A method for reconstructing consecutively missing seismic data based on recurrent feature reasoning

Authors

LI Zijuan, School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, ChinaFollow
CHANG Guangyao, School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, China
JIA Yongna, School of Artificial Intelligence, Hebei University of Technology, Tianjin 300401, ChinaFollow

Abstract

Objective Due to the constraints of natural environments like rapids, rifts, and high mountains, the acquired seismic data are often challenged by consecutive missing, affecting subsequent seismic data processing and geologic analysis. Hence, it is necessary to reconstruct the missing data through interpolation. Methods This study proposed a method for reconstructing consecutively missing seismic data based on recurrent feature reasoning. First, the missing seismic data undergo partial convolution operations, in which the weight of the convolution results is adaptively adjusted based on the proportion of valid feature map data in the receptive field, avoiding invalid convolution operations on consecutively missing seismic channels. Second, the missing parts are progressively reconstructed through recurrent feature reasoning. Partial convolution operations and recurrent feature reasoning are alternated until all missing data are reconstructed. Finally, the reconstructed features generated in each iteration are integrated through feature fusion, ensuring accurate reasoning. To enhance the model's ability to learn the texture details of consecutively missing areas, the texture loss and mean square error (MSE) functions are combined as a hybrid loss function to further increase the reconstruction accuracy. Results and Conclusions Key findings are as follows: (1) The proposed method based on recurrent feature reasoning can effectively reconstruct the consecutively missing seismic data, with the signal-to-noise ratio (SNR) increased to 28.15 dB on top of the original 14.89 dB for the missing data. (2) In multiple reconstruction experiments focusing on 30 to 80 consecutively missing seismic channels, the reconstruction results demonstrate that the proposed method outperforms the U-Net method in terms of assessment indices like SNR, structural similarity, and MSE. The effectiveness of the proposed method is further verified by the reconstruction effects of the proposed method tested on six different public datasets. (3) As revealed by the impacts of the size of the partial convolution kernel on the reconstruction results investigated through comparative experiments, the reconstruction results manifest a higher SNR and a shorter iteration time when the partial convolution kernel measures 3×3. The results of this study provide a novel approach for the reconstruction of consecutively missing seismic data.

Keywords

seismic data reconstruction, partial convolution, recurrent feature reasoning, hybrid loss function

DOI

10.12363/issn.1001-1986.24.02.0140

Recommended Citation

LI Zijuan, CHANG Guangyao, JIA Yongna, et al. (2024) "A method for reconstructing consecutively missing seismic data based on recurrent feature reasoning," Coal Geology & Exploration: Vol. 52: Iss. 9, Article 17.
DOI: 10.12363/issn.1001-1986.24.02.0140
Available at: https://cge.researchcommons.org/journal/vol52/iss9/17

Reference

[1] YU Siwei，MA Jianwei. Deep learning for geophysics：Current and future trends[J]. Reviews of Geophysics，2021，59(3)：e2021RG000742.

[2] WANG Yanghua. Seismic trace interpolation in the f-x-y domain[J]. Geophysics，2002，67(4)：1232−1239.

[3] 吴庚，刘财，刘殿秘，等. 连续缺失地震数据的高阶流式预测滤波插值方法[J]. 地球物理学报，2023，66(3)：1220−1231.

WU Geng，LIU Cai，LIU Dianmi，et al. Seismic data interpolation beyond continuous missing data using high-order streaming prediction filter[J]. Chinese Journal of Geophysics，2023，66(3)：1220−1231.

[4] CLAERBOUT J，ZHANG Lin. Wave Equation Resampling of Unevenly Spaced Traces[J]. Stanford Exploration Project，1997，75(17)：263−271.

[5] FOMEL S. Seismic reflection data interpolation with differential offset and shot continuation[J]. Geophysics，2003，68(2)：733−744.

[6] YANG Yi，MA Jianwei，OSHER S. Seismic data reconstruction via matrix completion[J]. Inverse Problems & Imaging，2013，7(4)：1379−1392.

