What is it about?
This study introduces a novel multidimensional discrimination approach to differentiate between mid-ocean ridge (MOR) and oceanic plateau (OP) settings using geochemical data. By applying linear discriminant analysis (LDA) and canonical analysis on a large database of elemental compositions, the research develops two highly effective models: ilrM, which utilizes log-ratio transformations of major elements, and ilrMT, which integrates both major and trace elements.
Featured Image
Photo by Chris Yang on Unsplash
Why is it important?
Understanding the geological differences between MOR and OP settings is crucial for reconstructing Earth's tectonic and magmatic history. Traditional bivariate and ternary discrimination diagrams have limitations in accuracy and robustness due to data constraints and post-emplacement geochemical alterations. This study overcomes these challenges by leveraging multidimensional analysis, improving classification success rates to over 90% in some cases. Additionally, the robustness of the approach ensures reliability even in the presence of analytical uncertainties and compositional changes over time.
Perspectives
From a machine learning perspective, this study highlights the power of multivariate statistical methods in geochemical classification. The use of LDA and canonical analysis demonstrates the effectiveness of dimensionality reduction techniques in improving classification performance. In addition, the study's integration of log-ratio transformations addresses common problems in compositional data analysis, such as closure effects and non-normality. Future developments could explore the application of deep learning models to further refine classification accuracy. In addition, the incorporation of unsupervised learning techniques, such as clustering algorithms, could help uncover hidden geochemical patterns and improve tectonic setting predictions. The study sets a precedent for integrating AI-driven approaches into the geosciences, paving the way for more sophisticated and automated classification systems in the future.
Lorena Díaz-González
Universidad Autonoma del Estado de Morelos
Read the Original
This page is a summary of: New discriminant-function-based multidimensional discrimination of mid-ocean ridge and oceanic plateau, Geoscience Frontiers, September 2020, Elsevier,
DOI: 10.1016/j.gsf.2020.01.005.
You can read the full text:
Contributors
The following have contributed to this page
