×

Full waveform inversion and the truncated Newton method. (English) Zbl 1357.86001

Authors’ abstract: Full waveform inversion (FWI) is a powerful method for reconstructing subsurface parameters from local measurements of the seismic wavefield. This method consists in minimizing the distance between predicted and recorded data. The predicted data are computed as the solution of a wave-propagation problem. Conventional numerical methods for the solution of FWI problems are gradient-based methods, such as the preconditioned steepest descent, the nonlinear conjugate gradient, or more recently the \(l\)-BFGS quasi-Newton algorithm. In this study, we investigate the desirability of applying a truncated Newton method to FWI. The inverse Hessian operator plays a crucial role in the parameter reconstruction, as it should help to mitigate finite-frequency effects and to better remove artifacts arising from multiscattered waves. For multiparameter reconstruction, the inverse Hessian operator also offers the possibility of better removing trade-offs due to coupling effects between parameter classes. The truncated Newton method allows us to better account for this operator. This method is based on the computation of the Newton descent direction by solving the corresponding linear system through the conjugate gradient method. The large-scale nature of FWI problems requires us, however, to carefully implement this method to avoid prohibitive computational costs. First, this requires working in a matrix-free formalism and the capability of efficiently computing Hessian-vector products. For this purpose, we propose general second-order adjoint state formulas. Second, special attention must be paid to defining the stopping criterion for the inner linear iterations associated with the computation of the Newton descent direction. We propose several possibilities and establish a theoretical link between the Steihaug-Toint method, based on trust regions, and the Eisenstat and Walker stopping criterion, designed for a method globalized by linesearch. We investigate the application of the truncated Newton method to two case studies. The first is a standard case study in seismic imaging based on the Marmousi model. The second is inspired by a near-surface imaging problem for the reconstruction of high-velocity structures. In the latter case, we demonstrate that the presence of large amplitude multiscattered waves prevents standard methods from converging, while the truncated Newton method provides more reliable results.
Editorial remark: This is a republication of [Zbl 1266.86002] with minor changes.

MSC:

86-08 Computational methods for problems pertaining to geophysics
86A15 Seismology (including tsunami modeling), earthquakes
35R30 Inverse problems for PDEs
65T60 Numerical methods for wavelets
94A11 Application of orthogonal and other special functions
94A08 Image processing (compression, reconstruction, etc.) in information and communication theory
49M15 Newton-type methods
90C06 Large-scale problems in mathematical programming
65K05 Numerical mathematical programming methods

Citations:

Zbl 1266.86002
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] P. Amestoy, I. S. Duff, and J. Y. L’Excellent, {\it Multifrontal parallel distributed symmetric and unsymmetric solvers}, Comput. Methods Appl. Mech. Engrg., 184 (2000), pp. 501-520. · Zbl 0956.65017
[2] M. Benzi, {\it Preconditioning techniques for large linear systems: A survey}, J. Comput. Phys., 182 (2002), pp. 418-477. · Zbl 1015.65018
[3] J.-P. Bérenger, {\it A perfectly matched layer for absorption of electromagnetic waves}, J. Comput. Phys., 114 (1994), pp. 185-200. · Zbl 0814.65129
[4] J. F. Bonnans, J. C. Gilbert, C. Lemaréchal, and C. A. Sagastizábal, {\it Numerical Optimization: Theoretical and Practical Aspects}, Universitext, Springer-Verlag, Berlin, 2006. · Zbl 1108.65060
[5] D. Borisov and S. C. Singh, {\it Three-dimensional elastic full waveform inversion in a marine environment using multicomponent ocean-bottom cables: A synthetic study}, Geophys. J. Internat., 201 (2015), pp. 1215-1234.
[6] C. Bunks, F. M. Salek, S. Zaleski, and G. Chavent, {\it Multiscale seismic waveform inversion}, Geophys., 60 (1995), pp. 1457-1473.
[7] E. Bŏzdag, J. Trampert, and J. Tromp, {\it Misfit functions for full waveform inversion based on instantaneous phase and envelope measurements}, Geophys. J. Internat., 185 (2011), pp. 845-870.
