This document summarizes research on the relationship between street network characteristics and road safety outcomes. The research found that cities with more dense, connected street networks with smaller blocks and more intersections had significantly fewer fatal and severe crashes compared to cities with less dense, more disconnected street patterns. Statistical analysis of crash data from 24 California cities showed that higher intersection density, higher connectivity, and other measures of a fine-grained network are strongly associated with fewer crashes and increased safety.
Z Score,T Score, Percential Rank and Box Plot Graph
Marshall - CNU Street Network Presentation
1. Street Netw orks,
Road Safety & Sustainability
W esley M arshall, P.E.
Norm W Garrick, PhD
an .
Center for Transportation & Urban
Planning
University of Connecticut
Sustainable Transportation Networks
Congress for the NewUrbanism XVII
June 13, 2009
Denver, Colorado
4. Fatalities: 8
VMT: 2 million miles
Fatalities per million VMT = 4
Population: 50,000
Fatalities per 100k pop. = 16
Fatalities: 8
VMT: 1 million miles
Fatalities per million VMT = 8
Population: 100,000
Population per 100k pop. = 8
5. 3,00
24 x
)
0
ns
io
i ll
(b
T
VM
2,000
20
15
10 1,000
Fatalities per million VMT
5 Population
2.5 x
00 0
192 193 194 195 196 197 198 199 200
5 5 5 5 5 5 5 5 5
7. Road Safety in the U.S.
30 60,000
)
ns
Fatalities per 100,000 population
io
Total No. of Fatalities
i ll
25
(b
T
VM
20 40,00
0
Fatalities per million VMT
15
10 20,00
0
5 Population
0 0
192 193 194 195 196 197 198 199 200
5 5 5 5 5 5 5 5 5
8. Road Fatalities per 100,000 Population by Country
United States
(Source: Organisation for Economic Co-operation and Development, OECD)
9. International Road Safety
Why is the U.S. falling behind
the rest of the world when it
comes to safety in the
transportation system?
(Source: Organisation for Economic Co-operation and Development, OECD)
10. Road Safety
When it comes to trying
to make our roads safer…
The focus tends to be
on finding the most
problematic locations
http://imgs.sfgate.com/c/pictures/2007/01/03/ba_octavia28_009_rad.jpg
and fixing them
www.streetsblog.org
11. California City Study
24 medium-sized
California cities
Cities selected to
represent a range
of traffic safety levels
Geographically
diverse with locally
generated traffic
12.
13. CALIFORNIA CITY COMPARISON
Less Safe Safer
Cities Cities
Population 59,845 65,719
(average)
Road Fatalities 771 257
3.0
(total over 11 years)
5.8 sk = 1.9
e Ri
per city per year
Fatalities per Re lativ
100,000 pop. 9.8 3.3
(per year)
Non-Highway
Road Fatalities 676 200
3.4
(total over 11 years)
sk = 1.5
e Ri
per city per year 5.1
Non-HW Fatalities Re lativ
per 100,000 pop. 8.6 2.5
(per year)
14. Why are these places
so different in terms of
safety outcomes?
15. CALIFORNIA CITY COMPARISON
Less Safe Safer
Cities Cities
Population
Density 2,673 per sq. mi. 5,736 per sq. mi.
