Post-Stack Seismic Data Conditioning and Attributes - Geophysics
Seismic Attributes – from Interactive Interpretation to Machine Learning (5-day)
A seismic attribute is any measure of seismic data that helps us better visualize or quantify features of interpretation interest. Seismic attributes fall into two broad categories – those that help us quantify the morphological component of seismic data and those that help us quantify the reflectivity component of seismic data. The morphological attributes help us extract information on reflector dip, azimuth, and terminations, which can in turn be related to faults, channels, fractures, diapirs, and carbonate buildups. The reflectivity attributes help us extract information on reflector amplitude, waveform, and variation with illumination angle, which can in turn be related to lithology, reservoir thickness, and the presence of hydrocarbons.
A human interpreter integrates the images provided by seismic attributes with well log and production data using an appropriate tectonic, stratigraphic, and/or diagenetic geologic models to infer lithology, porosity, and fractures and other features of interest. Because attributes quantitively measure the smaller-scale features and patterns seen by human interpreters, attributes provide critical input to modern machine learning analysis tools.
In this course, we will gain an intuitive understanding of the kinds of seismic features that can be identified by 3D seismic attributes, the sensitivity of seismic attributes to seismic acquisition and processing, and how ‘independent’ seismic attributes can be coupled through geology. Having learned the properties that each attribute measures, we will be able to choose appropriate attribute candidates for machine learning analysis to predict seismic facies, and when abundant well control exists, areas of better production, and zones of more costly drilling.
Although seismic attributes are well implemented in all of the larger interpretation software packages, licenses to such software is not generally available for use in public professional society courses. Geoscientists between jobs or engineering, data analytics, and other professionals wanting to move into geosciences may not have access to any interpretation software. To address this shortcoming participants will be provided an evaluation license of the AASPI software (described under http://mcee.ou.edu/aaspi/documentation.html ) for their desktop or laptop to carry out the class exercises, and ideally, to apply to their own data volumes.
|Module name||Topics addressed|
|Introduction||An overview of how seismic attributes fit within modern interpretation workflows.|
|Complex trace, horizon, and formation attributes||Theory, definition, and limitations of attribute based on the analytic (or complex trace) such as envelope and instantaneous frequency. Definition and use of attributes computed from a horizon, such as dip magnitude and horizon-based curvature as well as formation attributes computed between horizons, such as RMS amplitude and thickness.|
|Color and multiattribute display||Definition and interrelationship between RGB, CMY, and HLS color models. Best practices for multiattribute display.|
|Types of attribute displays||Definition and use of horizon slices, phantom horizon slices, stratal slices, Wheeler slices, and geobodies.|
|Spectral decomposition and thin-bed tuning||Theory, workflows, and advantages of the three most commonly used spectral decomposition algorithms (DFT, CWT, and matching pursuit). Their use in mapping “tuned” geologic features that fall at or below seismic resolution.|
|Spectral decomposition emerging technologies||The use of spectral components in bandwidth extension, Q estimation, and phase discontinuity mapping of unconformities.|
|Geometric attributes that map reflector configuration||A summary of volumetric dip and azimuth, curvature, reflector shapes, aberrancy, and reflector convergence (or nonparallelism).|
|Geometric attributes that map continuity and textures||A summary of coherence, multispectral coherence, amplitude or energy gradients and curvature, and gray level coocurrence matrix texture attributes.|
|Multispectral, multioffset, and multiazimuth coherence||The value of the separate and combined analysis of coherence and other attributes computed on separate and/or multiple input volumes|
|Attribute expression of tectonic deformation||Attribute expression of faulting and folding as seen on post stack volumes by coherence, curvature, and reflector rotation.|
|Attribute expression of clastic depositional environments||Attribute expression of fluvial/deltaic and deepwater systems as seen on post stack volumes by spectral decomposition, coherence, curvature, and refector convergence attributes. Attribute expression of differential compaction.|
|Attribute expression of carbonate deposition environments||Attribute expression of carbonate buildups and diagenesis as seen on post stack volumes by coherence, curvature, and texture attributes. Attribute expression of karst terrains.|
|Attribute expression of shallow stratigraphy and drilling hazards||Attribute expression of mass transport complexes, glide tracks, outrunner blocks, pock marks, glacial keel marks,and shale “dewatering” (syneresis) features, many of which when gas- or water-charged may become drilling hazards.|
|Attribute expression of igneous instrusion, extrusions, and basement||Attribute expression of volcanic mounds, sills, fractured basement, and lacoliths which can serve as or give rise to reservoirs. Impact of overlying igneous rocks on seismic data quality.|
|Seismic chronostratigraphy||An overview of workflows that generate 100s of horizons within a given formation or stratigraphic package; the use of Wheeler diagrams and simple thickness and RMS attributes to map areas of greater accomodation and deposition vs erosion.|
|Impact of acquisition and processing on seismic attributes||Value of long-offset, wide-azimuth, and dense seismic surveys in seismic data quality and attribute analysis.|
|Seismc attributes on depth-migrated data volumes||Peculiarities specific to computing seismic attributes from depth migrated volumes|
|Common pitfalls in seismic attribute interpretation||Review of artifacts including velocity pull-up/push-down, acquisition footprint, aligned reflectors across faults that give rise to mispicks, and how these same features appear on attribute volumes|
