Convolutional autoencoder with local wavefield characteristic constraint for suppressing near-surface seismic scattered waves

Abstract

In mountainous seismic exploration, complex near-surface environments cause strong wave scattering, reducing the signal-to-noise ratio and complicating data processing. Therefore, suppressing scattered waves is crucial. To address this issue, this study proposes a convolutional autoencoder constrained by local scattered wave characteristics to suppress near-surface scattered waves (NSWs) in full-wavefield seismic data. This method uses the scattered waves predicted by seismic interferometry as the network input, and the original seismic records containing true scattered waves as the label. Since there are differences in amplitude and phase between the predicted scattered waves and the true scattered waves, the network introduces local wavefield features of the scattered waves for constraint, and adds a smoothness regularization term in the loss function to ensure the continuity of the output waveform. After training, the network maps the energy of predicted scattered waves into the actual seismic records, thereby accurately extracting scattered waves. Finally, by subtracting the network output from the original records, clean data with scattered waves removed can be obtained. This method is a self-supervised learning strategy and does not require additional clean signal samples. During training, the weights of each item in the loss function can be dynamically adjusted to guide the network to focus on local scattered-wave features, avoid learning effective wave information, and ensure that only scattered-wave components are retained in the output. Practical application results show that this method can effectively suppress NSWs and improve the signal-to-noise ratio of seismic data.

Keywords

Near-surface scattered waves / Convolutional autoencoder / Interferometry / Wavefield characteristics

Acknowledgments

None.

Funding

This work is supported by the China National Petroleum Corporation Science and Technology Project, “Key Technologies for Seismic Physical Modeling, Acquisition, and Processing in the Tajik Basin” (2024DJ9905).

Conflict of interest

The authors declare that they have no conflict of interest.

Author contributions

Conceptualization: Daling Hou

Formal analysis: Daling Hou

Funding acquisition: Meng Li

Investigation: Daling Hou

Methodology: Daling Hou

Visualization: Daling Hou

Writing-original draft: Daling Hou

Writing-review & editing: Daling Hou, Jixiang Xu

Availability of data

The data used in this study are available from the corresponding author upon reasonable request.

References

[1] Wu X, Li Y, Zhang M, He J, Long Z. Causes and countermeasures for low signal-to-noise ratio in seismic data in complex areas (part 1) - surface scattered waves are the root cause of low signal-to-noise ratio in seismic data. Nat Gas Ind. 2012; 32(1):27-32. [Article in Chinese]. doi: 10.3787/j.issn.1000-0976.2012.01.005

[2] Shao J, Tang J, Sun C. Progress of seismic wave scattering theory and application. Prog Geophys. 2016; 31(1):334-343. doi: 10.6038/pg20160139

[3] Campman XH, Wijk Kvan, Scales JA, Herman GC. Imaging and suppressing near-receiver scattered surface waves. Geophysics. 2005; 70(2):V21-V29. doi: 10.1190/1.1884831

[4] Liu Y, Hu G, Fan T. Suppression of strong scattered waves using the transverse component. J Appl Geophys. 2015; 112:226-235. doi: 10.1016/j.jappgeo.2014.11.021

[5] Almuhaidib AM, Toksöz MN. Suppression of near-surface scattered body-to-surface waves: A steerable and nonlinear filtering approach. Geophys Prospect. 2016; 64(2):392-405. doi: 10.1111/1365-2478.12275

[6] Xu J, McLean JX, Song J. A near-surface seismic scattered wave separation method. Pet Explor Dev. 2014; 41(6):771-777. doi: 10.1016/S1876-3804(14)60091-4

[7] Halliday D, Bilsby P, West L, Kragh E, Quigley J. Scattered ground-roll attenuation using model-driven interferometry. Geophys Prospect. 2015; 63(1):116-132. doi: 10.1111/1365-2478.12165

[8] Xu J, Dong S, Cui H, Zhang Y, Hu Y, Sun X. Near-surface scattered waves enhancement with source-receiver interferometry. Geophysics. 2018; 83(6):Q49-Q69. doi: 10.1190/geo2017-0806.1

[9] Zhou P, Li P, Wang Y, Wei Z. Principle and Practice of Deep Convolution Neural Network. Beijing, China: Publishing House of Electronics Industry. 2020. p. 17-18.

