Machine learning-based re-classification of the geochemical stratigraphy of the Rajahmundry Traps, India
Hoyer P, Regelous M, Adatte T, Haase K (2022)
Publication Type: Journal article
Publication year: 2022
Journal
Book Volume: 428
Article Number: 107594
DOI: 10.1016/j.jvolgeores.2022.107594
Abstract
The lavas exposed at the Rajahmundry Traps have similar ages and geochemical compositions compared to those from the Ambenali and Mahabaleshwar formations of the Deccan Traps, which is why they are believed to represent their eastern extensions. However recent geochronological data for the lavas crop out in the Rajahmundry quarries and paleontological evidence from the expanded sediment-lava sequences preserved in the Krishna-Godavari Basin indicate that the lower Rajahmundry Flow could represent a flow of the Poladpur formation. Here we present new major and trace element data for lavas of the lower (five samples) and middle (three samples) Rajahmundry Flows exposed in the Gowripatnam Quarry. To test whether the lower Rajahmundry Flow show geochemical similarities to the Poladpur formation, we applied a machine learning-based classification model which is able to predict the most probable formation affiliation for an unknown sample. The classification method we used is the Random Forest algorithm and training as well as testing our model with new and published geochemical data from the Poladpur and Ambenali formations (n = 73) by using Repeated K-Fold Cross Validation yields a prediction accuracy of ~86%. According to our model results, the lavas of the lower Rajahmundry Flow were assigned to the Poladpur formation, while those from the middle Rajahmundry Flow show geochemical similarities to the Ambenali formation. Flows of each of the voluminous Deccan formations (Poladpur, Ambenali and Mahabaleshwar) are therefore present at Rajahmundry, confirming that these are an eastward extension of the Deccan Traps.
Authors with CRIS profile
Patrick Hoyer
Lehrstuhl für Endogene Geodynamik (LEG)
Marcel Regelous
Geozentrum Nordbayern
Karsten Haase
Lehrstuhl für Endogene Geodynamik (LEG)
Related research project(s)
Involved external institutions
Switzerland (CH)How to cite
APA:
Hoyer, P., Regelous, M., Adatte, T., & Haase, K. (2022). Machine learning-based re-classification of the geochemical stratigraphy of the Rajahmundry Traps, India. Journal of Volcanology and Geothermal Research, 428. https://doi.org/10.1016/j.jvolgeores.2022.107594
MLA:
Hoyer, Patrick, et al. "Machine learning-based re-classification of the geochemical stratigraphy of the Rajahmundry Traps, India." Journal of Volcanology and Geothermal Research 428 (2022).
BibTeX: Download