Example documentation in open_rasterio

[gabriel-abrahao]

This is my first time submitting an issue, so sorry if doing this wrong.

In the documentation for open_rasterio there's an example on how to generate 2D coordinates from a GeoTiff file using it:

xarray/xarray/backends/rasterio_.py

Lines 177 to 181 in 37522e9

	from affine import Affine
	da = xr.open_rasterio('path_to_file.tif')
	transform = Affine.from_gdal(*da.attrs['transform'])
	nx, ny = da.sizes['x'], da.sizes['y']
	x, y = np.meshgrid(np.arange(nx)+0.5, np.arange(ny)+0.5) * transform

However, this method does not really work because the da.attrs['transform'] attribute is not coded in the GDAL standard, but as affine.Affine expects directly (maybe it wasn't so in a previous version. So, at least for me (v0.16.2), getting the affine transformation right is done by simply replacing that line with the simpler:

transform = Affine(*da.attrs['transform'])

Again, sorry if this is not how this should be done, I just took a while to figure this out for my application and thought other people might benefit from it. If I can do something to fix it (if it's really broken) please let me know.

[scottyhq]

Thanks @gabriel-abrahao good catch! the docs are definitely outdated here and need to be changed. This has been brought up before here #3185 (comment)

You can either do what you suggested or simply transform = Affine(*da.transform)

xarray/xarray/tests/test_backends.py

Line 4468 in 12b4480

actual_transform = Affine(*da.transform)

A PR to update the documentation would be welcome. It'd be just a one-line fix here:

xarray/xarray/backends/rasterio_.py

Line 179 in d2582c2

transform = Affine.from_gdal(*da.attrs['transform'])

cc @snowman2 the docstring for open_rasterio has the same issue currently in rioxarray https://github.com/corteva/rioxarray/blob/63ce87f92835897d46aaa270f5ec07aca4e72cda/rioxarray/_io.py#L653

[snowman2]

Thanks for mentioning that. For rioxarray, the fix is:

transform = da.rio.transform()