Special Research Paper on “Applications of Data Science and Artificial Intelligence in Economic and Environmental Geology”

Split Viewer

Econ. Environ. Geol. 2024; 57(5): 499-512

Published online October 29, 2024

https://doi.org/10.9719/EEG.2024.57.5.499

© THE KOREAN SOCIETY OF ECONOMIC AND ENVIRONMENTAL GEOLOGY

Denoising Laplace-domain Seismic Wavefields using Deep Learning

Lydie Uwibambe, Jun Hyeon Jo, Wansoo Ha*

Department of Energy Resources Engineering, Pukyong National University, Busan 48513, South Korea

Correspondence to : *wansooha@pknu.ac.kr

Received: August 29, 2024; Revised: October 11, 2024; Accepted: October 11, 2024

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided original work is properly cited.

Abstract

Random noise in seismic data can significantly impair hydrocarbon exploration by degrading the quality of subsurface imaging. We propose a deep learning approach to attenuate random noise in Laplace-domain seismic wavefields. Our method employs a modified U-Net architecture, trained on diverse synthetic P-wave velocity models simulating the Gulf of Mexico subsurface. We rigorously evaluated the network’s denoising performance using both the synthetic Pluto velocity model and real Gulf of Mexico field data. We assessed the effectiveness of our approach through Laplace-domain full waveform inversion. The results consistently show that our U-Net approach outperforms traditional singular value decomposition methods in noise attenuation across various scenarios. Numerical examples demonstrate that our method effectively attenuates random noise and significantly enhances the accuracy of subsequent seismic imaging processes.

Keywords seismic data processing, deep learning, random noise attenuation, Laplace domain, full waveform inversion

Article

Special Research Paper on “Applications of Data Science and Artificial Intelligence in Economic and Environmental Geology”

Econ. Environ. Geol. 2024; 57(5): 499-512

Published online October 29, 2024 https://doi.org/10.9719/EEG.2024.57.5.499

Copyright © THE KOREAN SOCIETY OF ECONOMIC AND ENVIRONMENTAL GEOLOGY.

Denoising Laplace-domain Seismic Wavefields using Deep Learning

Lydie Uwibambe, Jun Hyeon Jo, Wansoo Ha*

Department of Energy Resources Engineering, Pukyong National University, Busan 48513, South Korea

Correspondence to:*wansooha@pknu.ac.kr

Received: August 29, 2024; Revised: October 11, 2024; Accepted: October 11, 2024

This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided original work is properly cited.

Abstract

Random noise in seismic data can significantly impair hydrocarbon exploration by degrading the quality of subsurface imaging. We propose a deep learning approach to attenuate random noise in Laplace-domain seismic wavefields. Our method employs a modified U-Net architecture, trained on diverse synthetic P-wave velocity models simulating the Gulf of Mexico subsurface. We rigorously evaluated the network’s denoising performance using both the synthetic Pluto velocity model and real Gulf of Mexico field data. We assessed the effectiveness of our approach through Laplace-domain full waveform inversion. The results consistently show that our U-Net approach outperforms traditional singular value decomposition methods in noise attenuation across various scenarios. Numerical examples demonstrate that our method effectively attenuates random noise and significantly enhances the accuracy of subsequent seismic imaging processes.

Keywords seismic data processing, deep learning, random noise attenuation, Laplace domain, full waveform inversion

    Fig 1.

    Figure 1.Architecture of a denoising U-Net.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 2.

    Figure 2.Sample velocity models used to generate training data.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 3.

    Figure 3.Inputs (top), labels (middle), and their profiles (bottom) from two validation samples. The profiles are extracted from the 51st shots indicated by dashed lines. The damping constants used are 10 s−1 and 2 s−1, and the grid sizes are 60.6 m and 70.6 m, respectively.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 4.

    Figure 4.Predicted results (top) and profiles of inputs (bottom) shown in Fig. 3.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 5.

    Figure 5.The Pluto velocity model (Stoughton et al., 2001).
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 6.

    Figure 6.Noisy, clean, and denoised wavefields (a) using the SVD and (b) U-Net for damping constant of 4, 6, 8, and 10 s−1.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 7.

    Figure 7.A shot gather from the Gulf of Mexico dataset.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 8.

    Figure 8.The original and denoised data using SVD and U-Net for damping constants of 4, 6, 8, and 10 s−1. The number of traces is 128, and the grid size is 80 m.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 9.

    Figure 9.Laplace-domain waveform inversion for the Pluto velocity model. (a) The initial velocity model and (b) the inversion result using clean data.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 10.

    Figure 10.Inversion results using (a) the noisy data, (b) SVD-denoised data, and (c) U-Net-denoised data.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 11.

    Figure 11.Error histories of Laplace-domain inversions.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 12.

    Figure 12.Laplace-domain waveform inversion of Gulf of Mexico field data. (a) The initial velocity model used in Laplace-domain waveform inversion, and inversion results from (b) the original noisy data, (c) SVD-denoised data, and (d) U-Net-denoised data.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 13.

    Figure 13.Error histories of Laplace-domain inversions.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Fig 14.

    Figure 14.The observed data and forward modeled data from the inversion results using the noisy data (Fig. 12b), SVD-denoised data (Fig. 12c), and U-Net-denoised data (Fig. 12d) for damping constants of 4, 6, 8, and 10 s−1. The number of traces is 408 and the grid size is 25 m.
    Economic and Environmental Geology 2024; 57: 499-512https://doi.org/10.9719/EEG.2024.57.5.499

    Table 1 . MSE losses of the noisy and denoised Pluto data calculated with the clean data.

    DataMSE loss
    Noisy data0.5870
    Denoised (SVD)0.0935
    Denoised (U-Net)0.0372

    Table 2 . MSE losses of the logarithmic forward-modeled data generated from the inversion results calculated with the observed Gulf of Mexico data.

    Data of FWIMSE loss
    Original noisy data2.6636
    Denoised (SVD)2.4736
    Denoised (U-Net)1.9878

    KSEEG
    Dec 31, 2024 Vol.57 No.6, pp. 665~835

    Stats or Metrics

    Share this article on

    • kakao talk
    • line

    Economic and Environmental Geology

    pISSN 1225-7281
    eISSN 2288-7962
    qr-code Download