Rationalising mesh dimension names #5908
Description
🐛 Bug Report
Given two valid CF-UGRID netCDF mesh files, the first of which has the following netCDF dimensions:
dimensions:
time = 4 ;
dim0 = 221184 ;
num_node = 221186 ;
num_vertices = 4 ;
The second of which has the following netCDF dimensions:
dimensions:
time = 22 ;
bnds = 2 ;
dim0 = 221184 ;
num_node = 221186 ;
num_vertices = 4 ;
I easily can use iris to load these files into as single cube list and then save to a single combined netCDF file as follows:
import iris
from iris.experimental.ugrid import PARSE_UGRID_ON_LOAD
with PARSE_UGRID_ON_LOAD.context():
cubes = iris.load(fnames)
iris.save(cubes, "combined.nc")Inspecting the dimensions of combined.nc gives:
dimensions:
num_node = 221186 ;
dim0 = 221184 ;
time = 22 ;
dynamics_face_N_nodes = 4 ;
bnds = 2 ;
num_node_0 = 221186 ;
dim0_0 = 221184 ;
time_0 = 4 ;
num_node_1 = 221186 ;
dim0_1 = 221184 ;
😢
Consequently, attempting to then load combined.nc results in the following traceback:
Traceback (most recent call last):
File "<snip>/script.py", line 36, in <module>
combined = iris.load(fname)
^^^^^^^^^^^^^^^^
File "<snip>/iris/lib/iris/__init__.py", line 326, in load
return _load_collection(uris, constraints, callback).merged().cubes()
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "<snip>/iris/lib/iris/__init__.py", line 294, in _load_collection
result = _CubeFilterCollection.from_cubes(cubes, constraints)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "<snip>/iris/lib/iris/cube.py", line 97, in from_cubes
for cube in cubes:
File "<snip>/iris/lib/iris/__init__.py", line 275, in _generate_cubes
for cube in iris.io.load_files(part_names, callback, constraints):
File "<snip>/iris/lib/iris/io/__init__.py", line 219, in load_files
for cube in handling_format_spec.handler(fnames, callback, constraints):
File "<snip>/iris/lib/iris/fileformats/netcdf/loader.py", line 634, in load_cubes
mesh_coords, mesh_dim = _build_mesh_coords(mesh, cf_var)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "<snip>/iris/lib/iris/experimental/ugrid/load.py", line 489, in _build_mesh_coords
mesh_dim = cf_var.dimensions.index(mesh_dim_name)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: tuple.index(x): x not in tupleThis issue is a direct result of the dimension name mangling performed by iris.save.
In this particular case, the mesh_dim_name is 'dim0', and the cf_var.dimensions are ('time_0', 'dim0_0'). Hence the traceback from cf_var.dimensions.index.
As a quick temporary workaround, I can monkey-patch iris.experimental.ugrid.load._build_mesh_coords with this weak-contract:
def monkey_patch(mesh, cf_var):
"""this is a workaround"""
element_dimensions = {
"node": mesh.node_dimension,
"edge": mesh.edge_dimension,
"face": mesh.face_dimension,
}
mesh_dim_name = element_dimensions[cf_var.location]
for i, dim_name in enumerate(cf_var.dimensions):
if dim_name.startswith(mesh_dim_name):
mesh_dim = i
break
else:
raise ValueError("oops!")
mesh_coords = mesh.to_MeshCoords(location=cf_var.location)
return mesh_coords, mesh_dimHowever, clearly we should address this issue in a more robust way.
Note that, I can supply data on request to recreate this issue.