Processing Techniques for Multiples in Seismic Data; Concepts, Applications, Trends

Course Description

Seismic data processing has traditionally used a subset of the data, the primaries, as the  signal part of the data from which to create images of the subsurface and to estimate  subsurface properties. Seismic multiples are another subset of the data, which are considered in most cases as noise to be removed, and in some cases as signal that  complements the primaries.  

In this course we cover a broad range of techniques that are used in seismic data  processing to remove multiples or to use multiples in imaging. For each of these  techniques we present the concepts, the key applications, the recent development  trends and we provide reading suggestions and exercises for self-study. 

By focusing on seismic multiples – their relation to primaries, their challenges and  opportunities in data processing - we provide an opportunity for students to gain a  deeper understanding of their data (primaries, multiples and other events) and of the  techniques they could use in processing to improve results.

Course Outline

The course is divided into 10 lectures, each of them being approximately 30-45  minutes. Within each lecture, concepts will be illustrated with synthetic and field data,  and exercises and reading suggestions will be provided for further study. 

Lecture 1: Introduction (part 1) 

• Summary of theoretical fundamentals 

o Acoustic and elastic media, free surfaces 

o Seismic wavefields in acoustic and elastic media 

o Scattering at different scales, and in particular at free surfaces o Classification of seismic events 

o Examples from seismology (videos and annotations) 

Lecture 2: Introduction (part 2) 

• Application domains and their datasets 

o Seismic surveys for oil and gas exploration, development, production o Seismic for CCUS 

o High-resolution shallow surveys for drilling hazards, windfarms, gas  hydrates 

o Seismology; global, local 

o Engineering applications 

• Seismic data processing methods 

o Separation of signal and noise 

o Parameter estimation problems 

o Imaging methods 

o Inversion methods 

• Canonical workflows and role of multiples 

Lecture 3: Lecture Multiple removal based on move-out  and dip discrimination 

• Introduction and principles 

• Common transforms in seismic data processing 

• Multiple removal using the parabolic Radon transform 

• Dip-discrimination basic workflow 

• Higher-dimensional transforms (beyond 2D) 

• Advances and developments in transform methods 

o Amplitude preserving, high-resolution transforms 

o Time-domain transforms 

o Joint transforms (demultiple in combination with one or several of  interpolation, de-alias, designature, linear noise removal, other denoise)

Lecture 4: Predictive deconvolution 

• Introduction, convolution and correlation concepts 

• Convolution and correlation concepts 

• Linear filtering, filter design techniques 

• Predictive deconvolution basics 

• Extending the predictive deconvolution concept 

Exercices, part 1 

Lecture 5: Principles of surface-related multiple  elimination 

• Relations between primaries and multiples for 1D media, normal incidence • Huygens’ virtual sources, Reciprocity theorems 

• Iterative SRME 

o Cases of 1D, 2D, 3D media 

• Other multiple prediction methods 

o Data-driven, model-based, or hybrid 

• Wavefield extrapolation methods 

o Classes of water-bottom related multiples 

o Model-based predictions of water-bottom related multiples 

o Relation to SRME 

Lecture 6: Practical aspects of surface-related multiple  elimination (part 1) 

• Data conditioning for SRME 

• Adaptive subtraction of predicted surface-related multiples 

Lecture 7: Practical aspects of surface-related multiple  elimination (part 2) 

• Shallow water case 

• Sea-bed acquisitions case 

o Up/down, down/down deconvolutions (1D media case) 

▪ P waves, pressure and Vz components 

▪ PS waves, pressure and radial components 

o SRME predictions using seabed and streamer data 

• Land data case

Exercices, part 2 

Lecture 8: Internal multiples 

• Introduction, historical perspective 

• Problem description 

• Extending the SRME concept to some classes of internal multiples and to 3D  data 

• Predicting all internal multiples using the Inverse Scattering Series • Novel approaches and perspectives using Marchenko theory 

• Model-based approaches to handling internal multiples 

Lecture 9: Using multiples as signal: concepts, synthetic  examples, state-of-art 

• Imaging primaries, imaging with multiples 

o Heuristic concepts 

o Notions of extended areal source, of crosstalk noise 

o Synthetic data examples 

o Further examples: VSP 

• The early imaging with multiples applications 

o Extended illuminations  

o Transforming multiples into primaries 

• SWIM (separated wavefield imaging with multiples) 

• Concepts/heuristics of multiples in VMB (tomography) and inversion 

Lecture 10: Using multiples as signal: recent advances 

• Scope (assumptions, advances, challenges and limitations for) 

o LS-RTM, free-surface multiples, joint primaries and FSM 

o FWI, high-frequency FWI 

o FWMOD, JMI (internal multiples reflections) 

o Marchenko redatuming, imaging, inversion 

Lecture 1: Introduction (part 1) 

