to_dataset_dict not setting time.encoding when decode_times=True

[klindsay28]

It looks like time.encoding is not being set when I read in multiple files with to_dataset_dict with decode_times=True.

When you write a dataset to netCDF with no time encoding, the resulting time units end up differing from the original datasets, which I find to be unexpected behavior.

Following pydata/xarray#1614, a potential behavior is to keep the encoding from the first file. I think this is already being done for attributes.

[andersy005]

@klindsay28, when you get time, can you try the master branch

pip install git+https://github.com/NCAR/intake-esm.git

and confirm that this is working properly?

[klindsay28]

@andersy005, this update appears to be working.

I spot checked a couple of CMIP6 datasets for which to_dataset_dict was previously not setting time.encoding, and time.encoding is being set with the updated intake_esm.

Thanks!