Mode Share
Driving 95.8% 84.1%
Walking 1.7% 5.4%
Biking 0.7% 4.1%
Transit 1.7% 6.6%
Avg. Year of
Incorporation 1932 1895
16. Davis Intersection Vehicle Mode % Fatal or
Density Share Severe Crashes
Pre 1940 211 / sq. mi 40.6% 1.6%
1940s 122 58.9% 3.9%
1950s 169 63.0% 2.6%
1960s 172 64.7% 2.3%
1970s 132 81.3% 3.0%
1980s+ 111 85.9% 3.0%
1940
1950
2000
1990
1980
1970
1960
19. How Do We Characterize Street Networks?
Connected
Dense Link to Node Ratio
Intersection Density
Gridded
Road Density Hierarchical
Block Size
Patterns
20. Characterizing Street Networks
There are 3 fundamental items of interest
in characterizing a street network…
i. Street Connectivity
ii. Street Network Density
iii. Street Patterns
24. Neighborhood Micro Network Citywide Macroscopic Network
Tree
Linear Grid
Tributary Radial
Tree
Grid
Adapted from Stephen Marshall, Streets & Patterns
26. NETWORK COMPARISON
Tree
Linear Grid
Tributary Radial
Tree
Avg. Year of
Development
1966 1965 1974 1966
Grid
Avg. Year of
Development 1950 Pre 1940 Pre 1940
30. STREET NETWORK COMPARISON
Safer Less Safe
Difference
Cities Cities
Intersection
106 per sq. mi. 63 per sq. mi. -40.6%
Density
Dead End
32 per sq. mi. 23 per sq mi. -28.1%
Density
% Dead Ends 23.2% 26.7%
Macro
7.5 per sq. mi. 4.9 per sq. mi. -34.7%
Node Density
% Major Nodes 7.1% 7.8%
Connectivity Measures
Link to Node Ratio 1.34 1.29 -3.7%
Connected Node Ratio 0.76 0.73 -4.0%
31. ROAD SAFETY COMPARISON
Safer Less Safe
Difference
Cities Cities
Fatal Crashes
3.3 per year 9.8 per year 197.0%
per 100,000 pop.
Severe Crashes
16.4 per year 18.4 per year 12.2%
per 100,000 pop.
Severity Risk 1.9% 68.4%
(% Fatal or Severe)
3.2%
Macro Road Fatal or
Severe per 100k pop. 16.4 per year 17.4 per year 6.1%
Severity Risk
(% Fatal or Severe)
1.9% 3.2% 68.4%
Micro Road Fatal or
Severe per 100k pop. 2.7 per year 4.6 per year 70.4%
Severity Risk
(% Fatal or Severe) 1.5% 3.1% 58.8%
32. SAFER CITIES – NETWORK DENSITY
Network
Density
Comparison
1 Sq. Mile
Grid Size 9x9 12x12 15x15
Block Length 660’ 480’ 375’
Intersection
Density 81 144 225
< 81 81-144 144-225 225+
Risk of
Injury 41.0% 38.5% 39.1% 37.7%
(non-highway)
Risk of
Severe Inj. 3.3% 1.9% 1.8% 1.5%
(non-highway)
Risk of
Fatality 0.5% 0.3% 0.2% 0.2%
(non-highway)
34. Statistical Analysis
What do we want to know?
How are street network
measures associated
(correlated) with
road safety outcomes?
35. Statistical Analysis
Built crash prediction models using
a generalized linear regression
Response Variables:
Model 1: Total No. of Crashes
Model 2: Total No. of Severe Injury Crashes
Model 3: Total No. of Fatal Crashes
36. Crash Model Results
Model 1 Model 2 Model 2 Model 3 Model 3 Model 4
Model 1
Severe Injury Severe
Any Injury Fatal & Fatal Crashes
Total Crashes Crashes
Total
Crashes Fatal Crashes
Variables
