Access data via common coordinate across all children of a DataTree

[HenrikAsmuth]

Hi,
I have a DataTree tree where my all data lies at the second child level. All datasets down there have the same coordinates x and y, as well as an additional multi-index idx. I would like to access the data across all children at the lowest level directly from tree, yet so far I am struggling to find a way to do this. tree.isel(idx=0) is working but the corresponding operation with .sel is not.
I briefly skimmed the discussions in #9077. Now I am wondering if I am just missing something when building my tree (do I need to explicitly pass the coordinates up from the children to the parents?), or if that functionality is simply not there, yet.
Thanks for you help!

[TomNicholas]

Hi @HenrikAsmuth, the coordinate inheritance feature works downwards from the root to the leaves, meaning that from the leaves you can access any coordinate variables or indexes defined closer to the root. It sounds like you might be expecting the opposite behaviour?

It would be easier to help you if you could provide a minimal example of what your tree looks like and what behaviour you are hoping for.

[HenrikAsmuth]

Hi Tom,

exactly, I would like to get the opposite behaviour. Here'e a minimal example:

data1 = xr.Dataset({'data': (['x', 'y'], np.array([[10, 11], [12, 13]]))},
                   coords={'x': np.array([0.0, 1.0]), 'y': np.array([0.0, 1.0])})
data1 = data1.stack(idx=[...])
data2 = xr.Dataset({'data': (['x', 'y'], np.array([[14, 15], [16, 17]]))},
                   coords={'x': np.array([0.5, 1.5]), 'y': np.array([0.5, 1.5])})
data2 = data2.stack(idx=[...])
data3 = xr.Dataset({'data': (['x', 'y'], np.array([[18, 19], [20, 21]]))},
                   coords={'x': np.array([0.25, 1.25]), 'y': np.array([0.25, 1.25])})
data3 = data3.stack(idx=[...])

dt = xr.DataTree()
dt["leaf1"] = data1
dt["leaf2"] = data2
dt["leaf3"] = data3

dt.isel(idx=2) #working
dt.sel(idx=2) #not working
dt.sel(x=0.0) #not working