Road centre line simplification principles for angular segment analysis

Angular segment analysis is one of the most fundamental analyses in space syntax practice that helps understand movement, land-use and other socio-economic patterns. It was initially applied in axial segment maps and later was used in road centre line maps as an attempt to overcome the 'segment problem' (Turner, 2005). Furthermore, the growing need to examine large urban systems has led to the wide use of road centre line maps instead of the previously hand-drawn axial maps. However, this transition to such datasets has lacked systematic studies on what is required to convert a road centre line map into a segment map, in order to produce reliable results of the angular segment analysis. To date, no consensual methodology has been developed within the space syntax community. This paper attempts to clarify what a road centre line segment represents spatially and suggests principles and rules to simplify a road centre line map to a segment map. Based on previous experience, the simplification mostly relies on the following two principles: reducing the number of nodes in the dual graph representation of a street network; optimising the angular change between adjacent nodes of the dual graph when space allows it. In addition to the above general principles, we discuss rules for special and complex cases, e.g. roundabouts, underpasses, bridges etc. To evaluate these rules and principles comparisons are carried out between traditional axial and RCL unsimplified and simplified segment maps, to develop a good understanding of how changes in dual graph representation of a street network can affect space syntax measure of 'choice'. Correlations of angular segment choice values are performed in order to evaluate which simplification technique can approximate better the axial representation of actual human activity. The results show that using a raw road centre line data set raises several inconsistencies in the analysis results, and the progressive application of the different simplification techniques brings these results closer to those of a traditional axial segment map, and thus to a better representation of socio-economic activity. The purpose of simplification is to minimise inconsistencies to ensure maximum accuracy in the results of angular segment analysis.