[7] OROPEZA V，SACCHI M. Simultaneous seismic data denoising and reconstruction via multichannel singular spectrum analysis[J]. Geophysics，2011，76(3)：V25−V32.

[8] 刘保童. 一种基于傅里叶变换的去假频内插方法及应用[J]. 煤田地质与勘探，2009，37(2)：63−67.

LIU Baotong. Dealiasing interpolation based on Fourier transform and its application[J]. Coal Geology & Exploration，2009，37(2)：63−67.

[9] KIM B，JEONG S，BYUN J. Trace interpolation for irregularly sampled seismic data using curvelet-transform-based projection onto convex sets algorithm in the frequency–wavenumber domain[J]. Journal of Applied Geophysics，2015，118：1−14.

[10] 李康楠，吴雅琴，杜锋，等. 基于卷积神经网络的岩爆烈度等级预测[J]. 煤田地质与勘探，2023，51(10)：94−103.

LI Kangnan，WU Yaqin，DU Feng，et al. Prediction of rockburst intensity grade based on convolutional neural network[J]. Coal Geology & Exploration，2023，51(10)：94−103.

[11] MOUSAVI S M，ELLSWORTH W L，ZHU Weiqiang，et al. Earthquake transformer：an attentive deep-learning model for simultaneous earthquake detection and phase picking[J]. Nature Communications，2020，11(1)：3952−3963.

[12] SIAHKOOHI A，LOUBOUTIN M，HERRMANN F J. The importance of transfer learning in seismic modeling and imaging[J]. Geophysics，2019，84(6)：A47−A52.

[13] YU Jiaxu，WU Bangyu. Attention and hybrid loss guided deep learning for consecutively missing seismic data reconstruction[J]. IEEE Transactions on Geoscience and Remote Sensing，2022，60：5902108.

[14] LI Xinze，WU Bangyu，ZHU Xu，et al. Consecutively missing seismic data interpolation based on coordinate attention unet[J]. IEEE Geoscience and Remote Sensing Letters，2022，19：3005005.

[15] HE Tao，WU Bangyu，ZHU Xu. Seismic data consecutively missing trace interpolation based on multistage neural network training process[J]. IEEE Geoscience and Remote Sensing Letters，2022，19：7504105.

[16] CHANG Dekuan，YANG Wuyang，YONG Xueshan，et al. Generative adversarial networks for seismic data interpolation[C]//SEG 2018 Workshop：SEG Maximizing Asset Value Through Artificial Intelligence and Machine Learning，Beijing，China，17–19 September 2018. Beijing，China. Society of Exploration Geophysicists and the Chinese Geophysical Society，2018：40–43.

[17] LIU Naihao，WU Lukun，WANG Jiale，et al. Seismic data reconstruction via wavelet-based residual deep learning[J]. IEEE Transactions on Geoscience and Remote Sensing，2022，60：4508213.

[18] 欧炳霖，曾同生，柳天成，等. 基于Huber-U-Net网络的地震数据重建与去噪[J]. 地球物理学进展，2023，38(6)：2540−2552.

OU Binglin，ZENG Tongsheng，LIU Tiancheng，et al. Seismic data reconstruction and de-noising based on Huber-U-Net network[J]. Progress in Geophysics，2023，38(6)：2540−2552.

[19] CHAI Xintao，GU Hanming，LI Feng，et al. Deep learning for irregularly and regularly missing data reconstruction[J]. Scientific Reports，2020，10：3302.

[20] LI Jingyuan，WANG Ning，ZHANG Lefei，et al. Recurrent feature reasoning for image inpainting[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle，WA，USA. IEEE，2020：7757–7765.

[21] LIU Guilin，REDA F A，SHIH K J，et al. Image inpainting for irregular holes using partial convolutions[M]//Lecture notes in computer science. Cham：Springer International Publishing，2018：89–105.

[22] SIMONYAN K，ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. CoRR，2014：1409−1556.