Example of using the `Streetscape` class¶

The class is designed for morphological streetscape analysis, focusing on generating nd analyzing streetscape measures based on sight points and sightlines. It attempts to capture the space from the pedestrian point of view.

For details and citation, refer to Araldi, A., Fleischmann, M., Fusco, G., Novotný, M., 2025. Streetscape morphometrics: Expanding momepy to analyze urban form from the street point of view. SoftwareX 31, 102242. https://doi.org/10.1016/j.softx.2025.102242.

import geopandas as gpd
import momepy
import numpy as np
import osmnx as ox
import rioxarray

Read all the data. Only streets and buildings are required.

streets = gpd.read_file(
    momepy.datasets.get_path("bubenec"), layer="streets"
).to_crs(5514)
buildings = gpd.read_file(
    momepy.datasets.get_path("bubenec"), layer="buildings"
).to_crs(5514)
plots = gpd.read_file(
    momepy.datasets.get_path("bubenec"), layer="plots"
).to_crs(5514)
dtm = rioxarray.open_rasterio(momepy.datasets.get_path("bubenec"), layer="dtm")

Mimic data on building category and height.

buildings["category"] = np.random.randint(0, 6, len(buildings))
buildings["height"] = np.random.randint(12, 30, len(buildings))

Initiate the class. This will dricectly compute builk of the sightline indicators based on streets and buildings.

sc = momepy.Streetscape(
    streets, buildings, category_col="category", height_col="height"
)

If you have plots and DTM, you can use two additional methods to compute additional variables.

sc.compute_plots(plots)
sc.compute_slope(dtm)

The resulting data can be extracted either on a street level:

street_df = sc.street_level()
street_df.head()

	N	n_l	n_r	left_os	right_os	os	left_os_std	right_os_std	os_std	left_os_mad	...	plot_WP_ratio	left_plot_WP_ratio	right_plot_WP_ratio	left_plot_seq_sb_index	right_plot_seq_sb_index	slope_degree	slope_percent	n_slopes	slope_valid	geometry
street_index
0	57	28	38	33.607135	38.163914	71.771049	14.794107	12.059253	13.436283	13.873261	...	0.170662	0.156054	0.184501	{96, 101, 103, 104, 82, 308, 87, 89, 92}	{102, 105, 309, 310, 311, 383}	2.809318	0.049073	57	True	LINESTRING (-743681.992 -1040957.169, -743809....
1	22	0	19	50.000000	17.497487	67.497487	0.000000	14.015687	9.794670	0.000000	...	0.255912	0.284263	0.230261	{242, 243}	{240, 238, 239}	0.166605	0.002908	22	True	LINESTRING (-743916.081 -1041162.952, -743899....
2	43	36	21	13.907200	35.792031	49.699231	16.207022	15.188099	15.613165	11.913767	...	0.200568	0.221176	0.181546	{78, 79, 80, 81, 82, 308}	{83, 87, 311, 89, 90}	1.122029	0.019587	43	True	LINESTRING (-743689.806 -1041115.822, -743698....
3	24	0	0	50.000000	48.865531	98.865531	0.000000	4.101247	2.869002	0.000000	...	0.156703	0.060853	0.248559	{380, 381, 382}	{385, 103}	2.735747	0.047794	24	True	LINESTRING (-743618.342 -1040934.607, -743621....
4	15	0	0	50.000000	50.000000	100.000000	0.000000	0.000000	0.000000	0.000000	...	0.068937	0.050164	0.058499	{390, 391, 398, 400, 401, 404}	{388, 391, 401, 403, 404}	1.505840	0.026294	15	True	LINESTRING (-743701.515 -1040870.813, -743693....

5 rows × 106 columns

It is a GeoDataFrame, so you can directly plot it.

ax = street_df.plot("os", figsize=(12, 12), legend=True)
buildings.plot(ax=ax).set_axis_off()

../_images/5a05bf09d22aeeb9b56d1bf4ed2e2205cbb746d9a21131fa6ed09730239012b1.png

Or for all individual sightline points.

