[FIXED] Buffer line across lon=180 causes banding geopandas matplotlib

Issue

I am trying to plot Polygon geometries across the 180 longitude when the map's central_longitude = 180; however there's an artificial banding occurring.

Is there a way to stop matplotlib from banding shown in the red circle?

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

lats = np.linspace(-75, 75, 100)
lons = np.linspace(150, 190, 100)
gdf = gpd.GeoDataFrame(
    crs=4326, geometry=gpd.points_from_xy(lons, lats)
)
gdf = gdf.to_crs({'init': 'epsg:3174'})
gdf['geometry'] = gdf['geometry'].buffer(90000)
gdf['route'] = 0
gdf = gdf.to_crs({'init': 'EPSG:4326'})

ax = plt.axes(projection=ccrs.PlateCarree(central_longitude=180))
ax.add_geometries(gdf['geometry'], crs=ccrs.PlateCarree())
ax.set_extent([120, 250, -75, 75], crs=ccrs.PlateCarree())
ax.gridlines(draw_labels=True, xlocs=[180])
ax.coastlines()

Solution

Reprojecting from epsg:3174 to EPSG:4326 jumbles the geometries crossing dateline. You can plot the geodataframe with epsg:3174 direcly as seen with this code:

import matplotlib.pyplot as plt
import geopandas as gpd
import cartopy.crs as ccrs
import numpy as np

lats = np.linspace(-75, 75, 100)
lons = np.linspace(150, 190, 100)
gdf = gpd.GeoDataFrame(
    crs=4326, geometry=gpd.points_from_xy(lons, lats)
)
# gdf2 = gdf.to_crs({'init': 'epsg:3174'})
gdf2 = gdf.to_crs(epsg=3174)
gdf2['geometry'] = gdf2['geometry'].buffer(90000)
gdf2['route'] = 0

ax = plt.axes(projection=ccrs.PlateCarree(central_longitude=180))
ax.add_geometries(gdf2['geometry'], crs=ccrs.epsg(3174))
ax.set_extent([120, 250, -75, 75], crs=ccrs.PlateCarree())
ax.gridlines(draw_labels=True, xlocs=[180])
ax.coastlines()

plt.show()

Answered By - swatchai