Seismic internal multiple suppression method with encoder-decoder convolutional network based on data augmentation
LIU Xiaozhou1, HU Tianyue1, LIU Tao2, WEI Zhefeng2, XIE Fei2, AN Shengpei2
1. School of Earth and Space Sciences, Peking University, Beijing 100871, China;
2. SINOPEC Petroleum Exploration and Production Research Institute, Beijing 100083, China
Abstract：Internal multiple suppression is a cutting-edge technology challenge in seismic data denoising,which is of great significance for obtaining high-quality data and understanding the real subsurface structure. Current suppression methods of internal multiples are time-consuming and have high requirements for manual parameter tuning,which may lead to internal multiple leakage when processing data with a low signal-to-noise ratio (SNR). Therefore,this paper proposes an internal multiple suppression method with the encoder-decoder convolutional neural network (CNN) based on data augmentation. First,we estimate the primaries and internal multiples from the raw data by the internal multiple suppression method based on virtual events to obtain primary labels. Then,we establish two augmented training datasets. On the one hand,the internal multiple data are augmented by the change in amplitude,polarity,and travel time of internal multiples in the training samples to raise the generalization ability of the internal multiple suppression network. On the other hand,the Gaussian noise augmented datasets are obtained after different levels of Gaussian noise are added to the raw data,which can improve the anti-noise performance of the network. Finally,a deep encoder-decoder network suitable for internal multiple suppression is built for neural network training and prediction by the combination of the advantages of the denoising CNN (DnCNN) and U-shaped fully connolutional network (U-Net). The tests on synthetic and field data indicate that the proposed method can effectively suppress internal multiples and protect primaries and has strong generalization ability and anti-noise performance,which can significantly improve computational efficiency.
WEGLEIN A B. Multiple attenuation: an overview of recent advances and the road ahead(1999)[J]. The Leading Edge,1999,18(1):40-44.
[2]
VERSCHUUR D J,BERKHOUT A J. Removal of internal multiples with the common-focus-point (CFP) approach:part 2 - application strategies and data examples[J]. Geophysics,2005,70(3):V61-V72.
[3]
WEGLEIN A B,GASPAROTTO F A,CARVALHO P M,et al. An inverse-scattering series method for attenuating multiples in seismic reflection data[J]. Geophysics,1997,62(6):1975-1989.
[4]
LUO Y,KELAMIS P G,FU Q,et al. Elimination of land internal multiples based on the inverse scattering series[J]. The Leading Edge,2011,30(8):884-889.
[5]
WANG M,HUNG B. 3D inverse scattering series method for internal multiple attenuation[C]. Exten-ded Abstracts of 76th EAGE Conference and Exhibition,2014,1-5.
[6]
MELES G A,LÖER K,RAVASI M,et al. Internal multiple prediction and removal using Marchenko autofocusing and seismic interferometry[J]. Geophy-sics,2015,80(1):A7-A11.
[7]
叶月明,赵昌垒,姚根顺,等. 数据驱动型层间多次波预测方法研究[J]. 石油地球物理勘探,2014,49(2):244-251.YE Yueming,ZHAO Changlei,YAO Genshun,et al. Study of data-driven interbed multiple prediction[J]. Oil Geophysical Prospecting,2014,49(2):244-251.
[8]
IKELLE L T. A construct of internal multiples from surface data only:the concept of virtual seismic events[J]. Geophysical Journal International,2006,164(2):383-393.
[9]
AN S P,HU T Y,LIU J H,et al. Scheme of applying virtual event method for internal multiples on land seismic data[C]. International Geophysical Confe-rence,Qingdao,China,2017,391-394.
[10]
刘嘉辉,胡天跃,彭更新. 自适应虚同相轴方法压制地震层间多次波[J]. 地球物理学报,2018,61(3):1196-1210.LIU Jiahui,HU Tianyue,PENG Gengxin. Suppres-sing seismic inter-bed multiples with the adaptive virtual events method[J]. Chinese Journal of Geophy-sics,2018,61(3):1196-1210.
[11]
LIU J H,HU T Y,PENG G X,et al. Removal of internal multiples by iterative construction of virtual primaries[J]. Geophysical Journal International,2018,215(1):81-101.
[12]
崔永福,刘嘉辉,陈猛,等. 虚同相轴方法及其在陆上地震层间多次波压制中的应用[J]. 石油地球物理勘探,2019,54(6):1228-1236,1245.CUI Yongfu,LIU Jiahui,CHEN Meng,et al. Land seismic peg-leg multiple attenuation with the virtual event method[J]. Oil Geophysical Prospecting,2019,54(6):1228-1236,1245.
[13]
陈德武,杨午阳,魏新建,等. 基于混合网络U-SegNet的地震初至自动拾取[J]. 石油地球物理勘探,2020,55(6):1188-1201.CHEN Dewu,YANG Wuyang,WEI Xinjian,et al. Automatic picking of seismic first arrivals based on hybrid network U-SegNet[J]. Oil Geophysical Prospecting,2020,55(6):1188-1201.
