Machine learning techniques in chemostratigraphy: A systematic literature review

Abstract

Chemostratigraphy, the analysis of geochemical signatures in a stratigraphic context, has gained significance in geosciences. Integration of chemostratigraphy with machine learning (ML) techniques has improved the analysis and correlation of depositional units based on their geochemical and isotopic composition. However, there is a lack of studies that comprehensively contextualize the current state of the art. To address this gap, a systematic literature review was conducted to analyze the use of ML in chemostratigraphy. For this purpose, six electronic databases were used to collect studies published after 2019. The data extraction process was conducted using the Parsifal software for study inclusion, exclusion, and quantitative evaluation. As a result, the literature search identified 1,198 publications, of which 35 were included in the review. Through this review, the most prevailing approaches were identified, providing a comprehensive overview of ongoing research aimed at enhancing chemostratigraphic interpretation. Additionally, some limitations are highlighted, such as scarcity of representative data and need for more interpretative methods. This study also points to open challenges, such as automation of interpretation and correlation process, and suggests future directions to advance in this research area. This systematic review contributes to the understanding of ML applications in chemostratigraphy and identifies avenues for further research and development in this interdisciplinary field, bridging the gap between geochemistry and ML methodologies.