geophystechГеофизические технологииRussian Journal of Geophysical Technologies2619-1563IPGG SB RAS10.18303/2619-1563-2022-1-99geophystech-216Research ArticleСтатьиИскусственная нейронная сеть, уменьшающая численную дисперсию, для постобработки результатов сейсмического моделированияDeep neural network reducing numerical dispersion for post-processing of seismic modeling resultsГадыльшинаК. А.GadylshinaK. A.

Младший научный сотрудник лаборатории вычислительной физики горных пород Института нефтегазовой геологии и геофизики СО РАН. Основные научные интересы: методы машинного обучения в приложении к решению задач геофизики.630090, Новосибирск, просп. Акад. Коптюга, 3

Koptyug Ave., 3, Novosibirsk, 630090

https://orcid.org/0000-0003-3544-4878ЛисицаВ. В.LisitsaV. V.

Доктор физико-математических наук, заведующий лабораторией вычислительной физики горных пород Института нефтегазовой геологии и геофизики СО РАН. Основные научные интересы: численные методы для моделирования физических процессов в пористых средах.630090, Новосибирск, просп. Акад. Коптюга, 3

Koptyug Ave., 3, Novosibirsk, 630090

LisitsaVV@ipgg.sbras.ruВишневскийД. М.VishnevskyD. M.

Научный сотрудник лаборатории вычислительной физики горных пород Института нефтегазовой геологии и геофизики СО РАН. Основные научные интересы: численное моделирование сейсмических волновых полей, высокопроизводительные вычисления.630090, Новосибирск, просп. Акад. Коптюга, 3

Koptyug Ave., 3, Novosibirsk, 630090

ГадыльшинК. Г.GadylshinK. G.

Кандидат физико-математических наук, старший научный сотрудник лаборатории вычислительной физики горных пород Института нефтегазовой геологии и геофизики СО РАН. Основные научные интересы: прямые и обратные задачи сейсмики, применение методов машинного обучения для повышения качества сейсмических данных.630090, Новосибирск, просп. Акад. Коптюга, 3

Koptyug Ave., 3, Novosibirsk, 630090

Институт нефтегазовой геологии и геофизики им. А.А. Трофимука СО РАНРоссияTrofimuk Institute of Petroleum Geology and Geophysics SB RASRussian Federation2022141220220199109Copyright © Гадыльшина К.А., Лисица В.В., Вишневский Д.М., Гадыльшин К.Г., 20222022Гадыльшина К.А., Лисица В.В., Вишневский Д.М., Гадыльшин К.Г.Gadylshina K.A., Lisitsa V.V., Vishnevsky D.M., Gadylshin K.G.This work is licensed under a Creative Commons Attribution 4.0 License.https://www.rjgt.ru/jour/article/view/216

В работе описан новый подход к сейсмическому моделированию, сочетающий традиционный конечно-разностный метод с глубоким обучением. На первом шаге генерируется обучающий набор данных, рассчитанный для ограниченного числа источников. Этот набор вычисляется путем конечно-разностного моделирования на мелкой пространственно-временной сетке. На втором этапе рассчитываются сейсмограммы для всей системы наблюдения, но с использованием грубой расчетной сетки. На третьем, заключительном этапе происходит обучение искусственной нейронной сети, уменьшающей численную дисперсию, и ее применение ко всему набору данных. На тестовом материале проверена работоспособность такого подхода в задаче подавления численной дисперсии и продемонстрировано десятикратное ускорение расчетов сейсмических волновых полей.

The article describes a new approach to seismic modeling that combines calculations using traditional finite difference methods with the deep learning tools. Seismograms for the training data set are calculated using a finite difference scheme with high-quality spatial and temporal discretization. A numerical dispersion mitigation neural network is trained on the training dataset and applied to inaccurate seismograms calculated on a raw grid with a large spatial spacing. The paper presents a demonstration of this approach for 2D model; it is showing a tenfold acceleration of seismic modeling.

