ISSN 1055- 1425
March 2010
This work was performed as part of the California PATH Program of the
University of California, in cooperation with the State of California Business,
Transportation, and Housing Agency, Department of Transportation, and the
United States Department of Transportation, Federal Highway Administration.
The contents of this report reflect the views of the authors who are responsible
for the facts and the accuracy of the data presented herein. The contents do not
necessarily reflect the official views or policies of the State of California. This
report does not constitute a standard, specification, or regulation.
Final Report for Task Order 6117
CALIFORNIA PATH PROGRAM
INSTITUTE OF TRANSPORTATION STUDIES
UNIVERSITY OF CALIFORNIA, BERKELEY
Seamless Travel: Measuring Bicycle and
Pedestrian Activity in San Diego County and its
Relationship to Land Use, Transportation,
Safety, and Facility Type
UCB- ITS- PRR- 2010- 12
California PATH Research Report
Michael G. Jones, Sherry Ryan, Jennifer Donlon,
Lauren Ledbetter, David R. Ragland, Lindsay Arnold
CALIFORNIA PARTNERS FOR ADVANCED TRANSIT AND HIGHWAYS
Institute of Transportation Studies
UC Berkeley Traffic Safety
Center
( University of California, Berkeley)
Year: 2010 Caltrans Task Order 6117
Seamless Travel:
Measuring Bicycle and Pedestrian Activity in
San Diego County and its Relationship to Land
Use, Transportation, Safety, and Facility Type
_________________________
( Measuring Bicycle and Pedestrian Activity in San Diego County and its
Relationship to Land Use, Transportation, Safety, and Facility Type)
Michael G. Jones 1
Sherry Ryan 2
Jennifer Donlon3
Lauren Ledbetter 4
David R. Ragland 5
Lindsay Arnold 6
1 Alta Planning + Design, Inc.
2 Ibid
3 Ibid
4 Ibid
5 UC Berkeley Traffic Safety Center
6 Ibid
Seamless Travel February 2010 1
Caltrans Task Order 6117
Seamless Travel:
Measuring Bicycle and Pedestrian Activity in
San Diego County and its Relationship to Land
Use, Transportation, Safety, and Facility Type
Abstract
This paper provides the data collection and research results for the Seamless Travel project. The
Seamless Travel Project is a research project funded by Caltrans and managed by the University
of California Traffic Safety Center, with David Ragland, PhD., as the Principal Investigator and
Michael Jones as the Project Manager. The project is funded by Caltrans Division of Innovation
and Research and is being conducted by the Traffic Safety Center of University of California
Berkeley and Alta Planning + Design.
Measuring bicycle and pedestrian activity is a key element to achieving the goals of the California
Blueprint for Bicycling and Walking ( the Blueprint) 7. Meeting these goals, which include a 50%
increase in bicycling and walking and a 50% decrease in bicycle and pedestrian fatality rates by
2010, and increases in funding for both programs, will require a quantifiable and defensible base
of knowledge. This research helps meet two of the Blueprint’s major strategic objectives: ( 1)
collecting data on volumes and facilities, and ( 2) determining the most cost- effective methods of
estimating bicycle and pedestrian collision rates.
Understanding why people walk or ride bicycles, how the type and quality of facility influences
these trips, and how adjacent land uses, density, access, roadway traffic volumes, and other items
impact walking or bicycling, are all critical to meeting the goals of the Blueprint. Good baseline
information on walking and bicycling is important to answer questions like that posed in the title
of this research: are Class I bike paths so attractive to potential commuters that they should be
given priority over Class II bike lanes, Class III bike routes, or other facilities?
Counts and surveys conducted throughout California since 2000 consistently show a
substantially higher demand for and use of Class I bike paths than on- street facilities. 8 Is this due
to inconsistent on- street systems, a lack of riding expertise by the public, perceived or real safety
concerns, recreational versus commuter use, high roadway traffic volumes and speeds, and/ or
other factors?
7 California Blueprint for Bicycling and Walking: Report to Legislature, California Department of Transportation, May 2002
8 Alta Planning + Design, staff experience on 62 bicycle and pedestrian plans in California since 1990
Seamless Travel February 2010 2
Caltrans Task Order 6117
This research is designed to ( a) evaluate existing bicycle and pedestrian data sources and
collection methods, ( b) conduct comprehensive counts and surveys of bicyclists and pedestrians
in a consistent manner using the National Bicycle & Pedestrian Documentation Project ( NBPD)
as a template 9, ( c) conduct counts and surveys using San Diego County ( with extensive historical
count information) as a model community, ( d) analyze how bicycle and pedestrian activity levels
relate to facility quality and factors such as land use and demographics, ( e) identify factors that
are highly correlated with increased bicycling and walking, ( e) provide methods for quantifying
usage and demand that will enhance research on benefits and exposure, and ( f) evaluate how the
transit- linkage ( bicycle and pedestrian connections to transit) can be improved.
This Report presents materials developed including a literature review, advisory committee
meeting input, project objectives, data collection methodology, results from the data collection
effort, analysis of correlations, trends, and patterns, conclusions on the accuracy and applicability
of the data, and recommendations on increasing walking and bicycling in California.
9 National Bicycle and Pedestrian Documentation Project, Jones, M., Buckland, L., Cheng, A., Transportation Research
Board, Aug. 2005
Seamless Travel February 2010 3
Caltrans Task Order 6117
FINAL REPORT
SEAMLESS TRAVEL:
Measuring Bicycle and Pedestrian Activity in San Diego County and its
Relationship to Land Use, Transportation, Safety, and Facility Type
PREPARED FOR
Task Order 6117
David R. Ragland, Traffic Safety Center ( TSC)
Michael G. Jones, Alta Planning + Design, Inc.
Seamless Travel February 2010 4
Caltrans Task Order 6117
University of California Traffic Safety Center – Institute of Transportation Studies
University of California – Berkeley, California 94730- 7360
Tel: ( 510) 642- 0655 Fax: ( 510) 643- 9922
Alta Planning + Design, Inc.
2560 Ninth Street, Suite 212
Berkeley, California 94710
Tel: ( 510) 540- 5008 Fax: ( 510) 540- 5039
Seamless Travel February 2010 5
Caltrans Task Order 6117
Table of Contents
EXECUTIVE SUMMARY ..................................................................... 9
1. INTRODUCTION................................................................. 17
Formation of Advisory Committee ............................................................................................................ 17
Project Objectives ............................................................................................................................... .......... 19
2. SYNTHESIS OF PUBLISHED RESEARCH ..................................... 25
Review of Existing Count and Survey Methods ....................................................................................... 25
Existing Data Sources ............................................................................................................................... ... 25
Pedestrian and Bicycle Research Efforts ................................................................................................... 26
National Bicycle and Pedestrian Documentation Project ....................................................................... 26
Count Methodologies ............................................................................................................................... .... 27
Pedestrian and Bicycle Travel Behavior Survey Methods ....................................................................... 29
Bicycling and Pedestrian Travel modeling ................................................................................................. 30
Four- Step Modeling Process ........................................................................................................................ 30
Non- Motorized Transportation Forecasting Efforts ............................................................................... 34
Conclusion ............................................................................................................................... ...................... 35
3. PRIMARY DATA COLLECTION ................................................ 37
Why San Diego County? ............................................................................................................................... 37
Count Methodology ............................................................................................................................... ...... 37
Automated Count Methodology ................................................................................................................. 38
Manual Counts and Surveys ......................................................................................................................... 48
Accuracy of the Count and Survey Data .................................................................................................... 49
4. Count and Survey Results ................................................... 51
Surveys ............................................................................................................................... ............................. 51
Automated Count results .............................................................................................................................. 63
Volume, Capacity, LOS Analysis ................................................................................................................. 63
Analysis of Hourly Counts ........................................................................................................................... 64
Analysis of Day of the Week Counts .......................................................................................................... 67
Analysis of Monthly Counts ......................................................................................................................... 69
Mode Split ............................................................................................................................... ....................... 72
Design Peak Period and Day ....................................................................................................................... 73
Manual Counts ............................................................................................................................... ............... 73
Summary of Count and Survey Findings ................................................................................................... 83
5. Development of A Predictive Model ....................................... 91
Purpose of a Bicycle/ Pedestrian Estimating Model ................................................................................ 91
The Bicycle and Pedestrian Demand Models ............................................................................................ 93
Seamless Travel February 2010 6
Caltrans Task Order 6117
Potential Variables ............................................................................................................................... ......... 93
Testing Multiple Variables ............................................................................................................................ 94
Modeling Bicyclist and Pedestrian Behavior ............................................................................................. 94
Modeling Approach # 1 ............................................................................................................................... 95
Modeling Approach # 2 ............................................................................................................................. 102
Modeling Approach # 3 ............................................................................................................................. 105
Pedestrian Model ............................................................................................................................... ........ 106
Bicycle Regression Model Development ................................................................................................ 112
6. References..................................................................... 117
Appendix A: Manual and Automatic Count Database
Appendix B: Training Manual
Appendix C: Instructions for Sending Future Data
Appendix D: Bicycle Model
Appendix E: Pedestrian Model
Appendix F: Summary of Comparison Surveys
Appendix G: Background Data for Analysis
Table of Figures
Figure 1: Comparison of Trip Purpose ........................................................................................................... 11
Figure 2: Historic Counts ............................................................................................................................... .. 13
Figure 3: Historic Percent Change ................................................................................................................... 13
Figure 4: Peak Hour Count Locations in San Diego County ..................................................................... 37
Figure 5: Yearly Count Locations in San Diego County .............................................................................. 41
Figure 6: Rose Canyon Bike Path, Mission Beach Boardwalk and Bayside Year- Long Automated
Count Locations ............................................................................................................................... ................. 43
Figure 7: University Avenue and Bayshore Bikeway Year- Long Automated Count Locations ............ 43
Figure 8: Number of Pedestrian and Bicycle Surveys Collected by Metropolitan Statistical Area ....... 52
Figure 9: Destination of Those Who Bicycle 1- 4 Times a Month ............................................................. 55
Figure 10: Preferred Bicycle Facilities ............................................................................................................. 56
Figure 11: Trip Purpose ............................................................................................................................... ..... 60
Figure 12: Hour of Day April - September .................................................................................................... 66
Figure 13: Hour of Day October- March ........................................................................................................ 66
Figure 14: Day of the Week ............................................................................................................................. 68
Figure 15: Month of Year ............................................................................................................................... . 70
Figure 16: Comparison of Monthly Volume .................................................................................................. 72
Seamless Travel February 2010 7
Caltrans Task Order 6117
Figure 17: Weekday AM Peak- Hour Bicycle Counts .................................................................................... 75
Figure 18: Weekend Midday Peak- Hour Bicycle Counts ............................................................................. 76
Figure 19: Weekday AM Peak- Hour Pedestrian Counts .............................................................................. 77
Figure 20: Weekend Midday Peak- Hour Bicycle Counts ............................................................................. 78
Figure 21: Comparison of Trip Purpose ......................................................................................................... 83
Figure 22: Historic Counts ............................................................................................................................... 85
Figure 23: Historic Percent Change................................................................................................................. 86
Figure 24: Pedestrian Activity at Count Locations .................................................................................... 100
Figure 25: Bicycle Activity at Count Locations .......................................................................................... 101
Figure 26: Pedestrian Model Results ............................................................................................................ 111
Table of Tables
Table 1: Comparison of Trip Purpose ............................................................................................................ 11
Table 2: Comparison of Pathway and On- Street Bicycling by Trip Purpose ........................................... 11
Table 3: Historic Bicycle Counts San Diego County 1985- 2008 ................................................................ 12
