Abstract:Due to the complex depositional environment of a turbidite reservoir in the Niuzhuang Delta, China, the traditional seismic facies classification is a challenge to perform accurately and continuously. Due to the thin turbidite layers in the reservoir, machine-learning based prediction of sandstone thickness is challenging. Inspired by the autoencoder, we propose an open-source deep learning workflow that combines unsupervised and supervised learning with jointed latent eigenvalues of the convolutional autoencoder and traditional seismic attributes for seismic facies classification and sandstone thickness prediction constrained by the facies distribution. First, we extract lower-dimensional latent eigenvalues as a category of novel seismic attributes from the seismic data using a convolutional autoencoder. To accurately and effectively extract lower-dimensional latent eigenvalues, we propose a hybrid loss function based on the mean squared error loss and the smooth L1 loss in this convolutional autoencoder. Then, we use principal component analysis to extract the first four principal components of these seismic lower-dimensional latent eigenvalues. Using unsupervised K-means, we cluster the first four principal components to form seismic facies. Finally, we take the first four principal components with the traditional seismic attributes as input and the sandstone thickness as labels for the random forest to predict the sandstone thickness distribution. The results of seismic facies and sandstone thickness distribution confirm the potential and advantages of the proposed workflow, which can speed up the identification of seismic facies with smoother boundaries, improve the prediction accuracy by 16% over than that of traditional seismic attributes, and provide more depositional insight for a turbidite reservoir of the Shahejie Formation in the Niuzhuang Delta, China.

Unsupervised seismic facies classification using deep convolutional autoencoder

Semi‐supervised Deep Autoencoder for Seismic Facies Classification

Recurrent Autoencoder Model for Unsupervised Seismic Facies Analysis

Unsupervised seismic facies analysis via deep convolutional autoencoders

Unsupervised Seismic Facies Classification Based on Multi-Feature Fusion Autoencoder

An Improved Deep Dilated Convolutional Neural Network for Seismic Facies Interpretation

Seismic facies recognition based on prestack data using deep convolutional autoencoder.

Unsupervised Seismic Facies Analysis Via Class-Imbalanced Deep Embedding Clustering

Seismic attribute analysis with a combination of convolutional autoencoder and random forest in a turbidite reservoir

Geology-constrained dynamic graph convolutional networks for seismic facies classification

Seismic Facies Recognition of Ultra-Deep Carbonate Rocks Based on Convolutional Neural Network

A Stronger Baseline for Seismic Facies Classification With Less Data

Seismic Facies Classification Algorithm Based on the EarthTransNet

Unsupervised Deep Learning for Facies Pattern Recognition on Borehole Images

Deep Classified Autoencoder for Lithofacies Identification

ADDCNN: an Attention-Based Deep Dilated Convolutional Neural Network for Seismic Facies Analysis with Interpretable Spatial–Spectral Maps

Supervised seismic facies analysis based on image segmentation

Improved Automating Seismic Facies Analysis Using Deep Dilated Attention Autoencoders

2D Multi-Spectral Convolutional Encoder-Decoder Model for Geobody Segmentation

Seismic Facies Recognition Based on Prestack Data Using Two-Dimensional Convolutional Auto-Encoder and Cluster Analysis

Semisupervised Multiattribute Seismic Facies Analysis