Сейсмическое моделированиечисленная дисперсияглубокое обучениеSeismic modellingnumerical dispersiondeep learningРазработки выполнены при Российского научного фонда (гранты №№ 22-11-00004 и 22-21-00738) и проекта ФНИ № 0266-2022-0022.ReferencesAbadi M., Agarwal A., Barham P., Brevdo E., Chen Z., Citro C., Corrado G., 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., Mané 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., Viégas F., Vinyals O., Warden P., Wattenberg M., Wicke M., Yu Y., Zheng X. TensorFlow: large-scale machine learning on heterogeneous distributed systems. – 2015, http://download.tensorflow.org/paper/whitepaper2015.pdf.Abadi M., Agarwal A., Barham P., Brevdo E., Chen Z., Citro C., Corrado G., 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., Mané 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., Viégas F., Vinyals O., Warden P., Wattenberg M., Wicke M., Yu Y., Zheng X. TensorFlow: large-scale machine learning on heterogeneous distributed systems. – 2015, http://download.tensorflow.org/paper/whitepaper2015.pdf.Chen J.-B. A 27-point scheme for a 3D frequency-domain scalar wave equation based on an average-derivative method // Geophysical Prospecting. – 2014. – Vol. 62 (2). – P. 258–277, doi: 10.1111/1365-2478.12090.Chen J.-B. A 27-point scheme for a 3D frequency-domain scalar wave equation based on an average-derivative method // Geophysical Prospecting. – 2014. – Vol. 62 (2). – P. 258–277, doi: 10.1111/1365-2478.12090.Collino F., Tsogka C. Application of the perfectly matched layer absorbing layer model to the linear elastodynamic problem in anisotropic heterogeneous media // Geophysics. – 2001. – Vol. 66. – P. 294–307, doi: 10.1190/1.1444908.Collino F., Tsogka C. Application of the perfectly matched layer absorbing layer model to the linear elastodynamic problem in anisotropic heterogeneous media // Geophysics. – 2001. – Vol. 66. – P. 294–307, doi: 10.1190/1.1444908.Fukushima K. Visual feature extraction by a multilayered network of analog threshold elements // IEEE Transactions on Systems Science and Cybernetics. – 1969. – Vol. 5 (4). – P. 322–333, doi: 10.1109/TSSC.1969.300225.Fukushima K. Visual feature extraction by a multilayered network of analog threshold elements // IEEE Transactions on Systems Science and Cybernetics. – 1969. – Vol. 5 (4). – P. 322–333, doi: 10.1109/TSSC.1969.300225.Gadylshin K., Silvestrov I., Bakulin A. Inpainting of local wavefront attributes using artificial intelligence for enhancement of massive 3-D pre-stack seismic data // Geophysical Journal International. – 2020. – Vol. 223 (3). – P. 1888–1898, doi: 10.1093/gji/ggaa422.Gadylshin K., Silvestrov I., Bakulin A. Inpainting of local wavefront attributes using artificial intelligence for enhancement of massive 3-D pre-stack seismic data // Geophysical Journal International. – 2020. – Vol. 223 (3). – P. 1888–1898, doi: 10.1093/gji/ggaa422.Glorot X., Bordes A., Bengio Y. Deep sparse rectifier neural networks // Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, PMLR. – 2011. – Vol. 15. – P. 315–323, https://proceedings.mlr.press/v15/glorot11a.html.Glorot X., Bordes A., Bengio Y. Deep sparse rectifier neural networks // Proceedings of the Fourteenth International Conference on Artificial Intelligence and Statistics, PMLR. – 2011. – Vol. 15. – P. 315–323, https://proceedings.mlr.press/v15/glorot11a.html.Guo X., Li W., Iorio F. Convolutional neural networks for steady flow approximation // KDD ’16: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. – San Francisco, CA, USA, 2016. – P. 481–490, doi: 10.1145/2939672.2939738.Guo X., Li W., Iorio F. Convolutional neural networks for steady flow approximation // KDD ’16: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. – San Francisco, CA, USA, 2016. – P. 481–490, doi: 10.1145/2939672.2939738.Kaur H., Fomel S., Pham N. Overcoming numerical dispersion of finite-difference wave extrapolation using deep learning / SEG Technical Program Expanded Abstracts. – 2019. – P. 2318–2322, doi: 