Digital Typhoon Dataset


1. IDENTIFICATION INFORMATION

Name Digital Typhoon Dataset
Edition V3-2025.R2
DOI doi:10.20783/DIAS.664
Metadata Identifier DigitalTyphoon20261031193551-DIAS20221121113753-en

2. CONTACT

2.1 CONTACT on DATASET

Name Asanobu KITAMOTO
Organization National Institute of Informatics
Address 2-1-2, Hitotsubashi, Chiyoda-ku, Tokyo, 101-8430, Japan
E-mail kitamoto@nii.ac.jp

2.2 CONTACT on PROJECT

2.2.1 Data Integration and Analysis System

Name DIAS Office
Organization Japan Agency for Marine-Earth Science and Technology
Address 3173-25, Showa-Cho, Kanazawa-ku, Yokohama-shi, Kanagawa, 236-0001, Japan
E-mail dias-office@diasjp.net

3. DOCUMENT AUTHOR

Name Asanobu KITAMOTO
Organization National Institute of Informatics
E-mail kitamoto@nii.ac.jp

4. DATASET CREATOR

Name Asanobu KITAMOTO
Organization National Institute of Informatics
E-mail kitamoto@nii.ac.jp

5. DATE OF THIS DOCUMENT

2026-10-31

6. DATE OF DATASET

  • creation : 2023-11-05

  • revision : 2026-10-16

7. DATASET OVERVIEW

7.1 Abstract

The "Digital Typhoon Dataset" is a meteorological satellite image and best-track dataset for typhoons in the Northern Hemisphere's Northwest Pacific (WP) and cyclones around the Southern Hemisphere's Australia (AU).

The meteorological satellite images are map-projected (using an azimuthal equidistant projection) within a 1250 km radius centered on the typhoon, utilizing multispectral images (Channels 1–4) from geostationary meteorological satellites Himawari-1 through 9. The best-track data is created by interpolating the best-track data published by the Japan Meteorological Agency (JMA) into hourly data.

As of October 2026, the latest version is V3-2025.R2, comprising 818,881 images for 1,680 typhoons (cyclones) from 1978 to 2025 seasons. This dataset is useful for machine learning, and the official dataset page also explains its applications in machine learning.

Change Log:

Version 1 to Version 2:

* Added data for the Southern Hemisphere (around Australia).

* Changed the map projection from an azimuthal equal-area projection to an azimuthal equidistant projection.

* Extended the time coverage up to the 2023/2024 season.

Version 2 to Version 3 (V3-2025.R2)

* Added data for Channels 2–4.

* Extended the temporal coverage up to the 2025 season.

* Fixed remaining data bugs present in V1/V2.

7.2 Topic Category(ISO19139)

  • climatologyMeteorologyAtmosphere

7.3 Temporal Extent

Begin Date 1978
End Date 2025
Temporal Characteristics Hourly

7.4 Geographic Bounding Box

North bound latitude 70
West bound longitude 100
Eastbound longitude 180
South bound latitude -70

7.5 Grid

7.6 Geographic Description

7.7 Keywords

7.7.1 Keywords on Dataset

Keyword Type Keyword Keyword thesaurus Name
theme ATMOSPHERIC PROCESSES > Tropical meteorology AGU
theme Earth Observation Satellites > GEOSTATIONARY SATELLITES GCMD_platform
theme Weather, Disasters, Climate GEOSS
discipline machine learning No_Dictionary

7.7.2 Keywords on Project

7.7.2.1 Data Integration and Analysis System
Keyword Type Keyword Keyword thesaurus Name
theme DIAS > Data Integration and Analysis System No_Dictionary

7.8 Online Resource

7.9 Data Environmental Information

7.10 Distribution Information

name version specification
HDF5 5
CSV RFC 4180

8. DATA PROCESSING

8.1 Data Processing (1)

8.1.1 General Explanation of the data producer's knowledge about the lineage of a dataset

8.1.2 Data Source

Data Source Citation Name Description of derived parameters and processing techniques used

9. DATA REMARKS

10. DATA POLICY

10.1 Data Policy by the Data Provider

10.2 Data Policy by the Project

10.2.1 Data Integration and Analysis System

If data provider does not have data policy, DIAS Terms of Service (https://diasjp.net/en/terms/) and DIAS Privacy Policy (https://diasjp.net/en/privacy/) apply.

If there is a conflict between DIAS Terms of Service and data provider's policy, the data provider's policy shall prevail.

11. LICENSE

CC-BY 4.0 : Creative Commons Attribution 4.0 International

12. DATA SOURCE ACKNOWLEDGEMENT

12.1 Acknowledge the Data Provider

The attribution of the dataset required for the CC BY license is as follows.

Digital Typhoon dataset (National Institute of Informatics) doi: https://doi.org/10.20783/DIAS.664

12.2 Acknowledge the Project

12.2.1 Data Integration and Analysis System

If you plan to use this dataset for a conference presentation, paper, journal article, or report etc., please include acknowledgments referred to following examples. If the data provider describes examples of acknowledgments, include them as well.

" In this study, [Name of Dataset] provided by [Name of Data Provider] was utilized. This dataset was also collected and provided under the Data Integration and Analysis System (DIAS), which was developed and operated by a project supported by the Ministry of Education, Culture, Sports, Science and Technology. "

13. REFERENCES

Asanobu Kitamoto, Victor Le Louarn, Noelia Falczuk, Ngoc Anh Tong, Anja Veleva, "Digital Typhoon Dataset V3: Benchmarking for Multi-spectral Extension and Analyzing Systematic Bias in Intensity Regression", arXiv:2610.21654, 2026.

Asanobu Kitamoto, Erwan Dzik, Gaspar Faure, "Machine Learning for the Digital Typhoon Dataset: Extensions to Multiple Basins and New Developments in Representations and Tasks", arXiv:2411.16421, 2024.

Asanobu KITAMOTO, Jared HWANG, Bastien VUILLOD, Lucas GAUTIER, Yingtao TIAN, Tarin CLANUWAT, "Digital Typhoon: Long-term Satellite Image Dataset for the Spatio-Temporal Modeling of Tropical Cyclones", NeurIPS 2023 Datasets and Benchmarks (Spotlight), 2023