Merge and non-scalar AuxCoords #3084
Description
Cubes in a cubelist are prevented from merging if they have non-scalar auxcoords with points (or bounds, presumably) that do not match. Example:
import iris
print iris.__version__
dimcoord = iris.coords.DimCoord(range(5), long_name='foo')
scalarcoords = [iris.coords.AuxCoord(x, long_name='bar') for x in [1, 2]]
auxcoords = [iris.coords.AuxCoord(range(x, x + 5),
long_name='wibble') for x in [3, 6]]
cubes = iris.cube.CubeList([
iris.cube.Cube([0] * 5, dim_coords_and_dims=[(dimcoord, 0)],
aux_coords_and_dims=[(auxcoord, 0), (scalarcoord, None)])
for auxcoord, scalarcoord in zip(auxcoords, scalarcoords)])
for cube in cubes:
print cube
print cubes.merge_cube()Output:
2.0.0
unknown / (unknown) (foo: 5)
Dimension coordinates:
foo x
Auxiliary coordinates:
wibble x
Scalar coordinates:
bar: 1
unknown / (unknown) (foo: 5)
Dimension coordinates:
foo x
Auxiliary coordinates:
wibble x
Scalar coordinates:
bar: 2
Traceback (most recent call last):
<snip>
raise iris.exceptions.MergeError(msgs)
iris.exceptions.MergeError: failed to merge into a single cube.
Coordinates in cube.aux_coords (non-scalar) differ: wibble.
Ideally merge would be able to stack the wibble coordinates together to create a 2D auxcoord spanning the new cube.
This came from an example discussed internally on Met Office Yammer, where cubes loaded from a pp-file each have a 2D surface_pressure coordinate that spans latitude and longitude. Attempting to merge the cubes (creating a time dimension) fail because the surface pressure changes with time.