Prediction and Modelling of Turbulence based on Machine Learning

JAXA Supercomputer System Annual Report February 2025-January 2026

Report Number: R25EACA39

Subject Category: JSS Inter-University Research

Responsible Representative: Masanobu Inubushi, Associate Professor, Tokyo University of Science
Contact Information: Masanobu Inubushi(inubushi@rs.tus.ac.jp)
Members: Susumu Goto, Masanobu Inubushi, Satoshi Matsumoto, Akane Okubo

Abstract

Turbulence models play essential roles in aerospace science and technology, such as flows around aircraft and of planetary atmospheres. They are rapidly empowered by machine learning methods and will be a crucial building block of aerospace science and technology in the near future. The present study aims to integrate physics and data-driven methods for turbulence modeling.

Reference URL

Please refer to https://x.com/MasaInubushi .

Reasons and benefits of using JAXA Supercomputer System

The reason to use JAXA Supercomputer System is that we can develop these methods based on training data of turbulent flows with high-resolution, numerical calculations requiring a massively parallel supercomputer.

Achievements of the Year

We studied fundamental properties of two-dimensional fluid motion, in particular the synchronisation and data-assimilation properties, and provided an applied-mathematical characterisation using numerical computations of conditional Lyapunov exponents. The results were published in the Journal of Fluid Mechanics (Inubushi & Caulfield, JFM, 2026). The paper was selected for the cover of JFM and was also introduced on the JSS3 website: https://www.jss.jaxa.jp/computer_engineering/29252/

In addition, through a collaborative study with Tokyo Denki University, we investigated the prediction of large-scale circulation reversals in turbulent thermal convection using neural networks, and the results were published in Physical Review Fluids (Katsumi–Inubushi–Yokoyama, Phys. Rev. Fluids, 2025).

Publications

- Peer-reviewed papers

Katsumi, Daigaku, Masanobu Inubushi, and Naoto Yokoyama. "Data-driven prediction of reversal of large-scale circulation in turbulent convection." Physical Review Fluids 10.5 (2025): 053501.

Masanobu Inubushi and Colm-cille P. Caulfield. "Synchronisation in two-dimensional damped-driven Navier–Stokes turbulence: insights from data assimilation and Lyapunov analysis" Journal of Fluid Mechanics 1027, A41 (2026).

- Invited Presentations

Masanobu Inubushi, Synchronization in turbulence and its significance for machine learning applications, 2025-05-15, Okinawa Convention Center, IUTAM Symposium on Machine Learning in Diverse Fluid Mechanics

Masanobu Inubushi, Synchronization in Turbulence and Its Significance for Data-Driven Approaches, 2025-05-29, 幕張メッセ, Japan Geoscience Union Meeting 2025

Usage of JSS

Computational Information

Process Parallelization Methods: MPI
Thread Parallelization Methods: OpenMP
Number of Processes: 16 - 64
Elapsed Time per Case: 30 Hour(s)

JSS3 Resources Used

Fraction of Usage in Total Resources*1(%): 0.05

Details

Please refer to System Configuration of JSS3 for the system configuration and major specifications of JSS3.

Computational Resources
System Name	CPU Resources Used (Core x Hours)	Fraction of Usage*2(%)
TOKI-SORA	1324009.80	0.06
TOKI-ST	16.39	0.00
TOKI-GP	0.00	0.00
TOKI-XM	0.00	0.00
TOKI-LM	0.00	0.00
TOKI-TST	0.00	0.00
TOKI-TGP	0.00	0.00
TOKI-TLM	0.00	0.00

File System Resources
File System Name	Storage Assigned (GiB)	Fraction of Usage*2(%)
/home	0.00	0.00
/data and /data2	0.00	0.00
/ssd	0.00	0.00

Archiver Resources
Archiver Name	Storage Used (TiB)	Fraction of Usage*2(%)
J-SPACE	0.00	0.00

*1: Fraction of Usage in Total Resources: Weighted average of three resource types (Computing, File System, and Archiver).

*2: Fraction of Usage：Percentage of usage relative to each resource used in one year.

ISV Software Licenses Used

ISV Software Licenses Resources
	ISV Software Licenses Used (Hours)	Fraction of Usage*2(%)
ISV Software Licenses (Total)	0.00	0.00

*2: Fraction of Usage：Percentage of usage relative to each resource used in one year.