Table 4: Manual and Automated Count Characteristics .............................................................................. 28
Table 5: Characteristics of General and Targeted Surveys .......................................................................... 29
Table 6: Selected Factors Influencing Non- Motorized Travel ................................................................... 33
Table 7: Methods for Modeling Non- Motorized Travel Demand ............................................................. 34
Table 8: Automatic County Technology Overview ...................................................................................... 40
Table 9: Passive Infrared Validation Counts JAMAR Scanner ................................................................... 44
Table 10: Active Infrared Validation Counts ................................................................................................. 46
Table 11: Bicycle Survey Respondent Locations and Percent of Total Volumes .................................... 53
Table 12: Bicycle Trip Purpose ........................................................................................................................ 54
Table 13: Frequency of Bicycle Riding ........................................................................................................... 54
Table 14: Reasons Preventing Respondents from Bicycle Riding More Often ....................................... 55
Table 15: Types of Facilities Respondents Enjoy ......................................................................................... 56
Table 16: Income Level of Bicycle Respondents .......................................................................................... 56
Table 17: Race/ Ethnicity of Bicycle Respondents........................................................................................ 57
Table 18: Gender of Bicycle Respondents ..................................................................................................... 57
Table 19: Number of Pedestrian Intercept Surveys by Location ............................................................... 58
Table 20: Walk Trip Purpose ............................................................................................................................ 59
Table 21: Frequency of Walking ...................................................................................................................... 59
Table 22: Reasons Preventing Respondents from Walking More Often .................................................. 60
Table 23: Quality of Pedestrian Facilities ....................................................................................................... 61
Seamless Travel February 2010 8
Caltrans Task Order 6117
Table 24: Income Level of Pedestrian Respondents .................................................................................... 61
Table 25: Race / Ethnicity of Pedestrian Respondents ............................................................................... 61
Table 26: Gender of Pedestrian Respondents ............................................................................................... 62
Table 27: Summary of 12- Month Counts San Diego County August 17 2007- August 16 2008 ........... 63
Table 28: Pathway Level of Service ................................................................................................................. 63
Table 29: Peak Periods by Mode and Season1 Automatic Count Locations2 ........................................... 64
Table 30: Hour of Day ............................................................................................................................... ...... 65
Table 31: Comparison of Weekday Hourly Counts ...................................................................................... 67
Table 32: Day of the Week San Diego County, 5 Locations, August 2007- July 2008 ............................ 68
Table 33: Comparison of Day of Week Counts ............................................................................................ 69
Table 34: Month of Year ............................................................................................................................... ... 70
Table 35: Comparison of Monthly Volume ................................................................................................... 71
Table 36: Comparison of Mode Split ( Bicycling/ Pedestrian) San Diego County/ 4 Other Pathways .73
Table 37: Monthly Adjustment Factors .......................................................................................................... 74
Table 38: Average Counts by Location ........................................................................................................... 79
Table 39: Summary Statistics Manual Counts ................................................................................................ 82
Table 40: Comparison of Trip Purpose .......................................................................................................... 83
Table 41: Comparison of Pathway and On- Street Bicycling by Trip Purpose ......................................... 84
Table 42: Historic Bicycle Counts San Diego County 1985- 2008 .............................................................. 85
Table 43: Dependent Variables Used in the Models .................................................................................... 94
Table 44: Independent Variables Considered for the Bicycle and Pedestrian Volume Models ............ 97
Table 45: Significant Differences in Means: Morning High and Low Pedestrian Count Locations ..... 99
Table 46: Pedestrian Generator Weights and Multipliers ......................................................................... 103
Table 47: Distance- Based Pedestrian Attractor Multipliers ...................................................................... 104
Table 48: Pedestrian Attractor and Generator Regression Model Results – Weekday AM Peak Counts
............................................................................................................................... ............................................ 105
Table 49: Pedestrian Volume Model ( Stepwise Method) .......................................................................... 106
Table 50: Residual Analysis of Stepwise Pedestrian Models .................................................................... 107
Table 51: Alternative Pedestrian Volume Model Specifications .............................................................. 108
Table 52: Alternative Pedestrian Volume Model Specifications with Refinement ............................... 109
Table 53: Alternative Bicycle Volume Model Specifications .................................................................... 112
Table 54: Previous Regression Modeling..................................................................................................... 114
Seamless Travel February 2010 9
Caltrans Task Order 6117
EXECUTIVE SUMMARY
The Seamless Travel Project, in coordination with the National Bicycle & Pedestrian
Documentation Project, is the largest and longest combined count and survey effort in the
United States focusing only on bicyclists and pedestrians. Using San Diego County as a case
study, the Seamless Travel Project is the first of its type to develop an extensive database of
count and survey data for use in analyzing and identifying factors that influence bicycling and
walking. While the bicycle and walk modes are studied together, it is recognized that they are
distinct from one another and they are always counted, surveyed, and analyzed separately. This
Final Report provides a review of the methodology along with count and survey results,
development of predictive models, model results, and information on how the count/ survey
results and models can be used by public agencies and transportation professionals.
Key findings include:
The Seamless Travel Project represents a significant advance in the non- motorized field
of research. Current and past research efforts have been limited by the lack of adequate data to
test and verify theories. The Seamless Travel Project is the largest study of bicyclist and
pedestrian behavior in the United States, with the largest number of manual count locations ( 80),
the first to use automatic count data collected over a 365- day period to adjust manual counts, the
first study to incorporate data from the National Bicycle & Pedestrian Documentation Project in
comparing results from around the country, the first to incorporate extensive survey results with
manual counts, and the first effort to date to create a predictive model that has been tested
against actual count results.
California should develop and implement a systematic bicyclist/ pedestrian count and
survey program. A systematic count and survey of bicyclists and pedestrians by Caltrans and
local agencies is an important step meeting the goals of the California Blueprint for Bicycling and
Walking ( the Blueprint) 10, Complete Streets policies, and other goals. The Seamless Travel study
provides specific materials ( Training Manual and Powerpoint) for how to conduct manual and
automatic machine counts, surveys, use of the data, and recommendations on how counts could
be institutionalized and funded. Counts and survey methods should be consistent with the
National Bicycle & Pedestrian Documentation Project.
Annual use should be the standard measurement for the bicycle and pedestrian modes.
Given the day to day and seasonal variability at many locations, and the fact that determining
peak hour capacity is not an overriding need, the use of annualized figures will allow a more
accurate comparison between locations.
Methods and conclusions based on data from San Diego County and the National
Bicycle & Pedestrian Documentation Project should be applicable to many community
types and locations. Compared to other modes where methods ( such as the ITE Trip
Generation Manual) and data collected from limited locations nationwide are accepted by all
agencies, there is no existing similar acceptance for the bicycle/ pedestrian field. The Seamless
Travel project and National Bicycle & Pedestrian Documentation Project represent the greatest
10 California Blueprint for Bicycling and Walking: Report to Legislature, California Department of Transportation, May 2002
Seamless Travel February 2010 10
Caltrans Task Order 6117
accumulation of data available today, and the data and methods should be applicable to a broad
range of communities nationwide. However, seasonal and other local variables do exist that
require additional efforts, especially year long machine counts.
Where peak hour volumes are needed to evaluate capacity, the standard ‘ Design Period
and Design Day’ on Class I and multi- use pathways should be as follows:
Maximum design load: 11am- 1pm, July, 4th
Weekday: 11am- 1pm, Mid- July, Tuesday, Wednesday, or Thursday ( non- holiday)
Weekend day: 11am- 1pm, Mid- July, Saturday ( non- holiday)
Class I and Multi use pathway capacity ranges between 15 and 270 persons per hour per
foot of pathway width. Free flow conditions suitable for higher bicycle commuting speeds are
represented at the lower end, while the maximum capacity range would require bicyclists to
dismount or ride very slowly. Both ends of the range require adequate separation between
directional flow, and preferably modes as well.
For planning purposes, the use of 120 persons per hour per foot of path width as the
maximum capacity is recommended to maintain adequate flows. Centerline separation
and supporting pathway management techniques ( signing, enforcement etc) on any pathway with
design day volumes over 10 persons per hour per foot of path width and pedestrian mode split
over 20%, or over 15 persons per hour per foot of path width and under 20% pedestrian mode
split are recommended. Design hour or day pedestrian volumes on sidewalks should conform
with the Highway Capacity Manual pedestrian level of service methodology, which is also used
to determine crosswalk capacities.
Bicycle and pedestrian volumes can be classified in ranges to facilitate mapping and
analysis. The recommended classification range is as follows:
Bicycle Volumes
Low 0- 20 per hour
Moderate 21- 60
High over 61
Pedestrian Volumes
Low 0- 40 per hour
Moderate 41- 100
High Over 100
The perception of the walk and bicycle trip making as recreational or discretionary is
unfounded. The walk and bicycle modes have significant ( and often the same) percentages of
work, school, or utilitarian trip making as household travel in general, and private vehicle trips
( see Table 1 and Figure 1). While funding for pedestrian and bicycle facilities is typically limited
to ‘ transportation’ functions only, funding for roadways, transit, and other systems make no such
distinction. The result is a potential funding bias against non- motorized facilities, as well as a
potential resistance to accommodate non- motorized modes in new projects despite adoption of
Complete Streets and other similar policies.
Seamless Travel February 2010 11
Caltrans Task Order 6117
Table 1: Comparison of Trip Purpose
All Households
( Percent) 1
Pedestrians2
( Percent)
Bicyclists2
( Percent)
Work, School, Utilitarian 27.5 21 12
Social, Recreational 27.1 24 71
Utilitarian, Personal ( shopping,
family/ personal business)
44.6 55 17
1. Bureau of Transportation Statistics, National Household Travel Survey, Fig 7, 2001
2. San Diego County survey results
Figure 1: Comparison of Trip Purpose
0
10
20
30
40
50
60
70
80
Work, School Social, Recreational Utilitarian, Personal
( shopping, family/ personal
business
Trip Purpose
Percent
All Households
Pedestrians
Bicyclists
Class I bike paths and multi- use paths in general serve as important transportation
facilities. The surveys of trip purpose combined with the year- long counts of four ( 4) bike
paths in San Diego County show ( see Table 2) these pathways alone are used by an estimated
691,969 bicyclists on work/ school/ utilitarian trips. This volume is 90% higher than the total
estimated annual volumes of all on- street bicycle trips counted at 69 of the 80 manual count
locations. It is likely that paths serve as important incubators for bicyclists learning or re- learning
how to ride bicycles as a transportation vehicle for short trips.
Table 2: Comparison of Pathway and On- Street Bicycling by Trip Purpose
Location Total Annual Use Transportation Trips1
Bayside Path 513,558 133,525
Gilman Path/ Rose Canyon 164,638 42,805
Strand Path 148,109 38,508
Boardwalk 1,835,426 477,131
Subtotal 2,661,426 691,969
On- Street Locations2 1,401,837 364,477
1. Defined as school, work, utilitarian trips
2. 69 of the 80 count locations, normalized to annual counts
Seamless Travel February 2010 12
Caltrans Task Order 6117
Bike lanes are not an indicator of bicycle use. Bicycle use on streets with bike lanes is similar
as streets without bike lanes. This does not mean that bike lanes do not attract or serve
bicyclists. Firstly, bike lanes have traditionally been installed where they are feasible rather than
where the highest existing uses are located. Secondly, all things being equal, bicyclists will choose
the best, most direct route with the best combination of topography, lane width, and traffic
volumes speeds available.
Location Determines Data. The location of the five ( 5) automatic counters drives the pattern
of data collected. Bicycle and pedestrian activity is affected by facility type ( pathways, sidewalks),
surrounding land use, weather, time of year, and many other factors. The data therefore
provides a ‘ snapshot’ of a limited range of possible activity patterns in San Diego County or in
any community. However, this data along with other year round data from around the country
starts to provide a picture of activity trends that can be used to frame parameters of activity.
Bicycle use in San Diego County based on historical counts back to 1987 has generally
been stable, and is increasing in the past year. Various agencies in San Diego including
SANDAG and Caltrans have conducted bicycle counts since 1985. Twelve ( 12) locations were
consistently counted between 1985 and 2008 ( 13 years). Initially the figures indicated a steep
decline in use at these 12 locations between 1985 and 1990. However, an in- depth analysis of
the figures shows that almost all of the decline was due to one location ( Site # 16:
College/ Montezuma). This location is next to the LRT station near San Diego State University,
which was completed during the count period, and may have impacted or changed bicycling
patterns in the area. Table 3 shows how, if this site is removed, volumes at the remaining 11
locations were stable from 1985- 2007. In all cases, volumes in the most recent count ( 2008)
have jumped between 40- 85%. The last column on Table 3 and Figure 2 shows the average
percent change of all 12 locations from 1985- 2008, showing a consistent increase during this
period except between 1990 and 1993.