[8] R. H. Byrd, P. Lu, J. Nocedal, and C. Zhu, {\it A limited memory algorithm for bound constrained optimization}, SIAM J. Sci. Statist. Comput., 16 (1995), pp. 1190-1208, . · Zbl 0836.65080
[9] C. Castellanos, L. Métivier, S. Operto, R. Brossier, and J. Virieux, {\it Fast full waveform inversion with source encoding and second-order optimization methods}, Geophys. J. Internat., 200 (2015), pp. 720-744.
[10] G. Chavent, {\it Analyse fonctionnelle et identification de coefficients répartis dans les équations aux dérivées partielles}, Ph.D. thesis, Université de Paris, Paris, 1971. · Zbl 0226.35006
[11] Y. Choi and C. Shin, {\it Frequency-domain elastic full waveform inversion using the new pseudo-Hessian matrix: Experience of elastic Marmousi 2 synthetic data}, Bull. Seismological Soc. Amer., 98 (2008), pp. 2402-2415.
[12] J. F. Claerbout, {\it Imaging the Earth’s Interior}, Blackwell Scientific, Oxford, UK, 1985.
[13] A. J. Devaney, {\it Geophysical diffraction tomography}, IEEE Trans. Geosci. Remote Sensing, GE-22 (1984), pp. 3-13.
[14] I. S. Duff and J. K. Reid, {\it The multifrontal solution of indefinite sparse symmetric linear systems}, ACM Trans. Math. Software, 9 (1983), pp. 302-325. · Zbl 0515.65022
[15] B. Dupuy, A. Asnaashari, R. Brossier, S. Garambois, L. Métivier, A. Ribodetti, and J. Virieux, {\it A downscaling strategy from FWI to microscale reservoir properties from high-resolution images}, The Leading Edge, 35 (2016), pp. 146-150.
[16] S. C. Eisenstat and H. F. Walker, {\it Choosing the forcing terms in an inexact Newton method}, SIAM J. Sci. Comput., 17 (1996), pp. 16-32, . · Zbl 0845.65021
[17] B. Engquist and B. D. Froese, {\it Application of the Wasserstein metric to seismic signals}, Commun. Math. Sci., 12 (2014), pp. 979-988. · Zbl 1305.86006
[18] I. Epanomeritakis, V. Akçelik, O. Ghattas, and J. Bielak, {\it A Newton-CG method for large-scale three-dimensional elastic full waveform seismic inversion}, Inverse Problems, 24 (2008), pp. 1-26. · Zbl 1142.65052
[19] A. Fichtner, B. L. N. Kennett, H. Igel, and H. P. Bunge, {\it Theoretical background for continental- and global-scale full-waveform inversion in the time-frequency domain}, Geophys. J. Internat., 175 (2008), pp. 665-685.
[20] A. Fichtner, B. L. N. Kennett, H. Igel, and H. P. Bunge, {\it Full waveform tomography for radially anisotropic structure: New insights into present and past states of the Australasian upper mantle}, Earth Planet. Sci. Lett., 290 (2010), pp. 270-280.
[21] A. Fichtner and J. Trampert, {\it Hessian kernels of seismic data functionals based upon adjoint techniques}, Geophys. J. Internat., 185 (2011), pp. 775-798.
[22] A. Fichtner and T. van Leeuwen, {\it Resolution analysis by random probing}, J. Geophys. Res. Solid Earth, 120 (2015), 2015JB012106, .
[23] D. N. Daescu, F. X. Le Dimet, and I. M. Navon, {\it Second-order information in data assimilation}, Monthly Weather Rev., 130 (2002), pp. 629-648.
[24] O. Gauthier, J. Virieux, and A. Tarantola, {\it Two-dimensional nonlinear inversion of seismic waveforms: Numerical results}, Geophys., 51 (1986), pp. 1387-1403.
[25] G. H. Golub, {\it Matrix Computation}, 3rd ed., Johns Hopkins Stud. Math. Sci., Johns Hopkins University Press, Baltimore, MD, 1996. · Zbl 0865.65025
[26] N. I. M. Gould, S. Lucidi, M. Roma, and Ph. L. Toint, {\it Solving the trust-region subproblem using the Lanczos method}, SIAM J. Optim., 9 (1999), pp. 504-525, . · Zbl 1047.90510
[27] B. Hustedt, S. Operto, and J. Virieux, {\it Mixed-grid and staggered-grid finite difference methods for frequency domain acoustic wave modelling}, Geophys. J. Internat., 157 (2004), pp. 1269-1296.
