Meta-attributes and Artificial Networking E-Book

Kalachand Sain

• Herausgeber: John Wiley & Sons
• Kategorie: Wissenschaft und neue Technologien
• Serie: Special Publications
• Sprache: Englisch

Seitenzahl: 350

Veröffentlichungsjahr: 2022

Beschreibung

Applying machine learning to the interpretation of seismic data Seismic data gathered on the surface can be used to generate numerous seismic attributes that enable better understanding of subsurface geological structures and stratigraphic features. With an ever-increasing volume of seismic data available, machine learning augments faster data processing and interpretation of complex subsurface geology. Meta-Attributes and Artificial Networking: A New Tool for Seismic Interpretation explores how artificial neural networks can be used for the automatic interpretation of 2D and 3D seismic data. Volume highlights include: * Historic evolution of seismic attributes * Overview of meta-attributes and how to design them * Workflows for the computation of meta-attributes from seismic data * Case studies demonstrating the application of meta-attributes * Sets of exercises with solutions provided * Sample data sets available for hands-on exercises The American Geophysical Union promotes discovery in Earth and space science for the benefit of humanity. Its publications disseminate scientific knowledge and provide resources for researchers, students, and professionals.

Leseprobe

Special Publications 76

META‐ATTRIBUTES AND ARTIFICIAL NETWORKING

A New Tool for Seismic Interpretation

Kalachand SainPriyadarshi Chinmoy Kumar

This Work is a co‐publication of the American Geophysical Union and John Wiley and Sons, Inc.

This edition first published 2022© 2022 American Geophysical Union

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.

Published under the aegis of the AGU Publications Committee

Matthew Giampoala, Vice President, PublicationsCarol Frost, Chair, Publications CommitteeFor details about the American Geophysical Union visit us at www.agu.org.

The right of Kalachand Sain and Priyadarshi Chinmoy Kumar to be identified as the authors of this work has been asserted in accordance with law.

Registered OfficeJohn Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA

Editorial Office111 River Street, Hoboken, NJ 07030, USA

For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.

Wiley also publishes its books in a variety of electronic formats and by print‐on‐demand. Some content that appears in standard print versions of this book may not be available in other formats.

Limit of Liability/Disclaimer of Warranty

While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

Library of Congress Cataloging‐in‐Publication Data

Names: Sain, Kalachand, author. | Kumar, Priyadarshi Chinmoy, author. | John Wiley & Sons, publisher. | American Geophysical Union, publisher.Title: Meta-attributes and artificial networking : a new tool for seismic interpretation / Kalachand Sain, Priyadarshi Chinmoy Kumar.Description: Hoboken, NJ : Wiley-American Geophysical Union, 2022. | Includes bibliographical references and index.Identifiers: LCCN 2021044593 (print) | LCCN 2021044594 (ebook) | ISBN 9781119482000 (cloth) | ISBN 9781119481911 (adobe pdf) | ISBN 9781119481768 (epub)Subjects: LCSH: Seismology–Data processing. | Neural networks (Computer science)–Scientific applications. | Artificial intelligence–Geophysical applications.Classification: LCC QE539.2.D36 S25 2022 (print) | LCC QE539.2.D36 (ebook) | DDC 551.2201/13–dc23/eng/20211119LC record available at https://lccn.loc.gov/2021044593LC ebook record available at https://lccn.loc.gov/2021044594

Cover Design: WileyCover Image: © local_doctor/Shutterstock

PREFACE

Surface measurements that identify and map subsurface geologic features in both onshore and offshore locations have multiple applications. These include hydrocarbon exploration; characterization of subsurface water, geothermal, and mineral resources; and understanding volcanic and seismo‐tectonic processes.

Seismic data are one type of surface measurements that can be used to better understand the characteristics of the subsurface. Over the past few decades, the attributes derived from seismic data have revolutionized interpretation of geologic structures, stratigraphic features, and reservoir properties. However, as the amount of seismic data collected by scientists keeps growing, so does the task of data processing and interpretation.

