[v0.2] Issue on migrating prepared cutout to new cutout format

[pypsa]

[v0.2] Issue on migrating prepared cutout to new cutout format

[coroa]

Fine, then I'll just re-raise the MergeError exception from xarray and we'll leave it at that, that is at least more descriptive than what you get now.

[coroa]

I think this is fixed by commit 193164d. I'd appreciate if you could test!

[euronion]

I think this is fixed by commit 193164d. I'd appreciate if you could test!

Here is the result

WARNING:atlite.utils:
Found an old-style directory-like cutout. It can manually be recreated using

cutout = atlite.Cutout("../cutouts/western-asia-2016-era5.nc",
                       module="era5",
                       time=slice("2016-01", "2016-12"),
                       x=slice(24.899999618530273, 64.19999694824219),
                       y=slice(44.099998474121094, 10.800000190734863)
cutout.prepare()

but we are trying to offer an automated migration as well ...

WARNING:atlite.utils:Automatic migration failed. Re-create the cutout with the command above!

---------------------------------------------------------------------------
UnboundLocalError                         Traceback (most recent call last)
<ipython-input-8-96088e8e5a64> in <module>
      5                     x=slice(25,64),
      6                     y=slice(44,11),
----> 7                     time="2016"
      8 )

~/miniconda/envs/atlite-v0.2/lib/python3.7/site-packages/atlite/cutout.py in __init__(self, name, data, cutout_dir, **cutoutparams)
     92             elif os.path.isdir(os.path.join(self.cutout_dir, self.name)):
     93                 data = utils.migrate_from_cutout_directory(os.path.join(self.cutout_dir, self.name),
---> 94                                                            self.name, self.cutout_fn, cutoutparams)
     95                 self.is_view = True
     96             else:

~/miniconda/envs/atlite-v0.2/lib/python3.7/site-packages/atlite/utils.py in migrate_from_cutout_directory(old_cutout_dir, name, cutout_fn, cutoutparams)
     85             logger.warning("Automatic migration failed. Re-create the cutout with the command above!")
     86 
---> 87     data.attrs['prepared_features'] = list(sys.modules['atlite.datasets.' + data.attrs["module"]].features)
     88 
     89     return data

UnboundLocalError: local variable 'data' referenced before assignment

I think the warning offered is helpful enough and provides all the commands a user needs to quickly get back into the game.
Providing a unprepared cutout instead is I think not a good solution:
Either the loading of the prepared cutout fails (preferred) or succeeds.
But importing a prepared cutout and recieving an unprepared cutout is not what I would expect as a user.

Maybe just throw an additional exception when the automigration / merging fails?

[coroa]

One must say, you do have a peculiar cutout there, with time-dependent roughness value in one month and an independent one in another?

MergeError: conflicting values for variable 'roughness' on objects to be combined:
first value: <xarray.Variable (y: 112, x: 132)>
dask.array<shape=(112, 132), dtype=float32, chunksize=(112, 132)>
Attributes:
    units:      m
    long_name:  Forecast surface roughness
second value: <xarray.Variable (time: 8784, y: 112, x: 132)>
dask.array<shape=(8784, 112, 132), dtype=float32, chunksize=(1440, 112, 132)>
Attributes:
    units:      m
    long_name:  Forecast surface roughness

[euronion]

Sorry to disappoint, but except for one details that is just a standard cutout downloaded with the cdsapi via the era5.py module. No time indepdendent/dependent roughness there:

> for dspath in pathlib.Path("../cutouts/western-asia-2016-era5/").glob("2016*.nc"):
>   with xr.open_dataset(dspath) as ds:
>       print(dspath,":", "time" in ds.roughness.coords, ds.roughness.size)

../cutouts/western-asia-2016-era5/201601.nc :  False 14784
../cutouts/western-asia-2016-era5/201602.nc :  False 14784
../cutouts/western-asia-2016-era5/201603.nc :  False 14784
../cutouts/western-asia-2016-era5/201604.nc :  False 14784
../cutouts/western-asia-2016-era5/201605.nc :  False 14784
../cutouts/western-asia-2016-era5/201606.nc :  False 14784
../cutouts/western-asia-2016-era5/201607.nc :  False 14784
../cutouts/western-asia-2016-era5/201608.nc :  False 14784
../cutouts/western-asia-2016-era5/201609.nc :  False 14784
../cutouts/western-asia-2016-era5/201610.nc :  False 14784
../cutouts/western-asia-2016-era5/201611.nc :  False 14784
../cutouts/western-asia-2016-era5/201612.nc :  False 14784

[coroa]

Ok, I don't understand the MergeError output then, but it also does not matter too much, anyway.