CODEWITHSUNDEEP
pythonpython-xarray
H. Müller
H. Müller

Reputation: 123

Merging Xarray's DataArrays

Let's assume we do have multiple values with assigned coordinates:

a = []
a.append(xr.DataArray([[5.]], dims=('x', 'y'), coords=dict(x=[1.],y=[2.])))
a.append(xr.DataArray([[1.]], dims=('x', 'y'), coords=dict(x=[2.],y=[2.])))
a.append(xr.DataArray([[2.]], dims=('x', 'y'), coords=dict(x=[1.],y=[1.])))

I now want to combine all DataArrays in a into one single new DataArray with combined coordinates. I.e. the three values 5, 1, 2 with the coordinates assigned. Not provided (missing) values could be filled with anything (0, None, …).

If I would do this operation along one axis, I'd use concatenate, but what should I do in the case of multiple dimensions? Thanks a lot in advance for your kind help!

Upvotes: 2

Views: 4428

Answers (1)

OriolAbril
OriolAbril

Reputation: 8763

You can use combine_by_coords. It works with Datasets, so you'll need to either use Datasets or set the name of the DataArrays:

a = []
a.append(xr.DataArray([[5.]], dims=('x', 'y'), coords=dict(x=[1.],y=[2.]), name="var")) 
a.append(xr.DataArray([[1.]], dims=('x', 'y'), coords=dict(x=[2.],y=[2.]), name="var")) 
a.append(xr.DataArray([[2.]], dims=('x', 'y'), coords=dict(x=[1.],y=[1.]), name="var"))

xr.combine_by_coords(a)
# Output
# <xarray.Dataset>
# Dimensions:  (x: 2, y: 2)
# Coordinates:
#   * x        (x) float64 1.0 2.0
#   * y        (y) float64 1.0 2.0
# Data variables:
#     var      (x, y) float64 2.0 5.0 nan 1.0

To avoid unexpected results and warnings during the conversion from DataArray to Dataset, I would recommend to explicitly convert the objects first to Dataset, then use combine_by_coords and eventually convert back to DataArray if desired.

Upvotes: 3

Related Questions