JAXA Supercomputer System Annual Report February 2025-January 2026

Report Number: R25ER2402

Subject Category: Space Technology

PDF (to be added)

• Responsible Representative: Osamu Ochiai, Director, Space Technology Directorate I, Earth Observation Research Center
• Contact Information: Misako Kachi(kachi.misako@jaxa.jp)
• Members: Misako Kachi, Shun Ohishi

Abstract

With the recent enhancement of ocean observation networks by satellites and Argo profiling floats, spatiotemporally higher-resolution temperature, salinity, and sea surface height have been observed. However, for example, the satellites cannot observe under rainy conditions, and the number of Argo float observations is still not sufficient to capture spatiotemporally short-scale variations over the global ocean. Data assimilation reproduces an accurate three-dimensional ocean analysis field without missing values by combining simulation and observations. In this study, using the JAXA Supercomputer System Generation 3 (JSS3), we aim to develop an ensemble Kalman filter (EnKF)-based ocean data assimilation system that assimilates satellite and in-situ observations at a daily interval and to create ocean analysis datasets.

Reference URL

Please refer to https://www.eorc.jaxa.jp/ptree/LORA/index.html .

Reasons and benefits of using JAXA Supercomputer System

The computation costs of high-resolution ensemble data assimilation using an ocean model and EnKF are expensive. Therefore, integration of an EnKF-based ocean data assimilation system becomes feasible with a high-performance computing infrastructure such as the JSS3.

Achievements of the Year

The LORA version 1.0 dataset covers the western North Pacific and Maritime Continent regions from August 2015 to January 2024. Because the ocean exhibits variability on interannual to decadal timescales, the limited temporal and spatial coverage of LORA constrains its applicability.

To address this limitation, this study extends previous ensemble Kalman filter (EnKF)-based ocean data assimilation systems to quasi-global and North Pacific domains with horizontal resolutions of 0.25 and 0.10 degree, respectively. Using these systems, we produced new analysis datasets since June 2002, when the ocean observing network was substantially enhanced by microwave satellite observations and by Argo profiling float measurements.

We validated the resulting eddy-permitting quasi-global analysis product, referred to as LORA-QG, by comparing it with eddy-permitting ocean reanalysis datasets produced by several international institutions. The results demonstrate that LORA-QG has sufficient accuracy for geoscientific research and practical applications. We are currently compiling these results into a manuscript for submission to an international journal.

Publications

- Peer-reviewed papers

- Almeida, A. P., H. M. J. Barbosa, S. R. Garcia, D. J. Gagne, K. Zhou, T. Kubota, T. Ushio, S. Otsuka, S. Pfreundschuh, and A. J. P. Calheiros, 2026: A regional benchmark for deep learning-based hourly precipitation nowcasting in Latin America. IEEE Access, doi:10.1109/ACCESS.2026.3670767 (accepted).

- Amemiya, A., and T. Miyoshi, 2026: Impact of reduced non-Gaussianity on analysis and forecast accuracy by assimilating every-30 s radar observation with ensemble Kalman filter: Idealized experiments of deep convection. Nonlin. Processes Geophys., 33, 1-16, https://doi.org/10.5194/npg-33-1-2026

- Chang, C., B. Mu, W. Han, Y. Ham, G. J. Zhang, A. Damiani, F. Ling, and T. Miyoshi, 2025: AI weather and climate prediction and applications. Bull. Amer. Meteor. Soc., 106, E2571-E2578, https://doi.org/10.1175/BAMS-D-25-0260.1

- Guerrieri, J. M., M. Pulido, T. Miyoshi, A. Amemiya, and J. J. Ruiz, 2026: Localization in the mapping particle filter. Nonlin. Processes Geophys., 33, 33-49, https://doi.org/10.5194/npg-33-33-2026

- Hascoet, T., V. Pellet, S. Oishi, and T. Miyoshi, 2026: Differentiable river routing for end-to-end learning of hydrological processes. J. Geophys. Res.: Machine Learning and Computation, 3, e2025JH000760, https://doi.org/10.1029/2025JH000760

