Local Station Catchments: Reconciling Theory with Reality
S Blainey, University of Southampton, UK; S Evens, Department for Transport, UK
UK National Rail Travel Survey data is used to investigate rail station catchments. Access distance, access mode and station choice were analysed and station choice models calibrated to enhance demand forecasts and assess inter-station abstraction.
In order to accurately forecast demand from new local railway stations it is necessary to define the catchment areas which such stations will serve. Most existing forecasting methodologies tend to use relatively simplistic catchment definition methods, usually based on the all-or-nothing allocation of population units to stations and constant distance decay functions. Catchment boundaries in such cases are usually chosen either by approximating observed results from small-scale surveys or by maximising demand model fit across a calibration case study, but there is no guarantee that such methods reflect actual travel behaviour. However, the UK National Rail Travel Survey (NRTS) provides an extremely large dataset on ultimate trip origins/destinations and station access/egress, which should permit a comprehensive analysis of actual station catchments and therefore allow a generalised procedure for estimating theoretical station catchments to be developed. This paper describes such an analysis, carried out for the North-East and Wales regions of the UK.
Firstly, NRTS data was used to assess the accuracy of various theoretical catchment definition methods, including crow-fly distance, road distance and road journey time buffer zones, with both overlapping and discrete boundaries. The results were disaggregated by Network Rail station category to allow variations between station types to be investigated. Extensive descriptive and exploratory statistical analysis of the NRTS data was then carried out. Inter-station variability was assessed, to establish the likelihood of a single generalised catchment boundary being suitable for all stations within a particular category. The relationship between access/egress mode and distance was then explored, along with the relative popularity of different access modes. One of the key aims of the analysis was to investigate station choice decisions, and the station chosen in each case was therefore ranked by access/egress distance from the trip origin/destination. These results were then disaggregated by station category, access distance (to both the nearest station and the station used), the proportion of total journey distance made up by access/egress, and station density.
The results from the descriptive analysis were used to inform the development and calibration of station choice models capable of providing catchment population inputs for rail demand models. A range of models at disaggregate and aggregate levels were calibrated for the same case study dataset, including ordinal regression, multinomial logistic regression and intervening opportunity models. In addition to their application in demand forecasting, the use of the models to assess abstraction by new stations was also investigated, by running the station choice models in the "before" and "after" situations and comparing the difference in the results. The outputs from the study can inform and enhance PDFH guidance on catchment definition for demand modelling, which is currently extremely basic, as well as forming the basis for guidance on forecasting access mode choice and station choice.
Association for European Transport