A deep learning-based method for separating up- and down-going waves in zero-offset vertical seismic profiles
WANG Teng-Yu1,2,3(), DENG Ding-Ding4(), ZHENG Duo-Ming1,2,3, LIU Yang4, ZHANG Zhen1, LUO Wen-Jun5
1. Research Institute of Exploration and Development, Tarim Oilfield Company, PetroChina, Korla 841000, China 2. R&D Center for Ultra-Deep Complex Reservoir Exploration and Development, CNPC, Korla 841000, China 3. Engineering Research Center for Ultra-deep Complex Oil and Gas Reservoir Exploration and Development, Xinjiang Uygur Autonomous Region, Korla 841000, China 4. State Key Laboratory of Petroleum Resources and Engineering, China University of Petroleum (Beijing), Beijing 102249, China 5. Beijing Borehole Logging Petroleum Technology Co., Ltd., Beijing 100085, China
Wavefield separation serves as a key step in processing the data of vertical seismic profiles (VSPs). Its accuracy directly influences seismic imaging, inversion of elastic parameters, lithology identification, and interpretation of hydrocarbon-bearing properties. Traditional methods face challenges in wavefield separation. For example, the median filtering requires manual intervention, often introducing errors and thus compromising separation accuracy; the FK filtering yields high accuracy but low efficiency. In contrast, deep learning techniques offer high automation, enabling both high accuracy and efficiency in wavefield separation. Hence, this study proposed a deep learning-based method for separating up- and down-going waves in zero-offset VSPs. First, the up- and down-going waves were separated through FK transform, generating a dataset. Second, a deep learning-based model, Unet++, was constructed for separating these waves in VSPs. Third, the relative down-going wavefield (obtained by subtracting the predicted up-going wavefield from the full wavefield) was incorporated into the loss function to mitigate the impacts of amplitude differences between up- and down-going waves on network updates. Moreover, the structural similarity index measure (SSIM) was employed as a regularization constraint to assist the network in learning the structural characteristics of the wavefield. The test results of actual VSP data demonstrate that the trained network can effectively learn the characteristics of the up- and down-going waves, achieving high accuracy and efficiency in wavefield separation.
王腾宇, 邓丁丁, 郑多明, 刘洋, 张振, 罗文君. 基于深度学习的零井源距VSP上、下行波分离方法[J]. 物探与化探, 2025, 49(6): 1319-1332.
WANG Teng-Yu, DENG Ding-Ding, ZHENG Duo-Ming, LIU Yang, ZHANG Zhen, LUO Wen-Jun. A deep learning-based method for separating up- and down-going waves in zero-offset vertical seismic profiles. Geophysical and Geochemical Exploration, 2025, 49(6): 1319-1332.
Hu D H, Yang S H, Yan Y P, et al. Extraction and application of seismic attributes from VSP data[J]. Geophysical Prospecting for Petroleum, 2004, 43(S1):83-85.
[2]
Suprajitno M, Greenhalgh S A. Separation of upgoing and downgoing waves in vertical seismic profiling by contour-slice filtering[J]. Geophysics, 1985, 50(6):950-962.
[3]
Wang X K, Chen J Y, Gao L, et al. An iterative zero-offset VSP wavefield separating method based on the error analysis of SVD filtering[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(8):1164-1168.
[4]
Matsushima J, Ali M Y, Bouchaala F. A novel method for separating intrinsic and scattering attenuation for zero-offset vertical seismic profiling data[J]. Geophysical Journal International, 2017, 211(3):1655-1668.
[5]
Gao L, Chen W C, Wang B L, et al. VSP wave field separation:An optimization method based on block relaxation and singular value thresholding[J]. Journal of Applied Geophysics, 2014, 104:156-162.
[6]
Jin C, Cao D P, Yin X Y. Joint waveform inversion with the separated upgoing and downgoing wavefields of VSP data[J]. Journal of Geophysics and Engineering,2019:gxz086.
[7]
Huang T, Yang G, Tang G. A fast two-dimensional Median filtering algorithm[J]. IEEE Transactions on Acoustics,Speech,and Signal Processing, 1979, 27(1):13-18.
[8]
Duncan G, Beresford G. Some analyses of 2-D median F-K filters[J]. Geophysics, 1995, 60(4):1157-1168.
[9]
Moon W, Carswell A, Tang R, et al. Radon transform wave field separation for vertical seismic profiling data[J]. Geophysics, 1986, 51(4):940-947.
