Converting column with LineString geometries to np.array

Question

I have a GeoDataFrame of a street network with the geometry of each row as a LineString. Say something like:

 type id tags geometry lanes bridge name highway
0 way 34953479 {'access': 'private', 'highway': 'service', 'i... LINESTRING (282025.712 6242798.110, 282029.830... 2.0 None Engineering Way service
1 way 34953670 {'access': 'private', 'highway': 'service', 'i... LINESTRING (281935.657 6242783.921, 281936.328... 2.0 None Science Way service

I want each vertex as a row in a np.array.

With

pt = [ np.array(line.coords) for line in rd.geometry ]
print(pt)

I get:

[array([[ 282025.71154782, 6242798.11011274],[ 282029.82990925, 6242798.7208016 ],[ 282033.94827067, 6242799.33149047],...

which is a list of np.array.

pt[0] is: array([[ 282025.71154782, 6242798.11011274],[ 282029.82990925, 6242798.7208016 ],[ 282033.94827067, 6242799.33149047],...

pt[1] is: array([[ 281935.65690061, 6242783.92136821],[ 281936.32837562, 6242780.16379144],[ 281936.99985063, 6242776.40621467],...

I want one np.array with the coordinates stacked.

How do I extract/convert a gdf of LineString geometry into one np.array where the coordinates are stacked

array([[ x, y],[ x, y ],[ x, y],...

Answer 1

There is a native NumPy's method concatenate():

Join a sequence of arrays along an existing axis.

import numpy as np
import geopandas as gpd
from shapely.geometry import LineString
d = {
 'attr1': ['line1', 'line2'],
 'geometry': [
 LineString([
 (350630.319649, 5333438.389906),
 (410389.192817, 5298093.202727),
 (466139.848883, 5303194.569949),
 (500391.885943, 5310482.237409),
 (522254.888322, 5300643.886338),
 (547761.72443, 5287890.468284),
 (594038.412798, 5288254.851657),
 (620638.399026, 5307931.553798),
 (643230.16815, 5355301.392285)
 ]),
 LineString([
 (465059.389802, 5337643.246403),
 (428840.7437, 5395427.193238)]
 )]
 }
gdf = gpd.GeoDataFrame(d, crs="EPSG:25832")
pt = list(gdf.geometry) #[<shapely.geometry.linestring.LineString object at 0x000001F3F082FD00>, <shapely.geometry.linestring.LineString object at 0x000001F3F0822CA0>]
# pt = gdf.geometry.tolist() 
pt_array = np.concatenate(pt)
print(type(pt_array))
print(pt_array)

The above code will result in:

<class 'numpy.ndarray'>
[[ 350630.319649 5333438.389906]
 [ 410389.192817 5298093.202727]
 [ 466139.848883 5303194.569949]
 [ 500391.885943 5310482.237409]
 [ 522254.888322 5300643.886338]
 [ 547761.72443 5287890.468284]
 [ 594038.412798 5288254.851657]
 [ 620638.399026 5307931.553798]
 [ 643230.16815 5355301.392285]
 [ 465059.389802 5337643.246403]
 [ 428840.7437 5395427.193238]]

• swiss_knight: As of 2024年09月16日 with numpy 1.26.4: pt_array = np.concatenate(pt) raises ValueError: zero-dimensional arrays cannot be concatenated.