Problem with Cartopy vectors on various projections

[valeriupredoi]

Not sure if this affects us or not, but better safe than sorry - thanks to @ajheaps who flagged this to me SciTools/cartopy#1179 - can someone who knows their diagnostic plots vectors and uses various projections with Cartopy have a look and see if their results are wonky, please? 🍺

[zklaus]

The only use of quiver in our code base that I could find is in esmvaltool/diag_scripts/emergent_constraints/lif1f2.py. There, it seems like this is not an issue because plot and data use the same projection (plate carrée). However, maybe this should be double checked? @katjaweigel and @axel-lauer have worked on that program. Could you two have a quick look?

[katjaweigel]

Yes, it uses the same projection (and at least the current map is rather close to the Equator than to the Pole), so this shouldn't be an issue in esmvaltool/diag_scripts/emergent_constraints/lif1f2.py

[zklaus]

Thanks @katjaweigel, closing. @valeriupredoi, please reopen if you still have doubts.

[valeriupredoi]

all good, was just about to do the same, cheers @katjaweigel 🍺

[lguez]

I think it is an issue for your code in lif1f2.py, even if the projection for axx is PlateCarree. quiver expects u /cos(phi) as an argument and you feed it with u. So I think you get wrong direction and wrong magnitude.