issues: 548607657

CF conventions define Cell Boundaries and specify how to encode the presence of cell boundary variables in dataset attributes.

To represent cells we add the attribute bounds to the appropriate coordinate variable(s). The value of bounds is the name of the variable that contains the vertices of the cell boundaries.

For example consider this dataset: http://esgf-data.ucar.edu/thredds/dodsC/esg_dataroot/CMIP6/CMIP/NCAR/CESM2/historical/r10i1p1f1/Amon/tas/gn/v20190313/tas_Amon_CESM2_historical_r10i1p1f1_gn_200001-201412.nc

python url = 'http://esgf-data.ucar.edu/thredds/dodsC/esg_dataroot/CMIP6/CMIP/NCAR/CESM2/historical/r10i1p1f1/Amon/tas/gn/v20190313/tas_Amon_CESM2_historical_r10i1p1f1_gn_200001-201412.nc' ds = xr.open_dataset(url) ds

gives <xarray.Dataset> Dimensions: (lat: 192, lon: 288, nbnd: 2, time: 180) Coordinates: * lat (lat) float64 -90.0 -89.06 -88.12 -87.17 ... 88.12 89.06 90.0 * lon (lon) float64 0.0 1.25 2.5 3.75 5.0 ... 355.0 356.2 357.5 358.8 * time (time) object 2000-01-15 12:00:00 ... 2014-12-15 12:00:00 Dimensions without coordinates: nbnd Data variables: time_bnds (time, nbnd) object ... lat_bnds (lat, nbnd) float64 ... lon_bnds (lon, nbnd) float64 ... tas (time, lat, lon) float32 ...

Despite the presence of the bounds attributes ```

print(ds.time.bounds, ds.lat.bounds, ds.lon.bounds) time_bnds lat_bnds lon_bnds ```

The variables time_bnds, lat_bnds, and lon_bnds are not decoded as coordinates but as data variables. I believe that this is not in accordance with CF conventions.

Instead, we should decode all bounds variables to coordinates.

I cannot think of a single use case where one would want to treat these variables as data variables rather than coordinates. It would be easy to implement, but it is a breaking change.

Not that this is just a proposal to move bounds variables to the coords part of the dataset. It does not address the more difficult / complex question of how to actually use the bounds for indexing or plotting operations (see e.g. #1475, #1613), although it could be a first step in that direction.

Full ncdump of dataset

``` xarray.Dataset { dimensions: lat = 192 ; lon = 288 ; nbnd = 2 ; time = 180 ; variables: float64 lat(lat) ; lat:axis = Y ; lat:bounds = lat_bnds ; lat:standard_name = latitude ; lat:title = Latitude ; lat:type = double ; lat:units = degrees_north ; lat:valid_max = 90.0 ; lat:valid_min = -90.0 ; lat:_ChunkSizes = 192 ; float64 lon(lon) ; lon:axis = X ; lon:bounds = lon_bnds ; lon:standard_name = longitude ; lon:title = Longitude ; lon:type = double ; lon:units = degrees_east ; lon:valid_max = 360.0 ; lon:valid_min = 0.0 ; lon:_ChunkSizes = 288 ; object time(time) ; time:axis = T ; time:bounds = time_bnds ; time:standard_name = time ; time:title = time ; time:type = double ; time:_ChunkSizes = 512 ; object time_bnds(time, nbnd) ; time_bnds:_ChunkSizes = [1 2] ; float64 lat_bnds(lat, nbnd) ; lat_bnds:units = degrees_north ; lat_bnds:_ChunkSizes = [192 2] ; float64 lon_bnds(lon, nbnd) ; lon_bnds:units = degrees_east ; lon_bnds:_ChunkSizes = [288 2] ; float32 tas(time, lat, lon) ; tas:cell_measures = area: areacella ; tas:cell_methods = area: time: mean ; tas:comment = near-surface (usually, 2 meter) air temperature ; tas:description = near-surface (usually, 2 meter) air temperature ; tas:frequency = mon ; tas:id = tas ; tas:long_name = Near-Surface Air Temperature ; tas:mipTable = Amon ; tas:out_name = tas ; tas:prov = Amon ((isd.003)) ; tas:realm = atmos ; tas:standard_name = air_temperature ; tas:time = time ; tas:time_label = time-mean ; tas:time_title = Temporal mean ; tas:title = Near-Surface Air Temperature ; tas:type = real ; tas:units = K ; tas:variable_id = tas ; tas:_ChunkSizes = [ 1 192 288] ; // global attributes: :Conventions = CF-1.7 CMIP-6.2 ; ... [truncated] ```

Output of xr.show_versions()

INSTALLED VERSIONS ------------------ commit: None python: 3.6.7 | packaged by conda-forge | (default, Jul 2 2019, 02:07:37) [GCC 4.2.1 Compatible Clang 4.0.1 (tags/RELEASE_401/final)] python-bits: 64 OS: Darwin OS-release: 16.7.0 machine: x86_64 processor: i386 byteorder: little LC_ALL: None LANG: en_US.UTF-8 LOCALE: en_US.UTF-8 libhdf5: 1.10.5 libnetcdf: 4.6.2 xarray: 0.14.0+19.gba48fbcd pandas: 0.25.1 numpy: 1.17.2 scipy: 1.3.1 netCDF4: 1.5.1.2 pydap: None h5netcdf: 0.7.4 h5py: 2.10.0 Nio: None zarr: 2.3.2 cftime: 1.0.3.4 nc_time_axis: 1.2.0 PseudoNetCDF: None rasterio: None cfgrib: 0.9.7.1 iris: None bottleneck: 1.2.1 dask: 2.4.0 distributed: 2.4.0 matplotlib: 3.1.1 cartopy: 0.17.0 seaborn: 0.9.0 numbagg: None setuptools: 41.2.0 pip: 19.2.3 conda: None pytest: 5.1.2 IPython: 7.8.0 sphinx: 1.6.5