Deep-learning based ocean bottom seismic wavefield recovery
- Ali Siahkoohi (School of Computational Science and Engineering, Georgia Institute of Technology) | Rajiv Kumar (Georgia Institute of Technology, and DownUnder GeoSolutions, Perth, Australia) | Felix J. Herrmann (School of Computational Science and Engineering, Georgia Institute of Technology)
- Document ID
- Society of Exploration Geophysicists
- SEG International Exposition and Annual Meeting, 15-20 September, San Antonio, Texas, USA
- Publication Date
- Document Type
- Conference Paper
- 2019. Society of Exploration Geophysicists
- Machine learning, Reconstruction, Reciprocity, Ocean-bottom node
- 5 in the last 30 days
- 6 since 2007
- Show more detail
Ocean bottom surveys usually suffer from having very sparse receivers. Assuming a desirable source sampling, achievable by existing methods such as (simultaneous-source) randomized marine acquisition, we propose a deep-learning based scheme to bring the receivers to the same spatial grid as sources using a convolutional neural network. By exploiting source-receiver reciprocity, we construct training pairs by artificially subsampling the fully-sampled single-receiver frequency slices using a random training mask and later, we deploy the trained neural network to fill-in the gaps in single-source frequency slices. Our experiments show that a random training mask is essential for successful wavefield recovery, even when receivers are on a periodic gird. No external training data is required and experiments on a 3D synthetic data set demonstrate that we are able to recover receivers for up to 90% missing receivers, missing either randomly or periodically, with a better recovery for random case, at low to midrange frequencies.
Presentation Date: Monday, September 16, 2019
Session Start Time: 1:50 PM
Presentation Time: 3:05 PM
Presentation Type: Oral
|File Size||7 MB||Number of Pages||6|
Bottou,L.,F. E.Curtis, andJ.Nocedal,2018,Optimization methods for large-scale machine learning:SIAM Review,60,223–311,doi:10.1137/16M1080173.
Cheng,J., andM. D.Sacchi,2015,Separation and reconstruction of simultaneous source data via iterative rank reduction:Geophysics,80,no.4,V57–V66,doi:10.1190/geo2014-0385.1.
Hornik,K.,M.Stinchcombe, andH.White,1989,Multilayer feedforward networks are universal approximators:Neural networks,2,359–366,doi:10.1016/0893-6080(89)90020-8.
Huang,L.,X.Dong, andT. E.Clee,2017,A scalable deep learning platform for identifying geologic features from seismic attributes:The Leading Edge,36,249–256,doi:10.1190/tle36030249.1.
Kumar,R.,H.Wason, andF. J.Herrmann,2015,Source separa- tion for simultaneous towed-streamer marine acquisition– -a compressed sensing approach:Geophysics,80,no.6,WD73–WD88,doi:10.1190/geo2015-0108.1.
Mandelli,S.,F.Borra,V.Lipari,P.Bestagini,A.Sarti, andS.Tubaro,2018,Seismic data interpolation through convo- lutional autoencoder:88th Annual International Meeting, SEG,Expanded Abstracts,4101–4105,doi:10.1190/segam2018-2995428.1.
Ovcharenko,O.,V.Kazei,D.Peter,X.Zhang, andT.Alkhalifah,2018,Low-frequency data extrapolation using a feed-forward ANN:80th Annual International Conference and Exhibition, EAGE,Extended Abstracts,doi:10.3997/2214-4609.201801231.
Rizzuti,G.,A.Siahkoohi, andF. J.Herrmann,2019,Learned iterative solvers for the Helmholtz equation:81st Annual International Conference and Exhibition, EAGE,Extended Abstracts,doi:10.13140/RG.2.2.13712.25609.
Siahkoohi,A.,M.Louboutin,R.Kumar, andF. J.Herrmann,2018b,Deep-convolutional neural networks in prestack seismic: Two exploratory examples:88th Annual International Meeting, SEG,Expanded Abstracts,2196–2200,doi:10.1190/segam2018-2998599.1.
Sun,H., andL.Demanet,2018,Low-frequency extrapolation with deep learning:88th Annual International Meeting, SEG,Expanded Abstracts,2011–2015,doi:10.1190/segam2018-2997928.1.
Veillard,A.,O.Morère,M.Grout, andJ.Gruffeille,2018,Fast 3D seismic interpretation with unsupervised deep learning: Application to a Potash Network in the North Sea:80th Annual International Conference and Exhibition, EAGE,Extended Abstracts,doi:10.3997/2214-4609.201800738.