Images for iris.tests.integration.plot.test_vector_plots.

[pp-mo]

Test plots for SciTools/iris#3158

[ajdawson]

The case for streamplot latlon_2d_polar looks wrong, the streamlines extend outside the map boundary.

[pp-mo]

@ajdawson the streamlines extend outside the map boundary.

I don't see why that is necessarily wrong ?

If you compare the 'non-latlon 2d' outputs ...
quiver
and streamplot
... you see that the coordinate range of the quiver plot (i.e. the coords used) is x=[0, 5] and y=[0, 4],
but the bounding rectangle of the corresponding streamplot is x=[0, 6] and y=[0, 5].
It basically just does that.

Pragmatically, I'm not at all sure how this could be fixed, anyway. Presumably it would need special support in cartopy, because I don't think the matplotlib command provides sufficient flexibility to allow you to control this aspect ?

[pp-mo]

Ok, on closer inspection I found that all this is completely wrong-headed because streamplot does not do what I thought ...
See : SciTools/iris#3158 (comment)

[pp-mo]

I've now ditched streamplot entirely.
Does this look ok now @ajdawson and/or @lbdreyer ?

[lbdreyer]

I'm happy with this. Just a shame about the streamplot