cartopy: higher resolution for great circle distance line

Question

I am trying to plot a great circle distance between two points. I have found an in the cartopy docs (introductory_examples/01.great_circle.html):

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

ax = plt.axes(projection=ccrs.Robinson())

ax.set_global()

ax.coastlines()

plt.plot([-0.08, 132], [51.53, 43.17], color='red',      transform=ccrs.Geodetic())
plt.plot([-0.08, 132], [51.53, 43.17], color='blue', transform=ccrs.PlateCarree())

plt.show()

which makes the following image:

great circle example

The thing is, in my own work, the two points are much closer together, and in a different projection (though I think that isn't important here). If I change this code to be a line in a smaller area, like so:

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

ax = plt.axes(projection=ccrs.Robinson())

ax.set_extent([-5, 55, 40, 55])

ax.coastlines()

plt.plot([-0.08, 50], [51.53, 43.17], color='red',      transform=ccrs.Geodetic())
plt.plot([-0.08, 50], [51.53, 43.17], color='blue', transform=ccrs.PlateCarree())

plt.show()

This makes the following image: shorter line

The red great circle line in this case looks crappy and looks like it is due to being too low resolution. How do I increase the number of points making up the great circle line?

Answer 1

This problem is due to the hard-coded threshold in the projection. Currently this is not a user controllable parameter, but you can work around that by defining your own sub-class:

class LowerThresholdRobinson(ccrs.Robinson):

    @property
    def threshold(self):
        return 1e3

If you use LowerThresholdRobinson() instead of ccrs.Robinson() when defining the axes this should eliminate the problem.