Metpy.advection returned all NaN values now

[rytam2]

Hello! I processed data from CMIP6 output and was trying to calculate temperature advection with metpy.calc.advection. I sucessfully calculated that about a month ago, but when I re-run the code today, all the values calculated with mpcalc.advection returned as NaN values.

dict_ds_ts,dict_ds_uas,dict_ds_vas are dictionaries with key as modelnames and value as a DataArray. I checked the values were mostly non-nans, so I am unsure where might have gone wrong.

import metpy.calc as mpcalc
for modelname in modelfilename_list:
    print(modelname)
    ds = mpcalc.advection(dict_ds_ts[modelname],u=dict_ds_uas[modelname], v=dict_ds_vas[modelname]) 
    dict_ds_tadv[modelname] = ds

My Metpy version is 1.3.1.

Any help is much appreciated!

[dopplershift]

Can you share the code you're using to access the data so we can try to reproduce the problem?

[rytam2]

Sure! Do you mind if I send you the jupyter notebook?

[dopplershift]

Unfortunately it seems that GitHub is stripping email attachments from replies. Can you try to attach it manually to a comment here? Notebooks aren't an accepted format as is, so you'll need to zip it before attaching.

[rytam2]

abrupt-4xCO2_CMIP6_data.zip
Sure thing! Let me know if this doesn't work

[dcamron]

Hey @rytam2! I don't have access to the grids you are using in interp_like, but it seems likely that the resulting lat/lon coordinates of your interpolated DataArrays are missing attrs['units'] on lat and lon, probably disappearing when you reassign and rename your coords in the provided code. This would hit a unit bug in our code resulting in NaN interpolations.

There's a fix for this incoming, but for the moment if you can assign .attrs['units'] = 'degrees' to your lat and lon coordinates, that should fix the calculation! Please let us know if that works, and thanks for the provided code.

[rytam2]

Hi @dcamron! Thank you for the suggestion! Unfortunately that doesn't seem to fix the problem, as there are still NaNs showing... Are there some other things I should try or look out in my code?