High performance computing systems offer a solution to the data processing part, as they allow processing of copious amounts of data quickly. However, data interpretation can still be challenging, particularly in areas that are geologically complex and when the data volumes are large. Thus, there is a need to automate the process, accelerate the interpretation, and reduce the need for intervention by human analysts.

The purpose of this book is to introduce a new approach to seismic interpretation based on artificial networking. Several workflows have been designed to amalgamate multiple seismic attributes to compute new attributes, known as “meta‐attributes” or “hybrid‐attributes.” The computation of meta‐attributes delimits the 3-D geometry of subsurface geologic features by their seismic properties and characteristics, and thus accelerates the interpretation. The exciting part of this semi‐automatic method is the fusion of human intelligence with machine intelligence over a small volume of data, followed by automatizing the process for the delineation of subsurface architecture from a huge volume of data, acquired on the surface.

The book is divided into three parts. Part I is dedicated to seismic attributes. Chapter 1 gives a brief overview of the history of seismic attributes and highlights the importance of attribute technology in exploration geophysics. Chapter 2 describes mathematical formulations of complex trace, structural, and stratigraphic attributes, and demonstrates how these can be extracted from seismic data for different purposes. Chapter 3 presents an interpretation task with ten examples of seismic cross‐sections and numerics that should be of use to those new to seismic interpretation in practices.

Part II focuses on meta‐attributes. Chapter 4 defines different meta‐attributes that can be used for delineating various subsurface geologic features. Chapter 5 introduces the fundamentals of artificial neural networks (ANN) including historical development of comparing the mathematical neuron with the biological neuron, multi‐layer perceptron (MLP) with feedforward and backpropagation algorithms for training, and types of neural networks. Chapter 6 introduces workflows for the computation of meta‐attributes from seismic data.

Part III presents a series of case studies with field examples to demonstrate the application of seismic meta‐attributes for automatic interpretation of 3-D geometry of subsurface geologic structures. Chapter 7 shows how gas chimneys can be picked up by the Chimney Cube Meta‐attribute from reflection seismic data. Chapters 8 looks at the distribution of thin faults through the computation of the Thin Fault Cube Meta‐attribute, while Chapter 9 demonstrates how a blend of Thinned Fault Cube and Fluid Cube Meta‐attributes can explain the hydrocarbon fluid leakage from a network of hard‐linked fault system. Chapters 10 and 11 explain how the Sill Cube Meta‐attribute can capture the sill network in basins of two different geologic set ups. Chapter 12 shows an amalgamation of Sill Cube and Fluid‐Cube meta‐attributes that can arrest the sill complexes and fluxed‐out magmatic fluids and explain their impact on forming structural folds in the overlying younger formations. Chapter 13 demonstrates the application of the Intrusion Cube Meta‐attribute for interpreting a buried volcano and other intrusive elements such as the sill webs, dyke swarms, and magmatic ascent into a complex tectonic regime. Finally, Chapter 14 highlights the use of the Mass Transport Deposit Meta‐attribute in deciphering the structural architecture and distribution of mass transport system from reflection seismic data.

Every chapter has a reference section to point readers to further literature, and there is a glossary at the beginning of the book explaining the main scientific terms in this field. There are also three appendices. Appendix A sheds light on the mathematical formulation of common series and transforms. Appendix B illustrates techniques of Dip-Steering. Appendix C presents answers to the interpretation and numerical tasks presented in Chapter 3. Users can make use of data sets available at www.wihg.res.in for hands‐on practice of interpretation and computation of meta‐attributes.

Machine learning is being applied across many fields including medical science, engineering, and astronomy. This book demonstrates its potential as a tool in the geosciences for automatic interpretation of 2-D and 3-D seismic data, a field of direct relevance to society and sustainable development. We hope that it will be of interest to specialists such as seismic interpreters in the petroleum industry as well as students and researchers who are new to seismic data analysis and interpretation.

This book is an outcome of dedicated research made by ourselves and our students on exploration seismology and their applications for subsurface imaging. We would like to convey our gratitude to our teachers and professors who nourished us from the beginning. We express deep gratitude to Prof. Harsh K. Gupta for his valuable guidance at several stages of our professional careers and personal development. We are also grateful to Prof. Sailesh Nayak for his suggestions and advice. The encouraging words of Prof. Ashutosh Sharma on the application of artificial intelligence to geosciences have inspired and motivated us in writing this book.

