Hierarchy between open source libraries addressing raster topic

Meta
1

Hello everybody,

I have several colleagues that struggle to understand the hierarchy between open source libraries addressing raster topic.

I tried to create a graph to summarize it (hereunder, very sketchy for now), but I was wondering if you knew if such a graph exists somewhere else and supported by people in charge of this wonderful ecosystem.

librairies_rasters.drawio
librairies_rasters.drawio911×511 9.19 KB

6 Likes
2

I don’t know of anything that matches your description exactly but think it’s a fantastic idea! The closest that comes to mind is the figure below that I think Joe Hamman made based on Stephan and Jake’s SciPy talk. It just doesn’t have explanations and is a subset of the relevant packages.

I recently got really into excalidraw and its obsidian plugin and made this diagram specifically for resampling, but again it’s not exactly what you’re asking for. I’d love to see this happen and enable its distribution, possibly through the Pangeo website. Maybe @brian-rose or @clyne have thoughts from the Pythia side since it’s fundamentally educational.

image
image1612×878 77.9 KB

image
image2122×734 136 KB

5 Likes
3

I don’t know of any particular existing figure that fits this need exactly, but it would make a great addition to Pythia Foundations if we converged on something.

The inconvenient truth is that there is nobody in charge of the ecosystem as a whole, which usually means that there’s a dearth of holistic documentation that bridges across multiple package. Foundations is trying to fill some of that gap for the geoscience community.

2 Likes
4

Hi there,

Thanks for these useful answers! I see that is an unanswered topic. If I do some updates to this graph, I’ll post it here, especially after my colleague’s feedback.

This is also where we see that the ecosystem itself is not mature yet (i.e. rioxarray not handling dask correctly besides being the official raster package of xarray).

1 Like
6

While I’m in this topic, do someone know if there is an overview somewhere of dask-compatible functions in terms of rasters? (i.e. something like a list all existing functions in GDAL and look at their python versions, either they are pure numpy like rasterio, xarray like rioxarray or dask-compatible like… ? xarray-spatial, odc-geo, geoutils… IMHO, factorization is desperately needed!)