Table 3: Historic Bicycle Counts San Diego County 1985- 2008
Year AM Counts1 Average % 2
AM
Counts
Average % 3 Average %
Change4
1985 1,022 414
1987 913 - 10 396 - 4 + 27
1990 659 - 28 395 0 - 2
1993 701 + 6 440 + 11 + 12
1997 541 - 33 410 - 7 + 12
2007 586 + 8 386 - 6 + 12
2008 823 + 40 713 + 85 + 30
1. AM Counts, weekdays 7am- 9am, adjusted seasonally, 12 locations
2. Count locations increased from 12 in 1985 to 80 in 2008
3. AM Counts, weekdays 7am- 9am, adjusted seasonally, 11 locations excluding College/ Montezuma
4. Average % change of all 12 locations from year to year
Seamless Travel February 2010 13
Caltrans Task Order 6117
Figure 2: Historic Counts
0
200
400
600
800
1,000
1,200
1985 1987 1990 1993 1997 2007 2008
Year
Counts
AM Counts
Figure 3: Historic Percent Change
- 5
0
5
10
15
20
25
30
35
1987 1990 1993 1997 2007 2008
Year
Percent Change
Average % Change
Mode split on Class I and multi- use pathways is highly related to regional and local
patterns, with bicycle mode splits ranging from 30% to 90% and pedestrian mode splits
from 10% to 70%. Predictive models should be able to identify a general mode split based on
adjacent demographics and land uses. Commuter paths located next to some kinds of land uses
may require the development of alternative routes, special delineation and/ or management to
preserve the ability to be used by bicyclists for commuting.
Seamless Travel February 2010 14
Caltrans Task Order 6117
Class I and multi- use paths in San
Diego County are used mostly by
bicycles. While this varies by location and
facility, bicyclists are the primary users of
the pathways counted in San Diego County.
Nationally, pedestrians outnumber bicyclists
on pathways 75% to 20% on average. Mode
split appears to be correlated with adjacent
land uses, regional bicycling patterns, and
quality of the bikeway network
Over the course of a year, there are no
distinct daily peak periods for
pedestrians and bicyclists. Unlike motor
vehicle traffic patterns, there is no sharp
commute pattern for either bicycle or
pedestrian mode regardless of facility type. Activity is evenly spread throughout the day, with
minor peaking patterns. This is likely due to the mix of recreational and utility/ work/ school
trips, and also an indication of the low proportion of commute trips overall. This finding is true
for locations with ( a) connections to mixed land uses ( residential, commercial, office), ( b)
recreational trips and destinations, and/ or ( c) visitor usage. This finding would not apply to
locations such as large employment centers with little/ no retail or restaurant uses, or near major
transportation hubs.
Actual day- to- day variability at many count locations may make forecasting difficult.
Actual day to day variability is largely related to the volumes ( higher volumes = less day to day
variability) and trip types ( recreational trips = higher variability). With many count locations
having very low volumes, any predictive model will need to accept a relatively high margin of
error. Also, validation counts would need to be conducted over a longer period of time during
the same month of year, or, adjusted using local automatic count machine data.
The 6am – 9pm period accounts for a consistent 95% of the total volumes. Bicycle and
pedestrian volumes gently taper off from about 6pm to 12 midnight. From 12 midnight to 6am
there is very little activity. Focusing on the 6am to 9pm period will capture a consistent
snapshot of the vast majority ( 95%) of activity. The exception may be count locations near large
entertainment centers or districts.
Bicyclists and pedestrians have nearly an identical daily pattern of use on multi- use
pathways. While bicyclists accounted for 55% of all users on the five ( 5) pathways, peaking
patterns were proportional with pedestrian volumes. This indicates trip purpose on pathways,
regardless of mode, is similar between bicyclists and pedestrians, and that the combined modes
can be used to analyze patterns.
Pedestrian volumes on sidewalks in some areas are highly consistent and spread evenly
throughout the day and evening, with little discernable peaking. The hourly pedestrian
volumes on University Avenue in the Hillcrest neighborhood of San Diego ( a higher density,
older neighborhood with good transit service) was extremely even on both weekdays and
weekdays, with virtually no change between about 10am and 12 midnight. This reflects the fact
Multi- use paths in San Diego County, such as the one
above in Chula Vista, are mostly used by bicyclists
Seamless Travel February 2010 15
Caltrans Task Order 6117
a neighborhood with a mix of residential and commercial uses produces nearly constant and
consistent walking volumes for most of the day. This will allow manual counts conducted
during any time of the year to be adjusted to an annual total figure. This finding is true for
locations with ( a) connections to mixed land uses ( residential, commercial, office), ( b)
recreational trips and destinations, and/ or ( c) visitor usage. This finding would not apply to
locations such as large employment centers with little/ no retail or restaurant uses, or near major
transportation hubs.
Peak periods on Class I and multi- use paths have a consistent annual peak period of
11am- 1pm, with minor variations. This will allow manual counts conducted during any time
of the year to be adjusted to an annual total figure. This finding is true for locations with ( a)
connections to mixed land uses ( residential, commercial, office), ( b) recreational trips and
destinations, and/ or ( c) visitor usage. This finding would not apply to locations such as large
employment centers with little/ no retail or restaurant uses, or near major transportation hubs.
Pedestrian volumes on sidewalks, while generally consistent, will have seasonal changes
in peak periods depending on the adjacent land uses. Peak periods on sidewalks for
pedestrians range from 1- 3pm on weekdays in the Fall/ Winter/ Spring to 9- 11pm in the
Summer. This finding is true for locations with ( a) connections to mixed land uses ( residential,
commercial, office), ( b) recreational trips and destinations, and/ or ( c) visitor usage. This finding
would not apply to locations such as large employment centers with little/ no retail or restaurant
uses, or near major transportation hubs.
Given the consistency in peaking patterns on Class I bike paths and multi- use paths and
sidewalks in the locations described, manual counts can be used to extrapolate annual
data. This assumes the count location has a moderate to high volume, is not predominately
recreational, and can be validated with counts conducted during the same period for at least two
( 2) days, or, validated with a local automatic count machine.
Bicycle and pedestrian count results can yield some unusual, unexpected results,
reflecting highly localized conditions. For example, the second highest month of activity on
the four ( 4) pathways was March, possibly due to the college and university break schedules.
Other unexpected results could be caused by events such as marathons or races, construction,
special events, pulses of patrons from nearby rail, transit or ferry operations, and sporting events.
Day of week volumes are consistent between modes and locations, both in San Diego
County and nationally. Over the course of a year, bicycle and pedestrian volumes by day of
week are nearly identical, with Saturday being the day with the highest activity, and weekends
being higher than weekdays. This breakdown is very consistent with national counts.
Monthly volumes appear to be highly related to regional conditions, especially weather.
The monthly pattern in San Diego County had both intuitive results ( July with the highest
volumes) and unusual results ( March had the second highest with 12%). Compared to other
locations in the country with more severe winters, use is relatively even over 12- months in San
Diego County. The need for automatic counters in different regions is apparent in order to
establish local monthly adjustment factors.
Seamless Travel February 2010 16
Caltrans Task Order 6117
The correlation between actual counts and variables is complex. An analysis of over 30
variables with the 80 bicycle and pedestrian count locations shows that while there are some
distinct patterns ( especially with pedestrian volumes), most variables are highly correlated with
each other ( and therefore not helpful) and there are significant numbers of ‘ outliers’ that cannot
be easily explained.
Population density and transit ridership are not the strongest indicators of walking.
Some variables commonly thought to be highly correlated to walking, such as population density
and transit ridership, turned out to be only mild indicators and much less effective than others
( such as employment density). If an agency’s goal is to create neighborhoods or corridors with
higher levels of walking, a mixture of employment and residential uses is critical.
Forecasting models cannot rely on multiple regression analysis. Multiple regression
analysis using computer- based programs provide very high ‘ Multiple R’ factors for some
variables, such as employment density for pedestrians. A closer examination of these outcomes
reveals that, in the best of cases, over 50% of the count locations had model estimates that were
off by more than 50 persons per hour, and many were incorrect by over 100 people/ hour. This
confirms published research that states that computer generated multiple regression models
produce artificially high outcomes and formulas that are not accurate enough for general use.
A model with refinement factors provides the best possible forecasting tool. Using the
multiple regression outcomes as a starting point, a refinement model with variables triggered by
specific thresholds of volumes helps to improve the forecasting accuracy of the bicycle and
pedestrian models. The models should be accurate enough with local adjustments ( especially for
monthly changes) to allow for estimates of use by location, exposure analysis, and other uses.
These refinements can be modified and expanded as more data is collected over time.
Seamless Travel February 2010 17
Caltrans Task Order 6117
1. INTRODUCTION
In 2006, Caltrans contracted with the Traffic
Safety Center of University of California
Berkeley and Alta Planning + Design to develop
a model for estimating bicycle and pedestrian
demand within San Diego County. The project
methodology includes conducting bicycle and
pedestrian counts and intercept surveys over a
two- year period throughout the county and
evaluating the effects that socio- demographic,
land use, and other variables have on walking
and biking rates within the county. The project
is funded by Caltrans Division of Innovation
and Research.
The research team identified trends in walking and bicycling; evaluated the relationship between usage
and facility quality, physical factors, and social factors; and reviewed the potential for using land- use and
infrastructure improvements to increase walking and bicycling. The product of this research will provide
Caltrans staff, local agency staff, advocates, elected officials, and others with the information and tools
needed to understand walking and bicycling rates, patterns, relationships, and trends within San Diego,
and may be useful to other areas of the state and country.
The Seamless Travel Project is the first large- scale test of count and survey methodology outlined by the
National Bicycle and Pedestrian Documentation Project ( NBPD). The NBPD is an annual bicycle and
pedestrian count and survey effort developed and managed by Alta Planning + Design in coordination
with the Institute of Transportation Engineers Pedestrian and Bicycle Council. The goals of the NBPD
are to establish a consistent national bicycle and pedestrian count and survey methodology, to establish a
national database of bicycle and pedestrian count information generated by these consistent methods
and practices and to use the count and survey information to begin analysis on the correlation between
various factors and bicycle and pedestrian activity.
FORMATION OF ADVISORY COMMITTEE
Local stakeholders and a Caltrans Technical Advisory Group were involved in developing the project
methodology and have been regularly updated on the progress of the Seamless Travel project.
Technical Advisory Group
This group met several times to discuss the progress of the project and provide direction. Members of
the group include:
Ann Mahaney, Project Manager, Caltrans HQ
Bob Justice, University Contract Manager, Division of Research & Innovation, Caltrans
Richard Haggstrom, Senior Transportation Engineer, Caltrans HQ
Counts and Surveys were conducted over a two- year period
Seamless Travel February 2010 18
Caltrans Task Order 6117
Ken McGuire, Bike Program Manager, Caltrans
David Ragland, Director, UC Berkeley Traffic Safety Center
Michael Jones, Principal, Alta Planning + Design
Lauren Buckland, Associate, Alta Planning + Design
Stakeholder Group
This group consists of all the members of the Technical Advisory Group listed above, as well as local
San Diego Stakeholders. The Local Stakeholder Group includes members from San Diego Association
of Governments ( SANDAG), City of San Diego, County of San Diego, Caltrans District 11, San Diego
State University and WalkSanDiego. The purpose of this group is to provide local knowledge and
advice.
Members of the group include all TAG members, as well as:
Brad Jacobsen, Associate Traffic Engineer/ Bicycle Program Coordinator, City of San Diego
Bob James, Bicycle and Pedestrian Coordinator, Caltrans, San Diego
Sherry Ryan, Associate Professor/ Planner, San Diego State University
Steve Ron, Project Manager, San Diego County DPW
Chris Schmidt, Senior Planner, Caltrans, D- 11
Stephan Vance, Senior Regional Planner, SANDAG
Andy Hamilton, WalkSanDiego
Kristen Mueller, WalkSanDiego
Meeting Schedule and Conference Presentation Dates and Summary
During the duration of the Seamless Travel Project the following meetings and presentations were held:
Date Meeting Summary
January 18, 2007 Stakeholder Meeting Kick- off meeting held with TAG and
stakeholder group to introduce all to the
project and to solicit information from the
stakeholders on work that has already been
done in San Diego County regarding
bicycle and pedestrian counts and surveys.
March 19, 2007 TAG Meeting The TAC reviewed the Statement of Work
for Seamless Travel through Task 5.
Review of selected count locations.
June 6, 2007 Stakeholder Meeting Michael Jones presented a PowerPoint
summarizing the count location selection
and initial count and survey data.
Seamless Travel February 2010 19
Caltrans Task Order 6117
Date Meeting Summary
June 6, 2007 California Bicycle Advisory
Committee
Lauren Ledbetter presented an update on
the Seamless Travel Project to the CBAC.
Comments regarding the methodology
were incorporated as appropriate into
project.
