Kazue Suzuki, Ph.D. / 鈴木香寿恵
Specially Appointed Associate Professor, Meiji University Graduate School
From detecting snowfall clouds in satellite imagery to anomaly detection with generative AI and dating ice cores through data assimilation — I connect machine learning and statistical modeling to observational data across polar and earth science, turning records too vast to read by hand into a clearer view of environmental change.
The starting point isn't "what can AI do" but "what remains unread in polar and earth science" — five research themes built from that question.
Using CNNs and semi-supervised image segmentation to detect snowfall clouds and atmospheric rivers in satellite cloud imagery, linking them to snowfall estimates over the Antarctic ice sheet.
Analyzing Polarized Micro-Pulse Lidar (PMPL) observations with generative models such as f-AnoGAN to automatically detect high-concentration aerosol events at Syowa Station, Antarctica.
Dating ice cores with Kalman filters and hierarchical Bayesian models, tracing atmospheric trajectories through sequential data assimilation, and building probabilistic typhoon-track models.
Bridging satellite and ground-based observations with machine learning to model heat- and trace-substance transport mechanisms in the polar regions (PI, Grant-in-Aid for Scientific Research (B)).
Designing and operating the observation systems that AI analysis depends on — including an automated snow-crystal imaging device — combined with snow sampling in Hokkaido and Antarctica. Also involved in a multi-sensor project reconstructing traditional weather-lore ("kanten-bouki") for disaster-precursor prediction.
A researcher who loves Antarctica and snow.
From a monsoon research lab to the National Institute of Polar Research. After my Ph.D., I met
statistics and started working on data assimilation. Today my work centers on
time-series modeling and machine learning.
After earning a Ph.D. in polar science from SOKENDAI, I worked at the National Institute of Polar Research and the Institute of Statistical Mathematics before moving into a research practice built on both statistics and machine learning — reading environmental change "from an Antarctic point of view." At Meiji University Graduate School I also teach data science, sharing practical statistical and machine-learning methods with undergraduate and graduate students.
At Meiji University Graduate School I teach "Fundamentals of Multivariate Analysis," "Data Analysis in the Humanities & Social Sciences," and "Data Science Seminar." "Fundamentals of Multivariate Analysis" is a required elective in the Basic Program of Meiji's Mathematics, Data Science, and AI Applied Fundamentals Program.
How to handle field data from meteorology, snow and ice science, and earth science with statistical models and machine learning — this lab's work should be useful both to data science majors and to natural-science students looking to extend their careers into AI. I'm also involved in organizing Meiji's data science competition and the Japanese Society of Snow and Ice's webinars for students.
For questions about analyzing polar observation data, applying AI/statistical methods, or discussing a research topic as a student, please feel free to reach out.
Kazue Suzuki
KAZUE SUZUKI LAB — POLAR × DATA SCIENCE