Petroleum Science >2026, Issue6: 3037-3058 DOI: https://doi.org/10.1016/j.petsci.2026.02.010
Reinterpretation of volcanic and hydrothermal influence on organic matter accumulation and paleo-depositional environment in a lacustrine shale system: Insight from interpretable machine learning model Open Access
文章信息
作者:Enze Wang, Xiaoxiao Ma, Maowen Li, Yingxiao Fu, Menhui Qian, Tingting Cao, Zhiming Li, Yue Feng, Zhijun Jin, Tong Li
作者单位:
投稿时间:
引用方式:Wang, E., Ma, X., Li, M., et al., 2026. Reinterpretation of volcanic and hydrothermal influence on organic matter accumulation and paleo-depositional environment in a lacustrine shale system: Insight from interpretable machine learning model. Petrol. Sci. 23 (6), 3037–3058. https://doi.org/10.1016/j.petsci.2026.02.010.
文章摘要
The significant resource potential of lacustrine shale oil has spurred interest in understanding organic matter enrichment mechanisms. However, conventional analytical methods are often constrained by multicollinearity and statistical limitations, hindering accurate identification of enrichment mechanisms under complex depositional conditions. Interpretable machine learning models present a viable alternative to address this issue. The 7th member of the Yanchang Formation (T3y7) is one of the key targets for lacustrine shale oil exploration and development in China. Nevertheless, the role of volcanic activity in organic matter accumulation within this unit remains controversial, hindering a comprehensive understanding of organic enrichment mechanisms in lacustrine systems. To address this challenge, this study integrates geochemical data from YYa borehole in the southern Ordos Basin and previously published datasets to reconstruct the paleoenvironmental settings of various lithofacies assemblages in the T3y7 shale. By leveraging published literature and new experimental data, this study develops an interpretable machine learning model based on random forest (RF), extreme gradient boosting (XGBoost), and support vector machine (SVM) to quantitatively determine the dominant controls on organic matter accumulation. Using the T3y7 shale as a case study, the influence mechanisms of both volcanic and hydrothermal activities on organic enrichment in lacustrine shale systems are further explored. Two distinct lithofacies assemblages were identified in the T3y7 shale: laminated black shale and massive dark-gray mudstone. These assemblages are characterized by significant differences in paleoclimate, salinity, redox conditions, primary productivity, terrigenous influx, and the intensity of volcanic and hydrothermal activities. Results from the interpretable machine learning model indicate that organic matter enrichment in the T3y7 shale is primarily governed by the coupling of high primary productivity and favorable preservation conditions, and decoupling of volcanic and hydrothermal activities was a significant factor in the development of different lithofacies in the T3y7. Importantly, the influence of volcanic activity on organic matter enrichment in lacustrine systems is nonlinear: while moderate volcanic activity promotes organic accumulation, intense volcanic activity tends to exert a detrimental effect. In contrast, hydrothermal activity appears to play a more consistently positive role in the formation of organic-rich shale in the T3y7. These findings highlight the need to distinguish between volcanic and hydrothermal processes in future studies in order to more objectively evaluate their respective impacts on organic-rich shale development in lacustrine settings. This study provides a robust framework for understanding the mechanisms of organic matter enrichment in shale deposited under complex geological conditions, and deepens our insight into the formation of organic-rich shale in lacustrine sedimentary systems influenced by volcanic and hydrothermal activity.
关键词
-
Lacustrine shale; Paleoenvironment reconstruction; Interpretable machine leaning; Volcanic and hydrothermal activity; Organic matter enrichment mechanisms; Ordos Basin