Automatic differentiation-based weighted instantaneous phase inversion

Abstract

Traditional full waveform inversion objective functions typically rely on discrepancies between waveforms, making full waveform inversion highly dependent on the initial model and the low-frequency components in the seismic data. The instantaneous phase of the seismic signal reflects the kinematic information of seismic waves and has a more linear relationship with the subsurface velocity structure, aiding in the retrieval of low-wavenumber components of velocity models. Conversely, waveform information is instrumental in achieving high-resolution inversion results while maintaining algorithmic stability. In addition, automatic differentiation provides an efficient and accurate mechanism for computing gradients by systematically applying the chain rule across the computational graph, yielding reliable derivative information for optimization. To balance the contributions of waveform and phase information in velocity inversion, within the framework of automatic differentiation, we integrate instantaneous phase and waveform information to propose an automatic differentiation-based weighted instantaneous phase inversion method. Numerical tests using the Marmousi model and the Overthrust model demonstrate that the proposed method achieves more accurate velocity inversion results.

Keywords

Full waveform inversion / Instantaneous phase / Weighted objective function / Automatic differentiation / Cycle skipping

Funding

This research was funded by the National Natural Science Foundation of China (Grant No. 42104116).

Conflict of interest

The authors declare they have no competing interests.

Author contributions

Conceptualization: Mengzi Wang, Yong Hu Formal analysis: Chunyuan Shi, Zhihan Zhang Investigation: Mengzi Wang, Yong Hu, Zhihan Zhang Methodology: Mengzi Wang, Yong Hu Software: Mengzi Wang, Zhihan Zhang Supervision: Mengzi Wang, Yong Hu Writing–original draft: Mengzi Wang, Chunyuan Shi Writing–review & editing: Xiaoyuan Liao

Availability of data

Data are available from the corresponding author upon reasonable request.

References

[1] Lailly P. The seismic inverse problem as a sequence of before stack migrations. In: Bednar JB, Robinson E, Weglein A, eds. Conference on Inverse Scattering—Theory and Application. Philadelphia, PA: SIAM; 1983: 206-220.

[2] Tarantola A. Inversion of seismic reflection data in the acoustic approximation. Geophysics. 1984; 49(8): 1259-1266. doi: 10.1190/1.1441754

[3] Fichtner A , Bunge HP , Igel H. The adjoint method in seismology: I. Theory. Phys Earth Planet Inter. 2006; 157(1—2): 86-104. doi: 10.1016/j.pepi.2006.03.016

[4] Virieux J , Operto S. An overview of full—waveform inversion in exploration geophysics. Geophysics. 2009; 74(6): WCC1-WCC26. doi: 10.1190/1.3238367

[5] Plessix RE. A review of the adjoint—state method for computing the gradient of a functional with geophysical applications. Geophys J Int. 2006; 167(2): 495-503. doi: 10.1111/j.1365—246X.2006.02978.x

[6] Luo JR , Wang BF. Initial model building in time domain elastic full waveform inversion using the instantaneous phase information. Prog Geophys. 2018; 33(6): 2435-2440. [In Chinese]. doi: 10.6038/pg2018CC0215

[7] Sun SY , Hu GH , He BH , Du ZY , Xu WC. Reflection waveform inversion and its application to onshore seismic data. Prog Geophys. 2021; 36(6): 2566-2572. [In Chinese]. doi: 10.6038/pg2021EE0462

[8] Li ZC , Wang ZY , Huang JP , Cui C. Multi—scale full waveform inversion based on gradient decomposition in wavenumber domain. Chin J Geophys. 2022; 65(7): 2693-2703. [In Chinese]. doi: 10.6038/cjg2022P0508

[9] Wang QQ , Song P , Hua QF , et al. Full waveform inversion based on Adam algorithm with optimized parameters. Chin J Geophys. 2023; 66(11): 4654-4663. [In Chinese]. doi: 10.6038/cjg2022Q0879

[10] Li JS , Liu WK , Liang YX , Wu JY , Li ZG. Time—domain full waveform inversion based on high—order amplitude information. Prog Geophys . 2023; 38(4): 1603-1609. [In Chinese]. doi: 10.6038/pg2023GG0129

[11] Kang PF , Hu Y , Liu RD , et al. Local—scale frequency—wavenumber domain phase inversion. Prog Geophys. 2025; 40(1): 155-165. [In Chinese]. doi: 10.6038/pg2025II0034

[12] Shah NK , Warner MR , Washbourne JK , Guasch L , Umpleby AP. A phase—unwrapped solution for overcoming a poor starting model in full wavefield inversion. In: Proceedings of the 74th EAGE Conference and Exhibition Incorporating EUROPEC 2012; Copenhagen, Denmark: Jun 4—7, 2012; European Association of Geoscientists & Engineers; 2012: cp—293—00460. doi: 10.3997/2214—4609.20148455

