Seismic data reconstruction method based on unsupervised residual network
MENG Hongyu1,2, YANG Huachen1,2, ZHANG Jianzhong1,2,3
1. Key Laboratory of Submarine Geosciences and Prospecting Techniques, MOE China, Qingdao, Shandong 266100, China;
2. College of Marine Geosciences, Ocean University of China, Qingdao, Shandong 266100, China;
3. Functional Laboratory of Marine Mineral Resources Evaluation and Exploration Technology, Pilot National Laboratory for Marine Science and Technology (Qingdao), Qingdao, Shandong 266100, China
Abstract:Seismic data collected in the field usually have the problem of missing seismic traces. Reconstructing such traces has always been a difficult problem in seismic data processing. The deep learning (DL) method currently used mainly adopts a supervised learning approach for seismic data reconstruction,that is,it needs to use complete seismic data as labels to train the network model. Nevertheless,accurate labels for measured field data are difficult to obtain,and the dependence on a large number of training samples affects the application of the depth learning method in seismic data reconstruction. Therefore,this paper proposes a seismic data reconstruction method of unsupervised deep learning based on a residual network. Instead of using complete seismic data as the training set to train the residual network,this method takes random data as the input of the residual network,with the seismic data containing the missing seismic traces as the expected output of the network. Through the back propagation of the error between the network predicted output and the expected output,the network parameters are iteratively optimized to minimize the error,obtain the residual network with the optimal parameters,and use the network to reconstruct the missing seismic data. During network parameter optimization,the local and translation invariant properties of convolution are leveraged to learn the similar features between seismic data neighborhoods at multiple scales with the convolution filter,and the learned prior features are presented in the network output. This method is used to reconstruct the regular and irregular missing traces in the seismic data simulated with the Marmousi model and the measured marine streamer data,and the results are compared with those of the traditional fast projection onto convex set-soft threshold (FPOCS-Soft) method. The comparison shows that the proposed unsupervised residual network method can effectively reconstruct missing seismic traces,offer results with high accuracy and continuity,and outperforms the FPOCS-Soft method in precision.
FOMEL S. Seismic reflection data interpolation with differential offset and shot continuation[J]. SEG Tech-nical Program Expanded Abstracts,2001,20:2045-2048.
[2]
SPITZ S. Seismic trace interpolation in the F-X domain[J]. Geophysics,1991,56(6):785-794.
[3]
PORSANI M J. Seismic trace interpolation using half-step prediction filters[J]. Geophysics,1999,64(5):1461-1467.
[4]
宜明理,严又生,魏新,等. F-K域抗假频道内插[J]. 石油物探,2001,40(2):36-41.YI Mingli,YAN Yousheng,WEI Xin,et al. Using anti-alias trace interpolation in F-K domain[J]. Geophysical Prospecting for Petroleum,2001,40(2):36-41.
[5]
DUIJNDAM A J W,SCHONEWILLE M A,HINDRIKS C O H. Reconstruction of band-limited signals,irregularly sampled along one spatial direction[J]. Geophysics,1999,64(2):524-538.
[6]
HERRMANN F J,HENNENFENT G. Non-parame-tric seismic data recovery with curvelet frames[J]. Geophysical Journal International,2008,173(1):233-248.
[7]
FOMEL S,LIU Y. Seislet transform and seislet frame[J]. Geophysics,2010,75(3):V25-V38.
[8]
IBRAHIM A,SACCHI M D. Simultaneous source separation using a robust radon transform[J]. Geophysics,2014,79(1):V1-V11.
[9]
GAN S W,WANG S D,CHEN Y K,et al. Dealiased seismic data interpolation using seislet transform with low-frequency constraint[J]. IEEE Geoscience and Remote Sensing Letters,2015,12(10):2150-2154.
[10]
TRICKETT S,BURROUGHS L,MILTON A,et al. Rank-reduction-based trace interpolation[C]. SEG Technical Program Expanded Abstracts,2010,29:3829-3833.
[11]
ZHANG D,ZHOU Y T,CHEN H M,et al. Hybrid rank-sparsity constraint model for simultaneous reconstruction and denoising of 3D seismic data[J]. Geophysics,2017,82(5):V351-V367.