|Direct and indirect estimates of fractures and horizontal stress||Use of curvature, impedance, and seismic anisotropy to map the orientation and intensity of natural fractures and/or horizontal stress. Calibration with lidar data and image logs. Measurement and interpretation of azimuthal anisotropy volumes.|
|Poststack seismic data conditioning||Spectral balancing, structure-oriented filtering and footprint suppression of poststack data volumes.|
|Prestack seismic data conditioning||Prestack structure-oriented filtering, nonhyperbolic moveout, and correction of NMO/migration stretch. Preconditioned least-squares migration and 5D interpolation.|
|Poststack impedance inversion||A hierarchal overview of inversion – emphasizing the assumptions and interpreter input to each process.|
|Prestack impedance inversion||Acoustic vs. elastic impedance inversion. Sequential vs. simultaneous inversion. The importance of using the misfit between measured and synthetic (predicted) gathers as a measure of confidence and as input to machine learning|
|Image enhancement and object detection||Algorithms that enhance faults and channels to generate computer “objects”. Lay- person’s explanation of modren ant-tracking, skeletonization, and level set algorithms that indicate the future of computer-assisted seismic interpretation.|
|Visual decision making and interactive multiattribute analysis||Review of multiattribute display, crossplotting, and geobodies. Principal component and independent component analysis dimensionality reduction techniques.|
|Statistical multiattribute analysis||Fundamentals of geostatistics, including kriging, kriging with external drift, colocated cokriging, sequential Gaussian simulation, and geostatistical impedance inversion.|
|Unsupervised multiattribute classification||Clustering algorithms including k-means, self-organizing maps (e.g. Stratimagic’s “waveform classification”) and generative topographic maps.|
|Supervised multiattribute classification||Artificial neural networks, probabilistic neural networks, random forest desicison trees, and support vector machine algorithms. The value of unsupervised vs. supervised neural networks.|
|Attributes and hydraulic fracturing of shale reservoirs||Review of the microseismic method and the relationship of microseismic events to surface seismic measurements. The use of prestack impedance inversion in predicting brittleness. Calibration using microseismic events and production logs.|
|The requirements of future E&P data integration – Shallow learning but deep thinking? Or Deep learning and shallow thinking||An overview of deductive reasoning as used in human-driven interpretation and inductive reasoning more common to conventional statistical analysis and more modern machine learning algorithms.|
|Seismic attributes and the road ahead||A summary of where we have been and we might be going with seismic attributes|
Who should attend?
- seismic interpreters who want to extract more information from their
- seismic processors and imagers who want to learn how their efforts impact subtle stratigraphic and fracture plays.
- sedimentologists, stratigraphers, and structural geologists who use large 3D seismic volumes to interpret their plays within a regional, basin-wide
- reservoir engineers whose work is based on detailed 3D reservoir models and whose data are used to calibrate indirect measures of reservoir
- Advanced knowledge of seismic theory is not required; this course focuses on understanding and practice.
Kurt Marfurt is an Emeritus Professor of Geophysics at the University of Oklahoma, where he mentors students and conducts research to aid seismic interpretation. Marfurt holds M.S. and Ph.D. degrees in Applied Geophysics from Columbia University in the City of New York and an A.B. in Physics and French from Hamilton College.
Marfurt experience includes 25 years as an academician, first at Columbia University, then later at the University of Houston and the University of Oklahoma. His career also includes 18 years in technology development at Amoco’s Tulsa Research Center working on a wide range of topics including seismic modeling, seismic imaging, VSP analysis, signal analysis, magnetotellurics, basin analysis, seismic stratigraphy, and seismic attributes. At OU, Marfurt led the Attribute-Assisted Seismic Processing and Interpretation (AASPI) consortium with the goal of developing and calibrating new seismic attributes to aid in seismic processing, seismic interpretation, and data integration using both interactive and machine learning tools. With colleagues, he has received several best paper and best presentation awards on seismic modeling, coherence, curvature, principal component analysis, and brittleness estimation. He is an honorary member of the SEG and in 2019 received the AAPG Robert R. Berg outstanding research award. Marfurt served as the 2006 EAGE/SEG and as the 2018 SEG Distinguished Short Course Instructor. He has taught continuing education short courses for the SEG and AAPG since 2003. From 1984-2013, he has served as either an associate or assistant editor for Geophysics. In 2013 he joined the editorial board of the SEG/AAPG journal Interpretation where he served as the Editor-in-Chief for 2016-2018, and subsequently as Deputy Editor-in-Chief for 2019-2021. He served as a Director-at-Large for the SEG from 2019-2022.
Results of unsupervised multiattribute classification using generative topographic mapping, co- rendered with coherence, over a turbidite system, offshore New Zealand. Input attributes included peak spectral frequency, peak spectral magnitude, curvedness, and GLCM entropy. (After Zhao et al., 2015; data courtesy of New Zealand Petroleum Ministry).
Phantom horizon slices 20 ms above the top Viola limestone through amplitude vs. azimuth (AVAz) anisotropy strike Ψ azimc modulated by its value B aniso
. Most-positive curvature is plotted against a gray scale and shows subtle faults. The survey in the NW has been hydraulically fractured while that in the SE has not. Note the compartmentalization of azimuth in the upper left survey, where curvature acts as fracture barriers. Note the stronger anisotropy (brighter colors)
in the SE survey which had not yet been hydraulically fractured. (Image courtesy S
of Shiguang Guo, OU).