Ubud

Gianyar · Small town

A village strung between two rivers. Most of its walking network is tagged path — rice-field and ridge tracks — rather than gang between houses, so the drive/walk gap here is as much its landscape as its settlement.

Drive

17.2 km · 60 intersections/km²

Fig. 1 · r = 800 m — The drivable network inside the sampling disc. Uniform ink, no road-class hierarchy, no basemap: the shape is the subject.

Drive H 2.637 · φ 0.68

Walk

27.4 km · 113 intersections/km²

Fig. 2 · r = 800 m — The walkable network at the same centre, the same disc, the same extract. Every difference from the drawing beside it is the tag mapping, not a different place.

Walk H 2.751 · φ 0.61

Delta

+10.3

km of network reachable only on foot · 37.4% of the walking network

+53
intersections per km²: −3.6%
Dead-end: +0.114
ΔH · orientation entropy

Both roses overlaid

Drive H 2.637 · φ 0.68Walk H 2.751 · φ 0.61

Drive
Walk
Both

Two readings, one table

Hairlines: every site in the set, both modes. Marks: this site. A position, not a value — neither end is better.

Ubud: driving and walking network metrics, their difference, and their position among 33 sites.
Metric	Drive	Walk	Delta	Position among 33 sites
Orientation entropynat	2.637	2.751	+0.114	Drive: position 2 of 33; Walk: position 2 of 33, when the set is sorted by this metric, smallest first.
H / H max	0.736	0.768	+0.032	Drive: position 2 of 33; Walk: position 2 of 33, when the set is sorted by this metric, smallest first.
φ orientation-order	0.676	0.614	−0.062	Drive: position 32 of 33; Walk: position 32 of 33, when the set is sorted by this metric, smallest first.
Circuitysampled	1.871	1.729	−0.142	Drive: position 32 of 33; Walk: position 31 of 33, when the set is sorted by this metric, smallest first.
Average node degree	2.07	2.12	+0.04	Drive: position 1 of 33; Walk: position 2 of 33, when the set is sorted by this metric, smallest first.
Four-way	3.3%	3.7%	+0.4%	Drive: position 6 of 33; Walk: position 2 of 33, when the set is sorted by this metric, smallest first.
Dead-end	45.3%	41.7%	−3.6%	Drive: position 32 of 33; Walk: position 30 of 33, when the set is sorted by this metric, smallest first.
Intersection density	60 /km²	113 /km²	+53 /km²	Drive: position 3 of 33; Walk: position 5 of 33, when the set is sorted by this metric, smallest first.
Median segment length	50 m	45 m	−6 m	Drive: position 29 of 33; Walk: position 29 of 33, when the set is sorted by this metric, smallest first.
Network length	17.2 km	27.4 km	+10.2 km	Drive: position 5 of 33; Walk: position 7 of 33, when the set is sorted by this metric, smallest first.

H is in nats — the natural-log unit of information. That is why its maximum is ln 36 ≈ 3.584 rather than log₂ 36.

Circuity is an estimate: network distance over straight-line distance, averaged across a seeded sample of node pairs rather than across all of them. Drive: 470 pairs sampled · 18 pairs unreachable.Walk: 454 pairs sampled · 36 pairs unreachable. Pairs with no route between them at all are excluded from the mean. A large count means this network is fragmented — a signal about the data, not about the place.

Each street segment’s bearing goes into one of 36 bins of 10°, weighted by length: a 500 m road counts ten times a 50 m gang. Bar length is the share of network length running that way, not a count of segments. The rose is always 180°-symmetric, because one segment traversed both ways differs by exactly 180°. Boeing 2019 §3

Fig. 3 · r = 800 m

What walking adds

Ink: the edges that exist only in the walking network. Hairline: the network both modes share. Membership is decided by the tag mapping — an edge counts as walk-only when the OSM way it came from is not admitted to the driving network.