[12]
OROPEZA V,SACCHI M. Simultaneous seismic data denoising and reconstruction via multichannel singular spectrum analysis[J]. Geophysics,2011,76(3):V25-V32.
[13]
WANG B F,ZHANG N,LU W K,et al. Deep-lear-ning-based seismic data interpolation:a preliminary result[J]. Geophysics,2019,84(1):V11-V20.
[14]
郑浩,张兵. 基于卷积神经网络的智能化地震数据插值技术[J]. 地球物理学进展,2020,35(2):721-727.ZHENG Hao,ZHANG Bing. Intelligent seismic data interpolation via convolutional neural network[J]. Progress in Geophysics,2020,35(2):721-727.
[15]
ALWON S. Generative adversarial networks in seismic data processing[C]. SEG Technical Program Expanded Abstracts,2018,37:1991-1995.
[16]
OLIVEIRA D A B,FERREIRA R S,SILVA R,et al. Interpolating seismic data with conditional generative adversarial networks[J]. IEEE Geoscience and Remote Sensing Letters,2018,15(12):1952-1956.
[17]
KAUR H,PHAM N,FOMEL S. Seismic data interpolation using deep learning with generative adversa-rial networks[J]. Geophysical Prospecting,2021,69(2):307-326.
[18]
杨冠雨,王璐,孟凡顺. 基于深度卷积神经网络的地震数据重建[J]. 地球物理学进展,2020,35(4):1497-1506.YANG Guanyu,WANG Lu,MENG Fanshun. Seismic data reconstruction via deep convolution neural network[J]. Progress in Geophysics,2020,35(4):1497-1506.
[19]
HE K M,ZHANG X Y,REN S Q,et al. Deep resi-dual learning for image recognition[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),Las Vegas,NV,USA,2016,770-778.
[20]
杨午阳,杨佳润,陈双全,等. 基于U-Net深度学习网络的地震数据断层检测[J]. 石油地球物理勘探,2021,56(4):688-697.YANG Wuyang,YANG Jiarun,CHEN Shuangquan,et al. Seismic data fault detection based on U-Net deep learning network[J]. Oil Geophysical Prospecting,2021,56(4):688-697.
[21]
张政,严哲,顾汉明. 基于残差网络与迁移学习的断层自动识别[J]. 石油地球物理勘探,2020,55(5):950-956.ZHANG Zheng,YAN Zhe,GU Hanming. Automatic fault recognition with residual network and transfer learning[J]. Oil Geophysical Prospecting,2020,55(5):950-956.
[22]
IOFFE S,SZEGEDY C. Batch normalization:accele-rating deep network training by reducing internal covariate shift[C]. Proceedings of the 32nd International Conference on International Conference on Machine Learning,2015,448-456.
[23]
HE K M,ZHANG X Y,REN S Q,et al. Delving deep into rectifiers:surpassing human-level perfor-mance on ImageNet classification[C]. 2015 IEEE International Conference on Computer Vision (ICCV),Santiago,Chile,2015,1026-1034.
[24]
KINGMA D P,BA L J. Adam:a method for stochastic optimization[C]. International Conference on Learning Representations (ICLR),2015,1-15.
[25]
王静,张军华,芦凤明,等. 构建三维深度监督网络的断层检测方法[J]. 石油地球物理勘探,2021,56(5):947-957.WANG Jing,ZHANG Junhua,LU Fengming,et al. Research on fault detection method based on 3D deeply supervised network[J]. Oil Geophysical Prospecting,2021,56(5):947-957.
[26]
袁崇鑫. 波动方程有限差分正演技术研究[D]. 四川成都:成都理工大学,2015.YUAN Chongxin. Finite Difference Wave Equation Forward Modeling Technology Research[D]. Chengdu University of Technology,Chengdu,Sichuan,2015.
[27]
马泽川,李勇,陈力鑫,等. 地震资料快速两步插值算法[J]. 石油地球物理勘探,2020,55(5):997-1004.MA Zechuan,LI Yong,CHEN Lixin,et al. Fast two-step interpolation algorithm for seismic data[J]. Oil Geophysical Prospecting,2020,55(5):997-1004.