- Huo, Z., Y. Liu, J. Taylor, Y. Zhou, A. Amemiya, H. Fan, and T. Miyoshi, 2025: Incremental analysis updates in a convective-scale ensemble Kalman filter using minute-by-minute phased array radar observations. J. Adv. Model. Earth Syst., 17, e2024MS004802, https://doi.org/10.1029/2024MS004802

- Konduru, R. T., R. Bale, M. Tsubokura, and T. Miyoshi, 2025: Transforming urban wind engineering by taming extreme weather strong winds over urban skylines with ultra-high-resolution simulations on supercomputer Fugaku. Proc. Supercomputing Asia Conf. (SCA '25), 79-87, https://doi.org/10.1145/3718350.3718353

- Konduru, R. T., J. Liang, S. Otsuka, and T. Miyoshi, 2026: Observing systems simulation experiments of hypothetical hourly global coverage of microwave satellite radiances: Imbalance and adaptive observation error inflation. J. Geophys. Res. Atmos., in press.

- Kotsuki, S., K. Shiraishi, and A. Okazaki, 2025: Ensemble data assimilation to diagnose AI-based weather prediction models: A case with ClimaX version 0.3.1. Geosci. Model Dev., 18, 7215-7225, https://doi.org/10.5194/gmd-18-7215-2025

- Matsugishi, S., Y.-W. Chen, K. Terasaki, H. Yashiro, S. Kotsuki, K. Kanemaru, K. Yamamoto, M. Satoh, T. Kubota, and T. Miyoshi, 2025: Intercomparison of NICAM-LETKF JAXA research analysis (NEXRA) version 2 and 3. SOLA, https://doi.org/10.2151/sola.2025-035

- Matsugishi, S., Y.-W. Chen, K. Terasaki, K. Kanemaru, S. Kotsuki, H. Yashiro, K. Yamamoto, M. Satoh, T. Kubota, and T. Miyoshi, 2025: NICAM-LETKF JAXA research analysis (NEXRA) version 2.0. Geosci. Data J., 12, e70011, https://doi.org/10.1002/gdj3.70011

- Miyoshi, T., 2026: A duality principle for chaotic systems: From data assimilation to efficient control. Nonlinear Dyn., 114, 105, https://doi.org/10.1007/s11071-025-12021-2

- Mulia, I., U. Shimada, N. Ueda, T. Miyoshi, and M. Maulana, 2025: Multi-horizon prediction of tropical cyclone intensity and its interpretability with temporal fusion transformer. Sci. Rep., 15, https://doi.org/10.1038/s41598-025-15522-7

- Ohishi, S., Y. Kobayashi, and T. Miyoshi, 2025: Including cross correlation between forecast and observation errors in an ensemble Kalman filter. Mon. Wea. Rev., 153, 1035-1043, https://doi.org/10.1175/MWR-D-25-0016.1

- Ohishi, S., T. Miyoshi, and M. Kachi, 2025: Deterministic and ensemble forecasts of the Kuroshio south of Japan. Ocean Dyn., 75, 92, https://doi.org/10.1007/s10236-025-01736-w

- Satoh, M., T. Kawabata, T. Miyakawa, M. Nakano, H. Yashiro, T. Miyoshi, L. Duc, P.-Y. Wu, T. Oizumi, Y. Maejima, J. Taylor, R. Yoshimura, K. Terasaki, Y. Yamada, R. Masunaga, T. Kawasaki, and M. Tanoue, 2025: Achievements in atmospheric sciences by the large-ensemble and high-resolution forecasting studies using the supercomputer Fugaku. Prog. Earth Planet. Sci., 12, 64, https://doi.org/10.1186/s40645-025-00730-6

