High-resolution seismic reservoir monitoring with multitask and transfer learning

Ahmed M. Ahmed, Ilya Tsvankin, Yanhua Liu

International Meeting for Applied Geoscience and Energy (IMAGE)

... or hydrocarbon production. This study leverages convolutional neural networks (CNNs), multitask learning (MTL), and transfer learning (TL) to accurately...

Unsupervised deep learning for seismic data reconstruction

Gui Chen, Yang Liu, Mi Zhang

International Meeting for Applied Geoscience and Energy (IMAGE)

... (RSVD) and projection onto convex sets (POCS) algorithms iteratively to reconstruct each frequency slice of the incomplete data in the f-x domain...

Self-parametrizing seismic data processing modules: An example on coherent noise suppression

Simone Re, Ran Bachrach, Massimo Clementi, Collin Wilson

International Meeting for Applied Geoscience and Energy (IMAGE)

... in the frequency-offset (FX) domain (Bilsby, 2015) thus enabling its application to uneven and irregular seismic acquisition layouts (i.e., non-uniform...

Abstracts: Revisiting Homomorphic Wavelet Estimation and Phase Unwrapping; #90173 (2015)

Roberto H. Herrera and Mirko van der Baan

Search and Discovery.com

.... The convolutional operator is denoted by  and η (t ) is the additive noise. In the frequency domain equation (1) can be expressed as: = W ( f ) R( f ) + Ν...

Extrapolated surface-wave dispersion inversion

Hongyu Sun, Laurent Demanet

International Meeting for Applied Geoscience and Energy (IMAGE)

... with respect to S-wave velocity of the model. With decreasing frequency, surface waves become sensitive to deeper velocity structures. The lower...

Convolutional neural networks as an aid to biostratigraphy and micropaleontology: a test on late Paleozoic microfossils

Rafael Pires De Lima, Katie F. Welch, James E. Barrick, Kurt J. Marfurt, Roger Burkhalter, Murphy Cassel, Gerilyn S. Soreghan

PALAIOS

...) Inception V3 CNN architecture reached a 3.5% top-5 error (frequency in which the model cannot predict the correct class as one of the top five most...

Relative geologic time generation based on 3D-CNNs and domain adaptation

Xin He, Bangli Zou, Yifeng Fei, Gang Yu, Dajun Li

International Meeting for Applied Geoscience and Energy (IMAGE)

... of the domain adaptation layer, the trained model exhibits good generalization ability to field seismic data. Furthermore, fine-tuning allows...

Towards flexible demultiple with deep learning

Mario Fernandez, Norman Ettrich, Matthias Delescluse, Alain Rabaute, Janis Keuper

International Meeting for Applied Geoscience and Energy (IMAGE)

... moveout to be considered multiple reflections in Mi+1 than in Mi . We build the training data through the convolutional model for a large number...

Seismic data interpolation via frequency-constrained 3D inception Unet

Yen Sun, Paul Williamson

International Meeting for Applied Geoscience and Energy (IMAGE)

... a second label – the data transformed into the frequency-wavenumber domain; this requires the addition of a Fourier Transform layer to the architecture...

Demultiple for Wide-Tow Broadband Acquisition in a Shallow Water Environment: A Case Study from the NW Shelf, Australia

Mike Hartley, Shuo Ji, Alex Browne

Petroleum Exploration Society of Australia (PESA)

... in shallow water, especially for outer cables. Rcvr depth(m) Shot: channel Frequency(Hz) Time (sec) Model base water-layer demultiple predicts...

Refining our understanding of the subsurface geology using deep learning techniques

Salma Alsinan, Philippe Nivlet, Hamad Alghenaim

International Meeting for Applied Geoscience and Energy (IMAGE)

... such as the confidence in both data and model domain as well as defining a better metric for accuracy when addressing transfer learning techniques. In future...

S/N RATIO AND BANDWIDTH CONSIDERATIONS WHEN UTILIZING SEISMIC DATA IN EXPLORING FOR SUBTLE TRAPS - EXAMPLES FROM THE KNOX PLAY

Edward R. Tegland, Exploration Development, Inc., S. Pikes Peak Dr., Parker, CO Patrick H. Bygott, Exploration Development, Inc., S. Pikes Peak Dr., Parker, CO

Ohio Geological Society

.... What is bandwidth? Bandwidth is the difference between the highest and lowest measurable frequency present in the data...

