Match projection of shapefile in cartopy - matplotlib

Question

I am trying to make a Choropleth map using matplotlib and cartopy for which I obviously need to plot a shapefile first. However, I did not manage to do so, even though a similar question has been asked here and here. I suspect either the projection or the bounds to be misspecified.
My shapefile has the projection
PROJCS["WGS_1984_UTM_Zone_32Nz",
GEOGCS["GCS_WGS_1984",
DATUM["WGS_1984",
SPHEROID["WGS_84",6378137,298.257223563]],
PRIMEM["Greenwich",0],
UNIT["Degree",0.017453292519943295]],
PROJECTION["Transverse_Mercator"],
PARAMETER["False_Easting",32500000],
PARAMETER["False_Northing",0],
PARAMETER["Central_Meridian",9],
PARAMETER["Scale_Factor",0.9996],
PARAMETER["Latitude_Of_Origin",0],
UNIT["Meter",1]]
and can be downloaded here where I am talking about vg250_2010-01-01.utm32w.shape.ebenen/vg250_ebenen-historisch/de1001/vg250_gem.shp
My code is
#!/usr/local/bin/python
# -*- coding: utf8 -*-
import cartopy.crs as ccrs
import cartopy.io.shapereader as shpreader
import matplotlib.pyplot as plt
fname = 'path/vg250_gem.shp'
proj = ccrs.TransverseMercator(central_longitude=0.0,central_latitude=0.0,
false_easting=32500000.0,false_northing=0.0,
scale_factor=0.9996)
municipalities = list(shpreader.Reader(fname).geometries())
ax = plt.axes(projection=proj)
plt.title('Deutschland')
ax.add_geometries(municipalities,proj,edgecolor='black',facecolor='gray',alpha=0.5)
ax.set_extent([32458044.649189778*0.9, 5556418.748046352*1.1, 32465287.307457082*0.9, 5564153.5456742775*1.1],proj)
plt.show()
where I obtained the bounds using the corresponding method from fiona. Python throws an error
Traceback (most recent call last):
File "***/src/analysis/test.py", line 16, in <module>
ax.set_extent([32458044.649189778, 5556418.748046352, 32465287.307457082, 5564153.5456742775],proj)
File "/usr/local/lib/python2.7/site-packages/cartopy/mpl/geoaxes.py", line 652, in set_extent
ylim=self.projection.y_limits))
ValueError: Failed to determine the required bounds in projection coordinates. Check that the values provided are within the valid range (x_limits=[-20000000.0, 20000000.0], y_limits=[-10000000.0, 10000000.0]).
[Finished in 53.9s with exit code 1]
This doesn't make sense to me. Also, experimenting with ccrs.UTM() gives me a plot showing a white area. I'd appreciate it if anyone can tell me how to fix this. Thank you!

Answer 1

I have found two issues. One is an incorrect specification of limits in your call to set_extent, the documentation specifies [x0,x1,y0,y1] should be the input, you seem to have given [x0,y0,x1,y1].
The other issue seems to be a limitation in cartopy, as best I can tell. It looks like projections outside the limits listed in the error message will always fail, and those limits are hardcoded. You can just edit the source (this line in their latest release), changing -2e7 to -4e7, likewise for the upper bound. After these fixes, your plot is generated without issue:
The new set_extent line:
ax.set_extent([32458044.649189778*0.975, 32465287.307457082*1.025,5556418.748046352*0.9, 556415,3.5456742775*1.1],proj)
You may also want to set central_longitude=9.0 in your TransverseMercator, that seems to be what's specified in your shapefile.
I would recommend contacting the developers about this, they might have a good reason for setting those bounds, or they might have a better workaround, or maybe they'll widen the bounds in a later release!
Update
Your bounds also seem to have been set based on only the first of the municipalities:
In [34]: municipalities[0].bounds
Out[34]: (32458044.649189778, 5556418.748046352, 32465287.307457082, 5564153.5456742775)
But the other elements have different bounds. You can get limits flushed to the actual drawing based on min/max values of the bounds of all municipalities.
bnd = np.array([i.bounds for i in municipalities])
x0,x1 = np.min(bnd[:,0]),np.max(bnd[:,2])
y0,y1 = np.min(bnd[:,1]),np.max(bnd[:,3])
ax.set_extent([x0,x1,y0,y1],proj)