Separation of simultaneous source wavefields using convolutional neural network

Zhehao Li, Hua-Wei Zhou, Kang Fu

International Meeting for Applied Geoscience and Energy (IMAGE)

... the unblended data while 𝑑𝑑 𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 is the deblended result. , (2) Figure 1: Denoising based convolution neural network (DnCNN...

Orthorhombic AVAZ inversion in the Laplace-Fourier domain

Lin Li, Jinmiao Zhang, Guangzhi Zhang, Zhenyu Zhu, Jiajia Zhang, Chao Li, Xinpeng Pan

International Meeting for Applied Geoscience and Energy (IMAGE)

...) derivative matrix, the T denotes the transpose of a matrix. symbol The convolution of the reflection coefficients R with the wavelets can be formulated...

Implementation of Seismic Data Quality Characterisation Using Supervised Deep Learning

Joshua Thorp, Krista Davies, Julien Bluteau, Peter Hoiles

Australian Petroleum Production & Exploration Association (APPEA) Journal

...) using the Google Tensorflow framework (Google LLC). It comprised six convolution layers plus one fully connected layer at each end. Each...

An integrated workflow for deep learning-accelerated seismic modelling of the Groningen gas field, the Netherlands

Haibin Di, Vanessa Simoes, Zhun Li, Cen Li, Anisha Kaul, Aria Abubakar

International Meeting for Applied Geoscience and Energy (IMAGE)

... structures that are a result of salt tectonics in the area, and a pseudo-3D CNN with dilated convolution is used here for salt body isolation. Figure 3d...

Joint 3D inversion of gravity and magnetic data using deep learning neural networks

Nanyu Wei, Dikun Yang, Zhigang Wang, Yao Lu

International Meeting for Applied Geoscience and Energy (IMAGE)

... take place at any stages of convolution, as long as the block sizes becomes the same (Figure 1b). In our work, we merge the branches at the end...

Deep learning decomposition for null and active space estimation for thin-bed reflectivity inversion

Kristian Torres, Mauricio D. Sacchi

International Meeting for Applied Geoscience and Energy (IMAGE)

... convolution via simple matrix-times-vector multiplication. The sought vector of model parameters m is the reflectivity, and d contains the seismic traces...

Value of elastic full-wavefield inversion in derisking clastic reservoirs in presence of noise

Partha Routh, Soumya Nayak, Jake Dorsett, Kaushik Banerji, Young Ho Cha, Vijay Gottumukkula, Gopal Palacharla, Vikas Gudipati, Ramesh (Neelsh) Neelamani, Seth Conway, Andrea Raphael

International Meeting for Applied Geoscience and Energy (IMAGE)

..., such that convolution modeling can be applied in a local window where the wavelet is stationary. If the processing can attenuate non-primary noises...

Upside down Rayleigh-Marchenko imaging: Towards a practical exact redatuming scheme for ocean bottom acquisitions

Ning Wang, Matteo Ravasi

International Meeting for Applied Geoscience and Energy (IMAGE)

...-dimensional convolution operator with particle velocity wavefield along the z-axis at the ΛR level. This means that the RM method needs a dense...

Probabilistic seismic interpolation with the implicit prior of a deep denoiser

Matteo Ravasi

International Meeting for Applied Geoscience and Energy (IMAGE)

... of a convolution, pσ (y) can be interpreted as a Gaussian-blurred version of p(x), with the level of blurring gσ (x) being directly proportional...

Fast viscoacoustic forward modeling method based on U-net Fourier neural operator

Wenbin Tian, Yang Liu

International Meeting for Applied Geoscience and Energy (IMAGE)

... learning approach by incorporating physical information into the structure of the convolution layer. This yields a deep learning architecture...

PP-PS depth-domain AVA inversion

Maud Cavalca, Robin P. Fletcher

International Meeting for Applied Geoscience and Energy (IMAGE)

... images in the time domain through convolution of the reflectivity with 1D wavelets, the above system accounts for illumination effects through...

The impact of the synthetic seismic data generation method on automated AI-based horizon interpretation

F. Vizeu, J. Zambrini, A. Canning

International Meeting for Applied Geoscience and Energy (IMAGE)

... convolution modeling and does not involve 3D wave equation simulations. This approach is suitable for postmigration challenges where seismic events...

