Reference point to align reprojected coordinates to

[eric542]

Hello. I would like to reproject an Xarray.Dataset using ds.rio.reproject while ensuring that multiple reprojected datasets result in aligned coordinates. An easy way to achieve this would be to pass on some align parameters to the function to provide some reference point that the reprojected x/y coordinates would align to, i.e. the resulting coords would be multiples of the resolution parameter (pixel size) which, if interpolated, would align with the reference point. Is this feasible with rio.reproject at all?... (e.g. datacube.Datacube.load has such a parameter available). Note: I would like to reproject multiple such datasets "in parallel" so I don't have a reference dataset that I could use as a template with reproject_match.

[snowman2]

It currently isn't possible, but the addition is welcome if someone wanted to add it.

Additional potential helpful references:

• https://odc-geo.readthedocs.io/en/latest/index.html
• Using Dask in reprojection #119

[Kirill888]

You can use odc-geo to compute exact "pixel grid compatible with some other image" that is offset in the world to cover some region with this method:

https://odc-geo.readthedocs.io/en/latest/_api/odc.geo.geobox.GeoBox.enclosing.html

you can then extract shape, crs and transform from that and pass to rioxarray for reprojection, or use reproject built into odc-geo (that one is lacking Dask support though)

[eric542]

Thanks for the feedback... much appreciated.