FloodCastBench: High- and Low-Fidelity Flood Inundation Simulations for Flood Forecasting (Australia, UK, Pakistan, Mozambique)

Provider
Qingsong Xu, Yilei Shi, Jie Zhao et al. (FloodCastBench); mirrored by mahajananhad
License
CC-BY-4.0 (source)
Size
14,870 files
Format
TIFF (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/mahajananhad/floodcastbench)
Access
Free and public. No account with the data provider is needed.
DOI
10.5281/zenodo.14017092
Category
Environmental Science
Source
https://doi.org/10.1038/s41597-025-04725-2

About this dataset

Flood modeling and forecasting benchmark from Xu, Shi, Zhao et al. ("FloodCastBench: A Large-Scale Dataset and Foundation Models for Flood Modeling and Forecasting", Scientific Data 12, 431, 2025), re-hosted unmodified from Zenodo record 14017092 (DOI 10.5281/zenodo.14017092) so regions can be downloaded selectively. The data are spatiotemporal flood inundation depth maps from hydrodynamic simulations of real flood events: high-fidelity runs at 30 m and 60 m resolution for Australia and the UK, and low-fidelity runs at 480 m for Pakistan and Mozambique, stored as one GeoTIFF of water depth per simulation timestep (filenames are integer simulation seconds; sort numerically). A Relevant data folder holds the digital elevation models (DEM), land use, rainfall forcing, initial conditions and world files for georeferencing. About 14,855 TIFF files, 27 GB. Read with rasterio or GDAL. Use it to train neural flood forecasting, physics-informed and CNN-LSTM inundation models. Released by the original authors under CC BY 4.0; cite Xu et al. 2025 (doi:10.1038/s41597-025-04725-2).

How to load this dataset in Python

from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="mahajananhad/floodcastbench", filename="High-fidelity flood forecasting/30m/Australia/0.tif", repo_type="dataset")
print(path)

Data structure

File types in sampled files: TIFF

Files

Example file: High-fidelity flood forecasting/30m/Australia/0.tif

Analyze this dataset without downloading it

On Scigantic, FloodCastBench: High- and Low-Fidelity Flood Inundation Simulations for Flood Forecasting (Australia, UK, Pakistan, Mozambique) 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 Environmental Science datasets