none 10.1190/geo2020-0017.1 Society of Exploration Geophysicists Society of Exploration Geophysicists 186 127903771 5981 2025121211572108000 10.1190 2025-12-12T16:57:28Z 2021-06-17T08:13:00Z 18 Geophysics 0016-8033 1942-2156 09 01 2021 86 5 Yonggyu Choi Hanyang University 1 , Department of Earth Resources and Environmental Engineering, Seoul 04763, Korea. E-mail: sja05152@hanyang.ac.kr ; joman90@hanyang.ac.kr ; ssjdoolly@hanyang.ac.kr ; jbyun@hanyang.ac.kr (corresponding author). http://orcid.org/0000-0003-2580-7866 Yeonghwa Jo Hanyang University 1 , Department of Earth Resources and Environmental Engineering, Seoul 04763, Korea. E-mail: sja05152@hanyang.ac.kr ; joman90@hanyang.ac.kr ; ssjdoolly@hanyang.ac.kr ; jbyun@hanyang.ac.kr (corresponding author). Soon Jee Seol Hanyang University 1 , Department of Earth Resources and Environmental Engineering, Seoul 04763, Korea. E-mail: sja05152@hanyang.ac.kr ; joman90@hanyang.ac.kr ; ssjdoolly@hanyang.ac.kr ; jbyun@hanyang.ac.kr (corresponding author). http://orcid.org/0000-0002-6253-4599 Joongmoo Byun Hanyang University 1 , Department of Earth Resources and Environmental Engineering, Seoul 04763, Korea. E-mail: sja05152@hanyang.ac.kr ; joman90@hanyang.ac.kr ; ssjdoolly@hanyang.ac.kr ; jbyun@hanyang.ac.kr (corresponding author). http://orcid.org/0000-0003-0445-0271 Young Kim YK Geophysics LLC 2 , Sequim, Washington 98382, USA. E-mail: young.kim@sk-houston.com . ABSTRACT The resolution of seismic data dictates the ability to identify individual features or details in a given image, and the temporal (vertical) resolution is a function of the frequency content of a signal. To improve thin-bed resolution, broadening of the frequency spectrum is required; this has been one of the major objectives in seismic data processing. Recently, many researchers have proposed and shown the applicability of machine-learning-based resolution enhancement and showed their applicability. However, because the performance of machine learning depends on what the model has learned, output from training data with features different from the target field data may be poor. Thus, we have developed a machine-learning-based spectral enhancement technique considering features of seismic field data. We used a convolutional U-Net model, which preserves the temporal connectivity and resolution of the input data, and we generated numerous synthetic input traces and their corresponding spectrally broadened traces for training the model. A priori information from field data, such as the estimated source wavelet and reflectivity distribution, was considered when generating the input data for complementing the field features. Using synthetic tests and field poststack seismic data examples, we showed that the trained model with a priori information outperforms the models trained without a priori information in terms of the accuracy of enhanced signals. In addition, our new spectral enhancing method was verified through the application to the high-cut-filtered data, and its promising features were evaluated through a comparison with well log data. 08 18 2021 09 01 2021 06 17 2021 V389 V408 Korea Institute of Energy Technology Evaluation and Planning http://dx.doi.org/10.13039/501100007053 20194010201920 Ministry of Science and ICT, South Korea http://dx.doi.org/10.13039/501100014188 2021R1A2C2014315 1 10.1190/crossmark-policy library.seg.org true 2020-01-14 2021-04-28 2021-05-05 2021-08-18 Korea Institute of Energy Technology Evaluation and Planning http://dx.doi.org/10.13039/501100007053 20194010201920 Ministry of Science and ICT, South Korea http://dx.doi.org/10.13039/501100014188 2021R1A2C2014315 10.1190/geo2020-0017.1 https://pubs.geoscienceworld.org/geophysics/article/86/5/V389/607243/Deep-learning-spectral-enhancement-considering https://pubs.geoscienceworld.org/seg/geophysics/article-pdf/86/5/V389/5433486/geo-2020-0017.1.pdf https://pubs.geoscienceworld.org/seg/geophysics/article-pdf/86/5/V389/5433486/geo-2020-0017.1.pdf