Seismic fault identification of deep fault-karst carbonate reservoir using transfer learning

Seismic fault identification is a critical step in structural interpretation, reservoir characterization, and well-drilling planning. However, fault identification in deep fault-karst carbonate formations is particularly challenging due to their deep burial depth and the complex effects of dissoluti...

Full description

Saved in:
Bibliographic Details
Main Authors: Hanqing Wang, Han Wang, Kunyan Liu, Jin Meng, Yitian Xiao, Yanghua Wang
Format: Article
Language:English
Published: KeAi Communications Co., Ltd. 2025-04-01
Series:Natural Gas Industry B
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2352854025000221
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850281151487279104
author Hanqing Wang
Han Wang
Kunyan Liu
Jin Meng
Yitian Xiao
Yanghua Wang
author_facet Hanqing Wang
Han Wang
Kunyan Liu
Jin Meng
Yitian Xiao
Yanghua Wang
author_sort Hanqing Wang
collection DOAJ
description Seismic fault identification is a critical step in structural interpretation, reservoir characterization, and well-drilling planning. However, fault identification in deep fault-karst carbonate formations is particularly challenging due to their deep burial depth and the complex effects of dissolution. Traditional manual interpretation methods are often labor intensive and prone to high uncertainty due to their subjective nature. To address these limitations, this study proposes a transfer learning–based strategy for fault identification in deep fault-karst carbonate formations. The proposed methodology began with the generation of a large volume of synthetic seismic samples based on statistical fault distribution patterns observed in the study area. These synthetic samples were used to pretrain an improved U-Net network architecture, enhanced with an attention mechanism, to create a robust pretrained model. Subsequently, real-world fault labels were manually annotated based on verified fault interpretations and integrated into the training dataset. This combination of synthetic and real-world data was used to fine-tune the pretrained model, significantly improving its fault interpretation accuracy. The experimental results demonstrate that the integration of synthetic and real-world samples effectively enhances the quality of the training dataset. Furthermore, the proposed transfer learning strategy significantly improves fault recognition accuracy. By replacing the traditional weighted cross-entropy loss function with the Dice loss function, the model successfully addresses the issue of extreme class imbalance between positive and negative samples. Practical applications confirm that the proposed transfer learning strategy can accurately identify fault structures in deep fault-karst carbonate formations, providing a novel and effective technical approach for fault interpretation in such complex geological settings.
format Article
id doaj-art-ed8f9eae47814bccaea2b95a1efb6638
institution OA Journals
issn 2352-8540
language English
publishDate 2025-04-01
publisher KeAi Communications Co., Ltd.
record_format Article
series Natural Gas Industry B
spelling doaj-art-ed8f9eae47814bccaea2b95a1efb66382025-08-20T01:48:26ZengKeAi Communications Co., Ltd.Natural Gas Industry B2352-85402025-04-0112217418510.1016/j.ngib.2025.03.006Seismic fault identification of deep fault-karst carbonate reservoir using transfer learningHanqing Wang0Han Wang1Kunyan Liu2Jin Meng3Yitian Xiao4Yanghua Wang5Petroleum Exploration and Production Research Institute, SINOPEC, Beijing, 102206, China; Resource Geophysics Academy, Imperial College London, London SW7 2AZ, UKPetroleum Exploration and Production Research Institute, SINOPEC, Beijing, 102206, China; Corresponding author.Petroleum Exploration and Production Research Institute, SINOPEC, Beijing, 102206, ChinaPetroleum Exploration and Production Research Institute, SINOPEC, Beijing, 102206, ChinaPetroleum Exploration and Production Research Institute, SINOPEC, Beijing, 102206, ChinaResource Geophysics Academy, Imperial College London, London SW7 2AZ, UKSeismic fault identification is a critical step in structural interpretation, reservoir characterization, and well-drilling planning. However, fault identification in deep fault-karst carbonate formations is particularly challenging due to their deep burial depth and the complex effects of dissolution. Traditional manual interpretation methods are often labor intensive and prone to high uncertainty due to their subjective nature. To address these limitations, this study proposes a transfer learning–based strategy for fault identification in deep fault-karst carbonate formations. The proposed methodology began with the generation of a large volume of synthetic seismic samples based on statistical fault distribution patterns observed in the study area. These synthetic samples were used to pretrain an improved U-Net network architecture, enhanced with an attention mechanism, to create a robust pretrained model. Subsequently, real-world fault labels were manually annotated based on verified fault interpretations and integrated into the training dataset. This combination of synthetic and real-world data was used to fine-tune the pretrained model, significantly improving its fault interpretation accuracy. The experimental results demonstrate that the integration of synthetic and real-world samples effectively enhances the quality of the training dataset. Furthermore, the proposed transfer learning strategy significantly improves fault recognition accuracy. By replacing the traditional weighted cross-entropy loss function with the Dice loss function, the model successfully addresses the issue of extreme class imbalance between positive and negative samples. Practical applications confirm that the proposed transfer learning strategy can accurately identify fault structures in deep fault-karst carbonate formations, providing a novel and effective technical approach for fault interpretation in such complex geological settings.http://www.sciencedirect.com/science/article/pii/S2352854025000221Seismic faultFault-karst carbonateU-NetTransfer learningAttention mechanism
spellingShingle Hanqing Wang
Han Wang
Kunyan Liu
Jin Meng
Yitian Xiao
Yanghua Wang
Seismic fault identification of deep fault-karst carbonate reservoir using transfer learning
Natural Gas Industry B
Seismic fault
Fault-karst carbonate
U-Net
Transfer learning
Attention mechanism
title Seismic fault identification of deep fault-karst carbonate reservoir using transfer learning
title_full Seismic fault identification of deep fault-karst carbonate reservoir using transfer learning
title_fullStr Seismic fault identification of deep fault-karst carbonate reservoir using transfer learning
title_full_unstemmed Seismic fault identification of deep fault-karst carbonate reservoir using transfer learning
title_short Seismic fault identification of deep fault-karst carbonate reservoir using transfer learning
title_sort seismic fault identification of deep fault karst carbonate reservoir using transfer learning
topic Seismic fault
Fault-karst carbonate
U-Net
Transfer learning
Attention mechanism
url http://www.sciencedirect.com/science/article/pii/S2352854025000221
work_keys_str_mv AT hanqingwang seismicfaultidentificationofdeepfaultkarstcarbonatereservoirusingtransferlearning
AT hanwang seismicfaultidentificationofdeepfaultkarstcarbonatereservoirusingtransferlearning
AT kunyanliu seismicfaultidentificationofdeepfaultkarstcarbonatereservoirusingtransferlearning
AT jinmeng seismicfaultidentificationofdeepfaultkarstcarbonatereservoirusingtransferlearning
AT yitianxiao seismicfaultidentificationofdeepfaultkarstcarbonatereservoirusingtransferlearning
AT yanghuawang seismicfaultidentificationofdeepfaultkarstcarbonatereservoirusingtransferlearning