Concatenate arrays to add timesteps

[hug-fr]

Let's say I have two datasets, both with the same two dimensions and the same n variables. One dimension is time and the other is gid.

Some variables depend on both but some others depend only on gid (for instance lat/lon). Between the tow datasets the gid coordinates are the same but the time coordinates are complementary.

What I would like is simply join these two datasets in order to have just one with all the time steps. I tried
new_ds = xr.concat ([ds1, ds2], dim = 'time')

But the result is not the expected one. The variables depending only on the gid dim now depend on both time and gid dimension.

For sur, I don't understand something and any help would be greatly appreciated

[mathause]

Can you try xr.combine_by_coords or xr.combine_nested - see also the docs: https://docs.xarray.dev/en/stable/user-guide/combining.html#combining-along-multiple-dimensions

[keewis]

the coords and data_vars parameters of concat allow controlling what is concatenated. Usually, setting them to "minimal" should be enough:

new_ds = xr.concat([ds1, ds2], dim="time", data_vars="minimal", coords="minimal")

[hug-fr]

Worked like a charm! Thank you very much

[hug-fr]

Thank you very much for your answers. I tried all the suggested methods and here are the results.

Dataset 1

Dimensions:     (time: 138, gid: 1047)
Coordinates:
  * time        (time) datetime64[ns] 1982-08-01T06:00:00 ... 1982-08-06T23:0...
  * gid         (gid) int64 5363 5364 5365 5444 5445 ... 9057 9058 9059 9060
Data variables: (12/14)
    lat         (gid) float64 ...
    lon         (gid) float64 ...

Dataset 2

Dimensions:     (time: 6, gid: 1047)
Coordinates:
  * time        (time) datetime64[ns] 1982-08-01 ... 1982-08-01T05:00:00
  * gid         (gid) int64 5363 5364 5365 5444 5445 ... 9057 9058 9059 9060
Data variables: (12/14)
    lat         (gid) float64 ...
    lon         (gid) float64 ...

Here is the result with combine_by_coords: did not provide the result I'm looking for

Dimensions:     (time: 144, gid: 1047)
Coordinates:
  * time        (time) datetime64[ns] 1982-08-01 ... 1982-08-06T23:00:00
  * gid         (gid) int64 5363 5364 5365 5444 5445 ... 9057 9058 9059 9060
Data variables: (12/14)
    lat         (time, gid) float64 46.48 46.47 46.47 ... 43.42 43.41 43.41
    lon         (time, gid) float64 7.498 7.602 7.706 ... 6.531 6.629 6.728

Here is the result with combine_nested with concat_dim = 'time': Uh! I don't really understand how the output happens with 6 additional time steps and all the dims applied to my lon/lat variables

Dimensions:     (time: 150, gid: 1047)
Coordinates:
  * time        (time) datetime64[ns] 1982-08-01 ... 1982-08-06T23:00:00
  * gid         (gid) int64 5363 5364 5365 5444 5445 ... 9057 9058 9059 9060
Data variables: (12/14)
    lat         (time, gid) float64 46.48 46.47 46.47 ... 43.42 43.41 43.41
    lon         (time, gid) float64 7.498 7.602 7.706 ... 6.531 6.629 6.728

Here is the result with concat using the "minimal" setting for both parameters as suggested: it looks like I got the expected result! Wouhou!

Dimensions:     (time: 144, gid: 1047)
Coordinates:
  * time        (time) datetime64[ns] 1982-08-01 ... 1982-08-06T23:00:00
  * gid         (gid) int64 5363 5364 5365 5444 5445 ... 9057 9058 9059 9060
Data variables: (12/14)
    lat         (gid) float64 46.48 46.47 46.47 46.48 ... 43.42 43.41 43.41
    lon         (gid) float64 7.498 7.602 7.706 5.411 ... 6.531 6.629 6.728