10.1190/segam2019-3207486.1.Kaur H., Fomel S., Pham N. Overcoming numerical dispersion of finite-difference wave extrapolation using deep learning / SEG Technical Program Expanded Abstracts. – 2019. – P. 2318–2322, doi: 10.1190/segam2019-3207486.1.Kragh E., Christie P. Seismic repeatability, normalized RMS, and predictability // The Leading Edge. – 2002. – Vol. 21 (7). – P. 640–647, doi: 10.1190/1.1497316.Kragh E., Christie P. Seismic repeatability, normalized RMS, and predictability // The Leading Edge. – 2002. – Vol. 21 (7). – P. 640–647, doi: 10.1190/1.1497316.Kragh E., Laws R. Rough seas and statistical deconvolution // Geophysical Prospecting. – 2006. – Vol. 54 (4). – P. 475–485, doi: 10.1111/j.1365-2478.2006.00549.x.Kragh E., Laws R. Rough seas and statistical deconvolution // Geophysical Prospecting. – 2006. – Vol. 54 (4). – P. 475–485, doi: 10.1111/j.1365-2478.2006.00549.x.Lei L., Yu L., Xiong Z., Han L., Wei X., Shifan Z. Convolutional recurrent neural networks based waveform classification in seismic facies analysis // SEG Technical Program Expanded Abstracts. – 2019. – P. 2599–2603, doi: 10.1190/segam2019-3215237.1.Lei L., Yu L., Xiong Z., Han L., Wei X., Shifan Z. Convolutional recurrent neural networks based waveform classification in seismic facies analysis // SEG Technical Program Expanded Abstracts. – 2019. – P. 2599–2603, doi: 10.1190/segam2019-3215237.1.Levander A.R. Fourth-order finite-difference P-SV seismograms // Geophysics. – 1988. – Vol. 53 (11). – P. 1425–1436, doi: 10.1190/1.1442422.Levander A.R. Fourth-order finite-difference P-SV seismograms // Geophysics. – 1988. – Vol. 53 (11). – P. 1425–1436, doi: 10.1190/1.1442422.Liu Y. Optimal staggered-grid finite-difference schemes based on least-squares for wave equation modelling // Geophysical Journal International. – 2014. – Vol. 197 (2). – P. 1033–1047, doi: 10.1093/gji/ggu032.Liu Y. Optimal staggered-grid finite-difference schemes based on least-squares for wave equation modelling // Geophysical Journal International. – 2014. – Vol. 197 (2). – P. 1033–1047, doi: 10.1093/gji/ggu032.Liu Y., Sen M.K. A new time–space domain high-order finite-difference method for the acoustic wave equation // Journal of Computational Physics. – 2009. – Vol. 228 (23). – P. 8779–8806, doi: 10.1016/j.jcp.2009.08.027.Liu Y., Sen M.K. A new time–space domain high-order finite-difference method for the acoustic wave equation // Journal of Computational Physics. – 2009. – Vol. 228 (23). – P. 8779–8806, doi: 10.1016/j.jcp.2009.08.027.Martin G.S., Wiley R., Marfurt K.J. Marmousi2: An elastic upgrade for Marmousi // The Leading Edge. – 2006. – Vol. 25 (2). – P. 156–166, doi: 10.1190/1.2172306.Martin G.S., Wiley R., Marfurt K.J. Marmousi2: An elastic upgrade for Marmousi // The Leading Edge. – 2006. – Vol. 25 (2). – P. 156–166, doi: 10.1190/1.2172306.Moseley B., Nissen-Meyer T., Markham A. Deep learning for fast simulation of seismic waves in complex media // Solid Earth. – 2020. – Vol. 11. – P. 1527–1549, doi: 10.5194/se-11-1527-2020.Moseley B., Nissen-Meyer T., Markham A. Deep learning for fast simulation of seismic waves in complex media // Solid Earth. – 2020. – Vol. 11. – P. 1527–1549, doi: 10.5194/se-11-1527-2020.Ronneberger O., Fischer P., Brox T. U-Net: convolutional networks for biomedical image segmentation // Medical Image Computing and Computer-Assisted Intervention. MICCAI 2015. Lecture Notes in Computer Science. – 2015. – Vol. 9351. – P. 234–241, doi: 10.1007/978-3-319-24574-4_28.Ronneberger O., Fischer P., Brox T. U-Net: convolutional networks for biomedical image segmentation // Medical Image Computing and Computer-Assisted Intervention. MICCAI 2015. Lecture Notes in Computer Science. – 2015. – Vol. 9351. – P. 234–241, doi: 10.1007/978-3-319-24574-4_28.Siahkoohi A., Louboutin M., Herrmann F.J. The importance of transfer learning in seismic modeling and imaging // Geophysics. – 2019. – Vol. 84 (6). – P. A47–A52, doi: 10.1190/geo2019-0056.1.Siahkoohi A., Louboutin M., Herrmann F.J. The