Spatial distribution of park-and-ride demand
Date
2014
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
Park-and-ride facilities have been effective hubs that connect private transportation options and public transit systems such as buses and trains. Because of their long construction and service time, accurate prediction about their usages is of the utmost importance because it is not only crucial for location and capacity decisions, but will also influence public images of transportation agencies if anything goes wrong. Reasonable description about park-and-ride demand spatial distribution is a critical part in its demand prediction. Previous researchers usually use catchment areas covering majority park-and-ride demand and assume uniform distribution within the catchment area. This leads to arbitrary choice of shape and size of catchment areas. It also does not allow for overlapped catchment areas which are very common in practice. This paper develops separate models for train-based and bus-based park-and-ride systems because of their different utilization patterns. More specifically, a combination of gravity model and mode choice model is proposed to describe train-based park-and- ride system, while Tobit model is used in bus-based system. Both models can be viewed as generalizations of catchment areas - different parameters will generate different sizes and shapes of catchment areas. They also allow for overlaps of different catchment areas. Empirical tests in Delaware show these two models have very good in-sample predictions as well.