A practical approach to automate end-to-end multi-stage iterative source separation with prior framework using machine learning

Rajiv Kumar, Yousif Izzeldin Kamil Amin, Riccardo Giro, Sunil Manikani, Nam Pham, Massimiliano Vassallo, Phillip Bilsby, Tao Zhao

International Meeting for Applied Geoscience and Energy (IMAGE)

... relies on the fact that the seismic signal of interest exhibits higher coherency and is sparse in the transform domain, whereas the interference noise...

Abstract: Cost Efficient Acquisition to Reduce Coarse Land 3D Line Spacings Through Beyond Nyquist Interpolation and Wavefield Reconstruction for Signal and Noise; #90187 (2014)

Bill Goodway

Search and Discovery.com

... not exceed Nyquist. Both authors concluded that the assumption of a smoothly varying linear model for the wavefield (or a plane wave decomposition...

Seismic Data Preconditioning for Improved Reservoir Characterization (Inversion and Fracture Analysis); #41347 (2014)

Darren Schmidt, Alicia Veronesi, Franck Delbecq, and Jeff Durand

Search and Discovery.com

... inversion schemes use well logs to construct the low frequency model to account for the missing low frequencies in the seismic. When the model has to fill...

Seismic Meta-Attributes and the Illumination of the Internal Reservoir Architecture of a Deepwater Synthetic Channel Model, #41267 (2014)

Staffan Van Dyke, Renjun Wen

Search and Discovery.com

...Seismic Meta-Attributes and the Illumination of the Internal Reservoir Architecture of a Deepwater Synthetic Channel Model, #41267 (2014) Staffan...

Conditioning Stratigraphic, Rule-Based Models with Generative Adversarial Networks: A Deepwater Lobe, Deep Learning Example; #42402 (2019)

Honggeun Jo, Javier E. Santos, Michael J. Pyrcz

Search and Discovery.com

... trend model, parameterized by gradients, orientations, mean, and standard deviation. Our deep learning-based, local data conditioning workflow consists...

3D velocity model building based upon hybrid neural network

Herurisa Rusmanugroho, Junxiao Li, M. Daniel Davis Muhammed, Jian Sun

International Meeting for Applied Geoscience and Energy (IMAGE)

... as an image are passed through some convolutional layers to estimate P-velocity model. This network is expected to learn from the features obtained from...

Abstract: Modeling of Seismic Signatures of Carbonate Rock Types, by B. Jan and Y. Sun, #90188 (2014)

Search and Discovery.com

Geostatistical Integration of Crosswell Data for Carbonate Reservoir Modeling, Mcelroy Field, Texas

William M. Bashore, Robert T. Langan, Karla E. Tucker, Paul J. Griffith

Special Publications of SEPM

... structures in order to be useful for the inversion process. The inversion is performed in the frequency domain, which requires the low-frequency model...

Implementation of Denoising Diffusion Probability Model for Seismic Interpretation

Fan Jiang, Konstantin Osypov, Julianna Toms

International Meeting for Applied Geoscience and Energy (IMAGE)

...: Aleatoric uncertainty; 5: Epistemic uncertainty. As generative model, diffusion process gains popularity and recognition in the image generation domain...

Noise analysis and ML denoising of DAS VSP data acquired from ESP lifted wells

Ge Zhan, Yao Zhao, Cheng Cheng, Josef Heim, Weihong Fei, Mike Craven, Scott Baker, Gilles Hennenfent

International Meeting for Applied Geoscience and Energy (IMAGE)

... developed a machine learning (ML) workflow that uses a deep convolutional U-Net architecture to model the ESP noise first and then subtract it from...

Representation Learning in Seismic Interpretation

Search and Discovery.com

Abstracts: Application of Neural Network Analysis and Post-Stack Inversion - Case Studies in Alberta; #90173 (2015)

Somanath Misra and Satinder Chopra

Search and Discovery.com

... the P-impedance from the post-stack data by way of model based inversion as well as neural network analysis. We are showing comparisons of the results...

Internal multiple elimination with an inverse-scattering theory guided deep neural network

Zhiwei Gu, Liurong Tao, Haoran Ren, Ru-Shan Wu, Jianhua Geng

International Meeting for Applied Geoscience and Energy (IMAGE)

... with the convolutional operation. Combining the CNN with the autoencoder can improve the feature extraction ability of the network model and have higher computational...

