How can one set cartopy's Gridline label styles (major_xticks and minor_xticks) of a set of geoaxes - subplots?

I am having serious difficulties in setting the Text styles of the xlabels and ylabels from the cartopy's gridlines.

When I follow the example given the cartopy's website, everything goes well, and I get the expected figure: .

Nevertheless, when I try to create multiple subplots, and each subplot with its own gridline (notice that they are all from the same geographical region, and thus, have the same gridline X and Y coordinates), I am unable to properly set the Gridlines' Text styles of its labels. In fact, some of the xlabels end up missing from the plot, as presented in this second example .

For sake of reproducibility, here is the code I used for the creation of Figure 2:

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

from cartopy.mpl.ticker import (LongitudeFormatter, LatitudeFormatter,
                                LatitudeLocator)

fig, axes = plt.subplots(2,2, subplot_kw={'projection':ccrs.Mercator()})

axes = axes.ravel()
    
for ax in axes:
        
    ax.coastlines()
    
    ax.set_extent([-120, -25, -45, 25])
    
    gl = ax.gridlines(crs=ccrs.PlateCarree(), draw_labels=True,
                      linewidth=2, color='gray', alpha=0.5, linestyle='--')
    gl.top_labels = False
    gl.left_labels = False
    gl.xlines = False
    
    gl.ylocator = LatitudeLocator()
    gl.xformatter = LongitudeFormatter()
    gl.yformatter = LatitudeFormatter()
    
    gl.xlocator = ticker.MaxNLocator(6)
    gl.ylocator = ticker.MaxNLocator(6)
    
    gl.xlabel_style = {'size': 12, 'color': 'k', 'rotation':45}
    gl.ylabel_style = {'color': 'red', 'weight': 'bold', 'size':6}
plt.tight_layout()
plt.show()

The present issue may also help other parties, as the present topic has been indirectly subject of third other posts: 1) How can I rotate gridline labels on a cartopy map?; 2) How to set curvilinear lat lon gridline labels parallel to gridlines with cartopy?;

Sincerely,

Solution

Questions:

How can I rotate gridline labels on a cartopy map?
How to set curvilinear lat/lon gridline labels parallel to gridlines with cartopy?

Answer to question 1

Difficulties found:

missing some/all longitude labels (bottom edges)
not achieve result when define some styling (text rotation)

From ref. document: https://scitools.org.uk/cartopy/docs/latest/matplotlib/gridliner.html

Cartopy's gridliner function is a new feature that will be evolving a lot in the future (mentioned in its documentation).

The current status is very good for most cylindrical types of projection with default style settings.

If one manipulates some label text's styles without proper settings of associate options, you don't get what you expect.

For gridlines' labels, the following rules apply: Extracted from help(gl._update_labels_visibility)

Labels are plotted and checked by order of priority, with a high priority for longitude labels at the bottom and top of the map, and the reverse for latitude labels.
A label must not overlap another label marked as visible.
A label must not overlap the map boundary.
When a label is about to be hidden, other angles are tried in the absolute given limit of max_delta_angle by increments of delta_angle of difference from the original angle.

Here are some of the options that should be set with proper values, default values often lead to unexpect results.

gridlines() options 'xpadding' and 'ypadding', can be set to get more space to plot label texts
gridlines.xlabel_style() keys
- 'size': default values = ??
- 'rotation': default value = 0 degrees
- 'ha': default is 'center' (clever use of text justification can avoid setting 'xpadding' and 'ypadding')

Here is the runnable code that generates the subplots of maps with all 4 sets of tick-labels on all sides of each map.

import matplotlib.pyplot as plt
import matplotlib.ticker as mticker
import cartopy.crs as ccrs
from cartopy.mpl.ticker import (LongitudeFormatter, LatitudeFormatter,
                                LatitudeLocator)

fig = plt.figure(figsize=(8, 10))  # set reasonable width, height in inches
axes = fig.subplots(2,2, subplot_kw={'projection':ccrs.Mercator()})
axes = axes.ravel()

# This plots all axes, adds most of map contents/components
#  some maps' marginal info will be plotted in second for-loop
for ax in axes:
    ax.coastlines()
    ax.set_extent([-120, -25, -45, 25])

    # plot grid lable text: left, right, bottom
    gl = ax.gridlines(crs=ccrs.PlateCarree(), draw_labels=True, 
                      linewidth=2, color='gray', alpha=0.5, linestyle='--')

    # manipulate the `gl` object as needed as follows
    # some of these are not accepted as options in the statment above (cause errors)
    gl.top_labels = False
    #gl.left_labels = True  #default already
    #gl.right_labels = True
    #gl.bottom_labels = True

    gl.xlines = True
    gl.ylines = True

    gl.ylocator = LatitudeLocator()

    gl.xformatter = LongitudeFormatter()
    gl.yformatter = LatitudeFormatter()

    gl.xlocator = mticker.MaxNLocator(6)
    gl.ylocator = mticker.MaxNLocator(6)

    # declare text size
    XTEXT_SIZE = 10
    YTEXT_SIZE = 10
    # to facilitate text rotation at bottom edge, ...
    # text justification: 'ha':'right' is used to avoid clashing with map's boundary
    # default of 'ha' is center, often causes trouble when text rotation is not zero
    gl.xlabel_style = {'size': XTEXT_SIZE, 'color': 'k', 'rotation':45, 'ha':'right'}
    gl.ylabel_style = {'size':YTEXT_SIZE, 'color': 'k', 'weight': 'normal'}

# Second round of plots, do the top marginal texts in particular
# also plot title text of each axes
for ix,ax in enumerate(axes):

    gl2 = ax.gridlines(crs=ccrs.PlateCarree(), draw_labels=True, color='none')
    gl2.left_labels = False
    gl2.right_labels = False 
    gl2.bottom_labels = False

    gl2.xlines = False  # already plotted in first loop, so we hide them
    gl2.ylines = False

    XTEXT_SIZE = 12  # ***intended larger value***
    # to facilitate text rotation on top edge, ...
    # text justification: 'ha':'left' is used here to avoid clashing with map's boundary
    # default of 'ha' is center, often causes trouble when text rotation is not zero
    gl2.xlabel_style = {'size': XTEXT_SIZE, 'color': 'blue', 'rotation':45, 'ha':'left'}

    # plot title
    ax.set_title("Subplot:"+str(ix))

plt.draw()  # draw/update enables better tight_layout adjustment
plt.tight_layout()  # without plt.draw(), real tight!!
plt.show()

Answer to question 2 (concepts only):

use gl._labels to get access to all label texts, you get them as a list
iterate for each of them as a 'text object'
get/inspect styles/attributes of the 'text object'
manipulate the text styles/attributes individually

Some comments

good question should focus on one topic only
your second question should be posted as a new question