Presentation Information
[2K0101-11-02]Machine Learning and Geostatistical Approaches to 3D Spatial Modeling of REY-bearing Deep Seafloor Mud Layers
○Vitor Ribeiro de Sa1, Katsuaki Koike (1. Kyoto University)
Chairperson: 木﨑 彰久（秋田大学）
Keywords:
Rare earth elements,3D modelling,Machine learning,Geostatistics,Deep-sea deposits
Rare earth elements and yttrium (REY) are essential for green technologies, particularly in magnets for electric vehicles, wind turbines, and high-performance electronics. With growing demand, deep-sea mining offers an alternative to conventional land-based sources, where REY grades vary and environmental impacts are significant. Among the most promising marine resources are REY-rich mud deposits. This study focuses on deep-sea REY-rich mud layers in a Japanese exclusive economic zone, where Tanaka et al. (2020) revealed exceptionally high total REY concentrations in the shallow seafloor layers. We used the dataset by Tanaka et al. (2020) composed of 501 piston core samples with 45 geochemical elements at 17 sites for the 814 km² target area. Biogenic calcium phosphate and phillipsite-bearing muds were rich in the samples. Given the challenge of modeling spatial continuity in the sparsely sampled, large-scale region, we applied principal component analysis to detect geochemical patterns and hierarchical clustering analysis to define geological domains. These were integrated with geostatistical simulations: turning bands simulation to estimate REY-grade distributions and pluri-Gaussian simulation to delineate layer geometry. The results highlight spatial features of high-grade REY zones and suggest control factors on the zone formation, highlighting the effectiveness of machine learning-assisted methods in deep-sea mineral resource assessment.Reference:Tanaka, E., Nakamura, K., Yasukawa, K., Mimura, K., Fujinaga, K., Iijima, K., Nozaki, T., Kato, Y., 2020. Chemostratigraphy of deep-sea sediments in the western North Pacific Ocean: Implications for genesis of mud highly enriched in rare-earth elements and yttrium. Ore Geol. Rev. 119. https://doi.org/10.1016/j.oregeorev.2020.103392