Variables Crashes Injury Crashes
Street NetworkNetwork Measures
Street
Street Pattern Pattern Type (categorical)
Street Type (categorical)
Intersection Density
Intersection Density
Macro-Intermediate Intersection Density
Macro-Intermediate Intersection Density
Dead End Node Density
Dead End Node Density
Link to Node Ratio
Link to Node Ratio
Curvilinear (0, 1)
Macro Road Characteristics
Street Level Data
Avg. # of Lanes
Avg. # of Lanes
Avg. Width of Outside Shoulder
Avg. Width of Outside Shoulder
Raised Median Median (0, 1)
Raised (0, 1)
Painted Painted Median (0, 1)
Median (0, 1)
% of Macro Road Length Length with On-Street Parking
% of Macro Road with On-Street Parking
% of Macro Road Length Length with Bike Lanes
% of Macro Road with Bike Lanes
% of Macro Road Length Length with Curbs
% of Macro Road with Curbs
Exposure
Exposure
VMT VMT
Proxy for Activity
Proxy for Activity
Miscellaneous
Miscellaneous
Distance from City Center
Distance from City Center
Avg. Income
Avg. Income
Adjacent Limited Access Highway
Mixed Land Uses
= Significant with Positive Association (More Crashes)
= Significant with Negative Association (Fewer Crashes)
= Not Significant
37. FULL NETWORK CRASH MODELS Total Crashes Severe Crashes Total Fatal Crashes
(Model 1) (Model 2) (Model 3)
% Change % Change % Change
Intersection Density
81 14.15% 20.05% 53.75%
144 (reference value) - - -
225 -15.64% -20.94% -42.48%
324 -31.48% -40.67% -70.74%
Link to Node Ratio
1.1 -14.29% -12.20% -28.21%
1.25 (reference value) - - -
1.4 16.67% 13.90% 39.29%
1.55 36.13% 29.73% 94.02%
Total No. of Lanes on Macro Roads
2 (reference value) - - -
4 65.17% 33.96% 34.15%
6 172.81% 79.46% 79.95%
Distance from City Center (miles)
0 41.48% 23.71% -12.86%
1 18.95% 11.23% -6.65%
2 (reference value) - - -
3 -15.93% -10.09% 7.12%
4 -29.32% -19.17% 14.75%
% of Macro Road Length with On-Street Parking
0% (reference value) - - -
50% 18.26% 19.49% -
100% 39.86% 42.79% -
% of Macro Road Length with Bike Lanes
0% (reference value) - - -
50% - - -14.29%
100% - - -26.53%
38. CITYWIDE MACRO CRASH MODELS Total Crashes Severe Crashes Total Fatal Crashes
(Model 4) (Model 5) (Model 6)
% Change % Change % Change
Intersection Density
81 7.85% 13.43% 39.52%
144 (reference value) - - -
225 -9.26% -14.96% -34.83%
324 -19.43% -30.23% -61.38%
Link to Node Ratio
1.1 - - -24.30%
1.25 (reference value) - - -
1.4 - - 32.10%
1.55 - - 74.50%
Total No. of Lanes on Macro Roads
2 -45.82% -30.54% -23.08%
4 (reference value) - - -
6 84.56% 43.96% 30.01%
Distance from City Center (miles)
0 51.26% 35.28% -12.88%
1 22.99% 16.31% -6.66%
2 (reference value) - - -
3 -18.69% -14.02% 7.14%
4 -33.89% -26.08% 14.78%
% of Macro Road Length with On-Street Parking
0% -18.12% -15.10% 19.93%
50% (reference value) - - -
100% 22.13% 17.78% -16.62%
% of Macro Road Length with Bike Lanes
