Donglin Lin , Zhaodong Xi , Shuheng Tang , Gary G. Lash , Yang Chen , Zhifeng Yan
{"title":"Lithofacies types and formation mechanisms of Carboniferous - Permian shales: Insights from big data and machine learning","authors":"Donglin Lin , Zhaodong Xi , Shuheng Tang , Gary G. Lash , Yang Chen , Zhifeng Yan","doi":"10.1016/j.earscirev.2025.105099","DOIUrl":null,"url":null,"abstract":"<div><div>Carboniferous-Permian shale deposits around the world are known to contain abundant shale gas resources making them critical for increasing global shale gas reserves and production. Lithofacies analysis is crucial for identifying and predicting “sweet spots” targets. This study employed total organic carbon (TOC) data from 8166 samples, mineral content data from 4524 samples, and major and trace element data from 814 samples of Carboniferous and Permian shales worldwide. The aim of the present study is the generation of a classification scheme of the studied shale samples and elucidation of the conditions under which they accumulated. Random Forest and Artificial Neural Networks methods were employed to identify those factors that exerted greatest control on development of shale lithofacies and to explore the implications of lithofacies types on the exploration and development of Carboniferous and Permian shale gas. Our results, based on big data statistical and deconvolution analytical methods, a lithofacies classification scheme was proposed uses a TOC of 5.3 % as a boundary, and combined with a ternary diagram of siliceous‑carbonate-clay minerals. Seven main lithofacies were developed in the Carboniferous and Permian shales, which are organic-rich siliceous shale (Ss-H), organic-rich mixed shale (Ms-H), organic-rich argillaceous shale (CMs-H), low-organic matter siliceous shale (Ss-L), low-organic matter mixed shale (Ms-L), low-organic matter argillaceous shale (CMs-L), and low-organic matter calcareous shale (Cs-L). The development of a particular shale lithofacies at a specific time interval appears to have been largely controlled by paleoclimate, paleoproductivity, as well as terrigenous input. Ss-H appears to be the most promising shale lithofacies type for hydrocarbon exploration and development of Carboniferous and Permian shale gas. These organic and silica-rich deposits appear to have accumulated under warm, moist paleoclimate conditions, moderate paleoproductivity, and in association with increased volcanic activity. The results of this study provide theoretical guidance for shale lithofacies research as well as for the exploration and development of Carboniferous and Permian shale gas.</div></div>","PeriodicalId":11483,"journal":{"name":"Earth-Science Reviews","volume":"264 ","pages":"Article 105099"},"PeriodicalIF":10.8000,"publicationDate":"2025-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Earth-Science Reviews","FirstCategoryId":"89","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0012825225000601","RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"GEOSCIENCES, MULTIDISCIPLINARY","Score":null,"Total":0}
Lithofacies types and formation mechanisms of Carboniferous - Permian shales: Insights from big data and machine learning
Carboniferous-Permian shale deposits around the world are known to contain abundant shale gas resources making them critical for increasing global shale gas reserves and production. Lithofacies analysis is crucial for identifying and predicting “sweet spots” targets. This study employed total organic carbon (TOC) data from 8166 samples, mineral content data from 4524 samples, and major and trace element data from 814 samples of Carboniferous and Permian shales worldwide. The aim of the present study is the generation of a classification scheme of the studied shale samples and elucidation of the conditions under which they accumulated. Random Forest and Artificial Neural Networks methods were employed to identify those factors that exerted greatest control on development of shale lithofacies and to explore the implications of lithofacies types on the exploration and development of Carboniferous and Permian shale gas. Our results, based on big data statistical and deconvolution analytical methods, a lithofacies classification scheme was proposed uses a TOC of 5.3 % as a boundary, and combined with a ternary diagram of siliceous‑carbonate-clay minerals. Seven main lithofacies were developed in the Carboniferous and Permian shales, which are organic-rich siliceous shale (Ss-H), organic-rich mixed shale (Ms-H), organic-rich argillaceous shale (CMs-H), low-organic matter siliceous shale (Ss-L), low-organic matter mixed shale (Ms-L), low-organic matter argillaceous shale (CMs-L), and low-organic matter calcareous shale (Cs-L). The development of a particular shale lithofacies at a specific time interval appears to have been largely controlled by paleoclimate, paleoproductivity, as well as terrigenous input. Ss-H appears to be the most promising shale lithofacies type for hydrocarbon exploration and development of Carboniferous and Permian shale gas. These organic and silica-rich deposits appear to have accumulated under warm, moist paleoclimate conditions, moderate paleoproductivity, and in association with increased volcanic activity. The results of this study provide theoretical guidance for shale lithofacies research as well as for the exploration and development of Carboniferous and Permian shale gas.
期刊介绍:
Covering a much wider field than the usual specialist journals, Earth Science Reviews publishes review articles dealing with all aspects of Earth Sciences, and is an important vehicle for allowing readers to see their particular interest related to the Earth Sciences as a whole.