10:00 AM - 10:15 AM
[MIS12-15] Deep-sea mineral resource for rare earth elements formed in the Hothouse ocean: Insights from deep learning-based microfossil observation
Keywords:seafloor sediments, mineral resources, REE-rich mud, microfossils, deep learning, ichthyoliths
Deep-sea mud enriched in rare-earth elements (REE), termed as “REE-rich mud,” is expected to be a promising seafloor mineral resource [1]. A decade of survey revealed that the mud with world’s highest REE concentration lies in the pelagic area of western North Pacific Ocean [2, 3]. Since the major host of REE is known to be fish debris [3, 4], deciphering fish productivities in response to global environmental changes is important for understanding the genesis and distribution of the resource. Previous studies have revealed that there have been multiple timings for the formation of highly REE enriched layers [5, 6], and the 1st (youngest) REE enrichment was triggered by global cooling during the Eocene–Oligocene climate transition [4]. However, the depositional ages of the older REE peaks remain uncertain.
Microfossils of fish teeth and denticles, called “ichthyoliths,” are only the microfossils that are commonly observed from REE-rich mud [4, 7]. However, previous manual observation required enormous time and efforts of skilled observers, which has hampered an efficient data accumulation. To overcome this problem, we have developed an effective method for ichthyolith observation using object detection, a kind of deep learning techniques [7, 8].
Empowered by the deep learning-based system, we observed more than 40,000 ichthyoliths from two piston cores obtained around Minamitorishima island and revealed that the second youngest REE enrichment occurred in early Eocene. In contrast to the 1st REE peak, which has deposited during the global cooling, the 2nd REE peak was formed in the hottest period in the Cenozoic era [9]. In the presentation, we discuss the mechanism how the Hothouse ocean could produce the valuable seafloor mineral resource in the pelagic realm.
References : [1] Kato et al. (2011). Nat. Geosci. 4, 535–539. [2] Iijima et al. (2013) Geochem. J. 50, 557–573. [3] Takaya et al. (2018) Sci. Repts. 8, 5763. [4] Ohta et al. (2020) Sci. Repts. 10, 9896. [5] Tanaka et al. (2020) Ore Geol. Rev. 119, 103392. [6] Yasukawa et al. (2023) Paleoceanogr. Paleoclimatol., 38, e2023PA004644. [7] Mimura et al. (2022) Appl. Comput. Geosci. 16, 100092. [8] Mimura et al. (2024) Earth Space Sci. 11, e2023EA003122. [9] Westerhold et al. (2020) Science 369, 1383-1387.
Microfossils of fish teeth and denticles, called “ichthyoliths,” are only the microfossils that are commonly observed from REE-rich mud [4, 7]. However, previous manual observation required enormous time and efforts of skilled observers, which has hampered an efficient data accumulation. To overcome this problem, we have developed an effective method for ichthyolith observation using object detection, a kind of deep learning techniques [7, 8].
Empowered by the deep learning-based system, we observed more than 40,000 ichthyoliths from two piston cores obtained around Minamitorishima island and revealed that the second youngest REE enrichment occurred in early Eocene. In contrast to the 1st REE peak, which has deposited during the global cooling, the 2nd REE peak was formed in the hottest period in the Cenozoic era [9]. In the presentation, we discuss the mechanism how the Hothouse ocean could produce the valuable seafloor mineral resource in the pelagic realm.
References : [1] Kato et al. (2011). Nat. Geosci. 4, 535–539. [2] Iijima et al. (2013) Geochem. J. 50, 557–573. [3] Takaya et al. (2018) Sci. Repts. 8, 5763. [4] Ohta et al. (2020) Sci. Repts. 10, 9896. [5] Tanaka et al. (2020) Ore Geol. Rev. 119, 103392. [6] Yasukawa et al. (2023) Paleoceanogr. Paleoclimatol., 38, e2023PA004644. [7] Mimura et al. (2022) Appl. Comput. Geosci. 16, 100092. [8] Mimura et al. (2024) Earth Space Sci. 11, e2023EA003122. [9] Westerhold et al. (2020) Science 369, 1383-1387.