Abstract:Fault identification plays a vital role in geological structure interpretation, reservoir characterization, and reservoir evaluation. The traditional process for fault identification is label intensive. Deep learning-based, automatic fault detection from seismic data has attracted extensive attention in recent years. However, using a 2D neural network (i.e., single seismic data slice) leads to the loss of correlation between seismic slices, and using a 3D neural network (i.e., 3D seismic data cube) increases the workload of fault labeling and the difficulty of training. To overcome the drawbacks of 2D and 3D neural networks, in this work, we introduce a channel attention mechanism in the original U-net architecture. The new model, which is referred to as 2.5D channel attention U-net (2.5D CAU-net), takes four adjacent seismic slices as inputs instead of the entire 3D seismic data cube. The efficacy of 2.5D CAU-Net for fault detection is demonstrated using publicly available synthetic seismic datasets and real seismic field datasets. We explored the influence of cropping the original seismic data into various sizes on fault identification. Experimental results show that the 2.5D CAU-net can effectively utilize the correlation information among adjacent seismic slices to improve fault detection performance. For different seismic data cropping strategies, results indicate that the more extensive size cropping method allows the network to obtain more contextual information and more complete fault features, improving the efficacy of fault picking. However, the cropping approach with a larger size leads to a smaller number of labeled examples. When the amount of training data is not large enough, the larger size cropping approach tends to lead to model over-fitting, which will decrease the performance of the fault recognition, especially in a 3D neural network. Overall, 2.5D CAU-net can improve the efficiency of fault recognition with less annotation effort and computational resources.

A Two-Stage Convolutional Neural Network for Interactive Channel Segmentation From 3-D Seismic Data

Seismic Lithology/Fluid Prediction via a Hybrid ISD-CNN

Seismic Fault Detection with Convolutional Neural Network

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

A deep convolutional encoder-decoder neural network in assisting seismic horizon tracking

Seismic stratigraphy interpretation by deep convolutional neural networks: A semisupervised workflow

A Spatially Coupled Data-Driven Approach for Lithology/Fluid Prediction

3-D Seismic Fault Detection Using Recurrent Convolutional Neural Networks With Compound Loss

KG-Unet: a knowledge-guided deep learning approach for seismic facies segmentation

Learning Cross-Channel Representations for Semantic Segmentation

Deep learning for high-resolution multichannel seismic impedance inversion

High-resolution Seismic Impedance Inversion Integrating the Closed-Loop Convolutional Neural Network and Geostatistics: an Application to the Thin Interbedded Reservoir

CSDN: Combing Shallow and Deep Networks for Accurate Real-time Segmentation of High-definition Intravascular Ultrasound Images

Seismic resolution improving by a sequential convolutional neural network

Ground-truth-free Deep Learning for 3-D Seismic Denoising and Reconstruction with Channel Attention Mechanism

Fault-Seg-LNet: A method for seismic fault identification based on lightweight and dynamic scalable network

Supervised seismic facies analysis based on image segmentation

Connect the Dots: In Situ 4D Seismic Monitoring of CO2 Storage with Spatio-temporal CNNs

Automatic Geological Fault Identification from Seismic Data Using 2.5D Channel Attention U-net

Review of Artificial Intelligence for Oil and Gas Exploration: Convolutional Neural Network Approaches and the U-Net 3D Model

Optimizing seismic-based reservoir property prediction: a synthetic data-driven approach using convolutional neural networks and transfer learning with real data integration