WMS_example: "ValueError: 'x' must be finite, check for nan or inf values" with new Jupyter env

[tlvu]

New Jupyter env in PR Ouranosinc/PAVICS-e2e-workflow-tests#121 (see relevant changes there).

19:17:05  Notebook cell execution failed
19:17:05  Cell 7: Cell execution caused an exception
19:17:05  
19:17:05  Input:
19:17:05  warnings.filterwarnings("ignore", category=cartopy.io.DownloadWarning)
19:17:05  fig = plt.figure(figsize=(8, 6))
19:17:05  ax = fig.add_axes([0, 0, 1, 1], projection=cartopy.crs.Mollweide())
19:17:05  ax.imshow(
19:17:05      data,
19:17:05      origin="upper",
19:17:05      extent=(-180, 180, -90, 90),
19:17:05      transform=cartopy.crs.PlateCarree(),
19:17:05  )
19:17:05  ax.coastlines()
19:17:05  plt.show()
19:17:05  
19:17:05  Traceback:
19:17:05  
19:17:05  ---------------------------------------------------------------------------
19:17:05  ValueError                                Traceback (most recent call last)
19:17:05  Cell In[8], line 4
19:17:05        2 fig = plt.figure(figsize=(8, 6))
19:17:05        3 ax = fig.add_axes([0, 0, 1, 1], projection=cartopy.crs.Mollweide())
19:17:05  ----> 4 ax.imshow(
19:17:05        5     data,
19:17:05        6     origin="upper",
19:17:05        7     extent=(-180, 180, -90, 90),
19:17:05        8     transform=cartopy.crs.PlateCarree(),
19:17:05        9 )
19:17:05       10 ax.coastlines()
19:17:05       11 plt.show()
19:17:05  
19:17:05  File /opt/conda/envs/birdy/lib/python3.10/site-packages/cartopy/mpl/geoaxes.py:318, in _add_transform.<locals>.wrapper(self, *args, **kwargs)
19:17:05      313     raise ValueError(f'Invalid transform: Spherical {func.__name__} '
19:17:05      314                      'is not supported - consider using '
19:17:05      315                      'PlateCarree/RotatedPole.')
19:17:05      317 kwargs['transform'] = transform
19:17:05  --> 318 return func(self, *args, **kwargs)
19:17:05  
19:17:05  File /opt/conda/envs/birdy/lib/python3.10/site-packages/cartopy/mpl/geoaxes.py:1331, in GeoAxes.imshow(self, img, *args, **kwargs)
19:17:05     1329 from cartopy.img_transform import warp_array
19:17:05     1330 original_extent = extent
19:17:05  -> 1331 img, extent = warp_array(img,
19:17:05     1332                          source_proj=transform,
19:17:05     1333                          source_extent=original_extent,
19:17:05     1334                          target_proj=self.projection,
19:17:05     1335                          target_res=regrid_shape,
19:17:05     1336                          target_extent=target_extent,
19:17:05     1337                          mask_extrapolated=True,
19:17:05     1338                          )
19:17:05     1339 alpha = kwargs.pop('alpha', None)
19:17:05     1340 if np.array(alpha).ndim == 2:
19:17:05  
19:17:05  File /opt/conda/envs/birdy/lib/python3.10/site-packages/cartopy/img_transform.py:192, in warp_array(array, target_proj, source_proj, target_res, source_extent, target_extent, mask_extrapolated)
19:17:05      186 # XXX Take into account the extents of the original to determine
19:17:05      187 # target_extents?
19:17:05      188 target_native_x, target_native_y, extent = mesh_projection(
19:17:05      189     target_proj, target_res[0], target_res[1],
19:17:05      190     x_extents=target_x_extents, y_extents=target_y_extents)
19:17:05  --> 192 array = regrid(array, source_native_xy[0], source_native_xy[1],
19:17:05      193                source_proj, target_proj,
19:17:05      194                target_native_x, target_native_y,
19:17:05      195                mask_extrapolated)
19:17:05      196 return array, extent
19:17:05  
19:17:05  File /opt/conda/envs/birdy/lib/python3.10/site-packages/cartopy/img_transform.py:278, in regrid(array, source_x_coords, source_y_coords, source_proj, target_proj, target_x_points, target_y_points, mask_extrapolated)
19:17:05      274 else:
19:17:05      275     # Versions of scipy >= v0.16 added the balanced_tree argument,
19:17:05      276     # which caused the KDTree to hang with this input.
19:17:05      277     kdtree = scipy.spatial.cKDTree(xyz, balanced_tree=False)
19:17:05  --> 278     _, indices = kdtree.query(target_xyz, k=1)
19:17:05      279 mask = indices >= len(xyz)
19:17:05      280 indices[mask] = 0
19:17:05  
19:17:05  File _ckdtree.pyx:795, in scipy.spatial._ckdtree.cKDTree.query()
19:17:05  
19:17:05  ValueError: 'x' must be finite, check for nan or inf values

[huard]

See SciTools/cartopy#2199

Possible solutions from the thread:

• pin scipy<1.11
• Install pykdtree

[huard]

I confirm that installing pykdtree resolves the issue.