BICYCLE ROUTE CHOICE Gps Data Collection And Travel Model .

BICYCLE ROUTE CHOICEgps data collection and travel modeldevelopment—year 1 (2012–13)FINAL PROJECT REPORTbyQing Shen, P.I.*; Peng Chen*; Peter Schmiedeskamp*; Alon Bassok*; Suzanne Childress†*University of Washington†Puget Sound Regional CouncilforPacific Northwest Transportation Consortium (PacTrans)USDOT University Transportation Center for Federal Region 10University of WashingtonMore Hall 112, Box 352700Seattle, WA 98195-2700

DisclaimerThe contents of this report reflect the views of the authors, who are responsible for the factsand the accuracy of the information presented herein. This document is disseminated underthe sponsorship of the U.S. Department of Transportation’s University TransportationCenters Program, in the interest of information exchange. The Pacific NorthwestTransportation Consortium and the U.S. Government assumes no liability for the contentsor use thereof.ii

Technical Report Documentation Page1. Report No.2012-S-UW-00192. Government Accession No.015381094. Title and SubtitleBicycle Route Choice: GPS Data Collection and Travel Model Development3. Recipient’s Catalog No.5. Report DateSeptember 15, 20146. Performing Organization Code7. Author(s)Qing Shen, Peng Chen, Peter Schmiedeskamp, Alon Bassok, Suzanne Childress9. Performing Organization Name and AddressPacTransPacific Northwest Transportation ConsortiumUniversity Transportation Center for Region 10University of Washington More Hall 112 Seattle,WA 98195-27008. Performing Organization Report No.19-62508310. Work Unit No. (TRAIS)Department of Civil and EnvironmentalEngineeringUniversity of Washington201 More HallSeattle, WA 9810511. Contract or Grant No.DTRT12-UTC10Puget Sound Regional Council101 Western AvenueSuite 500Seattle, WA 98104(List matching agency name and address)12. Sponsoring Organization Name and AddressUnited States of AmericaDepartment of TransportationResearch and Innovative Technology Administration13. Type of Report and Period CoveredResearch 9/1/2012-7/31/201414. Sponsoring Agency Code15. Supplementary NotesReport uploaded at www.pacTrans.org16. AbstractBicycle use is being promoted for a variety of social benefits. Because of the benefits associated with bicycling, jurisdictions across the centralPuget Sound region and the nation have been investing in improvements to bicycle infrastructure. Academic and professional literature providesa basis for generally understanding bicycling behavior. However, less is known about the benefits of one facility type over another, or thepotential inducement of new bicycle users when a policy intervention improves bicycling conditions.This study will rely on GPS bicycle trace data collected by the Puget Sound Regional Council through the CycleTrack mobile application. Theobjectives of the study include improving the Puget Sound Regional Council’s travel demand model to include bicycle route choice andassignment, which will allow for policy analysis and an improved understanding of the tradeoffs between facilities, the relationship betweenutilitarian and recreational bicycling, and an analysis of the utility of a number of bicycle facilities that will become operational over the courseof the study.17. Key Words18. Distribution StatementNo restrictions.Bicycle, data, GPS19. Security Classification (of thisreport)Unclassified.20. Security Classification (of thispage)Unclassified.Form DOT F 1700.7 (8-72)21. No. of Pages22. PriceNAReproduction of completed page authorizediii

ContentsExecutive Summary / Introduction1 Literature Review1.1 Weather . . . . . .1.2 Household size . .1.3 Vehicle availability1.4 Trip end amenities .1.5 Attitude . . . . . .1.6 Distance . . . . . .1.7 Slope . . . . . . . .1.8 Transit . . . . . . .1.9 Air quality . . . . .1.10 Bicycle facilities . .1.11 Pavement quality .1.12 Sidewalks . . . . .1.13 Turns . . . . . . . .1.14 Traffic signals . . .1.15 Traffic volume . . .1.16 Traffic speed . . . .1.17 Parking . . . . . .1.18 Summary . . . . .ix.11122234455667778892 Study Site / Data112.1 Study Site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 Methods3.1 Sampling and Recruitment . . . . . . . . . . . . . . . .3.2 Data Collection . . . . . . . . . . . . . . . . . . . . . .3.3 CycleTracks Data Cleaning and Bad Trace Identification3.4 Choice Set Generation . . . . . . . . . . . . . . . . . .15151518264 Results315 Discussion356 Conclusions and Recommendations37iv

Bibliography39A CycleTracks Survey Tool Interface45B Geospatial Data Extract Transform and Load Procedures49C Identifying GPS traces on a roadway network55D Identification of Bad CycleTracks GPS Traces (Spatialite)57E Example Spatial Intersection Operation in SQL (Spatialite)61v

List of Figures2.1Map of the Central Puget Sound Region (Puget Sound Regional Council 2014a) . . . . 123.13.23.33.43.53.6A clean GPS trace . . . . . . . . . . . . . . . . .A nearly clean GPS trace paralleling a facility . .Urban canyon effect in an otherwise clean trace .GPS stopped for part of a trace . . . . . . . . . .Apparent “teleportation” where GPS loses its fix .Trip logged with negligible distance traveled . . .4.14.2Population Pyramids for Central Puget Sound Region vs. CycleTracks Sample . . . . . 32Breakdown of trips by purpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32.A.1 Personal information and trip recording screens from IOS CycleTracks Application .A.2 Lock screen, trip purpose, and trace screens from IOS CycleTracks Application . . .A.3 Main screen, personal information, and trip recording screens from Android CycleTracks Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .A.4 Trip purpose and completed trace screens from Android CycleTracks Application . .vi.202122232425. 45. 46. 47. 48

