sss8, Santiago, 01-04-2012
Rafael H. M. Pereira
Frederico R. B. de Holanda
Valério A. S. de Medeiros
Ana Paula B. G. Barros
The use of SS in urban transport analysis
limits and potentials
Institute of Applied Economic Research
Brazil: overview
Brazil 2010
Population:
Total - 192 milions
Urban -159 milions (83.7%)
5,564 Municipalities
38 cities over 500,00 habitants
16 cities over 1 milion habitants
Brazil: overview
Brasilia 2010
Population figures:
1. Pilot Plan = 209,855
2. Federal District = 2,570,160
3. Conurbation = 3,276,966
4. Direct influence area = 3,451,043
Study aim and scope
 To explore the potentials and limits of applying
SS to the analysis of urban configurations so as
to provide urban environments with greater
transportation efficiency.
 Case study: Federal District (FD - Brazil)
+ its 19 administrative regions
Study aim and scope
2000 - 2009
Increasing motorization ratio (FD)
 Number of Vehicles for 100 Inhabitants
Population
2,70 % a. a.
Car fleet
7,14 % a. a.
45
40
42,8
35
30
25
20
25,9
15
10
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
Source: Denatran and IBGE
Shortcomings (transport studies)
Traditional syntax approach
 Macro-traffic structures (rail, metro) are not captured
 Fails to consider some street features that greatly
influence urban transportation performance






road capacity (number of lanes)
Direction of traffic flows
Pavement conditions
Topographic variations
“Obstacles” (impedance) – i.g. traffic lights, speed bumps, etc
Metric length
 ignores the global extension of the road system as a whole
Shortcomings (transport studies)
“Obstacles” - impedance
(a)
(b)
Same level of Global integration (Rn) = 3,13374
Source: Denatran and IBGE
Shortcomings (transport studies)
Metric length
(a)
5 Km
(b)
Same level of Global integration (Rn) = 3,13374
Source: Denatran and IBGE
10 Km
Material and Methods
Linear regression (Ordinary Least Squares - OLS)*
Urban Configuration  Urban Transport Performance
Configurational Variables:
Average Travel Time spent on
urban trips
- Topological Integration (Rn, R3)
- Mean Depth (Rn, R3 step)
- Topo-geometric measures: Length Wgt and Metric step
* few observations (20)
Material and Methods
 Origin-Destination Survey conducted in the
Federal District (Brazil) in 2000
 Information for every trip on a typical work day in 2000
 Filter: car, utility vehicle and taxi
 *Average travel time for the trips within each AR and the
Federal District (1,000,198 trips)
 20 axial/ segment maps
- Federal District (FD)
- 19 R.A.’s
FD Axial Map
Source: MEDEIROS (2006)
Material and Methods
RA Recanto das Emas
Rn
Rn Length Wgt
To
ta
l(
D
La F )
g
Ta o S
gu ul
at
in
ga
G
ua
Sã B rá
o ras
Se íli
ba a
La sti
go ão
Sa No
r
m
am te
R
ia
ch ba
ia
o
F
Sa un
nt do
a
So Ma
br ria
ad
in
C ho
ru
N
ze
úc
P
iro
le
l
o ana
Ba
l
nd tina
ei
r
C ant
ei
e
lâ
nd
ia
G
R
ec Br am
an az a
lâ
to
da ndi
a
s
Em
a
C
an Pa s
da ran
ng oá
ol
ân
di
a
Results
20
Tempo (min)
RN metric
16
12
8
4
0
Results
Configurational variables
Mean depth with Global topological radius Rn
Performance variable
Average Travel Time (ATT)
Mean depth with Global topological radius Rn (weighted by segment
length)
ATT
Mean depth with Local topological radius R3
ATT
Mean depth with Local topological radius R3 (weighted by segment
length)
ATT
Mean depth (100 meter radius)
ATT
Mean depth (500 meter radius)
ATT
Mean depth (1,000 meter radius)
ATT
Mean depth (5,000 meter radius)
ATT
Mean depth (10,000 meter radius)
ATT
Mean depth (50,000 meter radius)
ATT
Global Integration with topological radius Rn
ATT
Rn Global topo-geometric Integration (weighted by segment length)
ATT
Local Integration with radius R3
ATT
R3 Local topo-geometric Integration (weighted by segment length)
ATT
Statistics
R²
P-value
21,8%
0,0380
38,9%
3,8%
0,0033
0,4094
0,3%
1,0%
1,3%
2,7%
2,1%
14,5%
30,5%
22,0%
0,8060
0,6801
0,6339
0,4848
0,5402
0,0978
0,0115
0,0370
58,0%
8,5%
0,0001
0,2128
0,5%
0,7664
Results
Local Measures
Not significant
Configurational variables
Mean depth with Global topological radius Rn
Performance variable
Average Travel Time (ATT)
Mean depth with Global topological radius Rn (weighted by segment
length)
ATT
Mean depth with Local topological radius R3
ATT
Mean depth with Local topological radius R3 (weighted by segment
length)
ATT
Mean depth (100 meter radius)
ATT
Mean depth (500 meter radius)
ATT
Mean depth (1,000 meter radius)
ATT
