Error: Multiple 'Z' axis dims were found in this dataset. Please drop the unused dimension(s) before performing grid operations.

[mzelinka]

Question criteria

• I added a descriptive title here.
• I searched the xCDAT GitHub Discussions to find a similar question and didn't find it.
• I searched the xCDAT documentation and/or Xarray documentation.

Describe your question

I am reading in a dataset and trying to regrid it. I am getting the error noted in the title, despite the fact that I am explicitly setting only one coordinate to be the 'Z' axis.

Are there are any possible answers you came across?

No.

Minimal Complete Verifiable Example (MVCE)

import xcdat as xc
import numpy as np

# Create a grid
LAT = np.arange(-89,91,2.0)
LON = np.arange(1.25,360,2.5)
output_grid = xc.regridder.grid.create_grid(lat=LAT, lon=LON)

# Read in data
path='/p/css03/esgf_publish/CMIP6/CFMIP/CNRM-CERFACS/CNRM-CM6-1/amip-p4K/r1i1p1f2/CFmon/clisccp/gr/v20190820/'
fn = path+'clisccp_CFmon_CNRM-CM6-1_amip-p4K_r1i1p1f2_gr_197901-201412.nc'
ds=xc.open_mfdataset(fn)

# Explicitly set plev7,tau coords to be 'Z' and 'tau'
ds.plev7.attrs['axis'] = 'Z'
ds.tau.attrs['axis'] = 'tau' # could use anything here, the error below still occurs

# Regrid
output = ds.regridder.horizontal('clisccp', output_grid, tool="xesmf", method="bilinear", extrap_method="inverse_dist")

Relevant log output

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In [34], line 13
     11 ds.tau.attrs['axis'] = None
     12 # Regrid:
---> 13 output = ds.regridder.horizontal('clisccp', output_grid, tool="xesmf", method="bilinear", extrap_method="inverse_dist")

File ~/miniconda3/envs/xclim2/lib/python3.10/site-packages/xcdat/regridder/accessor.py:322, in RegridderAccessor.horizontal(self, data_var, output_grid, tool, **options)
    317 except KeyError as e:
    318     raise ValueError(
    319         f"Tool {e!s} does not exist, valid choices {list(HORIZONTAL_REGRID_TOOLS)}"
    320     )
--> 322 input_grid = _get_input_grid(self._ds, data_var, ["X", "Y"])
    323 regridder = regrid_tool(input_grid, output_grid, **options)
    324 output_ds = regridder.horizontal(data_var, self._ds)

File ~/miniconda3/envs/xclim2/lib/python3.10/site-packages/xcdat/regridder/accessor.py:445, in _get_input_grid(ds, data_var, dup_check_dims)
    442 input_grid = ds.drop_dims(to_drop)
    444 # drops extra dimensions on input grid
--> 445 grid = input_grid.regridder.grid
    447 # preserve mask on grid
    448 if "mask" in ds:

File ~/miniconda3/envs/xclim2/lib/python3.10/site-packages/xcdat/regridder/accessor.py:91, in RegridderAccessor.grid(self)
     89 for axis in axis_names:
     90     try:
---> 91         data, bnds = self._get_axis_data(axis)
     92     except KeyError:
     93         continue

File ~/miniconda3/envs/xclim2/lib/python3.10/site-packages/xcdat/regridder/accessor.py:111, in RegridderAccessor._get_axis_data(self, name)
    106 def _get_axis_data(
    107     self, name: CFAxisKey
    108 ) -> Tuple[xr.DataArray | xr.Dataset, xr.DataArray]:
    109     coord_var = get_dim_coords(self._ds, name)
--> 111     _validate_grid_has_single_axis_dim(name, coord_var)
    113     try:
    114         bounds_var = self._ds.bounds.get_bounds(name, coord_var.name)

