1887

Abstract

Summary

We propose a local low-rank factorization approach to improve the efficiency and robustness of multidimensional deconvolution. The approach divides the sought-after Green’s function matrix into tiles, leveraging local rank characteristics to represent off-diagonal tiles as low-rank products and diagonal tiles as either low- or full-rank products, all controlled by user-specified accuracy requirements, trading off accuracy for storage and computational efficiency. The method is evaluated on a large-scale 3D synthetic example based on the EAGE/SEG Overthrust model, and results demonstrate significant improvements over benchmark solvers. In particular, as an extra benefit of rank-based filtering, the proposed method shows greater robustness to noise and sparse shot sampling, achieving higher signal-to-noise ratios and eliminating artifacts such as remnant multiples and near-offset noise. This work presents a promising solution for multidimensional deconvolution, particularly under challenging data conditions, and highlights the benefits of adaptive tile-based low-rank factorization for seismic data processing.

© EAGE Publications BV
Loading

Article metrics loading...

/content/papers/10.3997/2214-4609.202510292
2025-06-02
2026-02-09

  • From This Site

    /content/papers/10.3997/2214-4609.202510292
    dcterms_title,dcterms_subject,pub_keyword
    -contentType:Journal -contentType:Contributor -contentType:Concept -contentType:Institution
    10
    5
Loading full text...

Full text loading...

References

  1. Chen, F., Ravasi, M. and Keyes, D. [2023] Solving multi-dimensional deconvolution via a nuclear-norm regularized least-squares approach. 84th Annual International Conference and Exhibition, EAGE, Extended Abstracts, 2023(1), 1–5.
     [Google Scholar]
  2. Chen, F., Ravasi, M. and Keyes, D. [2024a] A factorization-based, tile low-rank parametrization for 3D multidimensional deconvolution. In: Fourth International Meeting for Applied Geoscience & Energy Expanded Abstracts. 1528–1532.
     [Google Scholar]
  3. Chen, F., Ravasi, M. and Keyes, D. [2024b] A reciprocity-aware, low-rank regularization for multidimensional deconvolution. Geophysics, 90(1), R1–R11.
     [Google Scholar]
  4. Hong, Y., Ravasi, M., Ltaief, H. and Keyes, D. [2023] Can tile low-rank compression live up to expectations? An application to 3D multidimensional deconvolution. In: Third International Meeting for Applied Geoscience; Energy Expanded Abstracts. 1528–1532.
     [Google Scholar]
  5. Kumar, R., Boiero, D., Bagaini, C. and Vassallo, M. [2022] Transform-Domain Multidimensional De-convolution Sparsity V/S Low-Rank. 83rd EAGE Annual Conference & Exhibition, 2022(1), 1–5.
     [Google Scholar]
  6. Ltaief, H., Hong, Y., Wilson, L., Jacquelin, M., Ravasi, M. and Keyes, D. [2023] Scaling the Memory Wall for Multi-Dimensional Seismic Processing with Algebraic Compression on Cerebras CS-2 Systems. ACM/IEEE.
     [Google Scholar]
  7. Ravasi, M., Hong, Y., Ltaief, H., Keyes, D. and Vargas, D. [2022] Large-scale Marchenko imaging with distance-aware matrix reordering, tile low-rank compression, and mixed-precision computations. Second International Meeting for Applied Geoscience & Energy, 2606–2610.
     [Google Scholar]
  8. Ravasi, M., Örnhag, M.V., Luiken, N., Leblanc, O. and Uruñuela, E. [2024] PyProximal - scalable convex optimization in Python. J. Open Source Softw., 9, 6326.
     [Google Scholar]
  9. Recht, B., Fazel, M. and Parrilo, P.A. [2010] Guaranteed Minimum-Rank Solutions of Linear Matrix Equations via Nuclear Norm Minimization. SIAM Review, 52(3), 471–501.
     [Google Scholar]
  10. Vargas, D., Vasconcelos, I., Ravasi, M. and Luiken, N. [2021] Time-Domain Multidimensional Decon-volution: A Physically Reliable and Stable Preconditioned Implementation. Remote Sensing, 13(18), 3683.
     [Google Scholar]
/content/papers/10.3997/2214-4609.202510292
Loading
Adaptive Blockwise Low-Rank Approximation for Robust Large-Scale 3D Multidimensional Deconvolution
Conference Proceedings 2025, 1 (2025); https://doi.org/10.3997/2214-4609.202510292
/content/papers/10.3997/2214-4609.202510292
/content/papers/10.3997/2214-4609.202510292
Loading

Data & Media loading...

Most Cited This Month Most Cited RSS feed

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error