We thank the Government of India’s Ministry of Earth Sciences, Ministry of Petroleum and Natural Gas, and the Department of Science and Technology for providing financial support to create geophysical research facilities.

Special thanks are due to Dr. Rituparna Bose at Wiley for her encouragement and valuable input from the book’s inception to the final phases of writing. Thanks are due to Dr. Jun Matsushima, three other anonymous reviewers, and a member of the AGU Books Editorial Board for their constructive comments and suggestions to improve the content of our book. We also thank Dr. Jenny Lunn of AGU and Ms. Layla Harden of Wiley for their support and help at different stages of book preparation.

We acknowledge the New Zealand Petroleum Minerals, the New Zealand Ministry of Economic Development, and the dGB Earth Sciences in The Netherlands for providing the valuable data and academic license of the software.

Sincere thanks are also due to our international collaborators including Prof. Tiago M Alves (Cardiff University, UK), Dr. Kamaldeen Olakunle Omosanya (Oasisgeokonsult, Norway), Prof. Nicolas Waldmann (University of Haifa, Israel), and Prof. Qiliang Sun (China University of Geosciences), whose scientific advice have helped our students in carrying out research in applied seismology.

Finally, we acknowledge our families. From the bottom of his heart, KS affectionately thanks his wife Tumpa and son Ritwik for their continuous inspiration and love in the journey of writing this book. PCK expresses thanks to Baba and Maa. We also thank the Almighty for heavenly blessings and guidance.

Kalachand Sain

Priyadarshi Chinmoy Kumar

Wadia Institute of Himalayan Geology, India

ABOUT THE AUTHORS

Kalachand Sain is the Director of the Wadia Institute of Himalayan Geology in Dehradun, India. Previously he was Chief Scientist at the CSIR‐National Geophysical Research Institute in Hyderabad, India. He has an MSc (Tech) in Applied Geophysics from IIT‐Indian School of Mines, Dhanbad, and a PhD in Controlled Source Seismology from CSIR-National Geophysical Research Institute, Hyderabad. He spent time as a post‐doctoral fellow at Cambridge University (UK) and Rice University (USA), and was a visiting scientist at the United States Geological Survey. His research interests include exploration of gas hydrates, imaging sub‐volcanic sediments, understanding evolution of sedimentary basins and earthquake processes, and providing geotectonic implications, including the Himalayan orogeny, and glaciological and landslides hazards. He has also built expertise in travel time tomography, AVO modelling, full‐waveform tomography, impedance inversion, pre‐stack depth migration, seismic attenuation and meta‐attributes, artificial intelligence, rock physics modelling, and interpretation of 2‐D/3‐D seismic data. He is a Fellow of all three Indian science academies and is the recipient of numerous medals and awards including the National Mineral Award, National Award of Excellence in Geo-sciences, J.C. Bose National Fellowship, Decennial Award & Anni Talwani Memorial Prize of Indian Geophysical Union, and Distinguished IIT‐ISM Alumnus Award.

Dr. Priyadarshi Chinmoy Kumar is a Scientist at Wadia Institute of Himalayan Geology (WIHG) in Dehradun, India. He received a MSc (Tech) in Geophysics with First Class Distinction from Andhra University and Ph.D. in Science (Geophysics) from the Academy of Scientific and Innovative Research, Hyderabad. His research interests include processing and interpretation of seismic data, design and development of workflows for computation of meta‐attributes, and basin studies. He has been recognized with a young scientist award by the Indian Geophysical Union, National Academy of Science, India and as an Associate of the Indian Academy of Sciences.

