TRCLC 16-10

Transportation Research Center for Livable Communities

A Constraint-Based Bicycle Origin-Destination Estimation Procedure

PIs: Ziqi Song (PI), Utah State University

Project Start and End Dates: September 1, 2016 to August 31, 2018

Summary:

This project proposes a bicycle O-D estimation procedure that is flexible and can be adjusted to different levels of data availability and quality. Bicycle data that are useful for bicycle O-D estimation are first explored. A constraint-based bicycle O-D estimation procedure that utilizes bicycle data from multiple sources is then proposed. A case study is also conducted to demonstrate the proposed methodology. The results demonstrate that in practice, the proposed bicycle O-D estimation procedure is a promising tool. The findings in this project, if could be further verified with more extensive data sets, have importation implications in bicycle facility assessment. Crowdsourcing data is demonstrated to be a useful data source in bicycle modeling and planning. The combination of traditional bicycle count data and emerging crowdsourcing data provides many �鶹��ý and challenges for future studies.

Problem:

Cycling is an active, green transportation mode that improves environmental sustainability and the livability of urban communities. Promoting cycling also has significant public health benefits. Despite the increasing importance of cycling as a transportation mode, it is often ignored in traditional transportation planning procedures. Origin-destination (O-D) matrix estimation methods have focused on estimating O-D matrices from link traffic counts for motorized vehicles, which are regularly collected by transportation agencies for traffic monitoring purposes. However, traffic monitoring of non-motorized traffic is not as comprehensive as motorized traffic monitoring in the United States. Hence, O-D matrix estimation methods developed for motorized traffic cannot be directly used to estimate bicycle O-D trip matrices.

Research Results:

This study has explored bicycle data that are useful for bicycle O-D estimation. There are multiple sources for bicycle data, including traditional bicycle count data, emerging crowdsourcing bicycle data, and data from bike-sharing programs. This study has introduced each of these bicycle data sources and discussed their applicability in bicycle O-D trip estimation.
This study has proposed a constraint-based bicycle O-D estimation procedure that consists of two major stages. The first stage generates an efficient bicycle route set that contains a set of Pareto optimal (non-dominated) routes. The second stage is a bicycle PFE model that is based on the path-size logit (PSL) route choice model. The procedure is flexible and can be adjusted to different levels of data availability and quality.
Based on the proposed methodology, this study has conducted a real-world case study in Salt Lake City, Utah. The results show that the proposed O-D estimation procedure can effectively utilize the bicycle count data and crowdsourcing data, i.e., the Strava data, to estimate bicycle O-D matrix. The Strava data have the potential to serve as an economical complement for official count programs.

Presentation

Final Report