August 7, 2007 ITE Annual Meeting,
Pittsburgh, PA
Lauren Ledbetter presented the Seamless
Travel methodology and preliminary data
collection efforts in “ Estimating Bicycle
and Pedestrian Demand”
September 18, 2007 TAG Meeting Michael Jones presented a PowerPoint
summarizing the project to- date, count and
survey methodology, preliminary count and
survey data, modeling options and next
steps.
January 16, 2008 Transportation Research Board
Annual Meeting
Lauren Ledbetter presented the Seamless
Travel methodology and the data collection
and survey results in “ Estimating Bicycle
and Pedestrian Demand in San Diego
County”
January 30, 2008 CalPed Meeting Michael Jones presented an update on the
Seamless Travel Project to the California
Ped Committee.
November 12, 2008 TAG Meeting Michael Jones presented a PowerPoint
summarizing the project to- date, count and
survey findings, inital modeling steps.
March 5, 2009 TAG Meeting Michael Jones presented a PowerPoint
summarizing the modeling outputs and
potential data uses..
February 3, 2010 TAG Meeting Michael Jones presented the findings,
conclusions, and potential applications
PROJECT OBJECTIVES
Background
One of the greatest challenges facing the bicycle and pedestrian field is the lack of documentation on
usage and demand. Without accurate and consistent information on demand and usage, it is difficult to
measure the positive benefits of investments in these modes, or to compare them to other transportation
modes such as the private automobile.
Existing data sources such as the U. S. Census Journey- to- Work, and the National Household Travel Survey11
document aspects of biking and walking ( mostly as they relate to work commute trips of employed
adults or national/ regional travel behavior). These resources miss much of the actual bicycling and
walking activity in our communities— such as trips made by students, utilitarian trips, and linked trips,
and they do not tell us where we could expect to find pedestrians/ bicyclists ( trip distribution) or how
11 U. S. Department of Transportation, Bureau of Transportation Statistics, 2000
Seamless Travel February 2010 20
Caltrans Task Order 6117
many pedestrians/ bicyclists we would find at any specific
location. The data sources also may not represent a true cross
section of user groups or provide sufficient detail on
background elements ( such as destinations and origins or
frequency) that could provide insight into behavior.
Locally, counts and surveys conducted by agencies around the
state and country are done with no consistent methodology that
would allow researchers to understand bicycle and pedestrian
activity trends and relationships to physical and social factors.
The result is a limited understanding of the role of bicycling and
walking as transportation modes, difficulty in projecting future
use, difficulty in measuring developing collision rates, and a lack
of understanding of how factors such as facility type, climate,
topography, land use, and income influence activity levels.
Without bicycle and pedestrian usage information,
transportation professionals may have difficulty justifying new
bicycle and pedestrian investments, may undercount bicycling
and walking in regional modeling efforts, and may undervalue
the transportation, safety, economic, and health benefits of
bicycle and pedestrian infrastructure.
Goals and Objectives
The key goals of the Seamless Travel Project are to:
( a) Evaluate existing bicycle and pedestrian data sources and collection methods
( b) Conduct comprehensive counts and surveys of bicyclists and pedestrians in a consistent manner
using the National Bicycle & Pedestrian Documentation Project12 as a template
( c) Conduct counts and surveys using San Diego County as a model community
( d) Analyze how bicycle and pedestrian activity levels relate to facility quality, and other factors such
as land use and demographics
( e) Identify factors that are highly correlated with increased bicycling and walking
( e) Provide methods for quantifying usage and demand that will enhance research on benefits and
exposure, and
( f) Evaluate how the transit- linkage can be improved.
At the completion of this project a report will be produced on trends in walking and bicycling; how
usage relates to items such as facility quality, physical factors, and social factors; and the potential for
land- use and infrastructure improvements to increase walking and bicycling. The research will provide
Caltrans staff, local agency staff, advocates, elected officials, and others with the information and tools
needed to understand walking and bicycling rates, patterns, relationships, and trends.
12 National Bicycle and Pedestrian Documentation Project, Jones, M., Buckland, L., Cheng, A., Transportation Research Board, Aug. 2005
What factors influence
bicycling and walking?
Seamless Travel February 2010 21
Caltrans Task Order 6117
The Seamless Travel Project is designed to meet these goals through the following objectives and
performance criteria.
Goal 1: Evaluate existing bicycle and pedestrian data sources and collection methods
Objective 1.1. Work closely with local agencies, staff, and organizations to maximize the
efficiency of the data collection and analysis process.
Objective 1.2. Evaluate existing bicycle and pedestrian data sources to determine the data quality,
methodology used, and suitability of using these sources for time- related analyses.
Objective 1.3. Use existing bicycle and pedestrian data sources and collection methods to inform
the data collection methods used in this research project.
Objective 1.4. Identify and evaluate automated and manual count techniques, and develop
recommendations on the best applications and their related advantages and limits.
Goal 2: Conduct comprehensive counts and surveys of bicyclists and pedestrians in a consistent
manner using the National Bicycle & Pedestrian Documentation Project as a guide
Objective 2.1. Utilize National Documentation Project’s ( NBPD) existing methods, forms,
training, dates and times, location requirements, surveys, and other materials as a starting point,
allowing research team to facilitate data collection.
Objective 2.2. Refine the NBPD methodology as needed to ensure that the other goals are met.
Objective 2.3. To the extent possible, structure the data collection methodology to allow
integration of bicycle and pedestrian data into pre- existing local, regional, or statewide modeling
efforts, including the NBPD.
Goal 3: Conduct counts and surveys using San Diego County as a model community
Objective 3.1. Work with a local stakeholders group to ensure that the count and survey
collection reflects local knowledge and stakeholder’s interests.
Objective 3.2. Ensure that the counts and surveys reflect a diversity of facility types, demographic
groups, economic groups, and land- use types.
Objective 3.3. Build on past count and survey efforts in San Diego County, to provide a database
and model that allows for the study of trends, patterns, and relationships, with applications for
the rest of the State.
Goal 4: Analyze how bicycle and pedestrian activity levels relate to facility quality, and other
factors such as land use and demographics
Objective 4.1. Use GIS data from SanGIS, SANDAG, the U. S. Census and other sources to
relate activity levels to land use, facility type, and demographics.
Objective 4.2. Utilize spot field visits and aerial maps to verify and categorize facility quality.
Objective 4.3. Collect representative trip type and demographic data using surveys to identify
non- physical factors that may affect bicycle and pedestrian activity levels.
Seamless Travel February 2010 22
Caltrans Task Order 6117
Goal 5: Identify factors that are highly correlated with increased bicycling and walking
Objective 5.1 Utilizing historic data and data collected during the research project, employ
regression analysis to identify any factors highly correlated with increased bicycling and walking.
Objective 5.2. Develop a methodology for rating and categorizing items that are related to bicycle
and pedestrian activity levels, including a methodology for categorizing qualitative factors such as
facility quality.
Goal 6: Provide methods for quantifying usage and demand that will enhance research on
benefits and exposure
Objective 6.1. Develop an Online Database that will allow all collected data to be studied by the
research team, Caltrans, local agencies, and other research institutions.
Objective 6.2. Using high correlation factors identified during the course of research, develop
Bicycle and Pedestrian Demand Models that can help predict bicycle and pedestrian activity
levels at specific locations, for use in planning, exposure and collision analysis, design, and
management of non- motorized facilities.
Objective 6.3. Develop a Technical Report that provides an overview of the research project,
objectives, methods used, summary of results in text and tabular format, analysis of correlations,
trends, and patterns, conclusions on the accuracy and applicability of the data, and
recommendations on increasing walking and bicycling in California.
Objective 6.4. Develop a Training Manual for use by Caltrans and local agencies for conducting
bicycle and pedestrian counts and surveys in their communities.
Objective 6.5. Develop a PowerPoint presentation summarizing the research, conclusions, and
recommendations of the research that can be used by Caltrans and other organizations for
presentations.
Goal 7: Evaluate how the transit- linkage can be improved
Objective 7.1. Develop a Summary Report that includes information about preferences for
different types of bicycle facilities, potential for increased transit- linked trips, estimations of
benefits, and meeting the specific objectives of the California Blueprint.
Objective 7.2. Include count and survey locations that are near transit stops and use transit stop
and route characteristics in analyzing the count and survey data.
The consistent, comprehensive data on walking and bicycling produced through the National
Documentation Project, which now has data from over 60 agencies nationwide, will allow researchers to
address the following:
 Trends in walking and bicycling
 Exposure data for collision analysis
 Preferences for facility types by users
 Role of walking and bicycling in local and regional transportation modeling efforts
Seamless Travel February 2010 23
Caltrans Task Order 6117
 Developer responsibilities for bicycle and pedestrian impacts and mitigations
 Land- use planning and urban design to support walking and bicycling
 Documentation of health, economic, and other benefits
 Adequate facility design to meet user needs
 Documentation of usage and benefits for funding.
Seamless Travel February 2010 24
Caltrans Task Order 6117
This page intentionally left blank
Seamless Travel February 2010 25
Caltrans Task Order 6117
2. SYNTHESIS OF PUBLISHED RESEARCH
REVIEW OF EXISTING COUNT AND SURVEY METHODS
Interest in bicycle and pedestrian modes as a small but important component of the multi- modal
transportation system has been growing since the adoption of the Intermodal Surface Transportation
Efficiency Act ( ISTEA) in the early 1990s. A combination of increased interest in resolving traffic
congestion, building livable communities and streets, supporting more active and healthy lifestyles,
enhancing pedestrian and bicyclist safety, and encouraging Safe Routes to Schools, has resulted in a
desire and need to accurately measure bicycling and walking rates, collision rates, and to understand why,
when, and where people walk or bicycle. Furthermore, standardized pedestrian and bicycle data
collection and analysis techniques are important factors for elevating the status of planning and funding
for these travel modes.
EXISTING DATA SOURCES
The lack of consistent data on bicycling and walking is commonly cited, and is probably the single
greatest impediment to being able to understand these modes. In 2000, the Bureau of Transportation
Statistics published a report summarizing the existing bicycle and pedestrian data sources and the
importance, quality and usefulness of this data. According to the report Bicycle and Pedestrian Data: Sources,
Needs & Gaps, national data is commonly available, but consistent state, regional and local data is not.
The report notes that data quality ranges from fair to poor ( Bureau of Transportation Statistics, 2000).
On a national level, the U. S. Census Journey- to- Work,
National Survey of Bicyclist and Pedestrian Attitudes and
Behavior ( NHTSA), and the National Household Travel
Survey provide the only readily available, consistent
bicycle and pedestrian count and survey
information. These sources provide good
background information on bicycling and walking,
but either ( a) provide information on a limited part
of these trips or ( b) provide national level data only.
Due to its data collection methodology, the U. S.
Census often undercounts the actual number of
walking and biking trips made in a locality. The
census data only counts commute trips, leaving out
the significant number of people who bicycle or
walk for recreation, to conduct personal business,
or to socialize. Additionally, the Census long- form, which is used to gather journey to work information,
requires that respondents choose only one mode. As a result, multi- modal trips, such as walking to
transit, are not counted as a walking trip ( California Department of Transportation, May 2002).
The National Household Travel Survey ( NHTS) provides useful information on household- based trip
making. The NHTS selects a random sample of U. S. households and asks each to complete a travel
diary. All types of trips are collected, not just commute trips, and every component of a multi- modal trip
is captured. However, the NHTS uses a smaller sample size than the U. S. Census, and is only useful at a
national level. Recently, the NHTS has expanded its add- on program, which allows states and
Bicyclists using an overcrossing
Seamless Travel February 2010 26
Caltrans Task Order 6117
metropolitan planning organizations to purchase additional sample surveys for their area. Caltrans
purchased an add- on for the San Diego area for 2008.
The National Survey of Bicyclist and Pedestrian Attitudes and Behavior ( NHTSA) provides detailed information
on walking and bicycling that compliments the NHTS and studies of aggregate ( area wide) walk and bike
trips. The NHTSA conducted telephone interviews of non- institutionalized people 16 years or older in
the summer of 2002. Participants provide information about their bicycling and walking behaviors in the
most recent 30 days. The data cannot estimate future activity but offers a summary of activity in the
summer months.
As with any survey that relies on a subset of a population, sampling error may affect the accuracy of the
Census and the NHTS data. Both the Census Long Form ( which collects the journey- to- work data) and
the NHTS use samples of the population, and may under represent or omit subgroups of the population.
