search

carbonplan / cmip6-downscaling

Climate downscaling using CMIP6 data
https://docs.carbonplan.org/cmip6-downscaling
MIT License
164 stars 23 forks source link

ERA5 metadata wrong? #291

Open arbennett opened 1 year ago

arbennett commented 1 year ago

I have been using the daily ERA5 data that was used for training for the downscaling and think the metadata might not be filled in correct, but could also be doing something wrong.

To open the data I'm simply grabbing all the zarr stores and just opening them all up with:

era5_daily_cat = intake.open_esm_datastore(
    'https://cpdataeuwest.blob.core.windows.net/cp-cmip/training/ERA5-daily-azure.json'
)
prcp_files = sorted(list(era5_daily_cat.search(cf_variable_name='pr').df['zstore']))
ds = xr.open_mfdataset(prcp_files, engine='zarr')

And then when I look at any of the variable metadata/attributes I am seeing them all as the U component of wind.

norlandrhagen commented 1 year ago

Hey there @arbennett, thanks for flagging this. It's been a known issue on our end, but it's good to have a record on github. I'll post here if we get a chance to update it. If you need the correct ERA5 metadata, the Pangeo era5 stores might be a good shot or the ERA5 dataset hosted on the AWS Open Data Registry.

Thanks again for the flag!