Advancing Petrology with Data-driven Approaches

The integration of machine learning (ML) with petrology is transforming our ability to analyze the complex chemical and physical records preserved magmatic, metamorphic and ore-forming systems, and their derived detritus. From data-driven thermometers and barometers that constrain the pressure and temperature conditions of magma storage, to unsupervised clustering of crystal cargo that unravels pre-eruptive dynamics, ML methods are opening avenues that traditional approaches cannot match. Machine learning petrology has grown into a broad methodological toolkit that is now underpinning everything from tectonic discrimination and resource exploration to automated mineral segmentation and the interrogation of geochemical and experimental databases at a scale previously out of reach. The frontier now lies in grounding these data-driven approaches within the thermodynamic and kinetic constraints that define rock forming systems.We invite original research articles, methodological contributions, and focused case studies that advance the application of ML - including traditional algorithms, deep learning, and emerging physics-informed or generative approaches - to problems in petrology.This Research Topic invites submissions that highlight both current applications and future advancements of ML in petrology. We welcome articles addressing, but not limited to, the following themes:• Development of physics-informed ML models tailored to petrology• Strategies for data fusion and synthesis in geosciences• Collaborations bridging machine learning with thermodynamics in petrological research• Case studies exemplifying successful automation and accuracy improvement through ML in petrology• Data-driven approaches to understand the evolution of mantle redox state and deep volatile cycleWe encourage submissions encompassing original research, reviews, case studies, and method development articles to enrich this compelling field of study. Topic editors: Gregor Weber, Vincenzo Stagno, Gary O'Sullivan Source: https://www.frontiersin.org/research-topics/80233/advancing-petrology-with-data-driven-approaches