[14]
周创,居兴国,李子昂,等. 基于深度卷积生成对抗网络的地震初至拾取[J]. 石油物探,2020,59(5):795-803.ZHOU Chuang,JU Xingguo,LI Ziang,et al. A deep convolutional generative adversarial network for first-arrival pickup from seismic data[J]. Geophysical Prospecting for Petroleum,2020,59(5):795-803.
[15]
付超,林年添,张栋,等. 多波地震深度学习的油气储层分布预测案例[J]. 地球物理学报,2018,61(1):293-303.FU Chao,LIN Niantian,ZHANG Dong,et al. Prediction of reservoirs using multi-component seismic data and the deep learning method[J]. Chinese Journal of Geophysics,2018,61(1):293-303.
[16]
常德宽,雍学善,王一惠,等. 基于深度卷积神经网络的地震数据断层识别方法[J]. 石油地球物理勘探,2021,56(1):1-8.CHANG Dekuan,YONG Xueshan,WANG Yihui,et al. Seismic fault interpretation based on deep convolutional neural networks[J]. Oil Geophysical Prospecting,2021,56(1):1-8.
[17]
YU S W,MA J W,WANG W L. Deep learning for denoising[J]. Geophysics,2019,84(6):V333-V350.
[18]
JAIN V,SEUNG H S. Natural image denoising with convolutional networks[C]. Proceedings of the 21st International Conference on Neural Information Processing Systems,2008,769-776.
[19]
RONNEBERGER O，FISCHER P，BROX T. U-Net：Convolutional networks for biomedical image segmentation[C]. Medical Image Computing and Computer-assisted Intervention，2015，234-241.
[20]
ZHANG K,ZUO W M,CHEN Y J,et al. Beyond a Gaussian denoiser:residual learning of deep CNN for image denoising[J]. IEEE Transactions on Image Processing,2017,26(7):3142-3155.
[21]
LECUN Y,BOTTOU L,BENGIO Y,et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE,1998,86(11):2278-2324.
[22]
MORENO-BAREA F J,STRAZZERA F,JEREZ J M,et al. Forward noise adjustment scheme for data augmentation[C]. 2018 IEEE Symposium Series on Computational Intelligence (SSCI),2018,728-734.
[23]
ZHANG H Y,CISSE M,DAUPHIN Y N,et al. Mixup:beyond empirical risk minimization[DB/OL]. 2017-10-25,https://arxiv.org/abs/1710.09412.
[24]
韩卫雪,周亚同,池越. 基于深度学习卷积神经网络的地震数据随机噪声去除[J]. 石油物探,2018,57(6):862-869.HAN Weixue,ZHOU Yatong,CHI Yue. Deep lear-ning convolutional neural networks for random noise attenuation in seismic data[J]. Geophysical Prospecting for Petroleum,2018,57(6):862-869.
[25]
高好天,孙宁娜,孙可奕,等. DnCNN和U-Net对地震随机噪声压制的对比分析[J]. 地球物理学进展,2021,36(6):2441-2453.GAO Haotian,SUN Ningna,SUN Keyi,et al. Comparative analysis of DnCNN and U-Net on suppression of seismic random noise[J]. Progress in Geophysics,2021,36(6):2441-2453.
[26]
王钰清,陆文凯,刘金林,等. 基于数据增广和CNN的地震随机噪声压制[J]. 地球物理学报,2019,62(1):421-433.WANG Yuqing,LU Wenkai,LIU Jinlin,et al. Random seismic noise attenuation based on data augmentation and CNN[J]. Chinese Journal of Geophysics,2019,62(1):421-433.
[27]
SIAHKOOHI A,VERSCHUUR D J,HERRMANN F J. Surface-related multiple elimination with deep learning[C]. SEG Technical Program Expanded Abstracts,2019,38:4629-4634.
[28]
LI Z X,GAO H T. Feature extraction based on the convolutional neural network for adaptive multiple subtraction[J]. Marine Geophysical Research,2020,41(2):1-20.
[29]
宋欢,毛伟建,唐欢欢. 基于深层神经网络压制多次波[J]. 地球物理学报,2021,64(8):2795-2808.SONG Huan,MAO Weijian,TANG Huanhuan. Application of deep neural networks for multiples atte-nuation[J]. Chinese Journal of Geophysics,2021,64(8):2795-2808.
[30]
匡伟康,胡天跃,段文胜,等. 基于自适应变步长波场延拓的可控层分阶层间多次波模拟[J]. 地球物理学报,2020,63(5):2043-2055.KUANG Weikang,HU Tianyue,DUAN Wensheng,et al. Modeling inter-layer multiples based on adaptive step-length-varying wavefield extrapolation[J]. Chinese Journal of Geophysics,2020,63(5):2043-2055.
