Seismic Multiple Removal Techniques: Past, Present and Future
The main objective of this course is to provide the audience with an overview of the techniques in seismic multiple removal, starting with the deconvolution-based methods from the 1960s, via the move- out discrimination techniques of the 1980s and ending up with wave-equation based methods from the 1990s and their 3D extensions as developed in the 2000s. Furthermore, the current challenges in multiple removal and their relation with seismic imaging and inversion are treated. A secondary objective is to discuss more general processing concepts such as high-resolution seismic data transforms (Fourier, Radon), adaptive filtering techniques, wave-equation based forward and inverse wave propagation and the processing of seismic data in different transform domains. For each method some brief description of the theory in terms of mathematics is given. However, the emphasis in this course is not to thoroughly treat the mathematics but to present some understanding of the workings of each method.
At the end of each lecture, a list of relevant articles in the open literature will be specified. The course is subdivided in 10 lectures, each of them being approximately 30-45 minutes. Within each lecture, examples of the described concepts on synthetic and field data will play an important role. Lecture 1: Multiples ... what’s the problem? • Classification of multiple reflections • Characteristics of multiples • Impact on seismic imaging and interpretation • Categories of multiple removal methods Lecture 2: Multiple removal based on move-out and dip discrimination • Principle of multiple removal by move-out discrimination • F-K and Radon transforms • Multiple removal by filtering in the FK or Radon domain • Towards high-resolution Radon transforms • Limitations of multiple removal by move-out discrimination • Multiple removal by target-oriented dip filtering Lecture 3: Predictive deconvolution • Convolution and correlation concept • Designing adaptive filters by least-squares optimisation • Predictive deconvolution basics • Extending the predictive deconvolution concept Lecture 4: Multiple removal by wave field extrapolation • Forward and inverse wave field extrapolation • Multiple prediction by wave field extrapolation • Application in the wave number and linear Radon domain Lecture 5: Principles of surface-related multiple elimination • Derivation of SRME for the 1D situation • Including the source characteristics • Iterative implementation of SRME • Formulation of SRME for the 2D and 3D situation • Relation between multiple prediction and subtraction methods Lecture 6: Practical considerations for surface-related multiple elimination • Effect of missing data on SRME • Interpolation of missing near offsets • Application of SRME in different data domains • Shallow water multiple removal strategy Lecture 7: Adaptive subtraction of predicted multiples • Least squares and L1-norm subtraction • Pattern recognition and other multiple subtraction techniques Lecture 8: Towards 3D multiple removal • Multiples in complex 3D environments • 3D SRME: theory and practice • 3D SRME: solutions via data interpolation Lecture 9: Internal multiple removal • Internal multiple removal by move-out discrimination • Extending the SRME concept to internal multiples • Internal multiple removal by inverse scattering Lecture 10: Removing or using multiples? • Transforming multiple into primaries • Estimation of primaries by sparse inversion • Including multiples in the migration process • Including multiples in the inversion process For the 2-days course , especially the second part of the course, will be more elaborated with extra topics being: • more elaborate discussion on adaptive subtraction techniques (Lecture 7) • more extensive explanation on internal multiple removal (Lecture 9) • including the recently developed EPSI (Estimation of Primaries by Spares Inversion) methodology (Lecture 10) • including an extensive discussion on using surface multiples in Imaging (Lecture 10)
The target audience is composed of people involved in seismic processing, imaging and inversion. The mathematical content is kept to a minimum level with a strong link to the involved physical concepts, amplified by graphical illustrations. The audience is expected to have prior knowledge at a B.Sc./M.Sc. level on processing concepts such as convolution, correlation and Fourier transforms and some basic knowledge on wave theory.
Participants should have a basic knowledge of: • Basic signal processing (convolution, correlation, Fourier transform); • Basic seismic processing (preprocessing, imaging); • Basic knowledge on the acoustic wave equation and wave propagation.
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. From 1992 - 1997 he worked under a senior research fellowship from the Royal Dutch Academy of Art and Sciences (KNAW). In 1997 he became assistant professor and since 1999 he is an associate professor at the DUT at the laboratory of Acoustical Imaging and Sound Control. He is the project leader of the DELPHI research consortium in the area of Multiple Removal and Structural Imaging. His main interests are seismic modeling, processing and migration techniques. In 1997 he received SEG’s J. Clarence Karcher award. He is a member of SEG and EAGE.