Classical and Modern Signal Processing for Seismic Applications

Course Description

This two-day course offers a comprehensive overview of modern seismic signal processing techniques, classical methods, and emerging innovations in sparsity, machine learning, and rank-reduction. Participants will explore practical algorithms and develop hands-on skills through Julia-based software prototypes.

Course Objectives: 

Upon completion of the course, participants will understand current technology for advanced signal processing including technology for 5D reconstruction, Compressive Acquisition, De-blending, Reconstruction of Dispersed source array data.

Course Outline: 

1. Review 

A short review of Fourier Analysis, Convolution, and Deconvolution. A review of linear algebra and concepts needed to tackle geophysical data processing problems.

2. Predictability of Signals in the Frequency (FX) and Time (TX) Domains and Projection Filters

This topic examines the inherent predictability of signals in both the frequency and time domains. We will explore the application of prediction and projection filters to isolate and enhance predictable signal components, improving overall signal analysis and interpretation.

3. Sparsity in Transform-Based Signal Processing: Applications in Denoising, Interpolation, and Deblending

We will discuss the concept of sparsity and its critical role in transform-based signal processing techniques, including compressive sensing. The focus will be on denoising, interpolation, and deblending applications to demonstrate how leveraging sparse representations can significantly improve signal quality and processing efficiency.

4. Rank-Reduction Methods for Seismic Signal Processing: Hankel and Tensor-Based Approaches

This section will cover various rank-reduction techniques applicable to seismic signal processing, specifically focusing on Hankel and tensor-based methods. We will delve into their practical applications and discuss methods for implicit-form rapid rank-reduction methods.

5. Innovative Signal Processing Paradigms: Integrating Machine Learning for Denoising and Reconstruction

Here, we will explore new paradigms in signal processing that incorporate machine learning techniques, such as Implicit Network Representation. The discussion will highlight how these methods can be utilized for advanced denoising and reconstruction tasks, showcasing their potential to transform traditional signal processing workflows.

6. Emerging ideas: Dispersed Source Array Processing via sparsity-constrained inversion

Regularization by denoising. Role of denoisers in the solution of inverse problems.

Participant's Profile

The course is designed for geophysicists, engineers, data scientists working in industry, and graduate students seeking to expand their expertise in seismic data processing. A basic background in signal processing and linear algebra is recommended but not required.

Prerequisites:

Participants should have previous knowledge of linear algebra, Fourier theory, and programming.

About the Instructor:

Mauricio D. Sacchi is a geophysicist and professor in the Department of Physics at the University of Alberta. He earned a Geophysics diploma from the National University of La Plata in 1988 before moving to Canada to pursue a PhD in Geophysics at the University of British Columbia, which he completed in 1996. He joined the University of Alberta faculty in 1997.

He is the author (with Tad Ulrych) of the book Information-Based Signal Processing, and has taught extensively in both graduate courses and industry short courses on seismic signal theory and inversion. His excellence has been recognized by numerous awards, including the CSEG Medal (2012), SEG Honorary Lectureship (2014), Distinguished Speaker (2015), a CSEG Symposium held in his honor (2018), and most notably the prestigious SEG Virgil Kauffman Gold Medal in 2019 for his outstanding contributions to geophysical exploration.

Beyond research, Sacchi has served as Editor-in-Chief of GEOPHYSICS (2015–2017), Associate Editor for IEEE Transactions on Geoscience and Remote Sensing, and is an active member of professional societies including CGU, CSEG, EAGE, IEEE, and SEG.