[28] K. A. Innanen, {\it Seismic AVO and the inverse Hessian in precritical reflection full waveform inversion}, Geophys. J. Internat., 199 (2014), pp. 717-734.
[29] M. Jannane, W. Beydoun, E. Crase, D. Cao, Z. Koren, E. Landa, M. Mendes, A. Pica, M. Noble, G. Roeth, S. Singh, R. Snieder, A. Tarantola, and D. Trezeguet, {\it Wavelengths of Earth structures that can be resolved from seismic reflection data}, Geophys., 54 (1989), pp. 906-910.
[30] N. Korta, A. Fichtner, and V. Sallares, {\it Block-diagonal approximate Hessian for preconditioning in full waveform inversion}, in Expanded Abstracts, 75th EAGE Conference and Exhibition incorporating SPE EUROPEC 2013, London, 2013.
[31] J. Krebs, J. Anderson, D. Hinkley, R. Neelamani, S. Lee, A. Baumstein, and M. D. Lacasse, {\it Fast full-wavefield seismic inversion using encoded sources}, Geophys., 74 (6) (2009), pp. WCC105-WCC116.
[32] P. Lailly, {\it The seismic inverse problem as a sequence of before stack migrations}, in Conference on Inverse Scattering: Theory and Application, J. B. Bednar, R. Redner, E. Robinson, and A. Weglein, eds., SIAM, Philadelphia, 1983, pp. 206-220.
[33] F. X. Le Dimet and O. Talagrand, {\it Variational algorithms for analysis and assimilation of meteorological observations: Theoretical aspects}, Tellus A, 38A (1986), pp. 97-110.
[34] J. Lee, J. Zhang, and C.-C. Lu, {\it Incomplete LU preconditioners for large scale dense complex linear systems from electromagnetic wave scattering problems}, J. Comput. Phys., 185 (2003), pp. 158-175. · Zbl 1017.65027
[35] X. Li, A. Aravkin, T. van Leeuwen, and F. Herrmann, {\it Fast randomized full-waveform inversion with compressive sensing}, Geophys., 77 (2012), pp. A13-A17.
[36] J. L. Lions, {\it Contrôle Optimal de Systèmes Gouvernés par des Équations aux Dérivées Partielles}, Dunod, Paris, 1968. · Zbl 0179.41801
[37] S. Luo and P. Sava, {\it A deconvolution-based objective function for wave-equation inversion}, SEG Technical Program Expanded Abstracts, 30 (2011), pp. 2788-2792.
[38] Y. Luo and G. T. Schuster, {\it Wave-equation traveltime inversion}, Geophys., 56 (1991), pp. 645-653.
[39] O. A. Marques, {\it Blzpack: Description and User’s Guide}, Tech. report, 1997.
[40] G. S. Martin, R. Wiley, and K. J. Marfurt, {\it Marmousi2: An elastic upgrade for Marmousi}, The Leading Edge, 25 (2006), pp. 156-166.
[41] L. Métivier, {\it Utilisation des équations Euler-PML en milieu hétérogène borné pour la résolution d’un problème inverse en géophysique}, ESAIM Proc., 27 (2009), pp. 156-170. · Zbl 1177.86015
[42] L. Métivier, F. Bretaudeau, R. Brossier, S. Operto, and J. Virieux, {\it Full waveform inversion and the truncated Newton method: Quantitative imaging of complex subsurface structures}, Geophys. Prospecting, 62 (2014), pp. 1353-1375.
[43] L. Métivier and R. Brossier, {\it The SEISCOPE optimization toolbox: A large-scale nonlinear optimization library based on reverse communication}, Geophys., 81 (2016), pp. F11-F25.
