Transportation leadership you can trust The Bicycle Investment Scenario Analysis Model A Web-based Sketch Planning Tool for Los Angeles County 2013 TRB Planning Applications Conference May 6, 2013 Michael Snavely, Cambridge Systematics Chris Porter, Monique Urban, David Jackson (Cambridge Systematics) Robert Cálix (Los Angeles County Metro) with presented by
37
Embed
Transportation leadership you can trust. The Bicycle Investment Scenario Analysis Model A Web-based Sketch Planning Tool for Los Angeles County 2013 TRB.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Transportation leadership you can trust.
The Bicycle Investment Scenario Analysis ModelA Web-based Sketch Planning Tool for Los Angeles County
2013 TRB Planning Applications Conference
May 6, 2013
Michael Snavely, Cambridge Systematics
Chris Porter, Monique Urban, David Jackson (Cambridge Systematics)Robert Cálix (Los Angeles County Metro)
with
presented by
Agenda
Background / Context
Model Structure
Trip Estimation Models
Application: User Interface
Conclusion / Results
2
Background – Cycling in Los Angeles
1897 - First Class I Bike Facility in US: California Cycle-Way (LA – Pasadena)
» Privately funded &: tolled (10¢ per trip)
» Replaced by Red Car prior to completion
» 1951: right-of-way becomes Arroyo Seco Parkway
Bicycling in LA County Today:
» Over 1000 planned bikeway miles
» Cycling visibility
• Mayor’s accident
• Advocacy presence
» Bikesharing (DTLA, SM, LB)
» Countywide demand for active transportation alternatives
3
Background - Policy Context
2008 – 2012: Countywide Congestion Mitigation Fee Pilot Study
» 1/3 of projects submitted by cities (close to $1B) are bicycle related
» No way to estimate benefits
2012: Metro Board calls for tools to estimate impacts
» 1) Sketch-planning tool (zonal)
• Must be web-based
• Accessible to all 89 jurisdictions
» 2) Travel demand model component (network)
• Late 20144
Regional Context – Existing Bicycle Mode Share
5Average bicycle work-trip mode share = 0.75%
Methodology
Methodology – Guiding Principles
Enable easy web-based access for cities
Complete within 1 year (no new data collection)
Impacts should be sensitive to local conditions
To extent possible, estimate impacts of:
» Bikeways,
» Bike Parking Facilities, and
» Bike Sharing Programs
Sketch-level scenario analysis
» Provide “order of magnitude” estimate of benefits
New Annual Bike TripsDue to bikeway investments (year 2035 vs. no build)
Methodology – Bikeway Trip Estimation Models
11
Work Trip Model Recreational Trip Model
Data Source ACS (2007 - 2011), LA County
Travel Purpose
Model Type
NHTS (2009) LA & Orange
Work Trips Strictly recreational (enjoyment/exercise)
Logistic regression 2-Step: binary logit + linear regression
New
Facilities
Logistic Regression Model
∆ Work Bike Trips
∆ Other Utilitarian
Trips(4:1)
Loca
l
Fact
or
s
New
Facilities
Binary Logit Model
Loca
l
Fact
or
s
∆ Persons who take at least 1 rec bike trip
Linear Regression Model
∆ # New Recreational
Bike Trips
Benefits Calculation
Work Trip Model
Bicycle Work Trip Model
ParameterWork Trip Model
CoefficientIntercept -4.82Dense Urban (Core, CBD & Urban Business District)
1.15
Other Urban Area 0.92Suburban 0.39Percentage of HH with Zero Vehicles 1.99
% of Roads with Grades Greater than 3% -0.614Mean Travel Time to Work (Drive Alone) -0.05
Miles of Class 1 (off-street) Bicycle Facilities per Sq. Mi.
0.09
Miles of Class 2 & 3 (on-street) Bicycle Facilities per Sq. Mi.
0.13
13
Estimated on census tract-level data using 2007-2011 ACS and land use/infrastructure data from LA County
Logistic regression
Trip Purpose Work
Other Utilitaria
nAvg 1-way length (mi) 3.8 2.3
Fraction all trips 20% 80%
Commute days/year
250
Key constants
Work Trip Model Sensitivity Tests
Scenario 1 – Increasing avg bikeway density to 2.5 mi/sq mi would raise bike commute share to 0.9%
Scenario 2 - Increasing to 5.0 mi/sq mi would raise bike commute share to ~1.2%
» Reasonable results
» Comparable to recent national study (Buehler & Pucher, 2012)
14
Facility DensityAverage mi/sq mi
Bicycle Mode Share –
Commute
Scenario Class 1Class 2&3 TOTAL Mean
Base 0.18 0.90 1.1 0.76%
Scenario 1 0.5 2.0 2.5 0.89%
Scenario 2 1.0 4.0 5.0 1.16%
Recreational Trip Model
Rec Model - Data Sources and Processing
Data sources
» 2009 NHTS person data (~10,000 L.A., Orange Co. residents)
• Information on bicycling activity in past week
• Sociodemographics and dependent variables
GIS Data processing
» Facility density by tract
» Proximity measures:
• Number and length of facilities within 1, 2, 5, 10 miles
16
Binary Logit Model (n ~ 10,000): Propensity to Bicycle for Recreation
17
Variable CoefficientConstant -2.52Sex and Age
Female -1.04 Age (continuous): Number of years over 44 -0.044
Education LevelLess than high school (base) -
High school or GED 0.41 Vocational/Associate's 0.54
Graduated college 0.68 Master's, Ph.D., or Professional Degree 0.71
Bicycle FacilitiesDistance to nearest bike trail: < 1 mile 0.20
1-2 miles 0.041 > 2 miles -
Facility density in home zone (miles per sq. mi.): Class 1 0.13
Class 2 & 3 0.056
Linear Regression Model (n ~ 600):Number of Weekly Recreational Bicycling Trips
18
Variable Coefficient
Constant 2.92
Sex and Age
Female -0.27
Age -0.029
Additional Years over 44 0.057
Education Level
High School or Associate's 0.50
College Graduate or above 0.79
Household Data
Household Children under 18 -0.12
Household Income ($100,000s) -0.33
Summary of Recreational ModelTwo-stage model
1. Identify individuals that bicycle for recreation
2. Compute the number of trips made by each individual
Key findings
» Demographics – greatest impact
» Bicycle facilities – also significant
Data limitations
» Estimation dataset: NHTS person data (disaggregate)
» Application dataset: census tract (aggregate)
» GIS processing, zonal aggregation19
Michael Snavely
Update and/or include a slide summarizing key takeaways for policy audience(e.g. model is sensitive to facility variables, but with caveats, etc.)
Application: User Interface
Bicycle Investment Scenario Analysis Model
Results
ResultsBenefits» Shows order-of-magnitude estimates based on best
available data» Sensitivity tests showed reasonable results» For first time, cities can justify funding cycling projects
based on local conditions» Relatively low cost implementation and ease of
calculation
Limitations» Zonal aggregation misses connectivity/network issues» Limited by small sample sizes (e.g. only 600 rec trips)» More research needed to validate ‘off-model’ methods» For sketch purposes only
Opportunities to improve/expand functionality» Next step: census block zonal aggregation» Update pending TDM estimation & data collection results» Add other project types