importance of transfer learning in seismic modeling and imaging // Geophysics. – 2019. – Vol. 84 (6). – P. A47–A52, doi: 10.1190/geo2019-0056.1.Tan S., Huang L. A staggered-grid finite-difference scheme optimized in the time–space domain for modeling scalar-wave propagation in geophysical problems // Journal of Computational Physics. – 2014. – Vol. 276. – P. 613–634, doi: 10.1016/j.jcp.2014.07.044.Tan S., Huang L. A staggered-grid finite-difference scheme optimized in the time–space domain for modeling scalar-wave propagation in geophysical problems // Journal of Computational Physics. – 2014. – Vol. 276. – P. 613–634, doi: 10.1016/j.jcp.2014.07.044.Virieux J. P-SV wave propagation in heterogeneous media: Velocity-stress finite-difference method // Geophysics. – 1986. – Vol. 51 (4). – P. 889–901, doi: 10.1190/1.1442147.Virieux J. P-SV wave propagation in heterogeneous media: Velocity-stress finite-difference method // Geophysics. – 1986. – Vol. 51 (4). – P. 889–901, doi: 10.1190/1.1442147.Virieux J., Operto S., Ben-Hadj-Ali H., Brossier R., Etienne V., Sourbier F., Giraud L., Haidar A. Seismic wave modeling for seismic imaging // The Leading Edge. – 2009. – Vol. 28 (5). – P. 538–544, doi: 10.1190/1.3124928.Virieux J., Operto S., Ben-Hadj-Ali H., Brossier R., Etienne V., Sourbier F., Giraud L., Haidar A. Seismic wave modeling for seismic imaging // The Leading Edge. – 2009. – Vol. 28 (5). – P. 538–544, doi: 10.1190/1.3124928.Virieux J., Calandra H, Plessix R.-E. A review of the spectral, pseudo-spectral, finite-difference and finite-element modelling techniques for geophysical imaging // Geophysical Prospecting. – 2011. – Vol. 59 (5). – P. 794–813, doi: 10.1111/j.1365-2478.2011.00967.x.Virieux J., Calandra H, Plessix R.-E. A review of the spectral, pseudo-spectral, finite-difference and finite-element modelling techniques for geophysical imaging // Geophysical Prospecting. – 2011. – Vol. 59 (5). – P. 794–813, doi: 10.1111/j.1365-2478.2011.00967.x.Wu H., Zhang B., Lin T., Cao D., Lou Y. Semiautomated seismic horizon interpretation using the encoder-decoder convolutional neural network // Geophysics. – 2019. – Vol. 84 (6). – P. B403–B417, doi: 10.1190/geo2018-0672.1.Wu H., Zhang B., Lin T., Cao D., Lou Y. Semiautomated seismic horizon interpretation using the encoder-decoder convolutional neural network // Geophysics. – 2019. – Vol. 84 (6). – P. B403–B417, doi: 10.1190/geo2018-0672.1.Wu X., Geng Z., Shi Y., Pham N., Fomel S., Caumon G. Building realistic structure models to train convolutional neural networks for seismic structural interpretation // Geophysics. – 2020. – Vol. 85. – P. WA27–WA39, doi: 10.1190/geo2019-0375.1.Wu X., Geng Z., Shi Y., Pham N., Fomel S., Caumon G. Building realistic structure models to train convolutional neural networks for seismic structural interpretation // Geophysics. – 2020. – Vol. 85. – P. WA27–WA39, doi: 10.1190/geo2019-0375.1.Yu S., Ma J. Deep Learning for Geophysics: Current and Future Trends // Reviews of Geophysics. – 2021. – Vol. 59 (3). – e2021RG000742, doi: 10.1029/2021RG000742.Yu S., Ma J. Deep Learning for Geophysics: Current and Future Trends // Reviews of Geophysics. – 2021. – Vol. 59 (3). – e2021RG000742, doi: 10.1029/2021RG000742.Yu S., Ma J.van, Wang W. Deep learning for denoising // Geophysics. – 2019. – Vol. 84. – P. V333–V350, doi: 10.1190/geo2018-0668.1.Yu S., Ma J.van, Wang W. Deep learning for denoising // Geophysics. – 2019. – Vol. 84. – P. V333–V350, doi: 10.1190/geo2018-0668.1.Zhu W., Mousavi S.M., Beroza G.C. Seismic signal denoising and decomposition using deep neural networks // IEEE Transactions on Geoscience and Remote Sensing. – 2019. – Vol. 57 (11). – P. 9476–9488, doi: 10.1109/TGRS.2019.2926772.Zhu W., Mousavi S.M., Beroza G.C. Seismic signal denoising and decomposition using deep neural networks // IEEE Transactions on Geoscience and Remote Sensing. – 2019. – Vol. 57 (11). – P. 9476–9488, doi: 10.1109/TGRS.2019.2926772.

The authors declare that there are no conflicts of interest present.