How to build a correct cartopy feature around the south pole?

[oliviermarti]

I want to build a cartopy feature for Antartica in palaeo geography. I have the lon/lat of the coastline.

I can't find a way to have a nice plot in any projection.

Here is an example that plots a simple circle (filled) around the south pole. It gives an ugly result in some projection.

Is there a trick to make this work ? How was build the LAND feature to produce such plots ?

Thanks for any idea

Olivier

import numpy as np
import matplotlib.pyplot as plt, matplotlib as mpl
import cartopy.crs as ccrs, cartopy
import shapely as shp

# Defines a circle around Antartica
nn = 20
lon_coast = np.linspace (start=-180, stop=180, num=nn)
lat_coast = np.ones (nn)*(-70)
coord_coast = np.stack ( [lon_coast, lat_coast] ).T

<img width="950" height="562" alt="Image" src="https://github.com/user-attachments/assets/c522e187-e665-4574-9a4b-206dccfeef7d" />

coord_coast = np.vstack ( [coord_coast, np.array([np.nan, np.nan])])

# Build the cartopy feature for the circle
coast_shp  = shp.Polygon (coord_coast)
coast_feat = cartopy.feature.ShapelyFeature (coast_shp, ccrs.PlateCarree (central_longitude=0), facecolor='green')

# Plot the circle with various projections
fig = plt.figure(figsize=(12,8) )

def my_fig (idx, proj) :
    pos=230+idx
    ax = fig.add_subplot (pos, projection=proj)
    ax.add_feature (coast_feat, facecolor='green', edgecolor='red')
    ax.coastlines ()
    ax.gridlines  ()

my_fig (1, ccrs.PlateCarree  ())
my_fig (2, ccrs.Orthographic (central_latitude=-90))
my_fig (3, ccrs.PlateCarree  (central_longitude=180))
my_fig (4, ccrs.Robinson     (central_longitude=0))
my_fig (5, ccrs.Robinson     (central_longitude=0))

[greglucas]

Your feature is a single line going from (-180 -> 180) and never closes back on itself or has any defined thickness. I think most of the features that need to wrap around and enclose a pole would have to follow an edge to the border and then go around. Something like this looks like it works for your case, where we go down to the south pole on the right edge, then over to the left edge and Shapely will close it with the original point to form the closed polygon.

coord_coast = np.vstack ( [coord_coast, np.array([(lon_coast[-1], -90), (lon_coast[0], -90)])])

[oliviermarti]

I've tried to use this kind of trick. But :

• In polar projections, this plots a spurious line between the South pole and the coast.
• In PlateCarree, this plots a line along the bottom edge (the South pole line). And a line between the Pole and the coast.

I'm trying to build a feature that would work with a large set a projections, and can't find a general trick.

But I see that the cartopy.feature.LAND has this very same problem : a line between the South pole and the coast, at 180°. It is short, as it lies in the Ross Sea, with a coastline very close to the pole.

This means that not even the specialists have solved this problem. Not good news :-(

But the ax.coastline () is neat in all projections. And so is ax.add_feature(cartopy.feature.COASTLINE) (the first one being probably an alias of the second).

How are the coastal features builds ? They use LINESTRING geometries, and not POLYGON.

I can solve the problem by creating two features :

• One with shapely.Polygon, and adding the south Pole. It's OK for filling in any projection.
• One with shapely.LineString, with a closing point (adding the firts point at the end), but no South pole point. It's OK for the coastline in any projection. The filling works for polar projections, but with poor results at the edge. It does not work for rectangular projections.

I'm still aiming for "one feature fits all". But with no clue :-((

Thank you for caring,

I join a simple script for the circle featureS around Antarctica, and a plot for paleo geography 80 Ma ago

Olivier

import numpy as np
import matplotlib.pyplot as plt, matplotlib as mpl
import cartopy.crs as ccrs, cartopy
import shapely as shp

# Defines a circle around Antartica
nn = 100 ; center=0 # center=0 is OK. Other values give a truncated land in rectangular projections with central_longitude != center

lon_coast = np.linspace (start=center-180, stop=center+180, num=nn)
lat_coast = np.ones (nn)*(-70)

coords = np.stack ( [lon_coast, lat_coast] ).T
# Add a point at South Pole for the land feature
coord_land  = np.vstack ( [coords, np.array([(lon_coast[-1], -90), (lon_coast[0], -90)])])
# Close the coastline for the coast feature
coord_coast = np.vstack ( [coords, np.array([lon_coast[-1], lat_coast[-1]])])

coast_shp  = shp.Polygon    (coord_land)
coast_line = shp.LineString (coord_coast)

land_feat  = cartopy.feature.ShapelyFeature (coast_shp , ccrs.PlateCarree(), )
coast_feat = cartopy.feature.ShapelyFeature (coast_line, ccrs.PlateCarree(), )

# Plot the circle with various projections
fig = plt.figure(figsize=(12,8) )

def my_fig (idx, proj) :
    pos=230+idx
    ax = fig.add_subplot (pos, projection=proj)
    ax.add_feature (land_feat, facecolor='green', edgecolor='none')
    ax.add_feature (coast_feat, facecolor='none', edgecolor='red')
    #ax.add_feature (cartopy.feature.LAND, facecolor='green', edgecolor='red')
    ax.coastlines ()
    #ax.gridlines  ()

my_fig (1, ccrs.PlateCarree  ())
my_fig (2, ccrs.Orthographic (central_latitude=-90))
my_fig (3, ccrs.PlateCarree  (central_longitude=180))
my_fig (4, ccrs.Robinson     (central_longitude=0))
my_fig (5, ccrs.Robinson     (central_longitude=180))
#my_fig (6, ccrs.Stereographic(central_latitude=-90))