[M-IS05] Global climate change driven by the Southern Ocean and the Antarctic Ice Sheet

[MIS05-P06] Application of AI-based classification to planktic foraminifera assemblage in the Southern Ocean

*Hiroki Matsui¹, Takuya Itaki², Minoru Ikehara³ (1.Graduate School of International Resource Sciences, Akita University, 2.Geological Survey of Japan, AIST, 3.Center for Advanced Marine Core Research, Kochi University)

Keywords:Automatic classification, Planktic foraminifera, Southern Ocean

Microfossils provide valuable information on past climate changes, although counting and picking of microfossils necessitate a large amount of effort and time. Recent development of a system for automated image collection and AI-based classification of microfossils thus can significantly contribute biostratigraphy and paleoceanography. Because Southern Ocean (SO) planktic foraminifera assemblage is characterized by generally low diversity (i.e., small number of species), it is suitable to develop AI-based classification. We utilize miCRAD (microfossil Classification and Rapid Accumulation Device) system to differentiate planktic foraminifera Globigerina bulloides, Neogloboquadrina pachyderma, other foraminifera species, and all other particles of a marine core in the SO. We report preliminary results of classification accuracy to the above four categories (81.3% on average). We discuss further improvements of classification model and potential application to SO paleoceanography.