ACRONYMS

AAA

Anomalous Amplitude Attenuation

AFV

Average Frequency Variance

ANN

Artificial Neural Network

AOM

Ascent of Magma

AVO

Amplitude Variation with Offset

BN

Biological Neuron

BOPD

Barrels of Oil Per Day

BSR

Bottom Simulating Reflector

BS

Background Steering

BSS

Basal Shear Surface

CB

Canterbury Basin

CC

Chimney Cube

CM

Chimney Migration

CNN

Convolutional Neural Network

CTA

Complex Trace Analysis

CWT

Continuous Wavelet Transform

DHI

Direct Hydrocarbon Indicator

DS

Detailed Steering

DSDF

Dip‐Steered Diffusion Filter

DSMF

Dip‐Steered Median Filter

DVA

Dense Velocity Analysis

EW

East–West

E&P

Exploration & Production

FC

Fault Cube

FlC

Fluid Cube

FEF

Fault Enhanced Filter

FF

Forced Fold

FZ

Fault Zone

GC

Gas Chimney

GRNN

Generalized Regression Neural Network

HC

High Chimney

HFBZ

High Frequency Blackout Zone

HN

Hopfield Networks

HNS

Human Neural System

HP

Hydrocarbon Probability

IC

Intrusion Cube

LFBZ

Low‐Frequency Blackout Zone

LVQ

Learning Vector Quantizer

LW

Long Window

MBKB

Meter Bellow Kelly Bush

MLP

Multi‐Layer Perceptron

MN

Mathematical Neuron

MNN

Modular Neural Network

MP

Most Positive

MSE

Mean Square Error

MTC

Mass Transport Complex

MTD

Mass Transport Deposit

MTDC

Mass Transport Deposit Cube

MVC

Main Vent Complex

MW

Mid Window

NN

Neural Network

nRMS

Normalized RMS

NS

North–South

NZ

New Zealand

NZP&M

New Zealand Petroleum & Minerals

PD

Potential Difference

PDF

Probability Density Function

PE

Processing Element

PNN

Probabilistic Neural Network

PPS

Plio‐Pleistocene Sequence

QC

Quality Check

Q‐factor

Quality Factor

RBF

Radial Basis Function

REF

Ridge Enhancement Filter

RGB

Red Green Blue

RHS

Right‐Hand Side

RMS

Root Mean Square

SC

Sill Cube

SD

Spectral Decomposition

SEG

Society of Exploration Geophysicists

SW

Short Window

SEG

Society of Exploration Geophysicists

S/N

Signal to Noise

SOF

Structure Oriented Filter

SOM

Self‐Organizing Maps

STFT

Short‐Time Fourier Transform

SV

Seismic Volume

SW

Short Window

TB

Taranaki Basin

TD

Total Depth

TFC

Thinned Fault Cube

TFL

Thinned Fault Likelihood

TWT

Two‐Way‐Time

UVQ

Uniform Vector Quantizer

VC

Volcanic Core

VSP

Vertical Seismic Profiling

LIST OF SYMBOLS AND OPERATORS

A

i

(

t

)

Instantaneous amplitude

A

rms

RMS amplitude

b

i

(

t

)

Instantaneous bandwidth

e

dip

Exaggerated dip

e

slope

Exaggerated slope

f

a

(

t

)

Average frequency

f

i

(

t

)

Instantaneous frequency

f

rms

(

t

)

RMS frequency

f

l

Fault likelihood

γ

Exaggeration factor

H

Hilbert Transform

H

−1

Inverse Hilbert Transform

≡

Identical to

k

Curvature

k

mean

Mean curvature

k

max

Maximum curvature

k

pos

Most positive curvature

k

neg

Most negative curvature

k

gauss

Gaussian curvature

μ

amp

Average of the amplitudes

φ

Azimuth

p

inline dip component

p

(

φ

)

Apparent dip of the inline component measured at an azimuthal angle

q

xline dip component

q

(

φ

)

Apparent dip of the xline component measured at an azimuthal angle

q

a

(

t

)

Average quality factor

q

i

(

t

)

Quality factor or instantaneous quality factor

R

Radius of curvature

r

pseudo

Pseudo relief

ρ

x

Inline lag cross‐correlation

ρ

y

Xline lag cross‐correlation

σ

a

(

t

)

Average decay rate

σ

amp

Standard deviation of amplitudes

σ

i

(

t

)

