A travel demand modeling framework based on OpenStreetMap

Abstract

Demand modeling is an important part of the setup of a traffic model for a city. All travel demand models rely on land use data as the demand for traveling fundamentally stems from activities occurring at different locations; however, many cities lack these data, or experience in estimating travel demand in their region. In response, this study develops a methodology for generating highly detailed land use data in the form of points of interest (POIs) specifically aimed at travel demand estimation purposes. The framework includes a procedure to extract, clean, enhance, and categorize freely available land use data from OpenStreetMap (OSM) into different POI categories, such as residences, schools, and shops. These residential and activity POIs, which are typical origins and/or destinations of trips, serve as the starting point for estimating travel demand. This paper demonstrates the framework’s utility through three case studies across different cities in Belgium. It validates the effectiveness of OSM-derived POIs for travel demand estimation by replicating Antwerp’s existing demand model, examines the POIs classification’s suitability for various travel demand purposes in Leuven, and assesses the transferability of correlations between OSM data and travel demand from Antwerp to Ghent. Beyond the applications illustrated in this paper, the framework provides opportunities for future research on the consistent disaggregation of existing zonal demand estimates and design-based research in which future demand is estimated given the development of POIs. The framework is openly available as a Python tool called Poidpy.