• Summary of theoretical fundamentals 

o Acoustic and elastic media, free surfaces 

o Seismic wavefields in acoustic and elastic media 

o Scattering at different scales, and in particular at free surfaces o Classification of seismic events 

o Examples from seismology (videos and annotations) 

Lecture 2: Introduction (part 2) 

• Application domains and their datasets 

o Seismic surveys for oil and gas exploration, development, production o Seismic for CCUS 

o High-resolution shallow surveys for drilling hazards, windfarms, gas  hydrates 

o Seismology; global, local 

o Engineering applications 

• Seismic data processing methods 

o Separation of signal and noise 

o Parameter estimation problems 

o Imaging methods 

o Inversion methods 

• Canonical workflows and role of multiples 

Lecture 3: Lecture Multiple removal based on move-out  and dip discrimination 

• Introduction and principles 

• Common transforms in seismic data processing 

• Multiple removal using the parabolic Radon transform 

• Dip-discrimination basic workflow 

• Higher-dimensional transforms (beyond 2D) 

• Advances and developments in transform methods 

o Amplitude preserving, high-resolution transforms 

o Time-domain transforms 

o Joint transforms (demultiple in combination with one or several of  interpolation, de-alias, designature, linear noise removal, other denoise)

Lecture 4: Predictive deconvolution 

• Introduction, convolution and correlation concepts 

• Convolution and correlation concepts 

• Linear filtering, filter design techniques 

• Predictive deconvolution basics 

• Extending the predictive deconvolution concept 

Exercices, part 1 

Lecture 5: Principles of surface-related multiple  elimination 

• Relations between primaries and multiples for 1D media, normal incidence • Huygens’ virtual sources, Reciprocity theorems 

• Iterative SRME 

o Cases of 1D, 2D, 3D media 

• Other multiple prediction methods 

o Data-driven, model-based, or hybrid 

• Wavefield extrapolation methods 

o Classes of water-bottom related multiples 

o Model-based predictions of water-bottom related multiples 

o Relation to SRME 

Lecture 6: Practical aspects of surface-related multiple  elimination (part 1) 

• Data conditioning for SRME 

• Adaptive subtraction of predicted surface-related multiples 

Lecture 7: Practical aspects of surface-related multiple  elimination (part 2) 

• Shallow water case 

• Sea-bed acquisitions case 

o Up/down, down/down deconvolutions (1D media case) 

▪ P waves, pressure and Vz components 

▪ PS waves, pressure and radial components 

o SRME predictions using seabed and streamer data 

• Land data case

Exercices, part 2 

Lecture 8: Internal multiples 

• Introduction, historical perspective 

• Problem description 

• Extending the SRME concept to some classes of internal multiples and to 3D  data 

• Predicting all internal multiples using the Inverse Scattering Series • Novel approaches and perspectives using Marchenko theory 

• Model-based approaches to handling internal multiples 

Lecture 9: Using multiples as signal: concepts, synthetic  examples, state-of-art 

• Imaging primaries, imaging with multiples 

o Heuristic concepts 

o Notions of extended areal source, of crosstalk noise 

o Synthetic data examples 

o Further examples: VSP 

• The early imaging with multiples applications 

o Extended illuminations  

o Transforming multiples into primaries 

• SWIM (separated wavefield imaging with multiples) 

• Concepts/heuristics of multiples in VMB (tomography) and inversion 

Lecture 10: Using multiples as signal: recent advances 

• Scope (assumptions, advances, challenges and limitations for) 

o LS-RTM, free-surface multiples, joint primaries and FSM 

o FWI, high-frequency FWI 

o FWMOD, JMI (internal multiples reflections) 

o Marchenko redatuming, imaging, inversion 

Exercises, part 3 

Participants’ Profile

This course is intended for professionals or advanced degree students who are looking  for a refresher on the topic of multiples in seismic processing and for an introduction to  the latest research and technology developments. Also, it is expected that the students  would be willing to deepen their knowledge of the subject through self-study readings, exercises and interactions with the instructors during online sessions. 

Prerequisites

Participants should have a basic knowledge of:  

• Signal processing (convolution, correlation, Fourier transform);  

• Seismic processing (preprocessing, imaging);  

• Properties of acoustic waves, wave propagation, wave equations.

About the Instructor

Dirk J. (Eric) Verschuur received his M.Sc. degree in 1986 and his Ph. D degree (honors)  in 1991 from the Delft University of Technology (DUT), both in applied physics.

Clément Kostov is currently a Consulting research geophysicist, based in France.

He joined Schlumberger (now SLB) in 1990 and worked at SLB in research and  technology development as technical expert and as research manager until 2020.  Clément's interest in geosciences started at the Ecole des Mines de Paris and developed  through graduate studies at Stanford University (PhD in Geophysics '90). 

Clément is a member of the SEG, EAGE and IEEE professional societies and currently  volunteers in editorial roles for EAGE’s First Break and Geophysical Prospecting journals.