Big Cartesian triangle example

[zmoon]

As somewhat of a uxarray newcomer, based on these vertices and units of meters, I would respectfully disagree that this is a big triangle (taken from here).

---

I think it may need to be made more clear that it's the normalized Cartesian coords that matter in this situation. Also latlon is not documented in uxarray.open_grid.

Also, I wondered if it might be worth adding at least an option to the Matplotlib collection methods, to support making that plot automatically. That is, at least in the case that the ax is set to plate carrée projection, allow the edges to be presented in a (pseudo-)plate carrée way instead of geodetic.

I could take on this issue.

[philipc2]

@rajeeja

[rajeeja]

Hey @zmoon

Thank you for bringing this up. Yes, feel free to create a pull request for fixing this. This was just meant to be a toy example so this could definitely use some cleanup and fixes.

[zmoon]

That is, at least in the case that the ax is set to plate carrée projection, allow the edges to be presented in a (pseudo-)plate carrée way instead of geodetic.

Thanks @rajeeja, what do you and @philipc2 think about this? When the ax is plate carrée should the transform on the collections be set so as well? Or at least optionally? Currently it's always set to geodetic. But with the special case of plate carrée ax, straight line edges make more sense to me (and more efficient too).

[rajeeja]

That is, at least in the case that the ax is set to plate carrée projection, allow the edges to be presented in a (pseudo-)plate carrée way instead of geodetic.

Thanks @rajeeja, what do you and @philipc2 think about this? When the ax is plate carrée should the transform on the collections be set so as well? Or at least optionally? Currently it's always set to geodetic. But with the special case of plate carrée ax, straight line edges make more sense to me (and more efficient too).

Regarding the plate carrée visualization, I agree this would be a good enhancement. When the axes projection is plate carrée, having the option to render edges as straight lines (using plate carrée transform) instead of geodesics makes perfect sense both visually and computationally. We might want to add an optional parameter like edge_transform='auto'|'geodetic'|'platecarree' to give users control, with perhaps 'auto' matching the axes projection when it's plate carrée. This maintains backward compatibility while providing the intuitive behavior for this common use case. Would you be interested in submitting a PR for these fixes? We should also document the latlon parameter and fix the misleading meter unit as this is dimentionless unit sphere.