Machine learning-based identification of marine and terrestrial Volcanic Rocks in the Tibetan Plateau

Abstract

Distinguishing between marine and terrestrial basalts is crucial for understanding geological processes, including plate tectonics, ocean–continent transition history, and paleoenvironmental changes. However, traditional geochemical methods for tectonic setting discrimination are often limited by issues such as data overlap and the difficulty in obtaining representative samples, making it challenging to accurately differentiate tectonic environments. Machine learning provides an effective approach to address these challenges in the context of large datasets and complex geological problems. In this study, advanced machine learning techniques are applied to global basalt geochemical data to develop a model specifically tailored for regional applications. This model, designed to differentiate between oceanic and continental tectonic environments, is then applied to the Tibetan Plateau basalt data, offering insights into the tectonic background of this specific region. The results show that both oceanic and continental basalt records existed in the region up to 90 million years ago, but the oceanic basalt record subsequently disappeared. This finding suggests that, for a prolonged period before the closure of the Neo-Tethys Ocean, oceanic volcanism was no longer occurring in the Tibetan Plateau. This discovery provides new evidence for the timing of the marine–terrestrial environmental transition in the Tibetan Plateau and is significant for understanding the tectonic evolution and geodynamic processes of the Tethys tectonic domain.