The spatial relationship between pedestrian flows and street characteristics around multiple destinations

• Pedestrian flows on non-pedestrianised streets are accounted for well by street characteristics of origins.
• Pedestrian flows on pedestrianised streets are accounted for well by street characteristics of routes to a nearest key destination.
• While stations are important as the single nearest destination for pedestrians, attractions are important for the multiple nearby destinations.
• Nearby key destinations are linked by shops in a city centre.
• Pedestrianisation currently makes limited contribution to pedestrian flows on links between nearby key destinations.

Accessibility improvement for pedestrians has received increasing attention in planning. However, pedestrian space is more likely to be designed only for individual streets to secure minimum easiness of walking, and little attention has been paid to developing a street network for pedestrians to walk around multiple destinations on a neighbourhood scale. There is also a lack of empirical analysis of how much pedestrian accessibility would vary depending on the characteristics of streets on routes to specific destinations. This paper is aimed at examining the spatial relationship between pedestrian flows by street type and various street characteristics around multiple destinations in a city centre. First, a literature review summarises what street characteristics should be considered in accessibility analysis for pedestrians. Then, a pedestrian flow model is developed in a way that measures accessibility with street characteristics of origins, destinations, and routes on multi-scales from on-street ones to neighbourhood-scale ones. A multiple regression model is made using data from the West End area in London, in which street characteristics are taken for routes from each street segment to nearby stations and attractions. As a result, this analysis found that the route characteristics to a single nearest station and attraction can account for pedestrian flows well, but route characteristics to multiple nearby stations and attractions do not improve the model fit. These results are more prominent for pedestrianised streets. Their implication may be that these destinations are currently not linked well for pedestrians, and pedestrianisation is required to contribute more to the linkage.