List of Tables1.1Variables affecting bicycle route choice and their anticipated impacts . . . . . . . . . .2.12.22.32.4Population of Central Puget Sound Region (Puget Sound Regional Council 2014b) .Trip-level variables and associated data types . . . . . . . . . . . . . . . . . . . .Person-level variables and associated data types . . . . . . . . . . . . . . . . . . .Supplementary geospatial data layers . . . . . . . . . . . . . . . . . . . . . . . . .3.1Person-level variables collected by CycleTracks . . . . . . . . . . . . . . . . . . . . . 17vii.911131314

Executive Summary / IntroductionBicycle use is being promoted for a variety of social benefits. Persons who choose to bicycle ona regular basis for commute or other purposes receive benefits in the form of lower obesity ratesand other health improvements, and in lower transportation costs (Dill 2009; Frank et al. 2006;Sallis et al. 2004). In addition, society as a whole benefits from an increase in cycling. As peopleshift modes from automobiles to bicycles, additional capacity is available on the roadways relievingcongestion. Further, as those who choose to bicycle become healthier, they present less of a burdenon the health care system, reducing overall health care costs (Sturm 2002).Because of the benefits associated with bicycling, jurisdictions across the central Puget Soundregion and the nation have been investing in improvements to bicycle infrastructure. The improvements range from inexpensive tools such as painted sharrows on arterials and road rechannelizationprojects, to more costly greenways and separated off-road facilities. While investments in bicycleinfrastructure are far less costly than traditional roadway or transit improvements, jurisdictions mustdecide how to best invest limited resources.Academic and professional literature provides a basis for generally understanding bicyclingbehavior. However, less is known about the benefits of one facility type over another, or the potentialinducement of new bicycle users when a policy intervention improves bicycling conditions (Dilland Gliebe 2008; Krizek, El-Geneidy, and Thompson 2007; Tilahun, Levinson, and Krizek 2007).Furthermore, analytical tools aimed at assessing travel demand, benefit-cost analysis, and travelbehavior cannot be improved to capture the benefits of bicycling in the absence of 1) data on theuse of bicycling facilities, and 2) a theoretical framework for understanding cyclist route choicedecisions. This project begins to address these gaps by collecting GPS trace data for bicycle usersix

in the central Puget Sound, and utilizing it for policy analysis and travel model improvements.This project was successful in using a GPS smartphone application, CycleTracks, written by theSan Francisco County Transportation Authority to collect revealed preference GPS data representing 2,750 trips taken by 165 unique bicyclists. The collected sample has limitations that precludemaking robust, generalizable conclusions. However, in addition to validating the collection mechanism, this project has achieved a number of aims as it has worked toward developing a statisticalmodel.This project has contributed to knowledge transfer by reviewing the literature to identify factorsimportant to bicyclists and their decision making; identifying from the literature an appropriateapproach to choice set generation; and selecting / beginning implementation of a robust and repeatable approach to the data processing that more closely integrates with the IT infrastructure of thePuget Sound Regional Council.As noted above, the data collection process did yield a dataset with limitations, in particular withregards to the non-random sampling methodology, and the somewhat smaller than expected samplesize. The limitations posed by the sample could be addressed through a future data collection effortusing the CycleTracks application. A recommendation of this report is to revisit the sampling approach in the future—perhaps teaming with the organizers of a region-wide bike-to-work event forrecruitment and data collection. Finally, while the choice set generation proved more problematicthan initially anticipated, the analytical approach identified and the data processing strategies begunin this project could be completed to carry out the analysis of such a future data set.x

Chapter 1 Literature ReviewThis literature review considers the factors that affect a person’s choice to use a bicycle. The reviewbegins with a number of items that affect bicycle mode choice, but are not specific to route decisions.There variables are, for the most part, exogenous to route-choice modeling efforts and are alsogenerally outside the control of engineering and urban planning interventions. This literature reviewthen turns its attentions to built environment and other factors that can be used within route-choicemodeling and are heavily influenced by design interventions.1.1WeatherTemperature and precipitation play a large role in whether people choose to use bicycles. Whenit is cold or very hot outside less people will use bicycles. Similarly, as precipitation increases,less people will use bicycles. This relationship is far more pronounced for choice users rather thancommuters (Miranda-Moreno and Nosal 2011). That is, a rainy Saturday will see fewer riders compared to a sunny one as compared to a regular commute day. However, because there is generallylittle variation in temperature and rain along differing routes between an origin and a destination,weather has an affect on mode choice rather than route choice. And, for the purposes of modelling, it is unlikely that aggregate travel models can account for the fluctuations in temperatureand precipitation.1.2Household sizeThe smaller the household, the more likely the persons residing there are to use bicycling as a modeof travel (Andrade and Kagaya 2012). This finding makes intuitive sense. Smaller households,1