Impact of Seismic Attenuation on Downhole Microseismic Moment Tensor Inversion

Donghong Pei, Natalia Verkhovtseva, Jon Doucette, Price Stark

Unconventional Resources Technology Conference (URTEC)

... is represented as the convolution of M and the gradient of a Green’s function Gnp,q: 3 3 un (t )   M pq * G np ,q (2) p 1 q 1 where * stands...

Viscoacoustic wave propagation: Spectral-element simulations with a novel GSLS-type wave equation

Yu Chen, Bin He, Hejun Zhu

International Meeting for Applied Geoscience and Energy (IMAGE)

...; Qk is the quality factors of bulk moduli, which is equal to quality factors of P wave velocity. The convolution term can be eradicated by defining...

The reflected waves distributed on 3D9C seismic data from ISO media

Yuan Yuan Yue, Zhiwen Deng, Hong Cao, Qianggong Song

International Meeting for Applied Geoscience and Energy (IMAGE)

... to the reflection coefficient, then is convolved with a 20Hz Ricker wavelet. The convolution results are shown in Figure 4. It can be seen that the position where...

Uncertainty quantification of single and multi-parameter full-waveform inversion through a variational autoencoder

Abdelrahman Elmeliegy, Mrinal Sen, Jennifer Harding, Hongkyu Yoon

International Meeting for Applied Geoscience and Energy (IMAGE)

...standard deviation 𝜎. This is done through a sequence of 2D convolution layer Fourth International Meeting for Applied Geoscience ...

A Physics-Guided Deep Learning Predictive Model for Robust Production Forecasting and Diagnostics in Unconventional Wells

Syamil Mohd Razak, Jodel Cornelio, Young Cho, Hui-Hai Liu, Ravimadhav Vaidya, Behnam Jafarpour

Unconventional Resources Technology Conference (URTEC)

.... The component 𝑂. consists of a one-dimensional (1D) convolution operation, leaky-ReLU non-linear activation function and a one-dimensional up-sampling...

Some Shallow Tectonic Consequences of Subduction and Their Meaning to the Hydrocarbon Explorationist: Hydrocarbons

Peter A. Montecchi

AAPG Special Volumes

... the resistance to diastrophism of the contiguous and yet-undisturbed portion of the layer. The development of the next convolution will start as soon...

Abstract: Q Estimation by a Match-filter Method; #90174 (2014)

Peng Cheng and Gary F. Margrave

Search and Discovery.com

... convolution, the minimization is taken over the range of possible Q values; ( the forward Q filter for a quality factor value Q and travel time ( ) and can...

Seismic Facies Segmentation Using Deep Learning; #42286 (2018)

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

Search and Discovery.com

... blocks represent a convolution operation followed by a ReLU activation and a batch normalization (Chevitarese et al., 2018b). (b) Semantic...

Understanding the Seismic Wavelet; Steven G. Henry; Search and Discovery Article #40028 (2001)

Search and Discovery.com

Geology of the Ferris-Seminoe Mountain Area

L. C. Carpenter, H. T. Cooper

Wyoming Geological Association

... beneath over-lapping Tertiary sediments. It is the authors' opinion that the Ferris-Seminoe anticline is a relatively minor convolution on the southern...

Broadband Marine Seismic, Does Acquisition Make a Difference?

Martin Bayly, Sandeep Kumar Chandola, Low Cheng Foo, Abdul Aziz Muhamad, Swee Leng Ng

Indonesian Petroleum Association

... the success of these methods and their potential pitfalls. The key determinant of the eventual bandwidth of surface seismic data is the convolution...

Looking for a simplified and generalized training set in ML applications for gravity modelling

Luigi Bianco, Ciro Messina, Maurizio Fedi

International Meeting for Applied Geoscience and Energy (IMAGE)

..., doi: https://doi.org/10.1029/2021JB022476. Traore, B. B., B. Kamsu-Foguem, and F. Tangara, 2018, Deep convolution neural network for image recognition...

Least Square Q-Kirchhoff Migration: Implementation and Application

Xiaodong Wu, Yi Xie, Joe Zhou, Dechun Lin, Yu Wang

Petroleum Exploration Society of Australia (PESA)

...) is obtained by convolution Bm1 . Q It is worth mentioning that because the solution space for B is large, some sparse constraints have to be imposed on (4...