- Invited Presentations

- 2025/4/9 Miyoshi, Takemasa; Otsuka, Shigenori; Liang, Jianyu; Goodliff, Michael; Saliou, Gwendal; Ouala, Said; Tandeo, Pierre, RIKEN's activities to integrate DA and AI/ML, Workshop on data assimilation: Initial conditions and beyond, Part of ECMWF's 50th anniversary celebrations, Germany, Bonn

- 2025/5/6 Miyoshi, Takemasa, RIKEN's activities to integrate data assimilation and AI/ML, Meteorological-Geophysical Colloquium, Austria, Vienna

- 2025/5/14 Miyoshi, Takemasa, RIKEN's activities to integrate data assimilation and AI/ML, Seminar (DWD), Germany, Offenbach

- 2025/5/16 Miyoshi, Takemasa, RIKEN's activities to integrate data assimilation and AI/ML, SFB-Seminar, Germany, Potsdam

- 2025/5/25 Otsuka Shigenori, Global precipitation nowcasting with ConvLSTM and adversarial training, Japan Geoscience Union Meeting 2025, Japan, Chiba

- 2025/6/13 Miyoshi, Takemasa, Big Data Assimilation Revolutionizing Numerical Weather Prediction Using Fugaku, SEMINAIRE Sciences du climat, France, Paris

- 2025/6/18 Miyoshi, Takemasa, Big Data Assimilation - Revolutionizing Weather Prediction, Seminar (Department of Physics and Astronomy "Augusto Righi"), Italy, Bologna

- 2025/7/28 Takemasa Miyoshi, RIKEN's activities to integrate data assimilation and AI/ML, AOGS 2025, Singapore

- 2025/8/7 Miyoshi, Takemasa, Forging the Future of Weather Prediction: From Supercomputers to Societal Impact, CONAGUA2025, Argentina, Buenos Aires

- 2025/11/16 Miyoshi, Takemasa, From Gordon Bell Finalist to the Osaka Expo: The Evolution of Real-Time Forecasting on Fugaku, SC25, United States, St. Louis

- 2026/1/26 Otsuka Shigenori, Rapid-update precipitation predictions with HPC and AI, SupercomputingAsia 2026/The International Conference on High Performance Computing in Asia-Pacific Region 2026, Japan, Osaka

- 2026/3/6 Takemasa Miyoshi, "Osaka Expo 2025 Weather on Fugaku: Synergizing Big Data Assimilation and AI", JAMSTEC International Weather and Climate Modeling(IWCM) Workshop 2026, Yokohama

- Oral Presentations

- 2025/4/27 Miyoshi, Takemasa; Otsuka, Shigenori; Liang, Jianyu; Goodliff, Michael; Saliou, Gwendal; Ouala, Said; Tandeo, Pierre, RIKEN's activities to integrate DA and AI/ML, EGU2025, Austria, Vienna

- 2025/5/13 Miyoshi, Takemasa, Can we control chaotic extreme weather? -- what we learned from Japan's moonshot program, Meteorological Colloquium, Germany, Munich

- 2025/5/25 Miyoshi, Takemasa; Ohishi, Shun; Liang, Jianyu; Goodliff, Michael; Otsuka, Shigenori; Tomita, Hirofumi; Kanemaru, Kaya; Yashiro, Hisashi; Terasaki, Koji; Okamoto, Kozo; Kotsuki, Shunji; Matsugishi, Shuhei; Satoh, Masaki; Bishop, Craig; Kubota, Takuji; Kachi, Misako, Advancement and AI integration of satellite data assimilation of clouds, precipitation, and the ocean, Japan Geoscience Union Meeting 2025, Japan, Chiba

- 2025/6/23 Goodliff Michael, Using Data Assimilation to Improve Data Driven Surrogate Models, IMT-Atlantique & RIKEN joint Data Assimilation workshop, France, Brest

- 2025/6/26 Tarumi Yuta, Deep Bayesian Filter for SPEEDY, an Intermediate-Complexity Atmospheric GCM, IMT-Atlantique & RIKEN joint Data Assimilation workshop, France, Brest

