Seismic facies analysis interprets depositional environment and facies types from the reflection seismic data, an important step in exploration and reservoir characterization. While machine learning methods, especially deep learning models such as convolutional neural networks (CNNs) have been applied to assist interpretation such as salt identification, significant challenges still remain for 3D multi-class seismic facies classification: complex data representation, limited labeled data for training, imbalanced facies class distribution and lack of rigorous performance evaluation mechanism in realistic settings. In this study, we investigate the feasibility of using a supervised CNN and a semi-supervised generative adversarial network (GAN) for 3D facies classification from seismic data and well logs, to overcome these challenges. We assess the performance with varying data representations and in situations with sufficient and limited well data respectively. Unlike previous studies, the 3D facies models in this work were generated from the structural information of an onshore field and calibrated on actual well log data and core analysis and therefore, both our analysis and results provide a realistic and meaningful implication for quantitative seismic interpretation.
Presentation Date: Wednesday, September 18, 2019
Session Start Time: 1:50 PM
Presentation Start Time: 3:30 PM
Location: 301B
Presentation Type: Oral
