NASA Flood Extent Detection Dataset: Sentinel-1 SAR VV/VH Scenes with Water Labels (Bangladesh and US floods, 2017-2019)

Provider
University of Alabama in Huntsville / NASA IMPACT (Gahlot et al. 2021)
License
CC-BY-4.0 (source)
Format
PNG, TIFF (in sampled files)
Stored at
Azure Blob Storage, https://radiantearth.blob.core.windows.net/mlhub/nasa-floods
Access
Free and public. No account with the data provider is needed.
DOI
10.34911/rdnt.ebk43x
Category
Earth Observation & Geospatial
Source
https://doi.org/10.34911/rdnt.ebk43x

About this dataset

Synthetic aperture radar (SAR) imagery and hand-labeled water masks for flood events, from the European Space Agency's Sentinel-1A and Sentinel-1B C-band dual-polarized (VV, VH) missions over areas of interest in the United States and Bangladesh, 2017 to 2019 (for example a series of Hurricane Florence area scenes from May to October 2018). Scenes were processed to radiometric and terrain-corrected (RTC) amplitude with the Hybrid Pluggable Processing Pipeline (HyP3) and converted to grayscale; annotators labeled open water, where specular reflection from still water darkens the image, and flood water spreading beyond normal ponds, lakes and rivers. Each event folder (for example bangladesh_20170314t115609/ or florence_20180510t231343/) holds the VV and VH GeoTIFFs with .msk side files and a rasterized label GeoTIFF and PNG aligned to the SAR scene; documentation.pdf describes the methods. Suited to SAR flood and water semantic segmentation. Read with rasterio and train with PyTorch or TorchGeo. Licensed CC BY 4.0; cite Gahlot, Gurung, Molthan, Maskey and Ramasubramanian (2021), Flood Extent Data for Machine Learning, Radiant MLHub, doi:10.34911/rdnt.ebk43x.

How to access this dataset in Python

import adlfs
fs = adlfs.AzureBlobFileSystem(account_name='radiantearth', anon=True)
fs.ls('mlhub/nasa-floods/')[:10]

Lists the files with anonymous access. Runs on any machine, with no cloud account.

Data structure

File types in sampled files: PNG, TIFF

Files

Example file: bangladesh_20170314t115609/bangladesh_20170314t115609_label.png

Analyze this dataset without downloading it

On Scigantic, NASA Flood Extent Detection Dataset: Sentinel-1 SAR VV/VH Scenes with Water Labels (Bangladesh and US floods, 2017-2019) 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 Earth Observation & Geospatial datasets