Support polar axes labels with cartopy

[zxdawn]

Description

Now, cartopy has supported the axes labels for polar projection.

But, proplot doesn't work with the same newest version of cartopy.

Steps to reproduce

import proplot as plot
fig, axs = plot.subplots(proj='npstere')
axs.format(coast=True, latlines=10, lonlines=30, lonlabels=True, labels=True, boundinglat=60)

Expected behavior (matplotlib with cartopy)

Actual behavior: (proplot with cartopy)

Equivalent steps in matplotlib

import matplotlib.pyplot as plt
import cartopy.crs as ccrs

ax = plt.subplot(projection=ccrs.NorthPolarStereo())
ax.set_extent([-180,180,90,60], ccrs.PlateCarree())
ax.coastlines()
ax.gridlines(draw_labels=True)

Proplot version

0.6.4

[bnjmnr]

I have the same issue (using 'nplaea' projection). Considering that I am new to Proplot, I couldn't find where this lack of labels is coming from. I guess it should be somewhere around the '_toggle_gridliner_labels' method, but I can't find anything more. If someone can give me some hint of what to change, I am happy to participate in solving the bug, if that saves some time.

[lukelbd]

Finally figured this out. By default, proplot gives "pole-centered" projections circle boundaries to match basemap and because (IMHO) circle boundaries looks nicer. Unfortunately, however, cartopy does not seem to support gridline labels for maps whose boundaries were modified with set_boundary.

So, this is a cartopy limitation -- when I repeat your example in cartopy with a circular boundary, the labels disappear:

import matplotlib.pyplot as plt
import cartopy.crs as ccrs
ax = plt.subplot(projection=ccrs.NorthPolarStereo())
ax.set_extent([-180,180,90,60], ccrs.PlateCarree())
path = plot.axes.geo._circular_boundary(100)  # internal proplot method used to make polar boundaries
ax.set_boundary(path, transform=ax.transAxes)
ax.coastlines()
ax.gridlines(draw_labels=True)
plot.show()

However I recently decided to add the rc option cartopy.circular that can be set to False to disable the circle boundary thing. This will let you have labels on your polar plots (alternatively, you can also use basemap as the backend -- basemap supports labels on circular polar plots).

Here's an example with non-circular boundaries and gridline labels:

import proplot as plot
plot.rc['cartopy.circular'] = False
fig, ax = plot.subplots(proj='npstere')
ax.format(coast=True, latlines=10, lonlines=30, lonlabels=True, labels=True, boundinglat=60)

I've added a note about this to the docs in the latest version.

[zxdawn]

@lukelbd Great! Thanks for coming back to polish proplot ;) Hope your Ph.D. career goes well.

[Thomas-Moore-Creative]

indeed, as said by @zxdawn - great to see you back @lukelbd, keeping proplot shiny!

[zxdawn]

@lukelbd The latitude grid labels aren't shown nicely.

If only one latitude label is shown, readers won't know the space between each latitude line.

import proplot as pplt
pplt.rc['cartopy.circular'] = False
fig, ax = pplt.subplots(proj='npstere')
ax.format(coast=True, latlines=10, latlabels=True, boundinglat=65)

[lukelbd]

@zxdawn In this case you probably want inlinelabels=True. The new user guide goes into more details about different cartopy labelling options.

 import proplot as plot
 plot.rc['cartopy.circular'] = False
 fig, ax = plot.subplots(proj='npstere')
 ax.format(coast=True, lonlabels=True, labels=True, boundinglat=60, inlinelabels=True)

[zxdawn]

@lukelbd Wow, thanks for the new updates, Excellent!

[zxdawn]

@lukelbd In the newest master branch (0.9.3.post44), this doesn't work anymore:

import proplot as pplt

fig, axs = pplt.subplots(proj='npstere')
axs.format(coast=True, lonlabels=True, labels=True, boundinglat=60, inlinelabels=True)

[lukelbd]

Hmm not sure what's going on there -- can't reproduce this on my machine. I get the following results with circular=False or circular=True (note you can also set rc['cartopy.circular'] to False or True).

As expected gridlines only fail with circular=True.

import proplot as pplt
fig = pplt.figure()
ax = fig.add_subplot(proj='npstere', circular=False)
ax.format(coast=True, labels=True, inlinelabels=True)

import proplot as pplt
fig = pplt.figure()
ax = fig.add_subplot(proj='npstere', circular=True)
ax.format(coast=True, labels=True, inlinelabels=True)

[lukelbd]

FYI I also decided to make the application of inlinelabels less confusing by implementing the following behavior (ed372d6):

• loninline=True implies lonlabels=True (unless lonlabels was specified otherwise).
• latinline=True implies latlabels=True (unless latlabels was specified otherwise).
• inlinelabels=True implies labels=True (unless labels was specified otherwise).

[zxdawn]

@lukelbd Strange ... I tested it on my laptop with the master branch and also got the empty label image.

Here's the package info:

matplotlib: 3.4.3
cartopy: 0.20.1

For the inline and label, it still sounds confusing to me.
Let's take the image below as an example:

How can I add lonlabels and enable only latinline, if they are the same?