Mean depth (5,000 meter radius)
ATT
Mean depth (10,000 meter radius)
ATT
Mean depth (50,000 meter radius)
ATT
Global Integration with topological radius Rn
ATT
Rn Global topo-geometric Integration (weighted by segment length)
ATT
Local Integration with radius R3
ATT
R3 Local topo-geometric Integration (weighted by segment length)
ATT
Statistics
R²
P-value
21,8%
0,0380
38,9%
3,8%
0,0033
0,4094
0,3%
1,0%
1,3%
2,7%
2,1%
14,5%
30,5%
22,0%
0,8060
0,6801
0,6339
0,4848
0,5402
0,0978
0,0115
0,0370
58,0%
8,5%
0,0001
0,2128
0,5%
0,7664
Results
Global Traditional Measures
Sig. < 4%
e
R² = 22%
Configurational variables
Mean depth with Global topological radius Rn
Performance variable
Average Travel Time (ATT)
Mean depth with Global topological radius Rn (weighted by segment
length)
ATT
Mean depth with Local topological radius R3
ATT
Mean depth with Local topological radius R3 (weighted by segment
length)
ATT
Mean depth (100 meter radius)
ATT
Mean depth (500 meter radius)
ATT
Mean depth (1,000 meter radius)
ATT
Mean depth (5,000 meter radius)
ATT
Mean depth (10,000 meter radius)
ATT
Mean depth (50,000 meter radius)
ATT
Global Integration with topological radius Rn
ATT
Rn Global topo-geometric Integration (weighted by segment length)
ATT
Local Integration with radius R3
ATT
R3 Local topo-geometric Integration (weighted by segment length)
ATT
Statistics
R²
P-value
21,8%
0,0380
38,9%
3,8%
0,0033
0,4094
0,3%
1,0%
1,3%
2,7%
2,1%
14,5%
30,5%
22,0%
0,8060
0,6801
0,6339
0,4848
0,5402
0,0978
0,0115
0,0370
58,0%
8,5%
0,0001
0,2128
0,5%
0,7664
Results
Topo-geometric measures
Improved
results with
larger
radius
Melhor estatística
quanto
maior
o Raio de ação
Configurational variables
Mean depth with Global topological radius Rn
Performance variable
Average Travel Time (ATT)
Mean depth with Global topological radius Rn (weighted by segment
length)
ATT
Mean depth with Local topological radius R3
ATT
Mean depth with Local topological radius R3 (weighted by segment
length)
ATT
Mean depth (100 meter radius)
ATT
Mean depth (500 meter radius)
ATT
Mean depth (1,000 meter radius)
ATT
Mean depth (5,000 meter radius)
ATT
Mean depth (10,000 meter radius)
ATT
Mean depth (50,000 meter radius)
ATT
Global Integration with topological radius Rn
ATT
Rn Global topo-geometric Integration (weighted by segment length)
ATT
Local Integration with radius R3
ATT
R3 Local topo-geometric Integration (weighted by segment length)
ATT
Statistics
R²
P-value
21,8%
0,0380
38,9%
3,8%
0,0033
0,4094
0,3%
1,0%
1,3%
2,7%
2,1%
14,5%
30,5%
22,0%
0,8060
0,6801
0,6339
0,4848
0,5402
0,0978
0,0115
0,0370
58,0%
8,5%
0,0001
0,2128
0,5%
0,7664
Results
Topo-geometric measures
Improved
results with
larger
radius
Melhor estatística
quanto
maior
o Raio de ação
Configurational variables
Mean depth with Global topological radius Rn
Performance variable
Average Travel Time (ATT)
Mean depth with Global topological radius Rn (weighted by segment
length)
ATT
Mean depth with Local topological radius R3
ATT
Mean depth with Local topological radius R3 (weighted by segment
length)
ATT
Mean depth (100 meter radius)
ATT
Mean depth (500 meter radius)
ATT
Mean depth (1,000 meter radius)
ATT
Mean depth (5,000 meter radius)
ATT
Mean depth (10,000 meter radius)
ATT
Mean depth (50,000 meter radius)
ATT
Global Integration with topological radius Rn
ATT
Rn Global topo-geometric Integration (weighted by segment length)
ATT
Local Integration with radius R3
ATT
R3 Local topo-geometric Integration (weighted by segment length)
ATT
Statistics
R²
P-value
21,8%
0,0380
38,9%
3,8%
0,0033
0,4094
0,3%
1,0%
1,3%
2,7%
2,1%
14,5%
30,5%
22,0%
0,8060
0,6801
0,6339
0,4848
0,5402
0,0978
0,0115
0,0370
58,0%
8,5%
0,0001
0,2128
0,5%
0,7664
Final Remarks
Future Studies
 Test other configurational measures
 Replication in other metropolitan areas
 Method: multivariate and/or multilevel analyses
Final Remarks
Regarding urban transport performance,
results suggest that:
 Global characteristics (rather than local ) are important
 Traditional topological measures do not help much…
 Topo-geometric measures play important role
 More integrated and compact road systems (in topological
and geometrical terms) tend to provide a more efficient
urban environment in terms of time spent in car trips
Less environmentally damaging in terms of energy use
and pollutant emissions
sss8, Santiago, 01-04-2012
Email
fredholanda44@gmail.com
rafael.pereira@ipea.gov.br
Thank you.