Modelling Cycling Flow for the Estimation of Cycling Risk at a Meso Urban Spatial Level

Abstract

Abstract One of the prevailing challenges in cycling research, or indeed any vulnerable road user research, is the availability of data to ascertain a representative level of ‘exposure’ or simply how much cycling there is – “when and where”. Therefore, it is difficult for researchers and ultimately local authorities to determine if changes in observed accident trends over time are due to increased accident risk, (users or environment becomes more unsafe) or if they are a function of the higher numbers of cyclists using the existing roads and routes resulting in more incidents, i.e. increased exposure. This paper describes the use of recently developed open source transport modelling software and an open source bike routing application to assign realistic cycling flows to the network and validation against observed network link flows. The cyclist flows then provide the ‘exposure’ variable to examine cyclist safety performance at macro and meso levels using global and local models. The results highlight the need for local level mobility-based exposure metric to describe cyclist safety performance and the superior ability of local models to describe safety performance of cyclists in urban contexts, where more frequently used population based, and global models mask urban spatial patterns and safety performance.