Skip to main content

Full waveform inversion with random shot selection using adaptive gradient descent

Abstract

Full waveform inversion (FWI) is a powerful yet computationally expensive technique that can yield subsurface models at high resolution. Randomly selected shots (mini-batches) can be used to approximate the misfit and the gradient of FWI, thereby reducing its computational cost. Here, we present a methodology to perform mini-batch FWI using the Adam algorithm, an adaptive optimization scheme based on stochastic gradient descent. It provides for stable model updates by smoothing the gradient across iterations and can also account for the curvature of the optimization landscape. We describe empirical criteria to choose the hyperparameters of the Adam algorithm and the optimal mini-batch size. The performance of the outlined scheme is illustrated on synthetic data from the Marmousi model and compared with conventional full-batch FWI. FWI with random shot selection and optimized by the Adam algorithm exhibits rapid convergence and yields superior results compared to conventional FWI implemented with the l-BFGS method.

This is a preview of subscription content, access via your institution.

Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11
Figure 12
Figure 13
Figure 14

Notes

  1. 1.

    https://github.com/bshekar/FWI-Adam.

  2. 2.

    https://github.com/devitocodes/devito.

  3. 3.

    https://www.tensorflow.org.

References

  1. Abadi M, Agarwal A, Barham P, Brevdo E, Chen Z, Citro C, Corrado G S, Davis A, Dean J, Devin M, Ghemawat S, Goodfellow I, Harp A, Irving G, Isard M, Jia Y, Jozefowicz R, Kaiser L, Kudlur M, Levenberg J, Mane D, Monga R, Moore S, Murray D, Olah C, Schuster M, Shlens J, Steiner B, Sutskever I, Talwar K, Tucker P, Vanhoucke V, Vasudevan V, Viegas F, Vinyals O, Warden P, Wattenberg M, Wicke M, Yu Y and Zheng X 2015 TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems; Software available from tensorflow.org.

  2. Brenders A, Dellinger J, Ahmed I, Diaz E, Gherasim M, Jin H and Vyas M 2020 The Wolfspar experience: Learnings from processing sparse, low-frequency seismic data; In: 90th Annual International Meeting, SEG, Expanded Abstracts.

  3. Bunks C, Salek F M, Zaleski S and Chavent G 1995 Multiscale seismic waveform inversion; Geophysics 60(5) 1457–1473.

    Article  Google Scholar 

  4. Byrd R H, Hansen S L, Nocedal J and Singer Y 2014 A stochastic quasi-Newton method for large-scale optimization; arXiv preprint arXiv:1401.7020.

  5. Datta D and Sen M K 2016 Estimating a starting model for full-waveform inversion using a global optimization method; Geophysics 81(4) R211–R223.

    Article  Google Scholar 

  6. Dellinger J, Ross A, Meaux D, Brenders A, Gesoff G, Etgen J, Naranjo J, Openshaw G and Harper M 2016 Wolfspar, an FWI-friendly ultralow-frequency marine seismic source; In: 86th Annual International Meeting, SEG, Expanded Abstracts.

  7. Diaz E and Guitton A 2011 Fast full waveform inversion with random shot decimation; In: 79th Annual International Meeting, SEG, Expanded Abstracts.

  8. Ouellet F G, Gloaguen E and Giroux B 2017 A stochastic L-BFGS approach for full waveform inversion; In: 87th Annual International Meeting, SEG, Expanded Abstracts.

  9. Fichtner A and Trampert J 2011 Hessian kernels of seismic data functionals based upon adjoint techniques; Geophys. J. Int. 185 775–798.

    Article  Google Scholar 

  10. Friedlander M P and Schmidt M 2012 Hybrid determinisitic-stochastic methods for data fitting; SIAM J. Sci. Comput. 34(3) A1380–A1405.

    Article  Google Scholar 

  11. Goodfellow I, Bengio Y and Courville A 2016 Deep Learning; MIT Press, Cambridge.

    Google Scholar 

  12. Hale D 2011 Structure-oriented bilateral filtering of seismic images; In: 81st Annual International Meeting, SEG, Expanded Abstracts.

