Wrong grid line labels with Lambert projection

[zxdawn]

Description

Now, the newest cartopy (0.18.0) supports adding grid line labels to the Lambert projection.

But, the labels shown by proplot are wrong.

I suppose proplot misses some keys:

x_inline=False, y_inline=False

Steps to reproduce

import proplot as plot
f, axs = plot.subplots(proj='lcc')
axs.format(coast=True,
           latlines=1, lonlines=1,
           labels=True,
           lonlim=(113, 119),
           latlim=(37, 41.5))

Expected behavior:

Actual behavior:

Equivalent steps in matplotlib

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

proj = crs.LambertConformal(central_latitude = 38.854,
                            central_longitude = 115.643,
                            standard_parallels = (30, 60),
                            false_easting = 64.,
                            false_northing = 67.)


axs = plt.axes(projection=proj)
axs.coastlines()
axs.set_extent([113, 119, 37, 41.5], crs=crs.PlateCarree())
gl = axs.gridlines(draw_labels=True, dms=True, x_inline=False, y_inline=False)
gl.top_labels = gl.right_labels = False

Proplot version

master branch

[lukelbd]

Thanks, I'll add support for "inline" keywords before the next release. I think I'll also add rc keywords, something like rc['geogrid.loninline'] and rc['geogrid.latinline'], and maybe will make the default "inline" setting to False because I don't like how the "inline" labels look in the proplot examples and your first example looks much nicer to me.

[zxdawn]

@lukelbd Agree. The default inline should be false ;) Thank you for your help.

[zxdawn]

@lukelbd Another idea: decreasing the inline font size in case users want to add the inline labels. Then, it will perform better than cartopy.

[zxdawn]

Strange result based on the geogrid-features branch:

Code:

import proplot as plot
f, axs = plot.subplots(proj='lcc')
axs.format(coast=True,
           latlines=1, lonlines=1,
           labels=True,
           lonlim=(113, 119),
           latlim=(37, 41.5))

[lukelbd]

Label rotatation and "inline" labels are now disabled by default (the branch wasn't finished yet).

Note I'm still keeping basemap support because basemap longitude/latitude labels look way nicer even compared to cartopy 0.18, and disabling "inline" labels sometimes causes labels on the edge to mysteriously disappear from where the gridlines intersect the boundary. If your labels don't look right basemap=True may be your best bet.

[zxdawn]

@lukelbd Actually, it's not the problem of label rotation.
As you can see in this example, lon and lat and flipped:

import proplot as plot
f, axs = plot.subplots(proj='lcc')
axs.format(coast=True,
           latlines=1, lonlines=1,
           labels=True,
           lonlim=(113, 119),
           latlim=(37, 41.5))

plot.show()

proplot version
0.6.1

[lukelbd]

I think that's because you're zoomed-in on a part of the projection where parallels are oriented vertically and meridians are oriented horizontally. This is more clear if you look at the region around 100 degrees East in the default, zoomed out version:

import proplot as plot
f, axs = plot.subplots(proj='lcc', axwidth=4)
axs.format(
    coast=True,
    labels=True,
    #lonlim=(113, 119),
    #latlim=(37, 41.5
)

For the behavior you're looking for, you need to center the projection around the longitude into which you will be "zooming." For example:

f, axs = plot.subplots(proj='lcc', proj_kw={'lon_0': 116}, axwidth=4)
axs.format(
    coast=True,
    labels=True,
    #lonlim=(113, 119),
    #latlim=(37, 41.5
)

[zxdawn]

@lukelbd Thanks!