Instantaneous decay rate

σ

r

(

t

)

Relative amplitude

〈∙〉

s

Structure‐oriented averaging

S

(

t

)

Sweetness

S

m

Similarity

S

dip

Dip‐steered similarity

ψ

s

Strike of a surface

θ

d

Dip angle or true dip of a surface

θ

i

(

t

)

Instantaneous phase

θ

x

Apparent dip in x‐direction

θ

y

Apparent dip in y‐direction

u

Amplitude in a seismic cube

x

(

t

)

Real component of seismic trace

y

(

t

)

Quadrature component seismic trace

GLOSSARY

Activation function

A mathematical function that maps the output of the neural network in terms of binary values.

Amplitude

The magnitude values of the seismic trace or trace envelope.

Amplitude change

The change in seismic amplitude over an interval in a given direction.

Azimuth

The angle measured clockwise from the geographic north in the direction of the maximum downward dip or slope.

Backpropagation

A process of computing the error in prediction to the actual in the backward direction.

Bandwidth

The breadth of the frequency power spectrum of a waveform.

Bright spot

A local high‐amplitude seismic anomaly that shows the presence of hydrocarbons. Also known as the direct hydrocarbon indicator.

Curvature

The degree of curvedness of a surface, and is defined as the curvature, i.e. how or to what extent a surface bends or curves.

Chimney (or gas chimney)

Vertical disturbances observed on seismic data. Such features are associated with chaotic reflections where amplitudes are weaker.

Co‐rendering

A process in which attribute maps are superimposed and color scales are adjusted to improve visualization of subsurface geologic features.

Coherency

A measure of lateral change in seismic response caused by the variation in structure, stratigraphy, or lithology.

Data conditioning

A process of improving signal‐to‐noise (s/n) such that the seismic data can be readily interpreted.

Depth slice

A process of slicing a 3D seismic cube at a particular depth level.

Detailed steering cube

A steering cube that is obtained by applying mild filtering.

Dip‐steering

A process of estimating dip‐azimuth information at each sample location of a seismic trace. The process generates a steering cube, called the dip‐steering volume or dip‐azimuth volume.

Example locations

Places that consist of targets and non‐targets.

Faults

Discontinuous structures that are generally associated with reflector terminations, vertical disturbances, and breaks in seismic reflectors.

Fault dip

The angle that the fault plane makes with the horizontal.

Fault throw

The vertical separation of a layer generated due to faulting.

Feedforward

The process of feeding a neural network in the forward direction.

Filtering

A process of removing unwanted information from the data.

Frequency shadow

Loss or wash‐out of higher frequencies leaving behind a lower frequency that generates a shadow beneath the reservoir.

Gas chimney

Upward migration and escaping of accumulated gas and appears as a wash‐out zone in seismic sections.

Instantaneous amplitude (IA)

Change in amplitude at a given instant of time, or it represents the magnitude of the sinusoid at a given time that represents a seismic trace.

Instantaneous phase (IP)

Change in phase at a given instant of time.

Instantaneous frequency (IF)

Change in frequency at a given instant of time.

Magmatic Sill

Tabular intrusive rocks with concordant surfaces showing concave upwards cross‐sectional geometries and discordant limbs.

Mass Transport Deposit

Unconsolidated sediments transported into the deep water environment under the influence of gravitational force due to slope failure or slope instability.

Meta‐attribute

A hybrid attribute generated by combining a set of other seismic attributes using a neural‐based approach.

Migration

A process carried out to move dipping reflectors to their correct position to produce accurate images of the subsurface.

Misclassification

A quality control parameter used to understand the wrong prediction made during classification.

Multi‐layer

A network consisting of several layers.

Multi‐layer perceptron

A network arrangement in which nodes or neurons in each layer are fully connected to the nodes or neurons of the consecutive layers.

Overtraining

A process in which neural training becomes non‐universal, meaning the network fails to differentiate between the target and non‐targets.

Perceptron

The building elements in a layer of a neural network.

Phase

The relative position along a seismic waveform that is independent of amplitude.

RGB blending