fixed probable typo for 'coordinate' entry in dimensions dictionary

[ifenty]

convention here seems to be

XC=dict(dims=["j", "i"], attrs=dict(
    standard_name="longitude", long_name="longitude",
    units="degrees_east", coordinate="YC XC")),

some instances fixed where "coordinate" was spelled "coordinates". May have no effect.

[rabernat]

Thanks so much for this Ian!

Can you think of a way to test that this feature is working properly? For example, perhaps some check in this test?

xmitgcm/xmitgcm/test/test_mds_store.py

Lines 93 to 98 in 7c5305c

	def test_read_grid(all_mds_datadirs):
	"""Make sure we read all the grid variables."""
	dirname, expected = all_mds_datadirs
	ds = xmitgcm.open_mdsdataset(
	dirname, iters=None, read_grid=True,
	geometry=expected['geometry'])

[rabernat]

Ian, can you help us understand what impact this PR has? I agree there is a typo, but I'm struggling to understand the consequences of that typo.

If this coordinates / coordinates entry actually has no effect, we should just remove it.

[ifenty]

@rabernat Consequences are probably zero with respect to any calculations because when the dataset is saved to netcdf, xarray creates a 'coordinate' attribute for data variables: a 'space separated list of auxillary coordinates' as per CF conventions. It seems that xmitgcm is specifing its own coordinate list (e.g., ['XC YC'] or ['XG YG']) which xarray might use instead of making its own list. Because of the typo, the saved netcdf dataset could have a variables with both a 'coordinates' attributes (misspelled coordiante added by xmitgcm) and a 'coordinate' attribute added by xarray. I believe this 'coordinates' attribute is used in some plotting software, such as panoply. Seems like it's ok to keep the code in 'xmitgcm'.

[rabernat]

Ok understood. I'd prefer to have some sort of test for this, but since it's a very minor change, we can merge this as is.

Thanks!