Min C, Wen G Q, Li X G, et al. Research progress and application prospect of interpretable machine learning in artificial intelligence in oil and gas industry[J]. Natural Gas Industry, 2024, 44(9):114-126.
Guo Z X, Ma B, Zhang S, et al. Research status and prospects of deep learning in oil and gas production prediction[J]. Natural Gas Industry, 2024, 44(9):88-98.
[12]
Wang B F, Zhang N, Lu W K, et al. Deep-learning-based seismic data interpolation:A preliminary result[J]. Geophysics, 2018, 84(1):V11-V20.
[13]
Di H B, Li Z, Maniar H, et al. Seismic stratigraphy interpretation by deep convolutional neural networks:A semisupervised workflow[J]. Geophysics, 2020, 85(4):WA77-WA86.
[14]
Cui R A, Cao D P, Liu Q, et al. VP and VS prediction from digital rock images using a combination of U-Net and convolutional neural networks[J]. Geophysics, 2021, 86(1):MR27-MR37.
Cova D, Liu Y, Ding C Z, et al. First break picking method based on artificial intelligence and apparent velocity constraint[J]. Oil Geophysical Prospecting, 2021, 56(3):419-435,409.
[16]
Han S, Liu Y J, Li Y B, et al. First arrival traveltime picking through 3-D U-Net[J]. IEEE Geoscience and Remote Sensing Letters, 2021, 19:8016405.
[17]
Wang W L, McMechan G A, Ma J W, et al. Automatic velocity picking from semblances with a new deep-learning regression strategy:Comparison with a classification approach[J]. Geophysics, 2021, 86(2):U1-U13.
[18]
Sun J, Slang S, Elboth T, et al. A convolutional neural network approach to deblending seismic data[J]. Geophysics, 2020, 85(4):WA13-WA26.
[19]
Wang S R, Hu W Y, Yuan P Y, et al. A self-supervised deep learning method for seismic data deblending using a blind-trace network[J]. IEEE Transactions on Neural Networks and Learning Systems, 2023, 34(7):3405-3414.
[20]
Feng Q K, Li Y. Denoising deep learning network based on singular spectrum analysis—DAS seismic data denoising with multichannel SVDDCNN[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 60:5902911.
[21]
Zhang M, Liu Y, Bai M, et al. Seismic noise attenuation using unsupervised sparse feature learning[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(12):9709-9723.
[22]
Deng D D, Liu Y, et al. A wavefield decomposition-guided deep-learning framework for removing smearing artifacts from VSP RTM[J]. Journal of Geophysics and Engineering, 2025, 22(4):1085-1095.
Sun Y H, Liu Y, Chen T S. Multi-wave amplitude-versus-offset inversion and reservoir fluid identification based on unsupervised deep learning[J]. Geophysical Prospecting for Petroleum, 2021, 60(3):385-394,413.
[24]
Sun Y H, Liu Y, Zhang M, et al. Inversion of low- to medium-frequency velocities and densities from AVO data using invertible neural networks[J]. Geophysics, 2022, 87(3):A37-A42.
[25]
Zhang H R, Chen T S, Liu Y, et al. Automatic seismic facies interpretation using supervised deep learning[J]. Geophysics, 2021, 86(1):IM15-IM33.
[26]
Tao B C, Yang Y Y, Zhou H L, et al. Deep learning-based upgoing and downgoing wavefield separation for vertical seismic profile data[J]. Geophysics, 2023, 88(6):D339-D355.
[27]
Cao D P, Chen X, Jia Y, et al. Upgoing and downgoing wavefield separation in VSP data using CGAN based on asymmetric convolution blocks[J]. Journal of Geophysics and Engineering, 2024, 21(5):1511-1525.
[28]
Zhou Z W, Siddiquee M M R, Tajbakhsh N, et al. UNet++:Redesigning skip connections to exploit multiscale features in image segmentation[J]. IEEE Transactions on Medical Imaging, 2020, 39(6):1856-1867.
[29]
Huber P J. Robust regression:Asymptotics,conjectures and Monte Carlo[J]. The Annals of Statistics, 1973, 1(5):799-821.
[30]
Zhang Y F, Thorburn P J, Xiang W, et al. SSIM: A deep learning approach for recovering missing time series sensor data[J]. IEEE Internet of Things Journal, 2019, 6(4):6618-6628.
