Calculating time-dependent accessibility levels to determine public transport gaps

One of the concerns that has aroused much scholarly attention in transport geography lately is the extent to which public transport provision enables the less privileged population segments, especially those without privately owned motorized vehicles, to participate in activities that are deemed normal within the society they live in. Various studies conducted in the domain of transport-related social exclusion have used geographical information systems (GIS) to unravel the linkage between social disadvantage, public transport needs and transit provision. However, much of the empirical work to date is static because it describes what is accessible by public transit from a particular origin at a single point in time yet does not consider the temporal variability in accessibility levels at multiple origins. This study contributes to this line of inquiry by proposing a methodology for identifying public transit gaps, a mismatch between the socially driven demand for transit and the supply provided by transit agencies. The methodology draws on the latest accomplishments in the field of modeling time-continuous, schedule-based public transport accessibility. General Transit Feed Specification (GTFS) datasets for bus, tram, metro and train as well as street network data were integrated in a GIS in order to compute a single multimodal network that accounts for all components of a public transport trip: ingress, the waiting time, the actual travel time through the transit network and egress. An automated Python script enabled to determine the temporal variability in accessibility through this network, which occurs as a consequence of fluctuations in operating frequencies across the diurnal cycle, and between weekdays and weekends. As a result, accessibility levels to key destinations are calculated at regular time intervals and synoptic metrics of these levels over various peak and off-peak time windows are computed for weekdays and weekends. As a result, a temporally reliable picture of accessibility by public transport is constructed. The obtained index of public transport provision is compared to a public transport needs index based on the spatial distribution of various socio-demographics, in order to determine public transport gaps as the spatial mismatches between these two indices. This method was tested for Flanders, which is the northern, Dutch-speaking region of Belgium, and proved to be effective in pinpointing areas that need considerable attention in public transport planning due to an overall high need for and underprovision of public transportation. Additionally, the high degree of time variability suggested the inadequacy of applying a single time of day for transit-based accessibility research.