figsize is not affecting the actual figure size of a matplotlib figure when a cartopy projection is applied about cartopy HOT 6 CLOSED

Are you running in a notebook or a script? Please provide a complete example.

from cartopy.

thanks for the reply. I run the code from VScode Jupyter notebook.

Below is the complete code. The plot parameters can be obtained from the attached json file
plot_params.json
The figure I got after running the code is also attached here

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

with open('plot_params.json', 'r') as f:
    plot_params = json.load(f)
# 
glon_e3d = plot_params['glon_e3d']
gclat_e3d = plot_params['gclat_e3d']
glon_test = plot_params['glon_test']
gclat_test = plot_params['gclat_test']
lon_secs = plot_params['lon_secs']
lat_secs = plot_params['lat_secs']
# 
proj = ccrs.LambertAzimuthalEqualArea(central_latitude = clat, central_longitude = clon)
fig,axes = plt.subplots(1,3, figsize=(10,10),subplot_kw={'projection':proj})
axs = axes.ravel()
# 
extent = [205, 225, 63, 69]
ax = axs[0]
ax.gridlines()
ax.set_extent(extent, crs=ccrs.PlateCarree())
ax.scatter(glon_e3d, gclat_e3d, transform = ccrs.Geodetic(), c = 'r', marker = '.', label = 'data points', s=10)
# 
ax = axs[1]
ax.gridlines()
ax.set_extent(extent, crs=ccrs.PlateCarree())
ax.scatter(glon_test, gclat_test, transform = ccrs.Geodetic(), c = 'r', marker = '.', label = 'data points', s=10)
# 
ax = axs[2]
ax.gridlines()
ax.set_extent(extent, crs=ccrs.PlateCarree())
ax.scatter(lon_secs, lat_secs, transform = ccrs.Geodetic(), c = 'r', marker = '.', label = 'data points', s=10)

plt.show()

when the number of axes in the subplots is 1, the figsize parameter affects the actual size of the figure without changing extent of the axis. However, this is not the case when the number of the axes is higher than 1. In addition despite the given width and heights of the fig are equal, the result figure doesn't show that.

from cartopy.

Cartopy's axes all have fixed aspect ratios so will not fill up the figure when arranged like this. The default backend used by Jupyter always resizes the figure to fit around the artists on it. I suspect you will see consistent behaviour if you do a more simple plot without Cartopy:

import matplotlib.pyplot as plt

fig, ax_arr = plt.subplots(1,3, figsize=(10,10))

for ax in ax_arr:
    ax.set_aspect('equal')

from cartopy.

thanks for the reply. you are right, I see the same issue with the above code. However, the figure behave as expected when I remove the ax.set_aspect('equal') line. I will leave the issue open in case some one suggests a general solution

from cartopy.

@habtie-phys Right, ax.set_aspect('equal') is equivalent to what every Cartopy-based plot is doing due to the nature of plotting on geographic projections.

There isn't a general solution to this problem currently given that Matplotlib doesn't have a dynamic layout engine for Axes. Instead Matplotlib provides various APIs for carving out rectangles from a figure, which Axes are then free to adjust within to suit their needs.

Closing since at best anything addressing this would have to happen in Matplotlib itself.

from cartopy.

@dopplershift thank you for the reply. Now I understand that this is not the right place for the issue to remain open.

from cartopy.