We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Shale Anisotropy Model Building Based on Deep Neural Networks.
- Authors
You, Nan; Li, Yunyue Elita; Cheng, Arthur
- Abstract
Seismic anisotropy parameters are essential in the processing and interpretation of modern array data with multicomponent, long offsets and wide azimuth acquisitions. Traditional well logs do not measure anisotropy in a vertical well and thus cannot provide the needed information. Conventional calibration‐based as well as recent inversion‐based rock physics modeling methods involve tuning parameters and subjective choices that are largely empirical and difficult to generalize. Here we present a machine learning approach to alleviate these problems. Since it is impossible to collect massive labeled field well log data, we generate paired synthetic data of features (porosity, density, vertical P and S wave velocities, P wave and shear moduli) and labels (bulk and shear moduli of rock matrices and aspect ratio of ellipsoidal cracks). By tuning hyperparameters we obtain an optimal fully connected neural network with four hidden layers that fits well with the synthetic data. The neural network is applied to published laboratory measurements and field well log data from a Chinese well and a U.S. well without any modification. We show that anisotropy models estimated by the deep neural network agree well with the inversion results and with the laboratory measurements. The neural network optimized by extensive training based on massive synthetic data removes the subjectivity in parameter selection, generalizes to different geological environments, and has the potential to provide real‐time anisotropy estimation while logging.Plain Language Summary: Seismic anisotropy, defined as the dependency of seismic wave velocity on propagation directions, is essential for understanding the Earth's subsurface structure. However, anisotropy Earth model building is challenging and ambiguous due to the lack of direct measurements and the subjectivity commonly needed in rock physics modeling. Here, we propose a machine learning approach using a deep neural network trained with artificial data generated from the Hudson‐Cheng model to predict shale anisotropy from conventional well logs. The well‐trained neural network is then used to predict anisotropy parameters for shale rock samples whose properties were measured in laboratories and published in previous studies. The agreement between the deep neural network prediction and these measurements validates both the chosen rock physics model and the neural network. The neural network is applied directly to two field well log data sets, producing consistent anisotropy models in different geological settings in China and the United States, and compared favorably with the existing inversion‐based results. Our neural network is shown to be very efficient and general and can be applied in real time while well logs are being acquired.Key Points: A DNN trained with synthetics is proposed to predict seismic anisotropy from conventional well logsOur DNN model generalizes well to lab measurements and field well logs from China and the United StatesOur DNN model has the potential to provide real‐time estimates of anisotropy while logging
- Publication
Journal of Geophysical Research. Solid Earth, 2020, Vol 125, Issue 2, p1
- ISSN
2169-9313
- Publication type
Article
- DOI
10.1029/2019JB019042