[13] Choi Y , Alkhalifah T. Unwrapped phase inversion with an exponential damping. Geophysics. 2015; 80(5): R251-R264. doi: 10.1190/geo2014—0498.1

[14] Luo JR , Wu RS , Gao FC. Time—domain full waveform inversion using instantaneous phase information with damping. J Geophys Eng. 2018; 15(3): 1032-1041. doi: 10.1088/1742—2140/aaa984

[15] Hu Y , Wu RS , Han LG , Zhang P. Joint multiscale direct envelope inversion of phase and amplitude in the time—frequency domain. IEEE Trans Geosci Remote Sens. 2019; 57(7): 5108-5120. doi: 10.1109/TGRS.2019.2896936

[16] Zhu HJ , Fomel S. Building good starting models for full—waveform inversion using adaptive matching filtering misfit. Geophysics. 2016; 81(5): U61-U72. doi: 10.1190/geo2015—0596.1

[17] Warner M , Guasch L. Adaptive waveform inversion: Theory. Geophysics. 2016; 81(6): R429-R445. doi: 10.1190/geo2015—0387.1

[18] Hu Y , Han LG , Yu JL , Chen RD. Time—frequency domain multi—scale full waveform inversion based on adaptive non—stationary phase correction. Chin J Geophys. 2018; 61(7): 2969-2988. [In Chinese]. doi: 10.6038/cjg2018L0421

[19] Liu JG , Xu KL. Automatic differentiation and its applications in physics simulation. Acta Phys Sin . 2021; 70(14): 149402. [In Chinese]. doi: 10.7498/aps.70.20210813

[20] Ketkar N , Moolayil J. Automatic Differentiation in Deep Learning. In: Deep Learning with Python. Berkeley, CA: Apress; 2021: 133-145. doi: 10.1007/978—1—4842—5364—9_4

[21] LeCun Y , Bengio Y , Hinton G. Deep learning. Nature. 2015; 521(7553): 436-444. doi: 10.1038/nature14539

[22] Goodfellow I , Bengio Y , Courville A. Deep Learning . Cambridge, MA: MIT Press; 2016.

[23] Nolan JF. Analytical differentiation on a digital computer [master’s thesis]. Cambridge, MA: Massachusetts Institute of Technology, Department of Mathematics; 1953. https://hdl.handle.net/1721.1/12297

[24] Speelpenning B. Compiling Fast Partial Derivatives of Functions Given by Algorithms [PhD thesis]. Urbana, IL: University of Illinois at Urbana—Champaign, Department of Computer Science; 1980. Report No. UIUC DS—R—80—1002. doi: 10.2172/5254402

[25] Kasim MF , Vinko SM. ξ—torch: differentiable scientific computing library. arXiv . Preprint posted online October 6, 2020. doi: 10.48550/arXiv.2010.01921

[26] Sambridge M , Rickwood P , Rawlinson N , Sommacal S. Automatic differentiation in geophysical inverse problems. Geophys J Int . 2007; 170(1): 1-8. doi: 10.1111/j.1365—246X.2007.03400.x

[27] Tan L , Brytik V , Baumstein A , Hinkley D. Verification of gradient and hessian computation for full wavefield inversion using automatic differentiation. In: Levin S, ed. SEG Technical Program Expanded Abstracts 2010; SEG International Exposition and 80th Annual Meeting; 17—22 October 2010; Denver, CO, USA. Society of Exploration Geophysicists; 2010: 2762-2766. doi: 10.1190/1.3513417

[28] Liao W. An accurate and efficient algorithm for parameter estimation of 2D acoustic wave equation. Int J Appl Phys Math. 2011; 1(2): 96-100. doi: 10.7763/IJAPM.2011.V1.19

[29] Richardson A. Seismic full—waveform inversion using deep learning tools and techniques. arXiv. Preprint posted online January 22, 2018. doi: 10.48550/arXiv.1801.07232

[30] Cao D , Liao W. A computational method for full waveform inversion of crosswell seismic data using automatic differentiation. Comput Phys Commun. 2015; 188: 47-58. doi: 10.1016/j.cpc.2014.11.002

[31] Zhu W , Xu K , Darve E. A general approach to seismic inversion with automatic differentiation. Comput Geosci. 2021; 151: 104751. doi: 10.1016/j.cageo.2021.104751

[32] Song C , Wang Y , Richardson A , Liu C. Weighted envelope correlation—based waveform inversion using automatic differentiation. IEEE Trans Geosci Remote Sens. 2023; 61: 1-11. doi: 10.1109/TGRS.2023.3300127

[33] Liu F , Li H , Zou G , Li J. Automatic differentiation—based full waveform inversion with flexible workflows. J Geophys Res Mach Learn Comput. 2025; 2(1): e2024JH000542. doi: 10.1029/2024JH000542