Source-linked AI summary
Street-based Topological Representations and Analyses for Predicting Traffic Flow in GIS
Bin Jiang, Chengke Liu
TL;DR
The paper asks whether axial maps adequately represent the street morphology associated with traffic flow, given concerns about their validity and construction. It develops street-based topological representations and tests them with Hong Kong street-network and AADT data, finding better traffic predictability than conventional axial maps and proposing an alternative GIS representation.
Problem
Axial maps are questioned for cognitive and computational limitations, motivating comparison with street-based representations for traffic-flow prediction.
Method
The paper develops street-based topological representations and algorithms, then compares them with an axial map using Hong Kong street-network and AADT datasets.
Results
Street-based representations provide better traffic predictability than axial lines, with predictability for some areas reaching 0.7.
Takeaways & Limitations
The authors propose street-based topological representations as an alternative GIS representation and topological analysis as a means for geographic knowledge discovery.
Abstract
from arXiv · showhide
It is well received in the space syntax community that traffic flow is significantly correlated to a morphological property of streets, which are represented by axial lines, forming a so called axial map. The correlation co-efficient (R square value) approaches 0.8 and even a higher value according to the space syntax literature. In this paper, we study the same issue using the Hong Kong street network and the Hong Kong Annual Average Daily Traffic (AADT) datasets, and find surprisingly that street-based topological representations (or street-street topologies) tend to be better representations than the axial map. In other words, vehicle flow is correlated to a morphological property of streets better than that of axial lines. Based on the finding, we suggest the street-based topological representations as an alternative GIS representation, and the topological analyses as a new analytical means for geographic knowledge discovery.
1. Introduction
The paper questions conventional axial-map representations for modeling traffic flow and introduces street-based topological representations as an alternative. It evaluates this approach through Hong Kong street-network and AADT experiments.
- Axial-map traffic modeling represents streets as intersecting axial lines and ranks them using graph-theoretic measures.
- Axial maps are criticized as cognitively unsound and difficult to generate consistently because they require manually drawing the least number of longest lines.
- Streets replace axial lines in the proposed topological representation, with named streets formed by merging same-name street segments.
- The paper contributes a street-based representation, traffic-prediction evidence favoring it over axial maps, and a research prototype with related algorithms.
- Experiments use the Hong Kong street network and Annual Average Daily Traffic datasets to compare street-based representations with axial maps.
2. Topological representations and analyses
The paper represents urban streets and their relationships as graphs, then analyzes connectivity, path length, integration, clustering, and related local or global measures. These topological abstractions are intended to reveal network structure beyond geometric representations.
- 2.1 Geometric versus topological representations of urban street networks: A fictional city illustrates GIS layers, axial maps, and street networks, with lines or streets represented as graph nodes and intersections as links.The axial-map example has 13 axial lines and 18 intersection nodes, whereas the street network has 7 streets and 11 nodes.
- 2.1 Geometric versus topological representations of urban street networks: The adopted primal graph encodes street-street relationships; dual graphs instead encode point-point relationships.
- 2.1 Geometric versus topological representations of urban street networks: Axial lines approximate linear space perceptually, whereas named streets provide a cognitive-based representation of interconnected streets.
- 2.2 Topological measures: A graph uses vertices and edges, with matrix entry r_ij equal to 1 when two objects intersect and 0 otherwise.
- 2.2 Topological measures: Connectivity is a node’s number of directly connected nodes, while path length measures its distance from all other nodes through graph geodesics.
- 2.2 Topological measures: Clustering coefficient measures how likely two neighbors are to be linked, using the ratio of actual to possible edges.For node A, four neighbors yield six possible friendships and one actual friendship, producing 1/6.
- 2.2 Topological measures: Local integration considers nodes within a limited step range, whereas global integration considers all other nodes; local and global measures support intelligibility analysis.
- 2.3 Why the topological representations and analyses?: Topological representations can expose hidden network structures or patterns that geometric representation and analysis do not illustrate.
3. Experiments and results
Using Hong Kong street-network and AADT data, the experiments compare axial-line, named-street, and natural-street topologies. Street-based representations exhibit network properties and traffic-flow predictability that generally outperform axial lines, with greater stability across urban morphologies.
- Data sources and topology formation: Hong Kong experiments compare an axial map with named-street and natural-street topologies using street-network and AADT data.The study constructs an axial map and multiple street-based topologies, then correlates their measures with observed traffic flow.
- Data sources and topology formation: Six natural-street topologies use continuity thresholds from 20 to 70 degrees, while the named-street algorithm produces 7488 named streets.The continuity threshold changes the number of natural streets and supports correlation comparisons across multiple topology sizes.
- Computing topological properties: Street-based topologies show small-world behavior, while the axial-line topology has a longer path length than its random counterpart and a weaker small-world property.Named streets and natural streets at a 60-degree threshold have path lengths close to their random counterparts.
- Computing topological properties: The axial-line, named-street, and natural-street degree distributions have exponents around 2.0, indicating scale-free properties across the three topologies.The paper identifies fewer than 1% of streets as forming a network backbone and highlights these highly connected streets.
- Predictability of traffic flow: Street options predict traffic flow better than axial lines in the 10 sampled areas, with predictability reaching 0.7 in some areas.The axial-lines option is the poorest, whereas named and natural streets improve predictability individually and in their overall mean.
- Predictability of traffic flow: Axial-line predictability declines sharply from grid-like to deformed-grid and irregular areas, whereas named- and natural-street predictability remains comparatively stable.The morphology-related contrast reinforces the overall finding that street-based representations outperform axial lines for traffic forecasting.
4. Conclusion
The paper concludes that street-based topological representations outperform conventional axial maps for traffic prediction and can reveal hidden urban structures. It proposes them as an alternative GIS representation for broader geographic analysis.
- Street-based topological representations and analyses are superior to conventional axial maps for predicting urban traffic flow.
- These representations can uncover hidden structures or patterns that conventional geometric-oriented representations may not reveal.
- The paper proposes street-based topological representations as an alternative GIS representation for street networks.
- Future work will examine wider applications of the proposed representations and analyses.