriminant analysis, with previously available seismic object detection workflows demonstrates the karst-specific nature of the tool. Results suggest that the karst multiattribute workflow presented is capable of approximating the structural boundaries of karst systems with more accuracy and efficiency than a human counterpart or previously presented seismic interpretation schemes. This objective technique, using solely 3D seismic reflection data, is proposed as a practical approach to mapping karst systems for subsequent hydrogeologic modeling.

Submarine landslide origin of a tsunami at the Black Sea coast: Evidence based on swath bathymetry and 3D seismic reflection data

2021 ◽  
pp. 1-34
Author(s):  
Ilya Tishchenko ◽  
Gabor Tari ◽  
Mohammad Fallah ◽  
Jonathan Floodpage
Keyword(s):  
Numerical Modeling ◽  
Black Sea ◽  
Water Depth ◽  
Seismic Reflection ◽  
Submarine Landslide ◽  
Data Sets ◽  
3D Seismic ◽  
Seismic Reflection Data ◽  
Swath Bathymetry ◽  
Reflection Data

Tsunami waves were observed along the Bulgarian Black Sea coastline on May 7, 2007. The maximum rise and fall of the sea-level were 1.2 versus 2.0 m, respectively, with wave oscillations between 4-8 minutes. At first submarine landsliding and then later on atmospheric disturbance were suggested as the cause of the tsunami. Numerical modeling by Gusev et al. (2019), assuming a landslide displacing 30 to 60 million m3 material on the slope with a thickness range of over 20-40 m, could produce similar characteristics of the recorded tsunami. In the recent model the landslide initiated on the shelf at a water depth of 100 m with a runout of about 20 km into 1000 m water depth. Subsequent and recent numerical modeling suggested that the failure may have initiated on the slope, anywhere between 200 to 1500 m seafloor depth. The runout of the transported sediments in these latest model was at 1850 m water depth. Just a few years after the tsunami, OMV and its joint venture partners, TOTAL and Repsol, acquired modern deepwater data sets in the same area where the submarine landsliding was assumed to occur. These data sets included multi-beam swath bathymetry area and acquisition of a 3D reflection seismic data. These data sets offer a possibility to establish the presence of speculative submarine landslide responsible for the tsunami, with its geometry and nature. Our results provide direct evidence for the occurrence of large, non-seismic, catastrophic sediment failures along the Bulgarian coast. In this study we illustrate Quaternary submarine landslides on 3D seismic reflection data immediately below than the one responsible for the 2007 event, besides we also briefly point out the potential interpretation pitfall related to sediment waves and mass transport complexes.

Migration‐based traveltime waveform inversion of 2-D simple structures: A synthetic example

Geophysics ◽  
2001 ◽  
Vol 66 (3) ◽  
pp. 845-860 ◽  
Author(s):  
François Clément ◽  
Guy Chavent ◽  
Susana Gómez
Keyword(s):  
Least Squares ◽  
Waveform Inversion ◽  
Domain Of Attraction ◽  
Optimization Methods ◽  
Data Sets ◽  
Seismic Reflection Data ◽  
Prestack Migration ◽  
Reflection Data ◽  
Full Wave ◽  
The Time Domain

Migration‐based traveltime (MBTT) formulation provides algorithms for automatically determining background velocities from full‐waveform surface seismic reflection data using local optimization methods. In particular, it addresses the difficulty of the nonconvexity of the least‐squares data misfit function. The method consists of parameterizing the reflectivity in the time domain through a migration step and providing a multiscale representation for the smooth background velocity. We present an implementation of the MBTT approach for a 2-D finite‐difference (FD) full‐wave acoustic model. Numerical analysis on a 2-D synthetic example shows the ability of the method to find much more reliable estimates of both long and short wavelengths of the velocity than the classical least‐squares approach, even when starting from very poor initial guesses. This enlargement of the domain of attraction for the global minima of the least‐squares misfit has a price: each evaluation of the new objective function requires, besides the usual FD full‐wave forward modeling, an additional full‐wave prestack migration. Hence, the FD implementation of the MBTT approach presented in this paper is expected to provide a useful tool for the inversion of data sets of moderate size.

scholarly journals Improved Interpretation of Deep Seismic Reflection Data in Areas of Complex Geology Through Integration With Passive Seismic Data Sets

2018 ◽  
Vol 123 (12) ◽  
pp. 10,810-10,830
Author(s):  
Michael Dentith ◽  
Huaiyu Yuan ◽  
Ruth Elaine Murdie ◽  
Perla Pina-Varas ◽  
Simon P. Johnson ◽  
...  
Keyword(s):  
Seismic Data ◽  
Seismic Reflection ◽  
Data Sets ◽  
Seismic Reflection Data ◽  
Deep Seismic Reflection ◽  
Reflection Data ◽  
Passive Seismic ◽  
Complex Geology

Shallow seismic reflection study of a glaciated valley

Geophysics ◽  
1998 ◽  
Vol 63 (4) ◽  
pp. 1395-1407 ◽  
Author(s):  
Frank Büker ◽  
Alan G. Green ◽  
Heinrich Horstmeyer
Keyword(s):  
Seismic Reflection ◽  
High Amplitude ◽  
Data Sets ◽  
Seismic Section ◽  
Data Set ◽  
Seismic Reflection Data ◽  
Northern Switzerland ◽  
Reflection Data ◽  
Shallow Seismic ◽  
Glaciated Valley