Walk-only length: 10.3 km
Share of walking network: 37.4%

Where these numbers come from

Drive

Nodes: 245
Edges: 254
Intersections (degree ≥ 3): 121
Disc area: 2.01 km²
Intersection density: 60 /km²
Rose weighted length: 34.3 km
Per-edge circuity: 1.029

Walk

Nodes: 458
Edges: 485
Intersections (degree ≥ 3): 228
Disc area: 2.01 km²
Intersection density: 113 /km²
Rose weighted length: 54.8 km
Per-edge circuity: 1.032

The intermediate values the column above is built from. Intersection density is the intersection count over the disc area; the weighted length is the denominator that turns a rose bar into a share. Two different circuity measures: the per-edge one compares each segment against its own chord, the sampled one compares distances between junctions. The first measures how much streets bend, the second how far out of your way the network takes you.

Rose as a table — 36 bins — Ubud

Share of network length per 10° bearing bin, for drive and walk.
Bin	Bearing range	P — Drive	P — Walk	−P·ln P	−P·ln P
00	355°–5°	2.41%	3.12%	0.0896	0.1082
01	5°–15°	15.26%	11.76%	0.2869	0.2517
02	15°–25°	7.43%	6.23%	0.1931	0.1730
03	25°–35°	0.83%	1.02%	0.0399	0.0468
04	35°–45°	0.18%	0.30%	0.0116	0.0173
05	45°–55°	0.10%	0.20%	0.0069	0.0127
06	55°–65°	0.38%	0.29%	0.0212	0.0170
07	65°–75°	0.06%	0.14%	0.0045	0.0091
08	75°–85°	0.33%	0.34%	0.0191	0.0192
09	85°–95°	2.75%	2.76%	0.0987	0.0991
10	95°–105°	11.53%	13.72%	0.2491	0.2725
11	105°–115°	6.34%	6.88%	0.1750	0.1841
12	115°–125°	1.26%	1.16%	0.0551	0.0515
13	125°–135°	0.36%	0.43%	0.0203	0.0232
14	135°–145°	0.19%	0.25%	0.0120	0.0150
15	145°–155°	0.14%	0.31%	0.0089	0.0180
16	155°–165°	0.11%	0.48%	0.0075	0.0258
17	165°–175°	0.33%	0.62%	0.0189	0.0314
18	175°–185°	2.41%	3.12%	0.0896	0.1082
19	185°–195°	15.26%	11.76%	0.2869	0.2517
20	195°–205°	7.43%	6.23%	0.1931	0.1730
21	205°–215°	0.83%	1.02%	0.0399	0.0468
22	215°–225°	0.18%	0.30%	0.0116	0.0173
23	225°–235°	0.10%	0.20%	0.0069	0.0127
24	235°–245°	0.38%	0.29%	0.0212	0.0170
25	245°–255°	0.06%	0.14%	0.0045	0.0091
26	255°–265°	0.33%	0.34%	0.0191	0.0192
27	265°–275°	2.75%	2.76%	0.0987	0.0991
28	275°–285°	11.53%	13.72%	0.2491	0.2725
29	285°–295°	6.34%	6.88%	0.1750	0.1841
30	295°–305°	1.26%	1.16%	0.0551	0.0515
31	305°–315°	0.36%	0.43%	0.0203	0.0232
32	315°–325°	0.19%	0.25%	0.0120	0.0150
33	325°–335°	0.14%	0.31%	0.0089	0.0180
34	335°–345°	0.11%	0.48%	0.0075	0.0258
35	345°–355°	0.33%	0.62%	0.0189	0.0314
Sum = H				2.637	2.751

The last column is each bin’s term in H = −Σ P(i)·ln P(i). Add the thirty-six of them and you get the H printed under the rose. An empty bin contributes zero.

Note

Ubud’s figures stand out in both directions: low entropy and high φ, meaning its streets run along a small number of repeated bearings — more regularly than several of the planned sites in this set. That is what is measured, and that is all of it.

What bounds the reading is what its walking network is made of. Most of it is tagged path — rice-field and ridge tracks rather than gang between houses. Footway coverage counts both the same, so the drive/walk gap here is partly landscape rather than settlement alone. Its comparison with an urban kampung has to be read with that in mind.

Download this site’s data (JSON, ODbL)

Data jalan © OpenStreetMap contributors, ODbL 1.0. Basis data turunan ini ditawarkan dengan lisensi yang sama. · Boeing, G. 2019. Urban Spatial Order: Street Network Orientation, Configuration, and Entropy. Applied Network Science 4(1):67.