Open Declan-Curran1 opened 3 months ago
Hello,
I was also interested in following up on @Declan-Curran1's comment. Is there any chance to get the constant variables file or a script to download it?
We would like to use this repository as a demo for applying diffusion models to weather data for our reading group at the Alan Turing Institute.
Thank you very much!
Thanks for finding this and sorry. I manually downloaded that file from the ERA5 Copernicus Store. I am currently away but at some point write a script to download it. For now I've added that file to the directory data/
. You could also pull it from Copernicus here: https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels?tab=form (select Other, then select geopotential and land-sea mask, and any date/time and request it as netcdf).
Thank you!
Hi Laura/Robbie, thanks for providing the accompanying code to this paper.
I have downloaded the data per the instructions in the readme which comes to files samples_{year}.nc for each year.
However, when I go to run TrainUnet.py or TrainDiffusion.py - I run into the same error:
It appears that the 'UpscaleDataset()' class in DatasetUS.py requires another .nc file 'ERA5_const_sfc_variables.nc' that isn't downloaded through the download script and is separate to the yearly data?