Issue on page /getting-started/UXarray in 10 Lines

[DrCMet]

Dear Developers,
I have been following along the UXarray in 10 Lines tutorial on my own machine, working with a Mediterranean model run from the latest MPAS tutorial.

Everything works until the plot. My code is below along with output. MSLP is the variable in my MPAS output rather than PSI. Am hoping you can help me troubleshoot the error?
Thanks,
Frank Colby
Professor of Meteorology
UMass Lowell

CODE:
import uxarray as ux
grid_path = "Mediterranean.init.nc"
data_path = "diag.2019-09-01_12.00.00.nc"
uxds = ux.open_dataset(grid_path, data_path)
print(uxds)
print(uxds.uxgrid)
print(uxds.uxgrid.node_lon)
print("before mslp")
print(uxds["mslp"])
print("before mean")
print(uxds["mslp"].mean())
print("after mean")
uxds["mslp"].plot()

OUTPUT:
(base) mpasuser@stratus:.../MPAS-Model/Mediterranean$ python testux.py
<xarray.UxDataset> Size: 4MB
Dimensions: (Time: 1, n_face: 7912, n_node: 16152,
nIsoLevelsT: 5, nIsoLevelsZ: 13)
Coordinates:

• Time (Time) datetime64[ns] 8B 2019-09-01T12:00:00
  Dimensions without coordinates: n_face, n_node, nIsoLevelsT, nIsoLevelsZ
  Data variables: (12/108)
  initial_time |S64 64B ...
  xtime (Time) |S64 64B ...
  mslp (Time, n_face) float32 32kB ...
  relhum_50hPa (Time, n_face) float32 32kB ...
  relhum_100hPa (Time, n_face) float32 32kB ...
  relhum_200hPa (Time, n_face) float32 32kB ...
  ... ...
  t_oml_initial (Time, n_face) float32 32kB ...
  t_oml_200m_initial (Time, n_face) float32 32kB ...
  h_oml (Time, n_face) float32 32kB ...
  h_oml_initial (Time, n_face) float32 32kB ...
  hu_oml (Time, n_face) float32 32kB ...
  hv_oml (Time, n_face) float32 32kB ...
  <uxarray.Grid>
  Original Grid Type: MPAS
  Grid Dimensions:
• n_node: 16152
• n_edge: 24063
• n_face: 7912
• n_max_face_nodes: 10
• n_max_face_edges: 10
• n_max_face_faces: 10
• n_max_node_faces: 3
• two: 2
  Grid Coordinates (Spherical):
• node_lon: (16152,)
• node_lat: (16152,)
• edge_lon: (24063,)
• edge_lat: (24063,)
• face_lon: (7912,)
• face_lat: (7912,)
  Grid Coordinates (Cartesian):
• node_x: (16152,)
• node_y: (16152,)
• node_z: (16152,)
• edge_x: (24063,)
• edge_y: (24063,)
• edge_z: (24063,)
• face_x: (7912,)
• face_y: (7912,)
• face_z: (7912,)
  Grid Connectivity Variables:
• face_node_connectivity: (7912, 10)
• edge_face_connectivity: (24063, 2)
• edge_node_connectivity: (24063, 2)
• node_face_connectivity: (16152, 3)
• face_face_connectivity: (7912, 10)
• face_edge_connectivity: (7912, 10)
  Grid Descriptor Variables:
• face_areas: (7912,)
• edge_face_distances: (24063,)
• edge_node_distances: (24063,)

<xarray.DataArray 'node_lon' (n_node: 16152)> Size: 65kB
array([41.219387, 40.87472 , 40.768932, ..., 27.331116, 22.93844 ,
5.047055], dtype=float32)
Dimensions without coordinates: n_node
Attributes:
standard_name: longitude
long name: Longitude of the corner nodes of each face
units: degrees_east
before mslp
<xarray.UxDataArray 'mslp' (Time: 1, n_face: 7912)> Size: 32kB
[7912 values with dtype=float32]
Coordinates:

• Time (Time) datetime64[ns] 8B 2019-09-01T12:00:00
  Dimensions without coordinates: n_face
  Attributes:
  units: Pa
  long_name: Mean sea-level pressure
  before mean
  <xarray.UxDataArray 'mslp' ()> Size: 8B
  array(101190.859375)
  after mean
  Traceback (most recent call last):
  File "/hdd/mpas-model/MPAS-Model/Mediterranean/testux.py", line 13, in
  uxds["mslp"].plot()
  File "/home/mpasuser/anaconda3/lib/python3.12/site-packages/uxarray/plot/accessor.py", line 329, in call
  return self.polygons(**kwargs)
  ^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/mpasuser/anaconda3/lib/python3.12/site-packages/uxarray/plot/accessor.py", line 416, in polygons
  gdf = self._uxda.to_geodataframe(
  ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/mpasuser/anaconda3/lib/python3.12/site-packages/uxarray/core/dataarray.py", line 197, in to_geodataframe
  raise ValueError(
  ValueError: Data Variable must be 1-dimensional, with shape 7912 for face-centered data.
  (base) mpasuser@stratus:.../MPAS-Model/Mediterranean$

[philipc2]

Hi @DrCMet

For our plotting functionality, we expect a single time-slice, since we've provided custom functionality to plot the data on unstructured grids.

I think doing the following should work if you data is a time-series on one of the grid elements (either nodes, faces, or edges)

uxds["mslp"].isel(Time=0).plot()

If you need the basic Xarray functionality, you can drop down by doing

uxds["mslp"].to_xarray().plot()

Let me know if this helps!

[DrCMet]

Hi Eric and Philip,
The first fix got rid of the error message but didn't produce a plot.

The second fix produced a plot, but of the mslp variable on the y axis and the n face number on the x axis. I guess I don't understand the syntax of the plot routine. Was expecting the plot of mslp with lat and lon as the axes, which is what shows up in the UXarray in 10 Lines section.

Here's the plot:

[DrCMet]

I am accessing my system through putty and xming and so have to add "from matplotlib import pyplot as plt and then use plt.show(block=True) to get the plot to occur. But don't think that would have changed the axes, would it?

Thanks,
Frank

[philipc2]

Our plotting API is written around HoloViews, so if you are not in a notebook, I would suggest looking here: https://holoviews.org/user_guide/Exporting_and_Archiving.html

We support Matplotlib workflows, but they require a bit of extra work. See here: https://uxarray.readthedocs.io/en/latest/user-guide/mpl.html