This is especially pertinent for bicycle commuting data, for which the mode share is usually less than
1%. 13
The quantity and quality of regional and local bicycle and pedestrian data vary. State, regional and local
data collection efforts are generally tailored to suit the specific needs of the community or project being
evaluated ( Greene- Roesel et al. 2007). The Bureau of Transportation Statistics notes that, “ While a few
cities and metropolitan planning organizations routinely conduct pedestrian and bicycle counts, most
collect them only sporadically for specific studies or do not collect them at all”( Bureau of Transportation
Statistics 2000). In California, it is common for metropolitan planning organizations or regional
transportation planning agencies to collect regional travel surveys. Though these surveys generally focus
on motor vehicle trips, most have a mode share component.
PEDESTRIAN AND BICYCLE RESEARCH EFFORTS
Despite the lack of coordination among agencies, it is recognized that developing a coherent bicycle and
pedestrian data collection system is important for non- motorized planning, project development,
encouragement activities, and funding. The Bureau of Transportation Statistics notes that “ certain types
of data, such as numbers of trips by facility and user type, are potentially useful to a wide range of user
groups; but coordination among these groups is required to establish standardized, mutually beneficial
data collection procedures.” To offset the high cost of collecting data, agencies are relying on innovative
solutions, such as automated count technology or incorporating non- motorized data collection into
existing traffic data collection procedures.
NATIONAL BICYCLE AND PEDESTRIAN DOCUMENTATION PROJECT
The National Bicycle & Pedestrian Documentation Project ( NBPD) is an effort led by Alta Planning +
Design, in collaboration with the ITE Pedestrian & Bicycle Council, in response to the lack of useful
data on walking and bicycling. While other modes such as motor vehicles have established conventions
to collect and use data ( such as trip generation for traffic modeling), the lack of consistent data for the
walking/ bicycling modes has made it difficult to justify funding, justify the allocation of capacity and
right- of- way, develop exposure rates, among other issues.
13 Using Journey to Work data from the U. S. Census 2000, the bicycle mode share for the United States is 0.40% and the bicycle mode share for
California is 0.80%.
Seamless Travel February 2010 27
Caltrans Task Order 6117
The concept for the NBPD is very simple:
1. Provide materials and directions to agencies to conduct consistent counts and surveys,
2. Provide standard count dates and times,
3. Provide a location where this information can be sent,
4. Make this information available to the public.
The count and survey materials and methods have been evolving as more groups and researchers learn
about the program, and determine their own unique needs for the information.
As NBPD moves forward it will have four basic primary applications: ( 1) safety – through exposure
analysis, ( 2) trip generation— as part of impact analysis, land use and transport policy, ordinances, etc.,
( 3) monitoring – identifying changes and trends in overall activity use, and ( 4) modeling – projecting
existing/ future activity, identifying the relationship between walking/ bicycling and land use, multi- modal
analysis, demographics, etc.
COUNT METHODOLOGIES
Bicycle and pedestrian counts are generally conducted either
through manual counts or through automated counts. Many
communities have combined manual counts with existing
motorized vehicle counts at little or no extra cost. Manual
counts are typically conducted by two counters per
intersection, though a third person may be needed at busier
intersections. Manual counts allow for collection of
additional information, including type of users, use of
helmets, turning movements and gender. ( Schneider, Patton
et al.) Manual count methods include using a tally sheet, an
electronic board, a non- electronic counting board with
periodic manual tallying, and using a handheld counter with
periodic manual tallying.
Automated technologies are useful in conducting longer- term counts and establishing daily, weekly, or
monthly variations in usage. With the exception of video playback systems, automated technologies
generally require fewer person- hours than manual counts. The most common automated technologies
used for non- motorized data collection are:
 Passive infrared ( detects a change in thermal contrast)
 Active infrared ( detects an obstruction in the beam)
 Ultrasonic ( emits ultrasonic wave and listens for an echo)
 Doppler radar ( emits radio wave and listens for a change in frequency)
 Video Imagining ( either analyzes pixel changes or data are played back in high speed and
analyzed by a person)
 Piezometric ( senses pressure on a material either tube or underground sensor)
 In- pavement magnetic loop ( senses change in magnetic field as metal passes over it)
Automated counter
Seamless Travel February 2010 28
Caltrans Task Order 6117
Most automated technologies work well for counting users that pass a specific point but, with the
exception of active infrared and time lapse video technologies, cannot easily distinguish between
bicyclists and pedestrians ( Beckwith and Hunter- Zaworski 1997; Wolter and Lindsey 2001). Time- lapse
video has been used in Davis, California to capture user type, demographic information, and behavior
( Schneider et al. 2005). The Massachusetts Highway Department successfully modified an active
infrared traffic sensor and developed custom software to count and classify bicyclists and pedestrians.
The sensor was able to accurately count 97% of bicyclists and 92% of pedestrians, and accurately
classified 77% of bicyclists ( Noyce and Dharmaraju 2002). A combination of technologies such as Eco-
Counter’s Eco- Multi, can also distinguish between types of users.
All automated count technologies have an error factor, with no- detection rates varying from 5% to 45%,
depending on environmental conditions and usage patterns ( Beckwith and Hunter- Zaworski 1997). Trail
counts in Indiana using infrared traffic counters found the infrared sensors systematically
underrepresented users by 15% ( Wolter and Lindsey 2001). A Portland, Oregon study tested the
accuracy of three types of pedestrian sensors: passive infrared, Doppler radar and ultrasonic. The
sensors were tested under a variety of conditions, and were found to have varying error rates and could
be susceptible to adverse weather conditions ( Beckwith and Hunter- Zaworski 1997). Comparing
automated counts with manual counts allows researchers to correct for inherent error rates.
Ultimately, the decision to use automated or manual count technologies depends on the duration of the
count effort, the existence of other ongoing count efforts, the type of data that are to be collected, the
number of person- hours available for data collection and analysis, and the overall budget of the count
effort. Automated count technologies have a higher start- up cost than manual count technologies,
though they generally require fewer person- hours than manual counts and can mean long- run cost
savings. Manual counts require more person- hours than automated counts, but can collect additional
characteristics of bicyclists and pedestrians. A summary of manual and automated counts characteristics
is provided in Table 4.
Table 4: Manual and Automated Count Characteristics
Manual Counts Automated Counts
Integrating pedestrian and bicycle counts with
existing motor vehicle counts can reduce costs
Field observations are labor- intensive, which may
limit the number of count locations
Observations have a higher level of accuracy, and
can be more complex than automated counting
methods ( i. e., can include behaviors and other
characteristics of users)
Technologies can significantly reduce labor costs
Settings and positioning of devices must be
adjusted to maximize accuracy
Placement should minimize interference with
pedestrians and bicyclists and potential for
vandalism
Most technologies work in rain and a wide variety
of temperatures
Many technologies allow for remote data
download
Most technologies do not count all types of non-motorized
users and few can be used to observe
behaviors
Source: ( Schneider, Patton et al. 2005)
Seamless Travel February 2010 29
Caltrans Task Order 6117
PEDESTRIAN AND BICYCLE TRAVEL BEHAVIOR SURVEY METHODS
Bicycle and pedestrian surveys are useful to understand why people are walking and bicycling, to collect
socio- demographic information, and to discern attitudes about walking, biking and facilities. Surveys are
generally conducted either as a sample of the general population, or targeted specifically to non-motorized
users. Surveys have been criticized for two common shortcomings. First, surveys frame the
questions and limit the possible responses, thus increasing the chance that unexpected responses will be
unrecorded or that questions will be misunderstood. Second, traditional survey collection methods, such
as travel diaries and phone recruitment can under represent certain population groups, such as the elderly
and the poor. Clifton and Handy ( 2001) recommend using focus groups to test survey reliability and
ensure they are worded so that the target audience understands the questions. Survey respondents
should be compared with the population being sampled, and underrepresented segments of the
population may need to be reached through different channels.
Schneider et al. ( 2005) summarize key differences in travel surveys based upon general population
sampling and targeted sampling. These findings are summarized in Table 5.
Table 5: Characteristics of General and Targeted Surveys
Samples of the General Population Targeted Surveys
Results of well- executed random- sample
surveys can represent the entire community
Results can provide baseline and follow- up data
for the community as a whole
Potential participants should be identified using
a random selection procedure
Survey instrument design and survey
distribution techniques are critical to achieving
a high response rate and representative results
Gathering and analyzing responses can be
labor- intensive
Agency can obtain detailed characteristics about
people who make non- motorized trips
Results can provide baseline and follow- up data
about non- motorized users
Differences between survey participants and the
overall population are important to recognize
Survey instrument design and survey distribution
logistics are critical to the quality of the survey
Labor costs can be high, unless volunteers are
recruited
Source: ( Schneider, Patton et al. 2005)
Short intercept surveys can be supplemented by longer take- home surveys. In 2002, the Rhode Island
Department of Transportation conducted user surveys on six bicycle paths, where groups of users were
intercepted and a short survey was administered to persons willing to participate. The on- path survey
asked for the participant's street address or email so a paper copy of a longer survey, or a web link to the
longer survey could be sent to the participant. The survey collected information on mode of access to
the path, time spent and distance traveled on the path, usage by time of day, day of week and season, and
use of the path for commuting ( Gonzalez et al. 2004).
To reduce costs, the Rhode Island survey used University of Rhode Island students and volunteers to
conduct the surveys. Students and volunteers were given detailed instructions on how to introduce
themselves, identify their purpose, and describe the two- phase survey. According to the summary
report, interviewers felt the experience was " pleasant" and that most people on the path were
" enthusiastic users" ( Gonzalez et al. 2004).
Seamless Travel February 2010 30
Caltrans Task Order 6117
Abraham et al. ( 2002) used a stated preference survey to determine cyclist’s route choice preferences.
The intention of the survey was to develop parameters that could be used in the City of Calgary’s travel
demand model. The survey was distributed by email to downtown bicyclists who had participated in a
prior survey and were willing to be contacted again. The survey found that bicyclists strongly preferred
off- street bicycle facilities and low- traffic residential roads.
The National Survey of Pedestrian and Bicyclist Attitudes and Behaviors conducted for the U. S.
Department of Transportation’s National Highway Traffic Safety Administration ( NHTSA) conducted
telephone interviews. Random phone surveys reach a more representative sample however it is limited
to participants with a phone and is expensive to administer. The survey found respondents did not use
multi- use paths and bike lanes because they were either not convenient or did not go where the bicyclist
wanted to go.
BICYCLING AND PEDESTRIAN TRAVEL MODELING
Recent research studying the link between walking and
environmental factors has found that certain
environmental factors such as land use and sidewalk
completeness are positively correlated with pedestrian
volumes ( Berke et al. 2007). However, these studies
have not clearly demonstrated a causal link between
environmental factors and pedestrian activity ( Handy
1991; Boarnet and Crane 2001). In an Austin, Texas
study Cao, et al. ( 2006) demonstrated that residential
self- selection plays a role in walking rates, especially in
utilitarian walking ( e. g. walking to the store). In other
words, people who prefer to walk to the store may
move to neighborhoods that are more walkable. There
is still a question about the causal link between walking
and the built environment. For planning purposes,
creating a built environment that supports walking should generally increase walking rates, though it may
do so in part by attracting “ walkers” to a neighborhood.
The Austin study suggests that recreational walking, like strolling, is affected by the residential built
environment, while utilitarian walking is more affected by the destination’s built environment ( e. g. store
quality and proximity).
FOUR- STEP MODELING PROCESS
Transportation models fall under two groups: aggregate models or disaggregate models. Aggregate
studies model travel behavior based on the characteristics of an area ( e. g. population density,
employment density, household income, facility type). Disaggregate studies model travel behavior from
the perspective of individual travel choices. These models apply individual characteristics and
preferences ( e. g. attitudes, trends related to gender or age) to a population with known characteristics to
predict travel behavior.
Market Street and 5th Avenue, San Diego
Seamless Travel February 2010 31
Caltrans Task Order 6117
Aggregate and disaggregate models differ in their ease of use and predictive abilities. Aggregate models
can be developed using readily available data and methods. Disaggregate models are more complicated
to develop and require custom data and survey collection, but are more effective at predicting travel
behavior ( Federal Highway Administration 1999).
Regional transportation modeling and forecasting began in the 1950s with the growing need to predict
and plan for expected increases in population, vehicle ownership and vehicle miles traveled. The passage
of the 1963 Federal Aid Highway Act institutionalized regional transportation planning by requiring that
urban areas employ a “ continuing, comprehensive and cooperative” transportation planning process.