Shallow seismic reflection data were recorded along two long (>1.6 km) intersecting profiles in the glaciated Suhre Valley of northern Switzerland. Appropriate choice of source and receiver parameters resulted in a high‐fold (36–48) data set with common midpoints every 1.25 m. As for many shallow seismic reflection data sets, upper portions of the shot gathers were contaminated with high‐amplitude, source‐generated noise (e.g., direct, refracted, guided, surface, and airwaves). Spectral balancing was effective in significantly increasing the strength of the reflected signals relative to the source‐generated noise, and application of carefully selected top mutes ensured guided phases were not misprocessed and misinterpreted as reflections. Resultant processed sections were characterized by distributions of distinct seismic reflection patterns or facies that were bounded by quasi‐continuous reflection zones. The uppermost reflection zone at 20 to 50 ms (∼15 to ∼40 m depth) originated from a boundary between glaciolacustrine clays/silts and underlying glacial sands/gravels (till) deposits. Of particular importance was the discovery that the deepest part of the valley floor appeared on the seismic section at traveltimes >180 ms (∼200 m), approximately twice as deep as expected. Constrained by information from boreholes adjacent to the profiles, the various seismic units were interpreted in terms of unconsolidated glacial, glaciofluvial, and glaciolacustrine sediments deposited during two principal phases of glaciation (Riss at >100 000 and Würm at ∼18 000 years before present).

Migration with reduced artifacts from internal multiple reflections

Geophysics ◽  
2020 ◽  
Vol 85 (4) ◽  
pp. A25-A29
Author(s):  
Lele Zhang
Keyword(s):  
Seismic Reflection ◽  
Computational Cost ◽  
Data Sets ◽  
Square Root ◽  
Multiple Reflections ◽  
Data Set ◽  
Seismic Reflection Data ◽  
Reflection Data ◽  
Recent Developments ◽  
Multiple Elimination

Migration of seismic reflection data leads to artifacts due to the presence of internal multiple reflections. Recent developments have shown that these artifacts can be avoided using Marchenko redatuming or Marchenko multiple elimination. These are powerful concepts, but their implementation comes at a considerable computational cost. We have derived a scheme to image the subsurface of the medium with significantly reduced computational cost and artifacts. This scheme is based on the projected Marchenko equations. The measured reflection response is required as input, and a data set with primary reflections and nonphysical primary reflections is created. Original and retrieved data sets are migrated, and the migration images are multiplied with each other, after which the square root is taken to give the artifact-reduced image. We showed the underlying theory and introduced the effectiveness of this scheme with a 2D numerical example.

Investigating fault continuity associated with geologic carbon storage planning in the Illinois Basin

2014 ◽  
Vol 2 (1) ◽  
pp. SA151-SA162 ◽  
Author(s):  
John H. McBride ◽  
R. William Keach ◽  
Eugene E. Wolfe ◽  
Hannes E. Leetaru ◽  
Clayton K. Chandler ◽  
...  
Keyword(s):  
Carbon Storage ◽  
Seismic Reflection ◽  
Oil Field ◽  
Illinois Basin ◽  
Storage Site ◽  
3D Seismic ◽  
Seismic Attribute ◽  
Seismic Reflection Data ◽  
Reflection Data ◽  
Attribute Analyses

Because the confinement of [Formula: see text] in a storage reservoir depends on a stratigraphically continuous set of seals to isolate the fluid in the reservoir, the detection of structural anomalies is critical for guiding any assessment of a potential subsurface carbon storage site. Employing a suite of 3D seismic attribute analyses (as opposed to relying upon a single attribute) maximizes the chances of identifying geologic anomalies or discontinuities (e.g., faults) that may affect the integrity of a seal that will confine the stored [Formula: see text] in the reservoir. The Illinois Basin, a major area for potential carbon storage, presents challenges for target assessment because geologic anomalies can be ambiguous and easily misinterpreted when using 2D seismic reflection data, or even 3D data, if only conventional display techniques are used. We procured a small 3D seismic reflection data set in the central part of the basin (Stewardson oil field) to experiment with different strategies for enhancing the appearance of discontinuities by integrating 3D seismic attribute analyses with conventional visualizations. Focusing on zones above and below the target interval of the Cambrian Mt. Simon Sandstone, we computed attribute traveltime slices (combined with vertical views) based on discontinuity computations, crossline-directed amplitude change, azimuth of the dip, shaded relief, and fault likelihood attributes. The results provided instructive examples of how discontinuities (e.g., subseismic scale faults) may be almost “invisible” on conventional displays but become detectable and mappable using an appropriate integration of 3D attributes. Strong discontinuities in underlying Precambrian basement rocks do not necessarily propagate upward into the target carbon storage interval. The origin of these discontinuities is uncertain, but we explored a possible strike-slip role that also explains the localization of a structural embayment developed in Lower Paleozoic strata above the basement discontinuities.

Sign in / Sign up

Export Citation Format

Share Document