Coordinating on new Xarray accessor functionalities for rasters and DEMs

[rhugonnet]

Hi all,

We are planning to make new Xarray accessors for our packages xDEM and GeoUtils that focus on raster analysis and DEM analysis, respectively (right now relying only on rasterio and geopandas), and I'm dropping a message so we can coordinate to avoid unnecessary overlap, and try to discuss where functionalities would fit the best! 😊
See: pydata/xarray#8040, pydata/xarray#8041 and corteva/rioxarray#687

For xDEM:
I think there's only a bit of overlap. We provide functionalities to derive surface terrain attributes, for which you also have methods in xarray-spatial, section "Surface"! You can find ours here: https://xdem.readthedocs.io/en/stable/terrain.html. We also have in mind to introduce some other attributes more specific to hydrology, glaciology and geomorphology.
Do you have other attributes planned for the near future? How do you think we should coordinate the development of those in terms of Xarray extension? As those are quite specific to DEMs, I can see them living in xDEM.

For GeoUtils:
I think we only have the "proximity" function that overlaps. In this package, our core idea is to simplify end-users consistently dealing with the three big problems of georeferenced data: (1) regular/irregular georeferenced grid, (2) nodata values and (3) pixel interpretation.
We have in mind to wrap a large set of (mostly existing) tools that are not always trivial to use on georeferenced raster/vector data (registration, spatial variography, zonal statistics). Here, we could potentially make xarray-spatial a dependency of GeoUtils, and simply point towards xarray-spatial for existing functionalities, and implement new things also in xarray-spatial if they fit, on a per-case basis? What do you think? 🙂

[brendancol]

@rhugonnet Thanks for the issue and would love to learn more about what yall are doing for GeoUtils. I checked out the terrain documentation which looked awesome!