Related Experiment Video
Updated: Jul 12, 2026
12:18
Pore-scale Imaging and Characterization of Hydrocarbon Reservoir Rock Wettability at Subsurface Conditions Using X-ray Microtomography
Published on: October 21, 2018
Lithology Identification of Metamorphic Basement Reservoirs in the J Oilfield Using a Linear Discriminant
Yifan Wang1,2, Keyu Liu1,2,3, Taixun Liu2
1State Key Laboratory of Deep Oil and Gas, China University of Petroleum (East China), Qingdao, Shandong 266580, China.
ACS Omega
|July 10, 2026
Summary
Accurate lithology identification in offshore metamorphic basement reservoirs is challenging due to limited core data. This study introduces a machine-learning framework using elemental and conventional well logs to improve lithology prediction accuracy.
Area of Science:
- Geoscience
- Machine Learning
- Petroleum Engineering
Background:
- Precise lithology identification is crucial for characterizing complex metamorphic basement reservoirs and fractured intervals.
- Offshore environments present challenges for lithology identification due to scarce and costly core data.
- Existing methods often struggle with the complexity of metamorphic basement lithologies.
Purpose of the Study:
- To develop a novel machine-learning-driven lithological identification framework for complex metamorphic basement reservoirs in offshore environments.
- To integrate elemental and conventional well logging data for enhanced lithology prediction.
- To address the challenge of limited core data in offshore exploration.
Main Methods:
- A hybrid classification approach combining Linear Discriminant Analysis (LDA) and Supervised Self-Organizing Map (SSOM).
- LDA was used for feature extraction, reducing dimensionality by identifying key elemental and log responses.
- SSOM was employed as a classifier, using core lithologies for supervised constraints to capture nonlinear relationships.
Main Results:
- The proposed LDA-assisted SSOM model achieved an overall accuracy of 81.4% and a blind test accuracy of 75.0%.
- The hybrid model significantly outperformed methods relying solely on Elemental Capture Spectroscopy (ECS) or conventional logs.
- Effective dimensionality reduction and capture of nonlinear log-lithology relationships were demonstrated.
Conclusions:
- The supervised LDA-assisted SSOM workflow provides a robust solution for precise lithology prediction in metamorphic basement reservoirs.
- Integration of elemental and conventional well log data enhances lithological identification accuracy.
- This machine-learning framework offers a valuable tool for offshore reservoir characterization with limited core data.
