A modification of the time-geographic framework to support temporal flexibility in ‘fixed’ activities

Abstract

The time-geographic distinction between fixed and flexible activities is widely acknowledged to be a somewhat arbitrary dichotomy but is still the current modus operandi for time-geographic calculations. This paper proposes a modification of the classic time-geographic framework to support temporal flexibility in ‘fixed’ activities. This modification is crucial for time-geographic calculations on public transit networks with a low frequency of service, which would otherwise return unstable results.

Keywords:

• time-geography
• mobility
• geocomputation

Acknowledgement

The author would like to thank the three anonymous reviewers for their suggestions towards improving this paper, as well as Melinda Morang, from ESRI, for her useful support with scripting issues.

Notes

1. 27.5% of full-time workers in the United States in 2004 (United States Department of Labor – Bureau of Labor Statistics 2014).

2. To determine what is a low or a high frequency of service, one may go with this description by Jansson (1993, p. 33), ‘for low frequencies, and given the availability of a reliable timetable, people prefer to plan their trips according to the table. For high frequencies they prefer to go to the stop or station spontaneously, rather than consult the time table’. Indeed, when the scheduled interval between two vehicles becomes smaller, the stochastic variation of the actual vehicle arrival time around the scheduled time represents a greater portion of this interval. To the point that arrival times appear as entirely stochastic to riders, who just show up, expecting to catch the bus immediately if they are lucky, to just miss one if they are unlucky, and to wait a specific time on average. Accordingly, Charleux (2014) performs time-geographic calculations with travel times on a high-frequency public transit network, incorporating average waiting times or maximum waiting times at stops. This strategy is not satisfying however to model travel times on low-frequency transit networks.