0% - - 20.42%
50% (reference value) - - -
39. LT TT RT GT
Intersection Density 90 140 130 160
Link to Node Ratio 1.09 1.15 1.18 1.24
Expected Total Crashes 290 202 275 213
Expected Severe Injury Crashes 5.5 3.8 4.1 5.2
Expected Fatal Crashes 1.2 0.9 1.1 1.0
(Non-HW Crashes)
LG TG RG GG
Intersection Density - 225 289 265
Link to Node Ratio - 1.34 1.37 1.40
Expected Total Crashes - 191 211 209
Expected Severe Injury Crashes - 3.1 3.3 3.1
Expected Fatal Crashes - 0.8 0.6 0.7
41. TT
MODE CHOICE MODEL Transit Mode Pedestrian Biking Mode Automobile
Share Mode Share Share Mode Share
(all other variables (all other variables (all other variables (all other variables
held at mean) held at mean) held at mean) held at mean)
Variables
BASELINE BY STREET PATTERN TYPE 3.66% 2.28% 1.71% 92.35%
Intersection Density
81 3.81% 1.94% 1.29% 92.96%
144 3.65% 2.30% 1.74% 92.31%
225 3.44% 2.85% 2.56% 91.15%
324 3.18% 3.69% 4.06% 89.07%
Link to Node Ratio
1.1 3.42% 2.40% 1.74% 92.44%
1.25 4.17% 2.05% 1.65% 92.13%
1.4 5.06% 1.75% 1.55% 91.63%
1.55 6.14% 1.50% 1.46% 90.91%
Total No. of Lanes on Macro Roads
2 4.48% 2.34% 1.72% 91.46%
4 2.80% 2.19% 1.68% 93.32%
6 1.74% 2.05% 1.63% 94.58%
Distance from City Center (miles)
0 3.30% 4.03% 3.18% 89.49%
1 3.48% 3.06% 2.36% 91.11%
2 3.65% 2.31% 1.74% 92.30%
3 3.82% 1.74% 1.27% 93.17%
4 3.98% 1.31% 0.93% 93.78%
42. TG
MODE
strian CHOICE MODEL
Biking Mode Automobile Transit Mode
Transit Mode Pedestrian
Pedestrian Biking Mode
Biking Mode Automobile
Automobile
Share Share Mode Share Share
Share Mode Share
Mode Share Share
Share Mode Share
Mode Share
variables (all other variables (all other variables (all other variables (all other variables (all other variables (all other variables
(all other variables (all other variables (all other variables (all other variables
mean) held at mean) held at mean) held at mean)
held at mean) held at mean)
held at mean) held at mean)
held at mean) held at mean)
held at mean)
Variables
28%
BASELINE BY STREET PATTERN 92.35%
1.71% TYPE 3.66%
4.18% 2.28%
3.93% 1.71%
3.39% 92.35%
88.51%
Intersection Density
94% 81 1.29% 92.96% 3.81%
5.94% 1.94%
4.69% 1.29%
2.72% 92.96%
86.64%
30% 1441.74% 92.31% 3.65%
5.10% 2.30%
4.35% 1.74%
3.00% 92.31%
87.55%
85% 2252.56% 91.15% 3.44%
4.19% 2.85%
3.93% 2.56%
3.38% 91.15%
88.50%
69% 3244.06% 89.07% 3.18%
3.27% 3.69%
3.47% 4.06%
3.91% 89.07%
89.35%
Link to Node Ratio
40% 1.11.74% 92.44% 3.42%
2.58% 2.40%
2.87% 1.74%
1.59% 92.44%
92.95%
05% 1.65%
1.25 92.13% 4.17%
3.49% 2.05%
3.50% 1.65%
2.55% 92.13%
90.46%
75% 1.41.55% 91.63% 5.06%
4.67% 1.75%
4.22% 1.55%
4.05% 91.63%
87.06%
50% 1.46%
1.55 90.91% 6.14%
6.16% 1.50%
5.01% 1.46%
6.32% 90.91%
82.52%
Total No. of Lanes on Macro Roads
34% 2 1.72% 91.46% 4.48%
7.15% 2.34%
6.38% 1.72%
3.80% 91.46%
82.66%
19% 4 1.68% 93.32% 2.80%
