Street networks as determinants of active travel: application, evaluation and development of network-based measures and methods

Abstract

Street networks are the skeletons of cities. They shape land use and urban form, allow for movement, and provide access to destinations. Well-connected and safe pedestrian and cycling networks encourage active travel – walking and cycling – and discourage car use. While urban density and diversity provide proximity to relevant daily destinations, they can hardly support active travel if these destinations are not accessible through the network on foot or by bike. Despite the importance of walkable and cyclable streets for sustainable transport, common area-based, aggregated connectivity measures are often too coarse, spatially confined, or variable in scale for an effective evaluation of pedestrian and cycling networks. Spatial network analysis offers an alternative with high-resolution, citywide measures of connectivity, accessibility, and density. However, there is limited empirical evidence on the relationship between network measures and active travel – i.e., walking or cycling mode choice, modal shares, and frequencies – not least due to the complexity of network analysis with its various measures and network representations rooted in different disciplines.

This thesis examines network measures and their associations with walking and cycling, sheds light on their theoretical backgrounds, links network analysis to urban form and transportation, and develops new network-based measures to address identified gaps. The main research question “To what extent do network measures of connectivity, accessibility, and density explain active travel behaviour?” is explored through several sub-research questions addressed in one review paper, four empirical papers, and one supplementary paper.

The review paper summarises measures, methods, and models for evaluating cycling network connectivity and its associations with cycling mode choice, modal shares, frequencies, and flows. One empirical paper examines the associations of “reach” – i.e., the number of streets, amount of floor area, population, or retail accessible within walkable distances through the network – and centrality with walking frequencies. To address the limitations of network analysis for predicting mode choice for specific origin-destination trips, this thesis introduces the novel measure “route betweenness”, which quantifies the connectivity of entire shortest routes while accounting for route directness and urban density. I apply this measure in two empirical papers on active travel to school (ATS) and in one paper on the effects of a pedestrian- and cyclist bridge on the probability of cycling.

All empirical papers reveal significant positive associations between network measures and active travel. The key findings are:

Frequent walking is strongly and positively associated with reach, more so than with centrality or area-based measures, indicating that street networks for walking should be evaluated not only at the neighbourhood scale but also within a wider urban network. ATS is strongly and positively associated with route betweenness, indicating that well-connected school routes passing through areas with dense networks, activities, and opportunities for social interaction support ATS. Route betweenness is strongly associated with cycling mode choice. However, its statistical effect may be so pronounced that it can lead to over- or under-predictions of cycling probabilities when estimating the impacts of infrastructure changes.

This thesis contributes to research by expanding network analysis from single network elements to entire routes, based on the theoretical foundations of established network measures. The presented research further demonstrates that reach is a fundamental metric for more prominent measures, such as closeness and betweenness centrality. By providing an overview of the strengths and limitations of relevant measures and linking network analysis to urban form and transportation, this thesis makes network analysis more accessible to urban planners. In conclusion, the findings strongly support the application of network analysis in urban morphology, accessibility, and transportation research as well as in urban planning practice. I hope that this thesis will inspire further research and practical applications.