[P-EM16] Heliosphere and Interplanetary Space

[PEM16-05] Data mining of cosmic-ray variation observed with the Global Muon Detector Network

*Masayoshi Kozai¹, Yuki Hayashi², Chihiro Kato², Kazuoki Munakata², Akira Kadokura^1,3, Ryuho Kataoka³, Shoko Miyake⁴ (1.Polar Environment Data Science Center, Joint Support-Center for Data Science Research, Research Organization of Information and Systems, 2.Shinshu University, 3.National Institute of Polar Research, Research Organization of Information and Systems, 4.National Institute of Technology, Ibaraki College)

Keywords:Galactic cosmic rays, Solar modulation, Space weather

Data-mining approaches are attempted to extract space weather insights from cosmic-ray observation. Ground-based muon observation is sensitive to anisotropies of cosmic rays at approximately 50 GeV and has been operated with great stability for more than 10 years. Space environment influences cosmic-ray anisotropy in various time scales. It will provide scientific topics for which data-mining approaches are leveraged. This data-driven study is expected to lead to a discovery of a new anisotropy profile and provide complementary results with traditional approaches based on visual inspections and physical interpretations. The Global Muon Detector Network (GMDN) started its operation with two-hemisphere detectors at Nagoya (Japan) and Hobart (Australia) in 1992. SaoMartinho da Serra (Brazil) and Kuwait detectors were installed in 2006. GMDN was completed in 2016 by expanding Kuwait detector to a comparable detection area (25 m^2) with Nagoya detector (36 m^2). Each detector records muon counting-rates in multiple directional channels. Temporal resolution is 1-10 minutes in each detector, and 1-hour data is typically used to study the space weather phenomena. Cosmic-ray anisotropies in space are derived by solving the geomagnetic effects and atmospheric propagation by numerical calculations. We are attempting statistical investigation of the anisotropy data accumulated by GMDN, aiming to demonstrate the advantages of the data-mining approaches including unsupervised machine learning or cluster analysis. In this presentation, we will report on the preliminary results.