Validating machine learning-based seismic property prediction through self-supervised seismic reconstruction

Tao Zhao, Haibin Di, Aria Abubakar

International Meeting for Applied Geoscience and Energy (IMAGE)

... inversion, one uses impedance from well logs to build a low-frequency initial model, then computes the misfit between measured and modeled seismic data...

Facies-constrained elastic full-waveform inversion for tilted orthorhombic media

Ashish Kumar, Ilya Tsvankin

International Meeting for Applied Geoscience and Energy (IMAGE)

... offset is 3.6 km and maximum offsetto-depth ratio for the bottom of the model is about 2.6. The Ricker wavelet with a central frequency of 10 Hz...

Deep learning velocity model building using an ensemble regression approach

Stuart Farris, Guillaume Barnier, Robert Clapp

International Meeting for Applied Geoscience and Energy (IMAGE)

... framework that uses a convolutional neural network (CNN) to form an ensemble of low wavenumber model predictions which can be integrated to form...

Correlating Versus Inverting Vibroseis Records: Recovering What You Put into the Ground; #41577 (2015)

Glen Larsen, Paul Hewitt, Art Siewert

Search and Discovery.com

...) based on work of Allen et al. (1998). In effect, spiking the trace reduces it to a phase only operator. The usual vibroseis convolutional model is: x...

High-fidelity GPR image super-resolution via deep-supervised machine learning

Kai Gao, Carly M. Donahue, Bradley G. Henderson, Ryan T. Modrak

International Meeting for Applied Geoscience and Energy (IMAGE)

... migration images. To achieve this task, we adopt an attention-based residual convolutional neural network as the backbone (Bi et al., 2021), which uses...

Deep Dix: Enhancing interval velocity model estimation through adversarial regularization

Joseph Stitt, Robert Clapp, Biondo Biondi

International Meeting for Applied Geoscience and Energy (IMAGE)

... that Convolutional Neural Networks (CNNs) have successfully generated mappings from low-frequency shot gathers to low-wavenumber Earth model...

Seismic impedance inversion via neural networks and linear optimization algorithm

Bo Zhang, Yitao Pu, Ruiqi Dai, Danping Cao

International Meeting for Applied Geoscience and Energy (IMAGE)

..., and a low frequency model. The loss function of PINNs is designed to minimize the difference between real seismograms and synthetic seismic...

From Chaos to Caves … An Evolution of Seismic Karst Interpretation at the Vorwata Field

Riangguna Eloni, M.R. Husni Sahidu, Ilham Panggeleng, Christopher S. Birt, Ted Manning

Indonesian Petroleum Association

...’ between layers. It is often preferable to transform the reflectivity data into the impedance domain because impedance can be used to approximate...

Transfer Learning Applied to Seismic Images Classification; #42285 (2018)

Daniel Chevitarese, Daniela Szwarcman, Reinaldo Mozart D. Silva, Emilio Vital Brazil

Search and Discovery.com

... point to adjust model's parameters. A poor initialization may lead to longer training sessions or to the inability of finding a solution. To address...

Transfer learning seismic and GPR diffraction separation with a convolutional neural network

Alexander Bauer, Jan Walda, Dirk Gajewski

International Meeting for Applied Geoscience and Energy (IMAGE)

...Transfer learning seismic and GPR diffraction separation with a convolutional neural network Alexander Bauer, Jan Walda, Dirk Gajewski Transfer...

Depositional Facies Identification in Wireline Log Patterns Using 1D Convolutional Neural Network (CNN) Deep Learning Algorithms

Galatio Giovani Prabowo, Muhammad Fahmi Ramdani, Abiyyu Daffa Revanzha, Brian Muara Sianturi, Natalia Angel Momongan

Indonesian Petroleum Association

... to use Python, generating dummy data, training data, and model testing. The chosen tool for this research is the Convolutional Neural Network (CNN...

Automated Data and ML Pipelines to Accelerate Subsurface Digitalization

Raj Kannan, Vikas Jain

Unconventional Resources Technology Conference (URTEC)

... change over time rendering an ML model unable to perform as well as it did when it was first trained. Automated escalations to both domain experts...