Emerging data sources and advanced microsimulation in transport modelling

Abstract

Transport modelling is an essential tool for policy makers to make informed transport planning decisions for the wellbeing of the society. However, most strategic transport models do not accurately incorporate the dynamics of population and households and their interactions, resulting in inaccurate forecasts of travel and land use demands. While it is highly crucial that strategic transport models provide accurate long-term forecasts that are vital for the appraisal of large transport infrastructure projects, short-term and medium-term forecasts are also integral for enhancing the efficiency of transport systems. Furthermore, in the modern, fast-paced world, emerging data sources are integral in both transport research and practice. For instance, data collected from smartphone travel surveys and electronic ticketing systems can be used to derive deep insights into people’s travel behaviour. Therefore, this thesis is set out to investigate how population evolution modelling can be improved in transport models, and also how emerging data sources can be leveraged to fill existing knowledge gaps. This thesis focuses on two main themes: (1) to explore new opportunities from emerging technologies and data sources and (2) to develop a dynamic microsimulation model that can provide a more accurate population input for a transport demand model and also to enhance the model development process. Using transit smart card data, The first main chapter explores the impacts of transit ridership and reliability on residential property values in Brisbane, Australia. Three frequently used spatial regression techniques in hedonic studies are used along with a simple linear regression model. The second main chapter presents the first analysis on how social media offers a cost effective means to recruit and engage smartphone travel survey participants. On top of that, recommendations on recruitment, marketing, sample representativeness, incentivisation, and deployment of the study are discussed. The third main chapter introduces an open source dynamic microsimulation toolbox for integrated urban modelling. This toolbox allows microsimulation models to be more easily developed and enables several well-known modelling packages in R to be integrated with the models. Finally, the last main chapter presents a demographic microsimulation model, built with the toolbox, that not only simulates the lifecycle of individuals but also the integration of new migrants into the population using a novel household alignment method.