Seismic characterization of deeply buried paleocaves based on Bayesian deep learning

Zhang, Guoyin; Lin, Chengyan; Liu, Ren; Shi-yin, LI; Cui, Shiti; Wang, Kaiyu; Sun, Yanyu

doi:10.1016/j.jngse.2021.104340

Cited by 7 publications

(3 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Manually labeling or interpreting seismic data could be extremely time-consuming and highly subjective. In addition, inaccurate manual interpretation, including mislabeled and unlabeled faults, may mislead the learning process (Pham et al, 2019;Wu et al, 2019;Wu et al, 2020;Alkhalifah et al, 2021;Zhang et al, 2022). To avoid these problems, we create synthetic seismic signals and corresponding labels based on the convolution model for training and validating our RNN model.…”

Section: Approach Datasetsmentioning

confidence: 99%

Deep learning based on self-supervised pre-training: Application on sandstone content prediction

Wang

Wang²,

Chen³

et al. 2023

Front. Earth Sci.

View full text Add to dashboard Cite

Deep learning has been widely used in various fields and showed promise in recent years. Therefore, deep learning is the future trend to realize seismic data’s intelligent and automatic interpretation. However, traditional deep learning only uses labeled data to train the model, and thus, does not utilize a large amount of unlabeled data. Self-supervised learning, widely used in Natural Language Processing (NLP) and computer vision, is an effective method of learning information from unlabeled data. Thus, a pretext task is designed with reference to Masked Autoencoders (MAE) to realize self-supervised pre-training of unlabeled seismic data. After pre-training, we fine-tune the model to the downstream task. Experiments show that the model can effectively extract information from unlabeled data through the pretext task, and the pre-trained model has better performance in downstream tasks.

show abstract

Section: Approach Datasetsmentioning

confidence: 99%

Deep learning based on self-supervised pre-training: Application on sandstone content prediction

Wang

Wang²,

Chen³

et al. 2023

Front. Earth Sci.

View full text Add to dashboard Cite

show abstract

“…The evaluation and prediction of paleokarst carbonate reservoirs are an important but challenging aspect of hydrocarbon exploration due to the extreme complexity and heterogeneity of karst systems, which are controlled by subaerial karstification and burial alteration after tens to hundreds of millions of years (Li R. et al, 2021;Zhang et al, 2022). Previous studies on paleokarst reservoir evaluation and prediction based on conventional paleogeomorphological reconstruction methods are commonly qualitative or semiquantitative, which can only indirectly reflect karst fluid hydrodynamics and cannot depict internal karst systems in line with the degree of karstification.…”

Section: Introductionmentioning

confidence: 99%

Multi-Factor Evaluation of Deep Karst Dolomite Reservoir Based on Paleogeomorphological Reconstruction, a Case Study From the 4th Member of the Dengying Formation in the Central Sichuan Basin, China

Zhou

Wang²,

Yang³

et al. 2022

Front. Earth Sci.

View full text Add to dashboard Cite

The evaluation of reservoirs is of great importance in exploration practices, and reservoir distribution is dependent on multiple geological factors. The carbonate platform of the fourth member (Z2dn4) of the Dengying Formation (Z2dn) in the Gaoshiti-Moxi area (GMA) was uplifted above sea level during episode II of the Tongwan movement and underwent prolonged karstification by meteoric freshwater in the terminal Sinian, leading to a strong heterogeneity in reservoirs. Paleogeomorphology reflects hydraulic gradients that can affect the intensities of surface karstification and may be approximated by terrain slope gradients. Karst products (especially solution vugs and caverns) exist prevailingly in the Z2dn4; their development mirrors the extents of internal dissolution. Based on paleogeomorphological reconstruction using the elevation method, the slope angles of single boreholes were calculated through the digital elevation model (DEM). In accordance with single-borewell Fullbore Formation MicroImager (FMI) logging data, four electrofacies (host rock, vug, fracture, and cave) were identified in boreholes possessing FMI data. The identified electrofacies were matched with corresponding conventional logging curves to obtain the training data. A model generalized for other boreholes without FMI data was established with the random forest algorithm and the thicknesses of the four electrofacies in each borehole were determined to characterize the extent of karstification. Slope gradients and electrofacies thicknesses, together with multiple physical parameters of reservoirs in boreholes, were utilized for factor analysis. Four factors were obtained and fused into one comprehensive score based on their ratios of variance proportions. The favorability of reservoirs was described by the comprehensive scores of each borehole. These analyses showed that the most advantageous reservoirs are distributed in the vicinity of the platform margin in the west of the GMA, where they enjoy several benefits. Deposition of porous mound-shoal complexes occurs predominantly in the platform margin where large-scale caves are more developed in the mixing karst zone of seawater and freshwater. Meanwhile, the transmeridional lateral migration of hydrocarbons from the regional hydrocarbon-generating center abutting the west of the GMA contributes to earlier entrapment and accumulation in the mound-shoal complexes in the platform margin.

show abstract

“…These options are based on the hypothesis that seismic from nearby areas have similar features. However, this may not be true in all cases, since there may be changes in the composition of the terrain (Mustafa and AlRegib, 2021;Rollmann et al, 2022;Zhang et al, 2022a).…”

Section: Motivationmentioning

confidence: 99%

Generalization of the Deep Learning Model for Natural Gas Indication in 2d Seismic Image Based on the Training Dataset and the Operational Hyper Parameters Recommendation

FERNANDO MARIN SEPULVEDA

View full text Add to dashboard Cite

I especially thank my advisors whose example, guidance and constant support have made me a better professional and human. Likewise, I thank my colleagues, both from Rio de Janeiro and São Luís, since their constant help has allowed me to advance. Finally, I thank my dear family, who at all times in my life have supported me and encouraged me to overcome any obstacle.To all of you my most sincere thanks.

show abstract

Seismic characterization of deeply buried paleocaves based on Bayesian deep learning

Cited by 7 publications

References 34 publications

Deep learning based on self-supervised pre-training: Application on sandstone content prediction

Deep learning based on self-supervised pre-training: Application on sandstone content prediction

Multi-Factor Evaluation of Deep Karst Dolomite Reservoir Based on Paleogeomorphological Reconstruction, a Case Study From the 4th Member of the Dengying Formation in the Central Sichuan Basin, China

Generalization of the Deep Learning Model for Natural Gas Indication in 2d Seismic Image Based on the Training Dataset and the Operational Hyper Parameters Recommendation

Contact Info

Product

Resources

About