File ~/miniconda3/envs/xclim2/lib/python3.10/site-packages/xcdat/regridder/grid.py:744, in _validate_grid_has_single_axis_dim(axis, coord_var)
    725 """Validates that the grid's axis has a single dimension.
    726 
    727 If the grid has multiple dimensions (e.g., "lat" and "latitude" dims), xcdat
   (...)
    741     If the grid has multiple dimensions.
    742 """
    743 if isinstance(coord_var, xr.Dataset):
--> 744     raise ValueError(
    745         f"Multiple '{axis}' axis dims were found in this dataset, "
    746         f"{list(coord_var.dims)}. Please drop the unused dimension(s) before "
    747         "performing grid operations."
    748     )

ValueError: Multiple 'Z' axis dims were found in this dataset, ['plev7', 'tau']. Please drop the unused dimension(s) before performing grid operations.

Environment

xCDAT version 0.6.1

INSTALLED VERSIONS

commit: None
python: 3.10.9 | packaged by conda-forge | (main, Feb 2 2023, 20:20:04) [GCC 11.3.0]
python-bits: 64
OS: Linux
OS-release: 3.10.0-1160.102.1.el7.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: ('en_US', 'UTF-8')
libhdf5: 1.12.2
libnetcdf: 4.8.1

xarray: 2023.4.2
pandas: 1.5.1
numpy: 1.23.4
scipy: 1.8.1
netCDF4: 1.6.1
pydap: None
h5netcdf: None
h5py: 3.8.0
Nio: None
zarr: None
cftime: 1.6.2
nc_time_axis: 1.4.1
PseudoNetCDF: None
iris: None
bottleneck: 1.3.5
dask: 2022.10.2
distributed: 2022.10.2
matplotlib: 3.6.2
cartopy: 0.21.0
seaborn: 0.13.1
numbagg: None
fsspec: 2022.11.0
cupy: None
pint: 0.20.1
sparse: 0.13.0
flox: None
numpy_groupies: None
setuptools: 65.5.1
pip: 22.3.1
conda: 23.9.0
pytest: None
mypy: None
IPython: 8.6.0
sphinx: None

Anything else we need to know?

No response

[pochedls]

Unverified, but I think cf-xarray might be using the tau attribute {"positive": "up"} to determine that tau is a vertical axis. I think adding this resolves the issue:

del ds.tau.attrs['positive']

[mzelinka]

Thanks Steve, that worked. Do you know why explicitly setting the coords / dims like I do in the example above does not work?

[pochedls]

@tomvothecoder – can you clarify this – should xCDAT change how cf-xarray is used (I guess we can modify options when calling cf-xarray?) or are you saying we should open a github issue with cf-xarray to request / suggest changes?

[tomvothecoder]

@tomvothecoder – can you clarify this – should xCDAT change how cf-xarray is used (I guess we can modify options when calling cf-xarray?) or are you saying we should open a github issue with cf-xarray to request / suggest changes?

The first option, although it might be tricky because we use ds.cf.coords (and ds.cf.axes) which I think interprets positive and the other attributes listed in that (link).

We can also live with it and maybe improve the error message somehow?

[pochedls]

It seems like we should improve the error message and link to the cf-xarray table you shared. I think this, because I suspect that changing this on xcdat's side will create more problems than it solves. Also, it seems like the dataset discussed in this thread shouldn't have tau labelled with {positive: up}. So maybe cf-xarray is doing the right thing?

[tomvothecoder]

It seems like we should improve the error message and link to the cf-xarray table you shared. I think this, because I suspect that changing this on xcdat's side will create more problems than it solves. Also, it seems like the dataset discussed in this thread shouldn't have tau labelled with {positive: up}. So maybe cf-xarray is doing the right thing?

I agree, improving the error message and linking the cf-xarray table is a good idea. I'll open up an issue on it.

I think cf-xarray is doing the right thing here by interpreting positive for the Z axis. I'm not sure if tau is supposed to have that attribute set or not.

[tomvothecoder]

#610