- 2025/6/26 Otsuka Shigenori, Data-driven approaches for precipitation nowcasting, IMT-Atlantique & RIKEN joint Data Assimilation workshop, France, Brest

- 2025/7/28 Goodliff Michael, Using Data Assimilation To Improve Data-driven Models, AOGS 2025, Singapore

- 2025/8/18 Miyoshi, Takemasa, Big Data Assimilation Revolutionizing Numerical Weather Prediction Using Fugaku, IA CORRIENTES, TECH WEEK, Argentina, Corrientes

- 2025/8/22 Miyoshi, Takemasa, How a Breakthrough in Chaos Theory Inspired a National Mission to Control Extreme Weather, Ciclo de Charlas de Actualizacion y Divulgacion Cientifica, Tecnologica y de Innovacion, Argentina, Corrientes

- 2025/9/29 Miyoshi, Takemasa, Chaos implies effective controllability but only after the predictability limit, ISDA2025, Australia, Melbourne

- 2025/9/29 Goodliff Michael, Impacts of Model Errors on Convergence of Iterative Data Driven Model Refinement with Data Assimilation, 11th International Symposium for Data Assimilation, Australia, Melbourne

- 2025/9/29 Tarumi Yuta, Deep Bayesian Filtering for High-Dimensional Atmospheric Models - Scalable Data Assimilation with SPEEDY, 11th International Symposium on Data Assimilation, Australia, Melbourne

- 2025/10/5 Miyoshi, Takemasa, Big data assimilation for weather prediction, 65th ISI World Statistics Congress, Netherlands, Hague

- 2025/10/20 Miyoshi, Takemasa, Advancement and AI integration of satellite data assimilation of clouds, precipitation and the ocean, FY2025 The Joint PI Meeting of JAXA Earth Observation Missions, Japan, Tokyo

- 2025/10/27 Goodliff Michael, Using Data Assimilation to Improve Data-Driven Models, AIP CREST Challenge Meeting, Japan, Tokyo

- 2025/11/17 Miyoshi, Takemasa; Otsuka, Shigenori; Liang, Jianyu; Goodliff, Michael; Tarumi, Yuta; Saliou, Gwendal; Ouala, Said; Tandeo, Pierre, RIKEN's activities to integrate data assimilation and AI/ML, WS on NHM 2025, Japan, Morioka

- 2025/11/20 Ohishi Shun, LETKF-based Ocean Research Analysis (LORA) for a quasi-global domain, The Joint PI Meeting of JAXA Earth Observation Missions FY2025, Japan, Tokyo

- 2025/12/15 Miyoshi, Takemasa, Harnessing the Butterfly Effect: A Duality-Based Framework for the Efficient Control of Extreme Weather, AGU2025, United States, New Orleans

- 2025/12/15 Miyoshi, Takemasa; Otsuka, Shigenori; Liang, Jianyu; Goodliff, Michael; Tarumi, Yuta; Saliou, Gwendal; Ouala, Said; Tandeo, Pierre, RIKEN's activities to integrate data assimilation and AI/ML, AGU25, United States, New Orleans

- 2026/1/25 Miyoshi, Takemasa; Otsuka, Shigenori; Liang, Jianyu; Goodliff, Michael; Tarumi, Yuta; Saliou, Gwendal; Ouala, Said; Tandeo, Pierre, Riken's Activities to Integrate Data Assimilation and AI/ML, AMS2026, 106th annual meeting, United States, Houston

- 2026/2/22 Ohishi Shun, Deterministic and ensemble forecasts of Kuroshio south of Japan, Ocean Sciences Meeting 2026, United Kingdom, Scotland