Since these beginnings, institutionalized transportation models have been modified to reflect changing
social patterns and new environmental regulations and conformance requirements. The model
commonly used today is the four- step Urban Transportation Model System ( UTMS) ( Pas 1995).
The UTMS takes transportation system characteristics and land- use system characteristics as inputs, uses
four sub- models to determine trip generation, trip distribution, trip mode choice and trip assignment,
and produces an estimate of the volume and speed of traffic on the transportation network. The four
sub- models are commonly run in the sequence described below ( Pas 1995; Meyer and Miller 2001).
Step 1: Trip Generation asks: “ How many trips?” and predicts the number of trips produced by and
attracted to each area of analysis. This number is calculated based on the land- use type, intensity of the
use, and the socioeconomic characteristics of the activities using the land.
Step 2: Trip Distribution asks: “ Where do trips go?” and links each trip generated in step one to an origin
and a destination. The gravity model is the most commonly used method for distributing trips. The
gravity model calculates the number of trips from an origin to a destination based on ( 1) the number of
trips leaving a destination, ( 2) the attractiveness of the destination, and ( 3) the difficulty ( friction) of
traveling from the origin to the destination.
Step 3: Trip Modal Split asks: “ How do people get there?” and predicts the percentage of travel that will
use each mode between origins and destinations. Mode choice is estimated in two common ways. The
first, an aggregate model, links the mode split to the characteristics of the transportation system ( e. g.
transit frequencies, relative speed of biking or walking vs. driving) and the characteristics of the users
( e. g. average auto ownership, age, average income). The disaggregate model is concerned with the travel
behavior of individuals. These models link an individual’s choice to the characteristics of all mode
choices available for that trip ( such as travel cost, travel time) and the characteristics of each individual
( such as auto ownership, average income).
Step 4: Trip Assignment asks: “ What route will people take?” This step predicts the route that each trip
will take from each origin to each destination. The model considers attributes of the route, including
travel time and distance, number of stops, aesthetic appeal, but travel time is the most commonly used
attribute.
The four steps described above represent a sequential decision making process: Should I make a trip?
Where should I go? Should I drive, walk, bike, or take the bus? What route should I take? This process
has been criticized as a “ highly unrealistic representation of traveler’s decision making,” but the intention
of the four- step model is not to model individual trip decisions, but to provide a “ pragmatic approach to
reducing the extremely complex phenomenon of travel behavior into analytically manageable
Seamless Travel February 2010 32
Caltrans Task Order 6117
components” ( Meyer and Miller 2001). Some four- step models switch the order of steps two and three,
performing the modal split before distributing the trips.
Historically, transportation modeling has been focused on highway or transit networks, and considers
just two modes: private vehicles and public transportation ( Sheppard 1995; Meyer and Miller 2001).
Factors that could influence the decision to walk or bike are not usually included in the four- step
process. When developing a non- motorized transportation model, or when incorporating non- motorized
transportation into a traditional four- step model, several factors should be considered, as outlined in
Table 6.
Though walking and bicycling are often lumped together, there are significant differences between the
two modes. Most models that are developed for forecasting non- motorized transportation are
developed specifically for bicyclists or pedestrians. Three of the most significant differences between the
two modes are:
( 1) Walking trips are generally shorter than bicycling trips. This may affect the spatial scale of analysis.
( 2) A large percentage of walking trips are trips to access other modes, including the automobile or transit.
Bicycle trips are generally stand- alone trips. Modeling should consider the fact that pedestrian
trips may not replace automobile trips, but may result from those trips. Conversely, the quality
of the walking environment may need to be considered in predicting transit mode shares.
( 3) The decision to ride a bicycle involves a greater conceptual leap than the decision to walk. Public health and
social marketing fields have shown that the decision to even consider riding a bicycle is a multi-staged
process involving a variety of interacting personal, social and environmental factors.
Attitudinal research is important for modeling and understanding pedestrian travel, but is
perhaps most significant for bicycle travel ( Federal Highway Administration 1999).
Methods for modeling non- motorized travel are more varied than those used for motor vehicle and
transit modeling. Methods range from comparative studies to incorporation into regional four- step
demand models. Several common types of models are described in Table 7.
Seamless Travel February 2010 33
Caltrans Task Order 6117
Table 6: Selected Factors Influencing Non- Motorized Travel
Variable Description
Link Characteristics Measurable characteristics of a link in a roadway or pathway network ( e. g.,
traffic volume, lane width, or pavement quality)
Link “ Friendliness” The overall acceptability of a link as a bicycle or pedestrian route – a
function of link characteristics. Also varies by user characteristics ( e. g.,
experiences vs. novice bicyclist.)
Network
Characteristics
Characteristics of a network of links ( e. g., connectivity) that determine its
overall acceptability or “ friendliness” to the user
Network
“ Friendliness”
A general measure of how acceptable the local road/ path network is for
bicycling or walking
Supporting Policies Other programs, policies, facilities, etc,. which affect the acceptability of
bicycling or walking ( e. g. bicycle parking, showers/ lockers, and
educational programs)
Population
Characteristics
Characteristics of the local population which relate to likelihood of
bicycling or walking ( e. g. socioeconomic characteristics or attitudes)
Climate/ Weather General propensity to walk or bicycle, as a function of climate/ weather.
This might be considered a constant for a given area/ region
Characteristics of
Other Modes
Relative travel times and costs of bicycling or walking vs. other modes, as
well as safety, comfort, or other factors which influence choice of mode.
Policy variables might include parking pricing, transit service
improvements, etc.
Land Use Density and distribution characteristics of population, employment,
shopping, and other activities which affect where people travel, how many
trips are generated, trip length, etc.
Topography Where it is significant, topography will influence the travel patterns of
pedestrians, with people selecting more level routes even when they are
less direct
Aesthetics Bicyclists and pedestrians will typically choose a route that is more
aesthetic ( shade trees, views, lower traffic), even if is not direct. In some
cases, bicyclists/ pedestrians will deliberately seek out these types of
facilities for recreation/ exercise
Transit Access Accessibility to transit especially impacts pedestrian trip making, since all
transit trips begin and end with a pedestrian trip
Source: ( Federal Highway Administration 1999)
Seamless Travel February 2010 34
Caltrans Task Order 6117
Table 7: Methods for Modeling Non- Motorized Travel Demand
Purpose/ Method Description
Demand
Estimation
Methods that can be used to derive quantitative estimates of demand
Comparison
Studies
Methods that predict non- motorized travel on a facility by comparing it to
usage and to surrounding population and land- use characteristics
Aggregate
Behavior Studies
Methods that relate non- motorized travel in an area to its local population, land
use, and other characteristics, usually through regression analysis
Sketch Plan
Methods
Methods that predict non- motorized travel on a facility or in an area based on
simple calculations and rules of thumb about trip lengths, mode shares, and
other aspects of travel behavior
Discrete Choice
Models
Models that predict an individual’s travel decisions based on characteristics of
the alternatives available to them
Regional Travel
Models
Models that predict total trips by trip purpose, mode, and origin/ destination,
and distribute these trips using a gravity ( time/ distance) formula across a
network of transportation facilities, based on land- use characteristics such as
population and employment and on characteristics of the transportation
network
Sources: ( Schwartz et al. 1999; Federal Highway Administration 1999)
Pas notes that “ even mathematical models of travel and related behavior implicitly employ subjective
judgments and reflect particular perspectives on human behavior”( Pas 1995). The FHWA recommends
that for both disaggregate and aggregate models, “ it is important to remember that decision making
ultimately occurs at the individual level and that a forecasting procedure should approximate the
individual decision- making process as closely as possible ( Federal Highway Administration 1999).
Additionally, the validity of model outputs is related to the quality of the data inputs.
Collecting high quality non- motorized bicycle and pedestrian data will allow modelers to more accurately
estimate walking and biking.
NON- MOTORIZED TRANSPORTATION FORECASTING EFFORTS
Forecasting models of bicycle and/ or pedestrian travel has been developed by several researchers and
groups nationwide since the Seamless Travel project started in 2007, with notable efforts in Portland,
Oregon ( Columbia River Crossing, CRC Transportation Planning Team, 2008) and in Alameda County,
California ( Traffic Safety Center, Schneider, Arnold, Ragland, 2008). Both of these modeling efforts
advanced the state of non- motorized forecasting by using extensive count data, which provides
significantly more realistic basis than previous efforts.
The Columbia River Crossing project was part of a major corridor study of a proposed new crossing of the
Columbia River between Portland, Oregon, and Vancouver, Washington. A model was developed to
forecast future bicycle and pedestrian trips across the new bridge using a combination of U. S Census
mode share, travel surveys, a bicycle trip study conducted by Portland State University, and travel
characteristics on a nearby bridge ( Hawthorne Bridge). The model uses total forecasted trips on the new
bridge from the regional travel demand model, and assigns a mode split to those forecast trips of five ( 5)
miles or less for bicycles ( 2 miles or less for pedestrians), based on local survey results. The model
Seamless Travel February 2010 35
Caltrans Task Order 6117
forecasts a 650% increase in pedestrian trips and a 150%
increase in bicycle trips. The assumption behind the
model is that a straight line correlation exists between
vehicle and bicycling/ walking trips based on travel
time/ trip length, assuming the quality of the facilities
remains the same or improves.
The Alameda County forecasting model developed by
the U. C. Berkeley Traffic Safety Center ( A Pilot Model for
estimating Pedestrian Intersection Crossing Volumes, 2008) is
based on pedestrian counts at 50 locations and specific
variables including total population within .5 mile radius,
employment within a .25 mile radius, number of
commercial retail properties within .25 miles, and the
presence of a regional transit station within .1 miles of
the count location. The ‘ r’ value for this combination
of variables was .987.
In referring to previous pedestrian modeling efforts
including the Space Syntax Model, the pedestrian model
created for Manhattan ( Cameron) and Milwaukee
( Benham and Patel), the study states that “ few studies to
date have used continuous counts to account for daily,
weekly, and seasonal variations in pedestrian activity or
capture the effects of weather and other factors on
pedestrian volumes.”
The study selected 50 intersections in a variety of settings for its count locations, eliminating locations in
low density areas due to the potential for high variability. Each leg of an intersection was counted
separately, with some pedestrians being counted more than once. Infrared counters were installed to
conduct 24- hour a day counts, and calibrated with manual counts. Counts were conducted over a 13-
week period. Over 40 different potential variables were considered and tested using GIS mapping tools
and regression analysis.
CONCLUSION
Each of the data sources and research efforts described in this chapter provides another piece in the
puzzle to understand bicyclist and pedestrian travel. It is clear from the research that there are three
basic types of data and forecasting tools:
Area Wide ( Aggregate) Trips:
Using household daily trip generation and available travel and demographic information, it is possible to
develop estimates of area wide ( or national) bicycle and walking trips. This information can be used for
area wide planning and other purposes, such as the Non- Motorized Transportation Pilot Project.
Forecasting future trips in Portland, Oregon
Seamless Travel February 2010 36
Caltrans Task Order 6117
Land Use Based Trips:
Travel estimating for vehicles ( using the ITE Trip Generation Manual) is based almost exclusively on this
type of analysis. This data is then used as part of the four step modeling process to create traffic models,
assess impacts, and measure Level of Service. ITE has initiated a land use based trip generation data
collection effort for walking and bicycling trips, but is application and use is unknown at this point.
Corridor or Specific Location Estimating
While the land use- based trip generation techniques described above are used as the basis for vehicle
traffic models which can provide estimates of specific location and corridor volumes, no such validated
model exists today for bicycling and walking trips. Advances have been made in some areas ( Columbia
River Crossings, Alameda County) but no model has yet been based on data collected for a long period
of time ( at least one year) and over a large geographical area for both modes.
The Guidebook on Methods to Estimate Non- Motorized Travel ( 1999, Vol. 1, Section 4) states that “ further
development of modeling techniques and data sources are needed to better integrate bicycle and
pedestrian travel into mainstream transportation models and planning activities.” This research effort
seeks to enhance the existing sources of bicycle and pedestrian data within the San Diego region.
Seamless Travel February 2010 37
Caltrans Task Order 6117
3. PRIMARY DATA COLLECTION
This chapter addresses the count and survey data collection effort conducted during Years One and Two
of the Seamless Travel Project.
WHY SAN DIEGO COUNTY?
