Scholarlink

<p>Recent breakthroughs in ML and DL are reshaping geoscientific workflows by enabling more accurate data processing and interpretation, reducing manual intervention, and enhancing predictive performance. This review synthesizes state-of-the-art applications of ML and DL across key geoscience domains, including lithological and stratigraphic mapping, hydrological and hydrogeological modelling, geophysical and geochemical exploration, geohazard assessment, structural and tectonic analysis, and mineral and ore prospecting. A bibliometric analysis of over 800 peer-reviewed studies published between 2015 and 2024 underscores the accelerating adoption of AI-driven techniques—particularly convolutional neural networks, ensemble learning, and hybrid models. The findings reveal that ML/DL models are especially effective in handling heterogeneous, multisource geospatial and geophysical datasets. Nonetheless, several challenges persist, including limited model interpretability, restricted transferability across geologic contexts, and the need for large, well-annotated training datasets. To overcome these barriers, the review highlights emerging research directions, including explainable AI (XAI), domain-adaptive learning, semi-supervised methods, physics-informed neural networks, and real-time geological modelling. By mapping current capabilities and future directions, this comprehensive review provides a foundational reference for advancing AI-driven geoscience. It outlines a methodological roadmap for fostering innovation, scalability, and scientific transparency in the field.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine and deep learning in geological applications: a review of advances, challenges, and future research directions

  • 期刊论文

  • 发表日期：

    2026年2月23日

查看全文
  • Nezar Hammouri,
  • Rami Al-Ruzouq,
  • Abdallah Shanableh,
  • Ratiranjan Jena,
  • Hamdan A. Hamdan,
  • Mohamed Barakat G. Gibril,
  • Daniel Moraetis,
  • Mohamed I. Abdel-Fattah

摘要

Recent breakthroughs in ML and DL are reshaping geoscientific workflows by enabling more accurate data processing and interpretation, reducing manual intervention, and enhancing predictive performance. This review synthesizes state-of-the-art applications of ML and DL across key geoscience domains, including lithological and stratigraphic mapping, hydrological and hydrogeological modelling, geophysical and geochemical exploration, geohazard assessment, structural and tectonic analysis, and mineral and ore prospecting. A bibliometric analysis of over 800 peer-reviewed studies published between 2015 and 2024 underscores the accelerating adoption of AI-driven techniques—particularly convolutional neural networks, ensemble learning, and hybrid models. The findings reveal that ML/DL models are especially effective in handling heterogeneous, multisource geospatial and geophysical datasets. Nonetheless, several challenges persist, including limited model interpretability, restricted transferability across geologic contexts, and the need for large, well-annotated training datasets. To overcome these barriers, the review highlights emerging research directions, including explainable AI (XAI), domain-adaptive learning, semi-supervised methods, physics-informed neural networks, and real-time geological modelling. By mapping current capabilities and future directions, this comprehensive review provides a foundational reference for advancing AI-driven geoscience. It outlines a methodological roadmap for fostering innovation, scalability, and scientific transparency in the field.