[10] Yu S, Ma J, Wang W. Deep learning for denoising. Geophysics. 2019;84:V333-V350. doi: 10.1190/geo2018-0668.1

[11] Yu S, Yang W, Li H, Wang H, Ma H. Scattered ground roll intelligent attenuation based on deep learning. Chin Sci Bull (Chinese Version). 2021; 66(18):2343-2354. doi: 10.1360/TB-2020-0564

[12] Liu W, Lü Q, Chen Q, et al. A synthetic denoising algorithm for full-waveform induced polarization based on deep learning. Geophysics. 2023; 88(1):WA267-WA279. doi: 10.1190/geo2022-0234.1

[13] Sui Y, Wang X, Ma J. Deep unfolding dictionary learning for seismic denoising. Geophysics. 2023; 88(1):WA129-WA147. doi: 10.1190/geo2022-0198.1

[14] Li X, Qi Q, Yang Y, Duan P, Cao Z. Removing abnormal environmental noise in nodal land seismic data using deep learning. Geophysics. 2024; 89(1):WA143-WA156. doi: 10.1190/geo2023-0143.1

[15] Song H, Gao Y, Chen W, Zhang X. Seismic data denoising based on convolutional autoencoder. Oil Geophys Prospect. 2020; 55(6):1210-1219+1160-1161. [Article in Chinese]. doi: 10.13810/j.cnki.issn.1000-7210.2020.06.006

[16] Shao D, Zhao Y, Li Y, Li Y. Noisy2Noisy: Denoise pre-stack seismic data without paired training data with labels. IEEE Geosci Remote Sens Lett. 2022; 19:1-5. doi: 10.1109/lgrs.2022.3145835

[17] Wu T, Meng T, Liu H, Li T. A seismic random noise suppression method based on self-supervised deep learning and transfer learning. Acta Geophys. 2024; 72:655-671. doi: 10.1007/s11600-023-01105-5

[18] Wang K, Hu T, Wang S, Wei J. Seismic multiple suppression based on a deep neural network method for marine data. Geophysics. 2022a; 87(4):V341-V365. doi: 10.1190/geo2021-0206.1

[19] Wang K, Hu T, Zhao B. An unsupervised deep neural network approach based on ensemble learning to suppress seismic surface-related multiples. IEEE Trans Geosci Remote Sens. 2022b; 60:1-14. doi: 10.1109/TGRS.2022.3225809

[20] Wang K, Hu T, Zhao B, Wang S. Surface-related multiple attenuation based on a self-supervised deep neural network with local wavefield characteristics. Geophysics. 2023a; 88(5):V387-V402. doi: 10.1190/geo2022-0599.1

[21] Wang K, Hu T, Zhao B, Wang S. An unsupervised learning method to suppress seismic internal multiples based on adaptive virtual events and joint constraints of multiple deep neural networks. IEEE Trans Geosci Remote Sens. 2023b; 61:1-18. doi: 10.1109/tgrs.2023.3243106

[22] Sun K, Li Z, Wang Y, Ma J, Dai X, Li B. Adaptive subtraction based on expanded multichannel U-Net with multi-pattern multiple model for surface-related multiple removal. IEEE Trans Geosci Remote Sens. 2025; 63:5912511. doi: 10.1109/tgrs.2025.3565358

[23] Qi Z, Li Z, Sun N, Li X, Li B, Wang Y. Semi-supervised interpretable FISTA-Net for adaptive subtraction and removal of seismic multiples. IEEE Trans Geosci Remote Sens. 2025; 63:5912214. doi: 10.1109/tgrs.2025.3564071