- Poster Presentations

- 2025/5/25 Miyoshi, Takemasa; Ohishi, Shun; Tomita, Hirofumi; Taylor, James; Goodliff, Michael; Liang, Jianyu; Konduru, Rakesh Teja; Satoh, Masaki; Matsugishi, Shuhei; Kotsuki, Shunji; Okazaki, Atsushi; Honda, Takumi; Okamoto, Kozo; Ikuta, Yasutaka; Terasaki, Koji; Yashiro, Hisashi; Kanemaru, Kaya; Yamazaki, Akira, A platform to design and pre-evaluate frequent satellite observing systems for innovating weather, ocean and land surface prediction, Japan Geoscience Union Meeting 2025, Japan, Chiba

- 2025/5/25 Ohishi Shun, Intercomparison and Ensemble of Coastal Ocean Prediction Models in Japan: A Case Study in the Goto-nada, Japan Geoscience Union Meeting 2025, Japan, Chiba

- 2025/5/25 Goodliff Michael, Using Data Assimilation to Improve Data-Driven Models, JpGU 2025, Japan, Chiba

- 2025/7/6 Tarumi Yuta, Deep Bayesian Filter for Bayes-Faithful Data Assimilation, International Conference on Machine Learning, Canada, Vancouver

- 2025/7/20 Ohishi Shun, Intercomparison and Ensemble of Coastal Ocean Prediction Models in Japan: A Case Study in the Goto-nada, Busan IAMAS-IACS-IAPSO Joint Assembly 2025, Korea, Busan

- 2025/11/25 Goodliff Michael, Using Data Assimilation to Improve Data-Driven Models, International Workshop on Hydrometeorological Predictions, Japan, Kobe

- 2025/11/25 Ohishi Shun, Deterministic and Ensemble forecasts of Kuroshio south of Japan, International Workshop on Hydrometeorological Predictions, Japan, Kobe

- 2025/12/22 Otsuka Shigenori, Global precipitation nowcasting using a ConvLSTM with adversarial training, RIKEN-Nagoya Univ. (Zhang - Sogabe Lab) Joint Workshop on Prediction Science, Japan, Nagoya

- 2026/1/26 Unashish Mondal, Evaluation of EarthCARE Cloud Profiling Radar Observations with a 3.5-km-Resolution NICAM Simulation, SCA/HPCAsia 2026, Japan, Osaka

- 2026/1/26 Tarumi Yuta, Deep Bayesian Filter for SPEEDY: an Intermediate-Complexity Atmospheric Circulation Model, SCA/HPCAsia 2026, Japan, Osaka

- 2026/1/26 Goodliff Michael, Using Data Assimilation to Improve Data-Driven Models, SCA/HPC Asia 2026, Japan, Osaka

- 2026/1/26 Ohishi Shun, Development of an ensemble ocean data assimilation system, SCA/HPCAsia 2026, Japan, Osaka

- 2026/1/26 Otsuka Shigenori, Global precipitation nowcasting with ConvLSTM and adversarial training, SupercomputingAsia 2026/The International Conference on High Performance Computing in Asia-Pacific Region 2026, Japan, Osaka

- 2026/1/26 Unashish Mondal, Evaluation of EarthCARE Cloud Profiling Radar Observations with a 3.5-km-Resolution NICAM Simulation, SCA/HPCASIA 2026, Japan, Osaka

- 2026/2/5 Goodliff Michael, Using Data Assimilation to Improve Data-Driven Models, RIKEN-Fudan University joint workshop on atmospheric science, DA and ML, Japan, Kobe

- 2026/3/2 Goodliff Michael, Comparing Ensemble Data Assimilation and Variational Data Assimilation Methods for Data-Driven Surrogate Model Generation, AIP Challenge Meeting, Japan, Tokyo

Usage of JSS

Computational Information

• Process Parallelization Methods: MPI
• Thread Parallelization Methods: OpenMP
• Number of Processes: 4 - 2048
• Elapsed Time per Case: 30 Minute(s)

JSS3 Resources Used

Fraction of Usage in Total Resources*1(%): 2.75

Details

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

*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

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

JAXA Supercomputer System Annual Report February 2025-January 2026