San Diego County was chosen as a model
community for two reasons. First, regular bicycle
counts were conducted throughout the county in
1985, 1987, 1990, 1993, and 1997. Count
locations remained the same from year- to- year,
with the addition of new count locations in later
years. The original set of count locations was
randomly selected from the existing and proposed
county bicycle network. This historic bicycle
count data can be used to test and evaluate the
counts and correlations identified by the Seamless
Travel Project. Second, San Diego County has an
extensive, frequently updated countywide GIS
database that is freely available. Historic GIS
information is also available, allowing a
comparison of historic bicycle counts to historic
land uses.
The research team worked closely with local
agencies, staff, and organizations to maximize the
efficiency of the data collection and analysis
process. Representatives from several local
agencies were invited to participate in a local stakeholder team. This team provided input into methods
and also provided valuable local expertise. The following agencies were represented: San Diego
Association of Governments, City of San Diego, San Diego County, WalkSanDiego, San Diego Bicycle
Coalition, Caltrans District 11 ( San Diego District) and Caltrans Headquarters.
COUNT METHODOLOGY
The Seamless Travel Project includes two ( 2007 and 2008) manual peak period counts at 80 locations
throughout San Diego County and one- year of automated 24- hour counts at five locations ( August 2007
to July 2008).
Count locations were based on ( a) historic count locations and ( 2) representative locations based on land
use ( urban, suburban, rural), demographics ( a full range of ethnic and income locations), and facility
types ( bike paths, streets with bike lanes, arterials, local streets). It was determined that a random sample
would require many more count locations than were possible given the project budget in order to cover
the range of desired land uses, demographics, and facility types. Instead, count locations were selected to
ensure that a variety of demographic and physical characteristics were represented. Using GIS analysis
Figure 4: Peak Hour Count Locations in
San Diego County
Seamless Travel February 2010 38
Caltrans Task Order 6117
and input from local stakeholders, a final set of 80 count locations ( 40 historic bicycle counts, 40 new
counts) was established. Count locations were chosen to represent:
 Presence and type of bicycle facilities, including no bicycle facility
 High pedestrian crash areas
 Areas identified for future smart growth
 Locations near transit stops ( trolley, bus, ferry)
 Locations near planned or recently completed bicycle and pedestrian projects
 Variety of land uses and demographics
All 17 jurisdictions within the county and the unincorporated county are represented in the count
locations. The count locations focus on the more populated, western half of the county. Error!
Reference source not found. displays the locations of the eighty peak period count locations across the
County of San Diego.
Peak period manual counts were conducted during the traditional peak hours ( AM weekday peak from 7
AM to 9 AM and midday weekend peak from noon to 2 PM) at all 80 count locations. Additional PM
peak ( 4 PM- 6 PM) manual counts were conducted in Year Two at 20 locations, with all 80 locations
counted at the conclusion of the study. The choice to count only one peak period for all locations was
due to budgetary constraints. The AM peak was chosen based on counts from the National Household
Travel Survey, Bay Area Travel Survey and southern California counts conducted by Alta that show
bicycle and pedestrian travel peaks at the same time during the AM peak, but during the PM peak,
pedestrian travel peaks earlier than bicycle travel.
AUTOMATED COUNT METHODOLOGY
In addition to peak- hour counts, the Seamless Travel Project collected automated year- long counts to
establish trends in bicycling and walking. After evaluating the various automated counting tools available
on the market, the research team decided to use a combination of passive infrared counters and active
infrared counters. Both count tools collect time- stamped data, contain their own power source, and
allow data to be downloaded to a computer for analysis. Active infrared counters allow bicyclists and
pedestrians to be classified. They are more challenging to install ( two units as opposed to one), but are
less expensive than passive infrared. Passive infrared counters do not classify bicyclists and pedestrians,
but only require one unit per installation. Passive infrared counters can classify counts by direction as
well.
Active infrared counters can be set up to detect the speed of travelers thereby allowing for an
approximate differentiation between bicyclists and pedestrians based upon assumed travel speeds for the
two modes. Two units are installed along a single corridor. One unit is set to trigger a count when the
traveler is moving at a low speed ( the pedestrian), while the other unit is calibrated to trigger when a
traveler is moving at a higher speed. The low- speed unit counts all pedestrians and bicyclists while the
higher speed unit counts only bicyclists. Pedestrian counts can be determined by subtracting the bicyclist
count from the combined count. The research team experimented in the field to determine the
Seamless Travel February 2010 39
Caltrans Task Order 6117
appropriate speed at which the two units will need
to be set, however the California Bicycle Advisory
Committee has suggested that 8 mph is a good
speed at which to start counting bicyclists.
Infrared counters have been shown to
consistently undercount pedestrians. Pedestrians
that walk side- by- side are generally counted as one
pedestrian. Undercounts range from 5 to 30
percent, but are generally consistent at a location
( Greene- Roesel et al. 2007). To calibrate the
infrared counters for the Seamless Travel Project,
the researchers compared manual counts to
automated counts to establish a correction factor
for each site.
One automated count location ( Mission Beach Boardwalk) was discovered to have very high and variable
error rates in 2008. Extensive manual counts were conducted to determine the cause for this, and to
develop an accurate correction factor. It was determined that the width of the Boardwalk ( 22 feet)
combined with extremely high volumes ( for example, 3,135 people were counted in one 2 hour period)
resulted in error rates as high as 70%. The infrared counters were unable to distinguish between so
many people walking/ riding side by side when they passed the counter.
Count locations for the year- long automated counts were more restricted than the peak- hour manual
counts. Due to the count technology chosen, only off- street areas could be used. Infrared counters
cannot easily be used to monitor on- street bikeways, as vehicles will trip the sensor. It was determined
that using a pneumatic tube counter for on- street bikeways could pose safety concerns, and might be
affected by buses and vehicles rolling over the tube.
Year- long automated counts were conducted at five sites. These sites were chosen to reflect a variety of
recreational, commuter, bicycle and pedestrian traffic. A map of count locations is shown in Figure 5.
Information collected from the year- long automated counts was used to evaluate hourly, daily, monthly
and seasonal trends in biking and walking.
Equipment Technology
The research team reviewed published literature on counting non- motorized travel and conducted
internet searches to determine the most suitable technology available for this project. Key criteria
guiding this review included equipment cost, ease of installment, and potential for differentiating
pedestrian and bicycle modes. Table 8 presents an overview of automatic count technology.
Pedestrians walking side- by- side can create inaccuracies in
automatic counters
Seamless Travel February 2010 40
Caltrans Task Order 6117
Table 8: Automatic County Technology Overview
Technology How it
Works
Differentiate
between
bikes and
peds?
Where can it
be used?
Can it be
moved to
other
locations?
Other
Considerations
Techn
ology
Cost
Passive
infrared
Detects a
change in
thermal
contrast
No Sidewalk,
path
Easily $, 2000-
3,000
Active
infrared
Detects an
obstruction
in the
beam
Yes Sidewalk,
path
Easily $ 800-
$ 7,000
Video
imaging
Analyzes
pixel
changes
Unknown Intended for
indoor use
Yes Difficult
detection
outdoors, no
bike/ ped
application yet
$ 1,200-
$ 8,000
Video
playback
Video
analyzed by
a person
Yes Anywhere Yes Difficult
detection at
night and bad
weather.
Considerable
staff time
$ 7,000
Piezometric
Tube
Senses
pressure on
tube
No Path, on-street
Easily Bicycles only.
Potential
tripping hazard
$ 1,600
Piezometric
Pad
Senses
pressure
No Sidewalk,
path
No $ 2,000-
3,000
In-pavement
magnetic
loop
detectors
Senses
magnetic
field
change as
metal
passes
No Path, on-street
No Requires
cutting into
pavement to
install
$ 2,000-
3,000
Based on review in 2007, two types of count equipment technology were purchased: an active infrared
counter manufactured by TrailMaster and a passive infrared counter manufactured by JAMAR. The
active infrared equipment includes a transmitter that emits an infrared pulsing beam and a receiver,
which detects the beam. When the infrared beam is broken by a walker or bicyclist, the receiver counts
an event. The infrared beam’s pulse rate is adjustable, and allows for the TrailMaster to be sensitive to
the length of time required for an object to break the infrared beam. A benefit of this technology is that
two TrailMasters can be installed in the field at one location, and then each set differently, one to record
Seamless Travel February 2010 41
Caltrans Task Order 6117
all events and the other to record only pedestrians14. This allows for an estimation of mode share along
a path. The TrailMasters were installed inside small electric boxes and attached to poles or trees near the
respective pathways or walkways.
The JAMAR Scanner employs passive infrared technology whereby a single piece of equipment emits a
beam that is broken by a heated object passing through it, such as a human or an automobile. Therefore,
when a walker or bikers passes through the beam, the equipment detects the heat and counts an event.
This technology cannot distinguish mode or speed, but can detect the direction of the traveler.
Figure 5: Yearly Count Locations in San Diego County
Count Site Locations
Five locations within San Diego County were selected as sites for conducting continuous, year- long 24-
hour counts ( Figure 5). The site selection was based upon the need to collect data from a mix of urban
environments and facility types, and to capture differences in commute versus recreational trip making.
A local signage company, Kitt Signs, was hired to retrofit off- the- shelf electric boxes to hold the
TrailMasters, as well as to install all of the equipment in the field. The JAMAR Scanners were not fitted
into electric boxes, as they are encased in heavy, weatherproof plastic casing.
Each of the sites and justification for selection are summarized below:
14 All fast- moving trail users, such as skateboarders, are recorded as bicyclists.
Seamless Travel February 2010 42
Caltrans Task Order 6117
Gilman Drive / Rose Canyon Bike Path: This site is located in the City of San Diego along a
relatively long and well- utilized bicycle path that connects coastal residential areas to significant
concentrations of high- tech, university- related, and retail/ service employment. The site was expected to
be dominated by bicycle trip- making with a strong emphasis on commuting. Two TrailMasters were
installed at the site to capture differences in pedestrian and bicycle mode shares.
Bayside Walk @ San Juan Place and Bayside Walk @ Ormund Place: This site is located in the
City of San Diego’s Mission Beach community along Mission Bay. The pathway is part of a relatively
long facility that goes around Mission Bay’s entire eastern bay. The location tends to have heavy
recreational usage by both bicyclists and walkers/ joggers, but is also utilized by residents for shopping
trips and to obtain other services in nearby Pacific Beach. Two TrailMasters were installed at adjacent
locations along the Bayside Walk to capture differences in pedestrian and bicycle mode shares.
The Boardwalk @ San Juan: This site is located in the City of San Diego’s Mission Beach community
along the Boardwalk. The pathway is part of a long beach area pathway system that runs adjacent to the
ocean and connects with other pathways around Mission Bay. The location tends to be heavily utilized
for recreational travel. A JAMAR Scanner was installed at the site. The site was selected in order to
capture the upper extent of pedestrian and bicycle demand in San Diego, as this is one of the most
heavily traveled non- motorized pathways in San Diego.
Bayshore Bikeway @ Avenida de las Arenas: This site is located in the City of Coronado along the
Bayshore Bikeway ( The Strand). Two TrailMasters were installed at this location. The pathway is part of
a relatively long facility that goes around San Diego Bay. The location serves recreational bicyclists and
was selected in part because it was recently completed.
University Avenue between 4th and 5th Avenues: This site is located in the City of San Diego within a
relatively older, pedestrian- oriented neighborhood with mixed land uses and high residential densities. A
JAMAR Scanner was installed at this location. The location was selected to represent urban pedestrian
travel where high levels of multi- purpose walking trips are made.
Figure 5 provides a citywide overview of the count locations, while Figure 6 and Figure 7 present a
more detailed view of counts locations and equipment installation at the respective sites.
Seamless Travel February 2010 43
Caltrans Task Order 6117
Figure 6: Rose Canyon Bike Path, Mission Beach Boardwalk and Bayside Year- Long Automated
Count Locations
Rose Canyon Bike Path
Moderately high activity, bike commuters/
recreational walkers and bikers
Collected mode split information
Mission Beach ( Boardwalk)
High activity area, mainly recreational, did not
collect mode split information
Mission Beach ( Bayside Boardwalk) ( not
shown)
Moderately high activity, mainly recreational,
collected mode split information
Figure 7: University Avenue and Bayshore Bikeway Year- Long Automated Count Locations
University Avenue ( sidewalk)
High pedestrian activity area, mainly utilitarian
urban travel, did not collect mode split
information
Bayshore Bikeway, Coronado
Moderate activity levels, mainly recreational
walkers and bikers, collected mode split
information
Seamless Travel February 2010 44
Caltrans Task Order 6117
Validation Methods
The research team verified the accuracy of the 24- hour counting equipment by conducting manual
counts while the machines were counting, and then comparing the manual count data to the machine
count data. The first validation count revealed several types of installation problems. For example, at
the University Avenue site, the Scanner was initially located too close to a business entrance and was
found to be counting inaccurately due to people entering and exiting the business. The Scanner was
shifted away from the business door and found to count with increased accuracy. The angle at which the
infrared beam is directed across a facility also proved to be an important factor in the count accuracy.