[44] L. Métivier, R. Brossier, Q. Mérigot, E. Oudet, and J. Virieux, {\it Increasing the robustness and applicability of full waveform inversion: An optimal transport distance strategy}, The Leading Edge, submitted (2016). · Zbl 1410.86018
[45] L. Métivier, R. Brossier, Q. Mérigot, E. Oudet, and J. Virieux, {\it Measuring the misfit between seismograms using an optimal transport distance: Application to full waveform inversion}, Geophys. J. Internat., 205 (2016), pp. 345-377. · Zbl 1410.86018
[46] L. Métivier, R. Brossier, Q. Mérigot, E. Oudet, and J. Virieux, {\it An optimal transport approach for seismic tomography: Application to {\it 3}D full waveform inversion}, Inverse Problems, submitted (2016). · Zbl 1410.86018
[47] L. Métivier, R. Brossier, S. Operto, and J. Virieux, {\it Acoustic multi-parameter FWI for the reconstruction of P-wave velocity, density and attenuation: Preconditioned truncated Newton approach}, in Expanded Abstracts, 85th Annual SEG Meeting (New Orleans), 2015.
[48] J. L. Morales and J. Nocedal, {\it Automatic preconditioning by limited memory quasi-Newton updating}, SIAM J. Optim., 10 (2000), pp. 1079-1096, . · Zbl 1020.65019
[49] J. L. Morales and J. Nocedal, {\it Enriched methods for large-scale unconstrained optimization}, Comput. Optim. Appl., 21 (2000), pp. 143-154. · Zbl 0988.90035
[50] J. Moré, {\it The Levenberg-Marquardt algorithm: Implementation and theory}, in Numerical Analysis, G. Watson, ed., Lecture Notes in Math. 630, Springer, Berlin, Heidelberg, 1978, pp. 105-116, .
[51] S. G. Nash and J. Nocedal, {\it A numerical study of the limited memory BFGS method and the truncated-Newton method for large scale optimization}, SIAM J. Optim., 1 (1991), pp. 358-372, . · Zbl 0756.65091
[52] S. G. Nash, {\it A survey of truncated Newton methods}, J. Comput. Appl. Math., 124 (2000), pp. 45-59. · Zbl 0969.65054
[53] J. Nocedal, {\it Updating quasi-Newton matrices with limited storage}, Math. Comp., 35 (1980), pp. 773-782. · Zbl 0464.65037
[54] J. Nocedal and S. J. Wright, {\it Numerical Optimization}, Springer, New York, 1999. · Zbl 0930.65067
[55] J. Nocedal and S. J. Wright, {\it Numerical Optimization}, 2nd ed., Springer, New York, 2006. · Zbl 1104.65059
[56] G. Nolet, {\it A Breviary of Seismic Tomography}, Cambridge University Press, Cambridge, UK, 2008. · Zbl 1166.86001
[57] S. Operto, R. Brossier, Y. Gholami, L. Métivier, V. Prieux, A. Ribodetti, and J. Virieux, {\it A guided tour of multiparameter full waveform inversion for multicomponent data: From theory to practice}, The Leading Edge, Special Section: Full Waveform Inversion (2013), pp. 1040-1054.
[58] S. Operto, A. Miniussi, R. Brossier, L. Combe, L. Métivier, V. Monteiller, A. Ribodetti, and J. Virieux, {\it Efficient {\it 3}-D frequency-domain mono-parameter full-waveform inversion of ocean-bottom cable data: Application to Valhall in the visco-acoustic vertical transverse isotropic approximation}, Geophys. J. Internat., 202 (2015), pp. 1362-1391.
[59] D. Peter, D. Komatitsch, Y. Luo, R. Martin, N. Le Goff, E. Casarotti, P. Le Loher, F. Magnoni, Q. Liu, C. Blitz, T. Nissen-Meyer, P. Basini, and J. Tromp, {\it Forward and adjoint simulations of seismic wave propagation on fully unstructured hexahedral meshes}, Geophys. J. Internat., 186 (2011), pp. 721-739.
[60] R. E. Plessix, {\it A review of the adjoint-state method for computing the gradient of a functional with geophysical applications}, Geophys. J. Internat., 167 (2006), pp. 495-503.
[61] R. E. Plessix and C. Perkins, {\it Full waveform inversion of a deep water ocean bottom seismometer dataset}, First Break, 28 (2010), pp. 71-78.
