Romain Hugonnet

In case it's relevant: blockwise has a problem right now, see https://github.com/GlacioHack/xdem/issues/584

@adebardo Can close this as well?

Perfect, thanks a lot! :slightly_smiling_face: Only thing left is to update the imports to be optional like this: https://github.com/GlacioHack/xdem/blob/e1e2065ea52a7db3c8e88045d743aa1f6efd08bb/xdem/fit.py#L47 And raise an error during `BlockwiseCoreg.apply()` if the import failed, like...

Right now the entire module requires SciKit-Learn to be imported. From experience this can create two problems: - If another module imports something from `blockwise.py` in the future, and the...

No problem I'll open an issue to add it later.

All done, thanks! :smile: Don't forget to add "Resolves #XXX" in the PR descriptions in the future, so that issues are closed/linked automatically :wink: (here I had to do it...

Good catch. This means RichDEM is also wrong on that formulation, as we test against it and it cites Z&T: https://github.com/GlacioHack/xdem/blob/34e5049ae76a858b9c95c8763d9763b298c56e3c/tests/test_terrain.py#L208 Should we then implement both the formulations of Zevenberg...

After more reading, Moore et al. (1991) introduces this specific term, but it is based on the 9-term 3x3 window fitting method of Zevenberg and Thorne (1987), which doesn't introduce...

Finally, convex in 3D seems to correspond more to a bulging landform which would contain points, and concave to a hole that wouldn't (see figures on Google), so I think...

Great! :star_struck: Tell me if you need more feedback on this PR :) Question: I heard that Ruff was a "drop-in replacement" for black, flake8, etc (so producing almost exactly...