gpd.to_file(): Z dimension not saved with ESRI driver

Issue

The Z dimension of geometries is not preserved when using gpd.to_file() using the ESRI driver.

Example

import geopandas as gpd
from shapely.geometry import Point

point = Point(0, 0, 500)
gdf = gpd.GeoDataFrame(geometry=[point])
print(gdf)

          geometry
0  POINT Z (0 0 500)

gdf.to_file('./foo')
# Try to read file back
gdf2 = gpd.read_file('./foo')
print(gdf2)

   FID     geometry
0    0  POINT (0 0)

Expected Behavior

I’d expect the Z dimension to be included.

[jorisvandenbossche]

Although I don’t fully understand how it works, as here (https://github.com/Toblerity/Fiona/issues/465) is stated that ‘3D Point’ should now be an alias for ‘Point’

[jorisvandenbossche]

I suppose the error is in inferring the schema (https://github.com/geopandas/geopandas/blob/master/geopandas/io/file.py#L111), in that it needs special casing for 3D data.

If you manually specify the schema, it seems to work:

In [167]: point = Point(0, 0, 500)
     ...: gdf = gpd.GeoDataFrame({'a': [1]}, geometry=[point])

In [168]: gdf
   a           geometry
0  1  POINT Z (0 0 500)

In [170]: gpd.io.file.infer_schema(gdf)
Out[170]: {'geometry': 'Point', 'properties': OrderedDict([('a', 'int')])}

In [175]: schema = {'geometry': '3D Point', 'properties': {'a': 'int'}}

In [176]: gdf.to_file("__test3D.shp", schema=schema)

In [177]: gpd.read_file("__test3D.shp")
   a           geometry
0  1  POINT Z (0 0 500)