Several of the counting machines had to be shifted to transmit at a 45- degree angle across the facility in
order to record people traveling side- by- side. This adjustment improved the accuracy of the machine
count.
Validation Results
This section summarizes results of the validation analysis by equipment type, first discussing validation
analysis results for the passive infrared equipment installed along The Boardwalk and University Avenue,
and then discussing validation analysis results for the active infrared equipment installed along the
Bayside Walk, the Bayshore Bikeway, and the Rose Canyon Bike Path.
Passive Infrared Counters
Table 9 presents results of the validation efforts at the two sites where passive infrared technology was
installed – The Boardwalk and University Avenue.
Table 9: Passive Infrared Validation Counts JAMAR Scanner
Location
First Validation Second Validation
Date &
Time of
First
Validation
Count
Total
Manual
Count
Total
Machine
Count
Percent
Diff.
Adjustment
Date of
Second
Validation
Count
Total
Manual
Count
Total
Machine
Count
Percent
Diff.
The
Boardwalk
7/ 13/ 07
( 2: 45 PM to
4: 00 PM)
580 400 - 31.0%
Reposition
at 45° angle
across
facility.
7/ 17/ 07
( 12: 30 PM
to 1: 30
PM)
427 337 - 21.1%
University
Avenue
7/ 13/ 07
( 8: 00 AM to
9: 15 AM)
62 58 - 6.5%
Reposition
away from
business
entrance.
7/ 23/ 07
( 9: 15 AM
to 10: 15
AM)
20 17 - 15.0%
Source: ( Alta Planning + Design, November 2007)
The Boardwalk Site: The JAMAR Scanner was initially mounted on a sign post facing west along the
north/ south running Boardwalk, with the infrared beam aimed directly across the pathway. The first
validation count was conducted on July 13, 2007, between 2: 45 PM and 4: 00 PM. The JAMAR Scanner
was found to be undercounting by approximately 31%. The Scanner was then repositioned to face
north- west, at a 45 degree angle across the pathway, in the hopes that the equipment would be more
sensitive to people walking next to each other. A second validation count was conducted on July 17,
2007 between 12: 30 PM and 1: 30 PM. The counts revealed that the machine position adjustment
improved the machine’s count to within approximately 21% of the manual count.
Seamless Travel February 2010 45
Caltrans Task Order 6117
University Avenue Site: The JAMAR Scanner was mounted on a street light pole facing north along
the east/ west running University Avenue, with the infrared beam aimed directly across the pathway. The
first validation count was done July 13, 2007 between 8: 00 AM and 9: 15 AM. This validation count
showed that the machine was counting within 6.5% of the manual count, however, Alta staff noticed
that the beam was aimed almost directly at a business storefront and that every time someone entered or
exited the store, the Scanner recorded an event. The Scanner was repositioned to face north- west, at a
45 degree angle across the sidewalk and away from the store entrance. The second validation count was
done on July 23, 2007 between 9: 15 AM and 10: 15 AM. The Scanner was then found to be counting
within 15% of the manual count.
Active Infrared Counters
Table 10 summarizes the validation analysis results for the active infrared counting machines installed at
the Bayshore Bikeway, the Bayside Walk, and the Rose Canyon Bike Path.
Rose Canyon Bike Path Site: The Rose Canyon Bike Path validation count was conducted June 6,
2007 between 3: 30 PM and 5: 45 PM. The north set of boxes ( one transmitter and one receiver) was set
to capture an event for objects moving at any speed, and the south set of boxes was set to capture events
for objects moving at the speed of a pedestrian. Both sets of boxes broadcast infrared beams directly
across the path. The machines set to capture all travelers undercounted by about 12%, while the
machines set to count pedestrians undercounted by about 25%.
Bayshore Bikeway: The Bayshore Bikeway validation count was conducted July 9, 2007 between 10: 15
AM and 12: 15 PM. The north set of boxes was set to capture events for objects moving at any speed,
while the south set of boxes was set to capture objects moving at a pedestrian’s typical speed. The two
sets of equipment were initially set so that their beams traversed the path at a 90 degree angle. The first
validation count showed that the south set of boxes was undercounting by approximately 92% and the
north boxes were undercounting by about 22%.
The southern boxes were repositioned to direct the beam at a 45 degree angle across the path. The
northern set of boxes was realigned to ensure proper readings. A second validation count was done on
July 13, 2007 between 10: 15 AM and 11: 30 AM, and showed undercounting by about 36% at the
southern location and by about 12% at the northern location.
The pulse rate setting was then adjusted at the southern location, along with finding a new location that
allowed for positioning the receiver and transmitter closer together. A third validation count was
conducted on July 16, 2007 between 9: 00 AM and 10: 15 AM at the southern location, and found that the
machine count was within about 8% of the manual count.
Seamless Travel February 2010 46
Caltrans Task Order 6117
Table 10: Active Infrared Validation Counts
TrailMaster
Location
First Validation Second or Third Validation
Date &
Time of
First
Validation
Count
Total
Manual
Count
Total
Machine
Count
Percent
Difference
Adjust-ment
Date &
Time of
Second or
Third
Validation
Count
Total Manual
Count
Total Machine
Count
Percent
Diff.
All Ped All Ped All Peds All Ped All Ped All Ped
Gilman
Drive/ Rose
Canyon Bike
Path
6/ 6/ 07
( 3: 30 PM to
5: 45 PM)
75 4 66 3 - 12.0 - 25.0 -- -- -- -- -- -- -- --
Bayshore
Bikeway @
Avenida de las
Arenas
7/ 9/ 07
( 10: 15 AM
to 12: 15
PM)
67 13 52 1 - 22.4 - 92.3
Reposition
at 45°
across
facility.
( 2nd
Validation
Count)
7/ 16/ 07
( 9: 15 AM
to 10: 15
AM)
80 11 70 15 - 12.5 36.4
-- -- -- -- -- -- --
Changed
Infrared
Beam
Pulse Rate
( 3nd
Validation
Count)
7/ 16/ 07
( 10: 30 AM
to 11: 30
AM)
12 11 - 8.3
Bayside Walk
@ Ormund
Place and @
San Juan
6/ 9/ 07
( 12: 30 PM
to 2: 30 PM)
444 101 366 21 - 17.6 - 79.2
Changed
Infrared
Beam
Pulse Rate
7/ 10/ 07
( 4 PM to 6
PM)
89 46 - 48.3
Source: ( Alta Planning + Design, November 2007)
Seamless Travel February 2010
Caltrans Task Order 6117
47
Bayside Walk Site: Two TrailMasters were
installed along the Bayside Walk site, with the
northern machine set to record events for objects
moving at any speed, and the southern machine
set to record events caused by objects moving at
the speed of a pedestrian. The validation counts
were conducted on June 9, 2007 between 12: 30
PM and 2: 30 PM, and showed that the northern
machine was counting within 17.6% of the
manual count, and the southern machine was
undercounting by about 75.3%.
Alta staff noticed that at this particular site
walkers were moving along at relatively high
speeds, and that it was unlikely that the machine
was recording these fast walkers. The pulse rate
of the southern machine was therefore reset in an effort to capture slightly higher speed walkers. Alta
staff also noticed a high presence of grouped walkers. Unfortunately, installation opportunities at this
location are limited, and the transmitter cannot be rotated to direct the infrared beam across the pathway
at a 45 degree angle. A second validation analysis was conducted on July 10, 2007 between 4 PM and 6
PM, showing that the southern machine was still undercounting by approximately 48.3%. Pedestrians
walking side by side continue to be an issue for the southern TrailMaster at this location.
Summary of Observations
The JAMAR Scanners are undercounting by approximately 15% to 21%. The machine at the higher
volume location, The Boardwalk, shows less accurate counts than the machine at the lower volume
location along University Avenue.
The TrailMasters are undercounting all travelers by approximately 12% to 18%. Again, machines at the
lower volume locations, the Rose Canyon Bike Path and the Bayshore Bikeway, are providing more
accurate count data than the machines at the higher volume locations along Bayside Walk in Mission
Beach.
The TrailMasters are undercounting pedestrians by approximately 25% to 48%, displaying a similar
inverse relationship between count accuracy and traffic volume. It should be noted that limitations in
installation opportunities at the Bayside Walk and San Juan Place in Mission Beach, which prohibit
directing the infrared beam at a 45 degree angle across the pathway, are resulting in the most inaccurate
machine counting of all study locations.
The TrailMasters appear to be slightly more accurate than the JAMAR Scanner in counting all travelers,
however the TrailMaster requires identification of count locations where equipment can be installed on
both sides of the pathway, while the JAMAR Scanner can be effectively installed in locations with
poles/ street lights on just one side of the pathway or sidewalk. In other words, the Scanner allows for
effective counting in urban environments, while the TrailMaster is more limited to counting along
pathways or trails, where trees or poles can be found along both sides of the facility.
A bicyclist at the Bayside Walk Site, at Santa Clara St.
Seamless Travel February 2010
Caltrans Task Order 6117
48
MANUAL COUNTS AND SURVEYS
Manual peak period counts were conducted at eighty ( 80) intersection locations across San Diego County
during the months of July and August 2007. Graduate students from San Diego State University were
hired and received training to conduct counts and collect survey information. Counters were instructed
to record a pedestrian or bicyclist at the intersection leg where the traveler approached the intersection.
Peak period counts were conducted at eighty intersections during a weekday ( Tuesday, Wednesday, or
Thursday) morning peak period ( 7 AM to 9 AM) and a weekend ( Saturday or Sunday) midday peak
period ( 12 PM to 2 PM). In addition, evening peak period counts ( 4 PM to 6 PM) were collected at a
sample of twenty intersections, which were selected to represent a geographic distribution of study
intersections.
Survey Methodology
In addition to conducting counts, the Seamless Travel Project collected surveys from user intercepts at
thirty- five of eighty peak- period count locations. The following sections describe survey pre- testing and
pilot testing, survey administration, and special modifications to the bicycle intercept survey approach.
Survey Pre- Testing and Pilot Testing
The surveys were pre- tested and pilot tested in the field to determine how easy it was for people to
understand and give answers. A pre- test was conducted on 14 individuals in Pacific Beach on June 15,
2007. The pre- test participants were asked to provide feedback on question wording, sentence structure
and overall input to make the survey more easily understood. As a result of pre- testing efforts, the
following changes were made:
 Added the Gym/ Recreation as a destination choice for Question 6
 Added “ Never” box as an option for Question 7
 Added “ Never” box as an option for Question 8
 Added “ Never” box as an option for Question 9, and
 Made minor grammatical corrections.
After pre- testing the survey, pilot tests were administered at the Rose Canyon Bike Path on June 21,
2007 between 5 PM and 6 PM. A total of 12 pilot test surveys were administered ( 8 bicycle and 4
pedestrian). The subjects took the surveys and had no issues with the phrasing or meaning of any
questions.
Survey Administration
Alta staff administered bicycle and pedestrian intercept surveys with the assistance of temporary
employees hired to expedite survey collection. Prior to administering surveys, Alta staff completed the
Collaborative Institutional Training Initiative training to conduct research involving human subjects.
One staff research assistant debriefed and trained the remaining surveyors in the field. On- site trainings
accentuated sensitivity to vulnerable populations, including exclusion of child subjects. On- site trainings
Seamless Travel February 2010
Caltrans Task Order 6117
49
also emphasized obtaining verbal consent from participants, acknowledging participants’ anonymity, and
their right to terminate participation at any time. Alta equipped temporary employees with written
material to orient them to the purpose and scope of the study, as well as an adaptable script for
recruiting participants.
Thirty- five of eighty study sites were selected to capture a variety of land use and population
characteristics. Multiple surveyors were fluent in Spanish enabling administration of the survey in largely
Hispanic communities.
Generally, surveyors were organized