1. Angular and Metric Distance in Road Network Analysis: A nationwide correlation study.
- Author
-
Serra, Miguel and Hillier, Bill
- Subjects
- *
STREETS , *ANGULAR distance - Abstract
Abstract There is an epistemological divide in the field of road network analysis, concerning the way network distance should be conceptualized. On one hand, the generality of studies in the field adopt metric distance as a self-evident choice. On the other, space syntax studies adopt a different distance concept, namely that of angular distance, which ignores actual physical lengths. Theoretically, these two distance concepts imply quite different assumptions. Analytically, they produce also different results. In this paper we assess the differences between these two network distance concepts, using a model of the UK's complete road network (2,031,971 nodes) and a very large dataset of vehicular movement counts (20,752 locations). We describe the statistical associations between observed vehicular flows and the betweenness centrality of the road-network nodes where such flows were measured, the latter calculated with metric and angular distance functions, across a number of increasing radii. Relations to road capacity are also discussed in principal roads where this is known. The geographical comprehensiveness of our model and the size of our movement sample allow us to state, with unprecedented statistical validity, the clear outperformance of angular distance over metric distance, in what concerns the strength of the studied correlations. This is true for all types of roads (urban and non-urban) and for all motorized vehicles, representing 99.05% of the total traffic; the remaining 0.95% correspond to pedal cycles, which yielded inconclusive results. We also demonstrate the existence of two distinct regimes of association between movement and centrality, conspicuous in cities but altogether absent in non-urban areas, which represent new evidence supporting space syntax's dual model of urban form. Highlights • There is a basic epistemological divide in the field of street and road network analysis, which concerns the way network distance should be conceptualized. • In this paper we address this divide, through a nationwide study assessing the differences between the correlations of vehicular movement flows and two types of network distance (metric and angular). • We use a nationwide network model (the UK mainland) and a very large dataset of vehicular movement count points (N = 20,752, distributed over the whole UK territory). • The geographical comprehensiveness of our model and the size of our movement sample allows us to state, with unprecedented statistical validity, the clear outperformance of angular distance over metric distance, in what concerns the strength of the studied correlations. • This result strongly supports the adoption of angular distance in the analysis of street and road networks and has important consequences for their modelling and theoretical understanding. [ABSTRACT FROM AUTHOR]
- Published
- 2019
- Full Text
- View/download PDF