First of all, your first image is Azimuthal Equidistant Projection. So that, it is quite different from your second plot (Orthographic projection). To get the plot (first image) like that using Cartopy requires some steps that are interesting to follow. Here is the code with comments that produces the output plot that I consider a good result.
import cartopy.crs as ccrs
import matplotlib.pyplot as plt
from matplotlib.path import Path
import matplotlib.path as mpath
import numpy as np
r_limit = 20037508 #from: ax.get_ylim() of full extent
# this makes circle for clipping the plot
pts = [] #unit circle vertices
cds = [] #path codes
numps = 32
for ix,ea in enumerate(np.linspace(0, 2*np.pi, numps)):
#print(ea)
xi = np.cos(ea)
yi = np.sin(ea)
pts.append([xi,yi])
if (ix==0):
# start
cds.append(1)
elif (ix==numps-1):
# close
cds.append(79)
else:
cds.append(4)
# make them np.array for easy uses
vertices = np.array(pts)
codes = np.array(cds)
# manipulate them to create a required clip_path
scale = r_limit*0.5
big_ccl = mpath.Path(vertices*scale, codes)
clippat = plt.Polygon(big_ccl.vertices[:, :], visible=True, fill=False, ec='red')
# create axis to plot `AzimuthalEquidistant` projection
# this uses specific `central_latitude`
ax = plt.axes(projection=ccrs.AzimuthalEquidistant(central_latitude=-23))
# add the clip_path
ax.add_patch(clippat)
# draw graticule (of meridian and parallel lines)
# applying clip_path to get only required extents plotted
ax.gridlines(draw_labels=False, crs=ccrs.PlateCarree(),
xlocs=range(-180,180,30), ylocs=range(-80,90,20), clip_path=clippat)
# specify radial extents, use them to set limits of plot
r_extent = r_limit/(2-0.05) # special to the question
ax.set_xlim(-r_extent, r_extent)
ax.set_ylim(-r_extent, r_extent)
ax.set_frame_on(False) #hide the rectangle frame
plt.show()
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…