Digital Typhoon Dataset

The "Digital Typhoon Dataset" is a dataset of meteorological satellite images and the best track of typhoons in the Northwest Pacific basin. The meteorological satellite imagery is created from infrared imagery (11 micrometers) from geostationary meteorological satellites Himawari 1 to 9 by a map projection (Lambert azimuthal equal-area) of a 1250 km radius centered on the typhoon. The best track is created from the best track data published by the Japan Meteorological Agency by interpolating it into hourly data. As of October 2023, it consists of 189364 images for 1099 typhoons from 1978 to 2022 and will be updated annually. This is a particularly useful dataset for machine learning, and the official dataset page explains its use for machine learning.

数字台风数据集（Digital Typhoon Dataset）是西北太平洋海域台风气象卫星图像与最佳路径的数据集。气象卫星图像由葵花1号至9号（Himawari 1-9）静止气象卫星的11微米红外图像生成，采用以台风为中心、半径1250公里的兰伯特方位等积投影（Lambert azimuthal equal-area）。最佳路径由日本气象厅（Japan Meteorological Agency）发布的最佳路径数据插值为小时数据生成。截至2023年10月，该数据集包含1978年至2022年期间1099个台风的189364张图像，并将每年更新。该数据集对机器学习尤为有用，其官方数据集页面说明了其在机器学习中的应用。