Easy way to calculate maximum across multiple variables in a dataset

[gewitterblitz]

Hi,

I have a spatiotemporal dataset (3D space and 1D time) with multiple variables (same dimensions and units) and I would like to calculate which variable has the highest value in each grid cell at each time step. Here's an example:

What's the best way to go about doing such an operation?

[dcherian]

Dataset.to_array is very useful for this kind of thing. then call idxmax on the DataArray.

In [2]: ds = xr.Dataset(
   ...:     {"a": ("x", [0, 1, 2, 3]), "b": ("x", [-1, 2, 0, 0]), "c": ("x", [0, 1, 4, 5])}
   ...: )

In [3]: ds.to_array("variable").idxmax("variable")
Out[3]:
<xarray.DataArray 'variable' (x: 4)>
array(['a', 'b', 'c', 'c'], dtype='<U1')
Dimensions without coordinates: x

[gewitterblitz]

Thanks, @dcherian, this was helpful!