[62] M. J. D. Powell, {\it A new algorithm for unconstrained optimization}, in Nonlinear Programming, Academic Press, New York, 1970, pp. 31-66.
[63] R. G. Pratt, {\it Inverse theory applied to multi-source cross-hole tomography. Part II: Elastic wave-equation method}, Geophys. Prospecting, 38 (1990), pp. 311-330.
[64] R. G. Pratt, {\it Seismic waveform inversion in the frequency domain. Part I: Theory and verification in a physical scale model}, Geophys., 64 (1999), pp. 888-901.
[65] R. G. Pratt, C. Shin, and G. J. Hicks, {\it Gauss-Newton and full Newton methods in frequency-space seismic waveform inversion}, Geophys. J. Internat., 133 (1998), pp. 341-362.
[66] R. G. Pratt and M. H. Worthington, {\it Inverse theory applied to multi-source cross-hole tomography. Part I: Acoustic wave-equation method}, Geophys. Prospecting, 38 (1990), pp. 287-310.
[67] Y. Saad, {\it Iterative Methods for Sparse Linear Systems}, 2nd ed., SIAM, Philadelphia, 2003. · Zbl 1031.65046
[68] Y. Saad, M. Yeung, J. Erhel, and F. Guyomarc’h, {\it A deflated version of the conjugate gradient algorithm}, SIAM J. Sci. Comput., 21 (2000), pp. 1909-1926, . · Zbl 0955.65021
[69] C. Shin, S. Jang, and D. J. Min, {\it Improved amplitude preservation for prestack depth migration by inverse scattering theory}, Geophys. Prospecting, 49 (2001), pp. 592-606.
[70] L. Sirgue, O. I. Barkved, J. Dellinger, J. Etgen, U. Albertin, and J. H. Kommedal, {\it Full waveform inversion: The next leap forward in imaging at Valhall}, First Break, 28 (2010), pp. 65-70.
[71] T. Steihaug, {\it The conjugate gradient method and trust regions in large scale optimization}, SIAM J. Numer. Anal., 20 (1983), pp. 626-637. · Zbl 0518.65042
[72] W. W. Symes, {\it Migration velocity analysis and waveform inversion}, Geophys. Prospecting, 56 (2008), pp. 765-790.
[73] C. Tape, Q. Liu, A. Maggi, and J. Tromp, {\it Seismic tomography of the southern California crust based on spectral-element and adjoint methods}, Geophys. J. Internat., 180 (2010), pp. 433-462.
[74] A. Tarantola, {\it Inversion of seismic reflection data in the acoustic approximation}, Geophys., 49 (1984), pp. 1259-1266.
[75] T. van Leeuwen and F. Herrmann, {\it Fast waveform inversion without source-encoding}, Geophys. Prospecting, 61 (2012), pp. 10-19.
[76] D. Vigh, K. Jiao, D. Watts, and D. Sun, {\it Elastic full-waveform inversion application using multicomponent measurements of seismic data collection}, Geophys., 79 (2014), pp. R63-R77.
[77] M. Warner and L. Guasch, {\it Adaptive waveform inversion–FWI without cycle skipping–Theory}, in Proceedings of the 76th EAGE Conference and Exhibition 2014, We E106 13.
[78] M. Warner, A. Ratcliffe, T. Nangoo, J. Morgan, A. Umpleby, N. Shah, V. Vinje, I. Stekl, L. Guasch, C. Win, G. Conroy, and A. Bertrand, {\it Anisotropic {\it 3}D full-waveform inversion}, Geophys., 78 (2013), pp. R59-R80.
[79] H. Zhu, E. Bŏzdag, D. Peter, and J. Tromp, {\it Structure of the European upper mantle revealed by adjoint tomography}, Nature Geosci., 5 (2012), pp. 493-498.
[80] H. Zhu and S. Fomel, {\it Building good starting models for full-waveform inversion using adaptive matching filtering misfit}, Geophys., 81 (2016), pp. U61-U72.
[81] H. Zhu, S. Li, S. Fomel, G. Stadler, and O. Ghattas, {\it A Bayesian approach to estimate uncertainty for full-waveform inversion using a priori information from depth migration}, Geophys., 81 (2016), pp. R307-R323.
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.