2.10% 2.19%
2.10% 1.68%
2.85% 93.32%
92.95%
05% 6 1.63% 94.58% 1.74%
0.57% 2.05%
0.64% 1.63%
1.97% 94.58%
96.82%
Distance from City Center (miles)
03% 0 3.18% 89.49% 3.30%
3.88% 4.03%
5.47% 3.18%
4.86% 89.49%
85.79%
06% 1 2.36% 91.11% 3.48%
4.12% 3.06%
4.19% 2.36%
3.63% 91.11%
88.06%
31% 2 1.74% 92.30% 3.65%
4.35% 2.31%
3.18% 1.74%
2.69% 92.30%
89.77%
74% 3 1.27% 93.17% 3.82%
4.57% 1.74%
2.41% 1.27%
1.99% 93.17%
91.04%
31% 4 0.93% 93.78% 3.98%
4.78% 1.31%
1.81% 0.93%
1.46% 93.78%
91.95%
43. GG
MODE CHOICE MODEL
strian Biking Mode Automobile Transit Mode Pedestrian Biking Mode Automobile
Share Share Mode Share Share Mode Share Share Mode Share
variables (all other variables (all other variables (all other variables (all other variables (all other variables (all other variables
mean) held at mean) held at mean) held at mean) held at mean) held at mean) held at mean)
Variables
3%
BASELINE BY STREET PATTERN88.51%
3.39% TYPE 9.00%
3.66% 8.79%
2.28% 4.09%
1.71% 78.13%
92.35%
Intersection Density
9% 812.72% 86.64% 8.93%
3.81% 5.08%
1.94% 2.84%
1.29% 83.15%
92.96%
5% 3.00%
144 87.55% 8.98%
3.65% 6.14%
2.30% 3.23%
1.74% 81.65%
92.31%
3% 3.38%
225 88.50% 9.01%
3.44% 7.81%
2.85% 3.79%
2.56% 79.39%
91.15%
7% 3.91%
324 89.35% 8.96%
3.18% 10.40%
3.69% 4.56%
4.06% 76.08%
89.07%
Link to Node Ratio
7% 1.59%
1.1 92.95% 8.40%
3.42% 9.93%
2.40% 3.21%
1.74% 78.47%
92.44%
0% 2.55%
1.25 90.46% 8.69%
4.17% 9.35%
2.05% 3.62%
1.65% 78.34%
92.13%
2% 4.05%
1.4 87.06% 8.99%
5.06% 8.80%
1.75% 4.08%
1.55% 78.13%
91.63%
1% 6.32%
1.55 82.52% 9.29%
6.14% 8.28%
1.50% 4.59%
1.46% 77.85%
90.91%
Total No. of Lanes on Macro Roads
8% 2 3.80% 82.66% 8.28%
4.48% 8.57%
2.34% 3.45%
1.72% 79.70%
91.46%
0% 4 2.85% 92.95% 10.16%
2.80% 9.09%
2.19% 5.27%
1.68% 75.48%
93.32%
4% 6 1.97% 96.82% 12.26%
1.74% 9.48%
2.05% 7.93%
1.63% 70.33%
94.58%
Distance from City Center (miles)
7% 0 4.86% 85.79% 8.39%
3.30% 11.10%
4.03% 5.28%
3.18% 75.23%
89.49%
9% 1 3.63% 88.06% 9.04%
3.48% 8.62%
3.06% 4.00%
2.36% 78.33%
91.11%
8% 2 2.69% 89.77% 9.65%
3.65% 6.62%
2.31% 3.00%
1.74% 80.72%
92.30%
1% 3 1.99% 91.04% 10.22%
3.82% 5.05%
1.74% 2.23%
1.27% 82.50%
93.17%
1% 4 1.46% 91.95% 10.75%
3.98% 3.82%
1.31% 1.65%
0.93% 83.78%
93.78%
44. VMT
LT TT RT GT
VMT in Block Group
per capita per day 66 28 27 51
LG TG RG GG
VMT in Block Group
per capita per day - 21 23 24
45. Effect on VMT?
24 x
)
ns
io
i ll
(b
T
VM
8 x
Population
2.5 x
192 193 194 195 196 197 198 199 200
5 5 5 5 5 5 5 5 5
46.
47. Ro ad S afe ty & Mo de
Cho ic e
Gettingof which will requires
All things right help
a more comprehensive
Redefining the Problem
inform our efforts toward creating:
approach that considers:
Street Network Density &
S afe r
Street Design
reStreet Connectivity
Mo Co S us tainable s ig n e s
mmunity De Plac
Alternative Modes
Street Patterns