Access to Destinations: How Close is Close Enough?
Existing urban and suburban development patterns and the subsequent automobile dependence that is associated with them are leading to increased traffic congestion and air pollution. In response to the growing ills caused by urban sprawl, there has been an increased interest in creating more “livable” communities in which destinations are brought closer to one’s home or workplace (that is, achieving travel needs through land use planning). While several reports suggest best practices for integrated land use-planning, little research has focused on examining detailed relationships between actual travel behavior and mean distance to various services. For example, how far will pedestrians travel to access different types of destinations? How can we know if the “one quarter mile assumption” that has become conventional wisdom in planning and designing communities is reliable? How far will bicyclists travel to cycle on a bicycle only facility? How far do people drive for their common retail needs? This research provides evidence on these and other closely related questions.
The approach taken in this research is to estimate sets of distance-decay functions to describe the impedance of travel distance or time across the transportation network. The concept of distance-decay, used widely in geography and spatial interaction modeling, including many transportation forecasting models, can be interpreted as measuring either the impedance to travel through a network or the willingness of individuals to travel various distances to access opportunities. An important feature of this work is the explicit effort to extend the set of distance-decay functions to as many combinations of modes and trip purposes as available data permit. The set of modes that are considered of interest in this study are auto (single-occupant and shared ride), public transit, bicycling and walking. The data sets used in the analysis include a general-purpose travel survey for the Twin Cities region, an on-board survey of users of the regional public transit system, and a survey of users of joint-use trail facilities in Hennepin County. Together, these data sets allow for a fairly comprehensive look at travel behavior by various modes and trip purposes.
The findings of the research, particularly as they relate to non-motorized modes (walking and bicycling), provide evidence that can supplement existing rules-of-thumb for pedestrian and bicyclist behavior. The findings suggest, for example, that substantial shares of pedestrian travel (perhaps one-quarter to one-third) exceed the often-cited threshold of one-quarter mile. Moreover, this finding appears to be invariant to trip purpose. Unless the segment of the population who reported these pedestrian trips are substantially different from those who either did not make utilitarian trips by the pedestrian mode or did not think to report them, this may be a welcome finding for pedestrian planning as it indicates a greater willingness to walk than is generally thought to be the case.
Results for bicycle travel data reveal a substantial difference in travel distances by trip purpose. Primary activities such as work and school often involved very long bike trips (up to 20-30 KM) for those who chose this mode. In contrast, more discretionary trips (e.g. for shopping, entertainment or recreation purposes) tended to be substantially shorter in length. It would be desirable for future studies of these types of behavior to target bicycling specifically in order to provide a large enough sample to further substantiate these findings.
The public transit trips examined in this study reveal significant differences in travel behavior across several types of stratification. Type of service (local bus, express bus, light rail), access mode and trip purpose all appear to affect trip length. The origin-destination data examined in this study allowed for a disaggregation of trips by segment (access, egress, and line-haul), which presents opportunities to ask further questions about how travelers view the relative importance of different parts of their trip (e.g. access).
Lastly, an important purpose of this research is to demonstrate how the decay curves for different travel modes and trip purposes can be used to provide measures of accessibility. While the calculation of accessibility measures for auto and transit modes are relatively straightforward using conventional travel demand models, we show how the use of a more disaggregate zonal structure using Census geography can allow for the generation of non-motorized accessibility measures at a spatial scale more closely aligned with actual bicycle and pedestrian travel behavior. While the results reported here are restricted to a relatively small sample area, the techniques should be scalable to larger geographic areas given full spatial data sets and appropriate computational hardware.