https://pubs.geoscienceworld.org/geophysics/article/86/5/v389/607243/deep-learning-spectral-enhancement-considering 10.1190/segam2017-17751158.1 Canning A. Moulière-Reiser D. Weiss Y. Malkin A. Philip E. Grinberg N. Teitel A. Reznikov M. Yehezkel V. , 2017, Neural networks approach to spectral enhancement: 87th Annual International Meeting, SEG, Expanded Abstracts, 4283–4286, doi: http://dx.doi.org/10.1190/segam2017-17751158.1. The Leading Edge Chopra 35 176 2016 10.1190/tle35020176.1 Spectral decomposition and spectral balancing of seismic data 0162-8828 IEEE Transactions on Pattern Analysis and Machine Intelligence Dong 38 295 2015 10.1109/TPAMI.2015.2439281 Image super-resolution using deep convolutional networks 10.1190/segam2018-2996943.1 Halpert A. D. , 2018, Deep learning-enabled seismic image enhancement: 88th Annual International Meeting, SEG, Expanded Abstracts, 2081–2085, doi: http://dx.doi.org/10.1190/segam2018-2996943.1. First Break Jenner 34 69 2016 10.3997/1365-2397.34.1.83801 Extending the useable bandwidth of seismic data with tensor-guided, frequency-dependent filtering 0016-8033 Geophysics Kallweit 47 1035 1982 10.1190/1.1441367 The limits of resolution of zero-phase wavelets Kuleshov V. Enam S. Z. Ermon S. , 2017, Audio super-resolution using neural nets: Presented at the 5th International Conference on Learning Representations (ICLR), Workshop Track, arXiv:1708.00853. 0196-2892 IEEE Transactions on Geoscience and Remote Sensing Li 2021 Deep learning for simultaneous seismic image super-resolution and denoising 0016-8033 Geophysics Liang 82 4 W1 2017 10.1190/geo2015-0572.1 Tutorial: Spectral bandwidth extension invention versus harmonic extrapolation Loffe S. Szegedy C. , 2015, Batch normalization: accelerating deep network training by reducing internal covariate shift: Proceedings of the 32nd International Conference on Machine Learning, PMLR 37, 448–456. 0016-8033 Geophysics Margrave 76 3 W15 2011 10.1190/1.3560167 Gabor deconvolution: Estimating reflectivity by nonstationary deconvolution of seismic data 0016-8033 Geophysics Peacock 34 155 1969 10.1190/1.1440003 Predictive deconvolution: Theory and practice 0016-8033 Geophysics Puryear 73 2 R37 2008 10.1190/1.2838274 Layer-thickness determination and stratigraphic interpretation using spectral inversion: Theory and application 0016-8033 Geophysics Robinson 32 418 1967 10.1190/1.1439873 Predictive decomposition of time series with application to seismic exploration 10.1007/978-3-319-24574-4_28 Ronneberger O. Fischer P. Brox T. , 2015, U-Net: Convolutional networks for biomedical image segmentation: Computer Vision and Pattern Recognition, arXiv:1505.04597. 0016-8033 Geophysics Rosa 56 1244 1991 10.1190/1.1443144 Processing via spectral modeling First Break Smith 26 97 2008 10.3997/1365-2397.26.1288.28410 Extending seismic bandwidth using the continuous wavelet transform The Journal of Machine Learning Research Srivastava 15 1929 2014 Dropout: A simple way to prevent neural networks from overfitting 10.1190/1.3255039 Stark T. J. , 2009, Frequency enhancement via an integer multiplier or just another GeoWizardry attribute: 79th Annual International Meeting, SEG, Expanded Abstracts, 1092–1096, doi: http://dx.doi.org/10.1190/1.3255039. 10.1190/segam2018-2997928.1 Sun H. Demanet L. , 2018, Low frequency extrapolation with deep learning: 88th Annual International Meeting, SEG, Expanded Abstracts, 2011–2015, doi: http://dx.doi.org/10.1190/segam2018-2997928.1. Inverse problem theory and methods for model parameter estimation Tarantola 2005 10.1137/1.9780898717921 0016-8033 Geophysics Ulrych 36 650 1971 10.1190/1.1440202 Application of homomorphic deconvolution to seismology 0016-8033 Geophysics Wang 71 3 V51 2006 10.1190/1.2192912 Inverse Q-filter for seismic resolution enhancement 0016-8033 Geophysics Widess 38 1176 1973 10.1190/1.1440403 How thin is a thin bed 0956-540X Geophysical Journal International Zhou 205 203 2016 10.1093/gji/ggv553 Enhancing the resolution of non-stationary seismic data using improved time-frequency spectral modelling