Beyond Centrality: Understanding Urban Street Network Typologies Through Intersection Patterns

TL;DR

This study introduces a geometry-centric intersection-angle metric to classify 3- and 4-way street intersections and combines it with traditional topological metrics to cluster cities at a regional scale. Applied to the San Francisco Bay Area, the approach reveals three typologies—Gridded, Orthogonal, and Organic—facilitated by an enhanced clustering pipeline that includes node-patterns and bearing features. The enhanced clustering outperforms a baseline model, offering finer differentiation of street networks and actionable insights for planning, evacuation, and traffic management. The work highlights the practical value of integrating intersection geometry into urban network characterizations and outlines limitations related to directed graphs and angle ordering for future improvements.

Abstract

The structure of road networks plays a pivotal role in shaping transportation dynamics. It also provides insights into how drivers experience city streets and helps uncover each urban environment's unique characteristics and challenges. Consequently, characterizing cities based on their road network patterns can facilitate the identification of similarities and differences, informing collaborative traffic management strategies, particularly at a regional scale. While previous studies have investigated global network patterns for cities, they have often overlooked detailed characterizations within a single large urban region. Additionally, most existing research uses metrics like degree, centrality, orientation, etc., and misses the nuances of street networks at the intersection level, specifically the geometric angles formed by links at intersections, which could offer a more refined feature for characterization. To address these gaps, this study examines over 100 cities in the San Francisco Bay Area. We introduce a novel metric for classifying intersections, distinguishing between different types of 3-way and 4-way intersections based on the angles formed at the intersections. Through the application of clustering algorithms in machine learning, we have identified three distinct typologies - grid, orthogonal, and organic cities - within the San Francisco Bay Area. We demonstrate the effectiveness of the metric in capturing the differences between cities based on street and intersection patterns. The typologies generated in this study could offer valuable support for city planners and policymakers in crafting a range of practical strategies tailored to the complexities of each city's road network, covering aspects such as evacuation plans, traffic signage placements, and traffic signal control.

Figures (9)

• Figure 1: a. The out-degree distribution of the top twenty cities in the Bay Area. b. Distribution of various geometric metrics across cities.
• Figure 2: 3-way (left) and 4-way intersection patterns (right) for nodes.
• Figure 3: The figure displays a zoomed-in portion of the road networks for selected cities.
• Figure 4: The figures show the distribution of metrics for baseline clusters.
• Figure 5: The figures display example cities from enhanced clustering. The green polar plots illustrate the street bearings of all links in the city.