Changing the name and value of a coordinate in-place

[James Penn]

Hi all, I'm looking for suggestions for a cleaner way of changing the coordinate axis of a DataArray I compute from NetCDF data.

I have a netCDF dataset with lat and latb coordinate labels. lat is cell centres (the coordinate axis of my data variables), latb is cell edges.  len(latb) = len(lat)+1. To calculate finite differences I need the difference across a cell which is given by the diff of the latb values, the result of which will be centered on lat points.

The problem is dlat = data.latb.diff('latb') has coordinate latb, but this latb is inconsistent with my original data as it is now 1 shorter. I want to remap it to lat which has the correct length and positional alignment with my dataset. The solution I came up with is:

dlat = xarray.DataArray(data.latb.diff('latb').values*np.pi/180, coords=[('lat', data.lat)])*a

but this is ugly, and fragile, as it assumes only one dimensional input.

Essentially, I want to wholesale replace the name and values of a coordinate axis in-place. Is there a better way to do this?

[Jody Klymak]

Bumping this, as I have exactly the same question.  Did you get an answer James?  

[Jody Klymak]

OK, here is how I did it.  Criticisms welcome, as I'm just figuring out how to do things...

```

dlat = dlat.rename({'latb':'lat'})

dlat['lat'].data = data['lat'].data

```

For some reason simply saying `dlat['lat']=data['lat']` doesn't work.

[James Penn]

Hi Jody,

I ended up doing as you suggest above, although I used the `data['lat'].values` property.  Perhaps `.data` is the more appropriate one, thanks.