point_df = sc.point_level()
point_df.head(10)

	geometry	left_os_count	left_os	left_sb_count	left_sb	left_h	left_hw	left_bc	right_os_count	right_os	...	right_plot_seq_sb_index	os_count	os	sb_count	sb	h	hw	bc	plot_seq_sb	plot_seq_sb_depth
street_index
0	POINT (-743682.367 -1040957.5)	1	50.000000	0	NaN	NaN	NaN	0.000000	1	50.000000	...	{383}	2.0	50.000000	0.0	NaN	NaN	NaN	0.000000	14.568200	19.240118
0	POINT (-743684.627 -1040959.485)	1	50.000000	0	NaN	NaN	NaN	0.000000	1	50.000000	...	{383}	2.0	50.000000	0.0	NaN	NaN	NaN	0.000000	10.090066	18.011884
0	POINT (-743686.886 -1040961.471)	1	50.000000	0	NaN	NaN	NaN	0.000000	1	27.593446	...	{383}	2.0	38.796723	1.0	27.593446	14.0	0.507367	2.491801	10.100149	15.658901
0	POINT (-743689.145 -1040963.457)	1	50.000000	0	NaN	NaN	NaN	0.000000	1	26.229691	...	{383}	2.0	38.114846	1.0	26.229691	14.0	0.533746	5.604209	9.996735	27.216748
0	POINT (-743691.405 -1040965.443)	1	50.000000	0	NaN	NaN	NaN	0.000000	1	25.412462	...	{383}	2.0	37.706231	1.0	25.412462	14.0	0.550911	6.825341	9.996961	32.570991
0	POINT (-743693.664 -1040967.428)	1	23.978541	0	NaN	NaN	NaN	6.557431	1	25.198986	...	{383}	2.0	24.588764	1.0	25.198986	14.0	0.555578	11.811294	9.997186	29.171224
0	POINT (-743695.923 -1040969.414)	1	14.720375	0	NaN	NaN	NaN	15.763951	1	26.764536	...	{383}	2.0	20.742455	1.0	26.764536	14.0	0.523080	16.722367	9.997412	25.070210
0	POINT (-743698.182 -1040971.4)	1	14.705134	0	NaN	NaN	NaN	16.541363	1	28.184169	...	{383}	2.0	21.444652	1.0	28.184169	14.0	0.496733	15.669026	9.997637	21.235336
0	POINT (-743700.442 -1040973.386)	1	14.689893	0	NaN	NaN	NaN	19.812513	1	29.961410	...	{383}	2.0	22.325652	1.0	29.961410	14.0	0.467268	13.593521	9.997863	24.073478
0	POINT (-743702.701 -1040975.371)	1	14.674652	0	NaN	NaN	NaN	19.807911	1	31.357056	...	{383}	2.0	23.015854	1.0	31.357056	14.0	0.446470	13.355549	9.998088	25.161555

10 rows × 34 columns

Again, it is a GeoDataFrame, this time with point geometry.

ax = point_df.plot("os", figsize=(12, 12), legend=True, markersize=5)
buildings.plot(ax=ax).set_axis_off()

../_images/d84909eb71104620a4248202799df8a2b175af6ddb6dc63b6ee4ec47361ec370.png

Using OpenStreetMap data¶

You can also use any other data. If you fetch buildings and streets from OSM, you can measure a subset of characters depending on those.

gdf = ox.features_from_place("Kahla, Germany", tags={"building": True})
buildings = ox.projection.project_gdf(gdf).reset_index()

streets_graph = ox.graph_from_place("Kahla, Germany", network_type="drive")
streets_graph = ox.projection.project_graph(streets_graph)
edges = ox.graph_to_gdfs(
    streets_graph,
    nodes=False,
    edges=True,
    node_geometry=False,
    fill_edge_geometry=True,
).reset_index(drop=True)

Initiate the class. This time without additional arguments as OSM does not contain category or reliable height.

sc = momepy.Streetscape(edges, buildings)

The measurememnt happens on class initialisation. Now you can simply extract it either on a street level:

street_df = sc.street_level()

ax = street_df.plot("os", figsize=(12, 12), legend=True)
buildings.plot(ax=ax, color="lightgray").set_axis_off()

../_images/245b5c6f3570d83f2826c480b4116be5231c367c90c6c2d7a3ec59d8f58468c7.png

Or on a point level, as before.

point_df = sc.point_level()

ax = point_df.plot("os", figsize=(12, 12), legend=True, markersize=0.1)
buildings.plot(ax=ax, color="lightgray").set_axis_off()

../_images/6cf5f42b0be5217ca3dbad85e7bf2769df44c4e64917392dd09fc5550fee6508.png

Example of using the Streetscape class¶

Using OpenStreetMap data¶

Example of using the `Streetscape` class¶