make xee compatible with rioxarray

[12rambau]

when dealing with rasters represented as xarray datasets, rioxarray have proven to be extremely useful to perform all sort of enoying tasks like reproject, window reading, parralalesiation, plottting etc... It shine with a vibrant community and I don't see any reason to reinvent the wheel here.

As rioxarray is a xarray extention, it's automatically loaded upon import. The only limitation to use it is to make sure that datasets generated by xee have all the metadata required by GDAL and rio to perform their magic.

Do you think it would make sense ? Happy to propose a PR in this direction.

[alxmrs]

I’m not so familiar with rioxarray. What metadata is required to be compatible with GDAL? Could you show me a (mock) code example where the two libraries would interact? I don’t understand what problem for the user that this solves where there isn’t a solution today. I’m not saying there isn’t a problem, I simply don’t know. Like, isn’t it a good thing that we don’t have to interface with GDAL client-side?

[12rambau]

TBH once I started using rasterio in a Python environment, GDAL stopped being an issue as the compilation is maintained and managed by rasterio wheel itself (the consequence is of course a very big wheel but you know the omelet, the eggs....). I have multiple use cases that requires lots of extra work when they are literally one-liner within rioxarray. The biggest challenges being reprojection, plotting and histogram matching.

I'll try to make you a small reproducible example during the the geo4good hackathon.

[alxmrs]

I’m warming up to this feature having seen more of how folks use Xee. I wrote a note on user expectations here:

https://github.com/google/Xee/issues/45#issuecomment-2372747409

Maybe it’s my first time experiencing “the death of the author.” :)

[alxmrs]

And I look forward to your prototype!