[24] Wang Z, Bovik AC. A universal image quality index. IEEE Signal Proc Let. 2005; 9(3):81-84. doi: 10.1109/97.995823

[25] Almadani M, Waheed U, Masood M, Chen Y. Dictionary learning with convolutional structure for seismic data denoising and interpolation. Geophysics. 2021; 86(5):V361-V374. doi: 10.1190/geo2019-0689.1

[26] Sun Y, Williamson P. Seismic data interpolation via frequency-constrained 3D inception Unet. In: Second International Meeting for Applied Geoscience and Energy; 2022. p. 1679-1683. doi: 10.1190/image2022-3751145.1

[27] Simon J, Fabien-Ouellet G, Gloaguen E, Khurjekar I. Hierarchical transfer learning for deep learning velocity model building. Geophysics. 2023; 88(1):R79-R93. doi: 10.1190/geo2021-0470.1

[28] Schuster GT. Theory of Daylight/Interferometric Imaging: Tutorial. In: 63rd EAGE Annual International Conference and Exhibition, Amsterdam, Netherlands; 2001. doi: 10.3997/2214-4609-pdb.15.A-32

[29] Schuster GT. Imaging the Most Bounce out of Multiples. In: 65th EAGE Conference and Exhibition, Stavanger, Norway; 2003. doi: 10.3997/2214-4609.201405725

[30] Schuster GT, Yu J, Sheng J, Rickett J. Interferometric/daylight seismic imaging. Geophys J Int. 2004; 157(2):838-852. doi: 10.1111/j.1365-246X.2004.02251.x

[31] Schuster GT. Fermat’s interferometric principle for multiple reflection tomography. Geophys Res Lett. 2005a; 32(12):L12303. doi: 10.1029/2005gl022351

[32] Schuster GT. Fermat’s interferometric principle for target-oriented traveltime tomography. Geophysics. 2005b; 70(4):U47-U50.doi: 10.1190/1.1997368

[33] Schuster GT, Zhou M. A theoretical overview of model-based and correlation-based redatuming methods. Geophysics. 2006; 71(4):103-110. doi: 10.1190/1.2208967

[34] Wapenaar K, Draganov D, Snieder R, Campman X, Verdel A. Tutorial on seismic interferometry: Part 1-basic principles and applications. Geophysics. 2010a; 75(5):75A195-75A209. doi: 10.1190/1.3457445

[35] Wapenaar K, Slob E, Snieder R, Curtis A. Tutorial on seismic interferometry: Part 2-Underlying theory and new advances. Geophysics. 2010b; 75(5):75A211-75A227. doi: 10.1190/1.3463440

[36] Wapenaar K, Ruigrok E, Neut J, Draganov D. Improved surface-wave retrieval from ambient seismic noise by multidimensional deconvolution. Geophys Res Lett. 2011; 38(1):L01313. doi: 10.1029/2010GL045523

[37] Curtis A, Halliday D. Source-receiver wave field interferometry. Phys Rev E Stat Nonlin Soft Matter Phys. 2010; 81(4):046601. doi: 10.1103/physreve.81.046601

[38] He K, Zhang K, Ren S. Deep Residual Learning for Image Recognition. In: Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. IEEE Computer Society; 2016. p. 770-778. doi: 10.1109/CVPR.2016.90

[39] Vaswani A, Shazeer N, Parmar N, et al. Attention is All You Need. In: Advances in Neural Information Processing Systems [Preprint]. Vol. 30; 2017. p. 5998-6008. doi: 10.48550/arXiv.1706.03762

[40] Wang Z, Bovik AC, Sheikh HR, Simoncelli EP. Image quality assessment: From error visibility to structural similarity. IEEE Trans Image Process. 2004; 13(4):600-612. doi: 10.1109/TIP.2003.819861

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