RolnickLab / climart

A benchmark dataset for Machine Learning emulation of atmospheric radiative transfer in weather and climate models (NeurIPS 2021 Datasets and Benchmarks Track)

Repository from Github https://github.comRolnickLab/climartRepository from Github https://github.comRolnickLab/climart

Damaged files for inputs 1995 and pristine outputs 1851

ArsamAryandoust opened this issue · comments

When downloading data and trying to load them with the h5py.File() class/function of the h5py package in python, there occurs an error.

All other data files work flawlessly. Several tests have been conducted to ensure there is no damaged file problems during download.

Hi @ArsamAryandoust, you're right there seems to be an error with some files (likely caused by our cloud hosting service.) We're working on regenerating the dataset and will be uploading those on the server soon!

@Venka97 @salvaRC Cool, it would be great if you give a short notice when this is done. Also, you may want to check out how my team and I are currently using your data here: https://github.com/ArsamAryandoust/EnergyTransitionTasks

I btw really enjoyed reading your paper. I find it really well written! Good job!

Hi @ArsamAryandoust, we'll keep you updated. Running the simulation from the climate model takes a while so it might take a few weeks.

Your repository looks really cool! I see that you're interested in downscaling too :)

I also worked on downscaling recently. If you would like to check it out, you can find it here: https://arxiv.org/pdf/2208.05424.pdf

Hi @Venka97. Thanks for the reference. That's indeed the dataset I am integrating!

I will close this issue and follow your update.