  13. Howard J and Gugger S 2020 Fastai: A layered API for deep learning; Information 11(2) 108.

    Article  Google Scholar 

  14. Jaiswal P, Zelt C, Bally A W and Dasgupta R 2008 2-D travel-time and waveform inversion for improved seismic imaging: Naga thrust and fold belt, India; Geophys. J. Int. 173(2) 642–658.

    Article  Google Scholar 

  15. Jaiswal P, Zelt C A, Dasgupta R and Nath K K 2009 Seismic imaging of the Naga thrust using multiscale waveform inversion; Geophysics 74(6) WCC129–WCC140.

    Article  Google Scholar 

  16. Kingma D P and Ba J 2014 Adam: A method for stochastic optimization; arXiv preprint arXiv:1412.6980.

  17. Krebs J R, Anderson J E, Hinkley D, Neelamani R, Lee S, Baumstein A and Lacasse M D 2009 Fast full-wavefield seismic inversion using encoded sources; Geophysics 74(6) WCC177–WCC188.

    Article  Google Scholar 

  18. Louboutin M, Lange M, Luporini F, Kukreja N, Witte P A, Herrmann F J, Velesko P and Gorman G J 2019 Devito (v3.1.0): An embedded domain-specific language for finite differences and geophysical exploration; Geosci. Model Dev. 12(3) 1165–1187.

    Article  Google Scholar 

  19. Luporini F, Lange M, Louboutin M, Kukreja N, Huckelheim J, Yount C, Witte P, Kelly P H J, Herrmann F J and Gorman G J 2018 Architecture and performance of devito, a system for automated stencil computation; CoRRabs/1807.03032.

  20. Matharu G and Sachhi M 2018 Source encoding in multiparameter full waveform inversion; Geophys. J. Int. 214(2) 792–810.

    Article  Google Scholar 

  21. Matharu G and Sachhi M 2019 A subsampled truncated-Newton method for multiparameter full-waveform inversion; Geophysics 84(3) R333–R340.

    Article  Google Scholar 

  22. McCandlish S, Kaplan J, Amodei D and Open AI Dota Team 2018 An empirical model of large batch training; arXiv preprint arXiv:1812.06162.

  23. Metivier L, Allain A, Brossier R, Merigot Q, Oudet E and Virieux J 2018 Optimal transport for mitigating cycle skipping in full-waveform inversion: A graph-space transform approach; Geophysics 83(5) R515–R540.

    Article  Google Scholar 

  24. Metivier L, Brossier R, Merigot Q, Oudet E and Virieux J 2016 Measuring the misfit between seismograms using an optimal transport distance: Application to full waveform inversion; Geophys. J. Int. 205 345–377.

    Article  Google Scholar 

  25. Metivier L, Brossier R, Operto S and Virieux J 2012 Second-order adjoint state methods for full waveform inversion; In: 74th Conference and Technical Exhibition: EAGE, Extended Abstracts.

  26. Neelamani R and Krohn C E 2008 Simultaneous sourcing without compromise; In: 70th Conference and Technical Exhibition: EAGE, Extended Abstracts.

  27. Nocedal J and Wright S J 1999 Numerical Optimization; Springer, Berlin.

    Book  Google Scholar 

  28. Operto S, Ravaut C, Improta L, Virieux J, Herrero A and DellAversana P 2004 Quantitative imaging of complex structures from dense wide-aperture seismic data by multiscale travel time and waveform inversions: A case study; Geophys. Prospect. 52(6) 625–651.

    Article  Google Scholar 

  29. Ovcharenko O, Kazei V, Kalita M, Peter D and Alkhalifah T 2019 Deep learning for low-frequency extrapolation from multioffset seismic data; Geophysics 84(6) R989–R1001.

    Article  Google Scholar 

  30. Perrone F and Sava P 2009 Comparison of shot encoding functions for reverse-time migration; In: 79th Annual International Meeting, SEG, Expanded Abstracts.

  31. Plessix R E 2006 A review of the adjoint-state method for computing the gradient of a functional with geophysical applications; Geophys. J. Int. 167(2) 495–503.

    Article  Google Scholar 

  32. Plessix R E and Cao Q 2011 A parametrization study for surface seismic full waveform inversion in an acoustic vertical transversely isotropic medium; Geophys. J. Int. 185(1) 539–556.

