This is my dataset for the Smoky Mountain Conference Data Challenge (https://smc-datachallenge.ornl.gov/).
You can see the source document at ./smcefr.md
(which contains markdown sources)
To convert to another format, use pandoc (apt install pandoc
), like so:
$ pandoc -s smcefr.md -o index.html
See the file ./dl-smcefr.py
for technical specs. The basic idea is to download EFR (Earth Full Resolution) from the Sentinel-3 satellite (via Copernicus), and reduce the multispectral data into standard RGB image formats, cropped to (1024, 1024, 3) and stored as a directory of PNG files (./data
)
The full data is available here (smcefr-full.tar.gz
), on Google Drive: https://drive.google.com/file/d/1HHSqd7LYi1npEqHD_frCgOSGa5VrvALy/view
Navigate to the GitHub releases, and download smcefr-mini.tar.gz
Then, you can expand the tar file, and have a directory of the PNG image dataset
To generate the dataset by yourself (WARNING: uses lots of network/disk access), run:
$ pip3 install pillow numpy netCDF4
$ python3 ./dl-smcefr.py