Introducing stochasticity into CNN-based property estimation from angle-stack seismic

Haibin Di, Tao Zhao, Aria Abubakar

International Meeting for Applied Geoscience and Energy (IMAGE)

... and perturbing with Gaussian noises ℕ(0,1) per prior rock property model. convolutional layer for reconstructing the fullstack seismic, and (iii) one...

Abstract: Deterministic Marine Deghosting: Tutorial and Recent Advances; #90224 (2015)

Mike J. Perz and Hassan Masoomzadeh

Search and Discovery.com

... is minimum phase. We can easily express this operator in the frequency domain as (2) 𝐺(𝑤) = 1 − 𝑟𝑒 𝑖𝑤𝑡 𝑑 , and its minimum phase inverse as: 1...

Incorporating Artificial Intelligence into Traditional Exploration Workflows in the Cooper-Eromanga Basin, South Australia

H. M. Garcia, W. G. "Woody" Leel Jr., M. Riehle, P. Szafian

International Meeting for Applied Geoscience and Energy (IMAGE)

... that are diagenetically similar (should have the same frequency decomposition response). The 3D convolutional neural network shows an unprecedented level...

First arrival enhancement by statics preserving filtering using surface-consistent constraints

Alejandro Quiaro, Mauricio D. Sacchi

International Meeting for Applied Geoscience and Energy (IMAGE)

... model, we are considering the effect of T, as applying a time shift 't"sr = -r(s)+-r(r) +noise on the otherwise predictable signal. In the frequency...

U-net based primary alignment

Ricard Durall, Ammar Ghanim, Norman Ettrich

International Meeting for Applied Geoscience and Energy (IMAGE)

... processing workflows. Misalignments are mainly caused by inaccuracies in the velocity model. Traditional approaches to event flattening typically involve...

Conditional image prior for uncertainty quantification in full-waveform inversion

Lingyun Yang, Omar M. Saad, Tariq Alkhalifah, Guochen Wu

International Meeting for Applied Geoscience and Energy (IMAGE)

... data. However, FWI results are effected by the limited illumination of the model domain and the quality of that illumination, which is related...

High-resolution seismic data processing method based on deep convolutional dictionary learning

Xiayu Gao, Qingyu Feng, Yaojun Wang, Bangli Zou, Yang Luo

International Meeting for Applied Geoscience and Energy (IMAGE)

... decomposition on the entire image, fully considering the local relevance of seismic data and strictly following the seismic data convolutional model...

Improved Resolution of Thin Turbiditic Sands in Offshore Sabah with Bandwidth Extension … A Pilot Study (Paper C11)

G. Yu, N. Shah, M. Robinson, N. H. Nghi, A. A. Nurhono, G. S. Thu

Geological Society of Malaysia (GSM)

... by a convolutional-like process in the CWT domain as illustrated in Figure 2. This effectively reshapes the wavelet and broadens the spectrum. Any...

Deep learning approach for denoising and resolution enhancement of poststack seismic data

Ruslan Malikov, Tatyana Yusubova, Izat Shahsenov

International Meeting for Applied Geoscience and Energy (IMAGE)

... complexity, frequency content, and signal-to-noise ratio (SNR). Synthetic seismic data without the added noise template is used to train the denoising model...

Abstract: CO2 Distribution Prediction Using Machine Learning Based Proxy Model in Geological Carbon Sequestration;

Zhi Zhong, Alexander Sun

Search and Discovery.com

... knowledge and assumptions on input data distributions. In particular, our cDC-GAN model is designed to learn cross-domain mappings between high...

3D CNN for channel identification in seismic volume

Haishan Li, Wuyang Yang, Xiangyang Zhang, Xinjian Wei, Xin Xu

International Meeting for Applied Geoscience and Energy (IMAGE)

... volumes with complex structure using an end-to-end 3D convolutional neural network. To train the network, we automatically generate a training dataset...

Facies Classification Based on Well Logs by Using an Convolutional Neural Network

Search and Discovery.com

Improving fault resolution from multiple angle stacks by latent feature analysis with deep learning

Fan Jiang, Konstantin Osypov

International Meeting for Applied Geoscience and Energy (IMAGE)

... of seismic exploration, combining multiple stacks, e.g. multi-angle, multi-azimuth, multi-frequency, of seismic data is becoming more and more common...