    Article  Google Scholar 

  33. Pratt R G 1999 Seismic waveform inversion in the frequency domain, Part I: Theory and verification in a physical scale model; Geophysics 64(3) 888–901.

    Article  Google Scholar 

  34. Prieux V, Operto S, Brossier R and Virieux J 2009 Application of acoustic full waveform inversion to the synthetic Valhall model; In: 77th Annual International Meeting, SEG, Expanded Abstracts.

  35. Rao Y and Wang Y 2017 Seismic waveform tomography with shot-encoding using a restarted L-BFGS algorithm; Sci. Rep. 7 8494.

    Article  Google Scholar 

  36. Richardson A 2018 Seismic full-waveform inversion using deep learning tools and techniques; arXiv preprint arXiv:1801.07232.

  37. Robbins H and Monro S 1951 Robust stochastic approximation approach to stochastic programming; Ann. Math. Stat. 22 400–407.

    Article  Google Scholar 

  38. Romero L A, Ghiglia D C, Ober C C and Morton S A 2000 Phase encoding of shot records in prestack migration; Geophysics 65 426–436.

    Article  Google Scholar 

  39. Shin C and Cha Y H 2009 Waveform inversion in the Laplace–Fourier domain; Geophys. J. Int. 173 922–931.

    Article  Google Scholar 

  40. Sirgue L and Pratt R G 2004 Efficient waveform inversion and imaging: A strategy for selecting temporal frequencies; Geophysics 69(1) 231–248.

    Article  Google Scholar 

  41. Smith L N 2017 Cyclical learning rates for training neural networks; In: 2017 IEEE Winter Conference on Applications of Computer Vision (WACV) 2017 464–472.

  42. Sun H and Demanet L 2020 Extrapolated full-waveform inversion with deep learning; Geophysics 85(3) R275–R288.

    Article  Google Scholar 

  43. Sun J, Niu Z, Innanen K A, Li J and Trad D O 2020 A theory-guided deep-learning formulation and optimization of seismic waveform inversion; Geophysics 85(2) 1MA–Z8.

    Article  Google Scholar 

  44. Tu N and Herrmann F J 2012 Least-squares migration of full wavefield with source encoding; In: 74th Conference and Technical Exhibition: EAGE, Extended Abstracts.

  45. Van Leeuwen T, Aravkin A Y and Herrmann F J 2011 Seismic waveform inversion by stochastic optimization; Int. J. Geophys. 2011 1–18.

    Article  Google Scholar 

  46. Van Leeuwen T and Herrmann F J 2013 Fast waveform inversion without source-encoding; Geophys. Prospect. 61 (Suppl. 1) 10–19.

    Article  Google Scholar 

  47. Versteeg R 1994 The Marmousi experience: Velocity model determination on a synthetic complex data set; Lead. Edge 13(9) 927–936.

    Article  Google Scholar 

  48. Virieux J and Operto S 2009 An overview of full-waveform inversion in exploration geophysics; Geophysics 74(6) WCC1–WCC26.

    Article  Google Scholar 

  49. Yang H, Jia J, Wu B and Gao J 2018 Mini-batch optimized full waveform inversion with geological constrained gradient filtering; J. Appl. Geophys. 152 9–16.

    Article  Google Scholar 

  50. Yang Y and Engquist B 2019 Improving optimal transport based FWI through data normalization; In: 89th Annual International Meeting, SEG, Expanded Abstracts 89 1315–1318.

Download references

Acknowledgements

We would like to thank the Associate Editor Dr Arkoprovo Biswas and two anonymous reviewers for their constructive feedback.

Author information

Affiliations

Authors

Contributions

Kuldeep formulated the research problem, wrote computer code to produce the examples and wrote the initial draft of the manuscript. Bharath Shekar provided research guidance and edited the manuscript.

Corresponding author

Correspondence to Bharath Shekar.

Additional information

Communicated by Arkoprovo Biswas

Appendix

Appendix

figurea

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Kuldeep, Shekar, B. Full waveform inversion with random shot selection using adaptive gradient descent. J Earth Syst Sci 130, 183 (2021). https://doi.org/10.1007/s12040-021-01679-y

Download citation

Keywords

  • Full waveform inversion
  • random shot selection
  • optimization