Benchmarking between Denoising Brute Stack Seismic Data Using Machine Learning and F-K Filter

Authors

  • Fadi Ghazi Department of Geology, College of Science, University of Baghdad, Baghdad, Iraq
  • Ali M. Al-Rahim Department of Geology, College of Science, University of Baghdad, Baghdad, Iraq

DOI:

https://doi.org/10.46717/igj.2025.58.2A.9

Keywords:

Signal processing; Machine learning; Signal-to-noise ratio improvement; FK filter; Denoising of seismic data

Abstract

The most common problem facing the obtaining of geophysical data, especially seismic data, is noise that obscures the details of the reflectors and thus leads to subsequent problems in the interpretation process. Therefore, it is necessary to develop machine learning techniques that help remove noise and preserve the signal without distortion, instead of conventional methods that cause problems such as overgeneralization or affecting the real signal. To confirm the machine learning models' reliability, comparisons were made to confirm their effectiveness in separating noise from the signal, moving beyond conventional methods. This paper is a comparative study conducted between the conventional filter (F-K) and the machine learning model (Auto-encoder) and discusses the strengths and weaknesses. It was found that the machine learning model is more accurate in removing noise, preserves the signal, and avoids the problems encountered by the conventional filter.

Downloads

Published

2025-07-31

Most read articles by the same author(s)

1 2 > >>