3D-NEXRAD: 3D Gridded WSR-88D Radar Sequences of US Severe Storms for Precipitation Nowcasting (GridRad-Severe, 2020-2022)

Provider
Ziye Wang, Yiran Qin, Lin Zeng, Ruimao Zhang
License
CC-BY-4.0 (source)
Size
17 files
Stored at
Hugging Face (https://huggingface.co/datasets/Ziyeeee/3D-NEXRAD)
Access
Free and public. No account with the data provider is needed.
Category
Climate & Weather
Source
https://arxiv.org/abs/2502.14895

About this dataset

Three-dimensional gridded weather radar observations of severe storm events across the United States, built for radar sequence prediction and weather nowcasting by Wang, Qin, Zeng and Zhang (ICLR 2025, arXiv:2502.14895). The data were re-collected and preprocessed from GridRad-Severe, the three-dimensional gridded NEXRAD WSR-88D radar product for severe events at the NCAR Research Data Archive (DOI 10.5065/2B46-1A97, CC BY 4.0). Each event is a sequence of 25 frames spanning 125 minutes (5-minute interval) with 512 x 512 horizontal pixels and 28 vertical levels (0.5 to 7 km every 0.5 km, then 7 to 22 km every 1 km), covering 2020 to 2022. Seven radar variables are provided: reflectivity, velocity spectrum width, azimuthal shear and radial divergence of the radial velocity, differential reflectivity (ZDR), specific differential phase (KDP) and copolar correlation coefficient. Files are yearly split tar.gz archives (nexrad-YYYY.tar.gz.00 and so on, about 141 GB); concatenate and extract them. Use it to train 3D radar extrapolation, storm evolution and Gaussian-splatting style spatiotemporal models. Released under CC BY 4.0; cite Wang et al. 2025 and GridRad-Severe.

How to load this dataset in Python

from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="Ziyeeee/3D-NEXRAD", filename="2020/nexrad-2020.tar.gz.00", repo_type="dataset")
print(path)

Data structure

Files

Example file: 2020/nexrad-2020.tar.gz.00

Analyze this dataset without downloading it

On Scigantic, 3D-NEXRAD: 3D Gridded WSR-88D Radar Sequences of US Severe Storms for Precipitation Nowcasting (GridRad-Severe, 2020-2022) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Climate & Weather datasets