Investigating the Impact of Network Structure on Road Traffic Crashes: A Macro-level Analysis

Abstract

Road traffic crashes (RTCs) are one of the significant externalities associated with transportation that pose a serious public health risk. While addressing the safety concerns of road transport, road network structure is often overlooked. Among the high-risk locations within a road network, intersections pose a particular concern due to the intricate and conflicting traffic movements. Intersection skewness (the angle at which an intersection deviates from 90 degrees) and intersecting road categories are considered to influence intersection safety significantly. While previous research has extensively explored the impact of skew angles and intersecting road categories at the intersection level, there is a lack of investigation at the ward/zonal level. This study aims (a) to estimate the skewness and heterogeneity of intersecting leg categories at the ward level and (b) to model the RTCs and the street network characteristics such as skewness, hierarchy, connectivity, and centrality along with covariates like ward-level population density and economic status. Utilizing data from OpenStreetMap, this study conducted the ward-level analysis for Delhi, India. This study, along with the proposed metrics, also quantified various network structure metrics that are commonly used in safety analysis, such as betweenness centrality, meshedness coefficient, average edge length, node density, and average circuity. Using three years of crash data from 2017-19, this study modelled the crash rate at the ward level for Delhi (289 wards) by developing a negative binomial regression model. The explanatory variables include the network structure metrics, nightlight intensity (indicative of economic status), and population density. It was observed that wards with a higher deviation from the conventional right-angle intersection experienced a higher rate of crashes. Moreover, the wards with more heterogeneous intersections had more crash rates. However, wards with more central (higher betweenness) road network structure had fewer crashes. Furthermore, wards with more circuitous roads (more network distance to travel from one point to another) had more crash rates. However, a negative association was found between the RTCs and ward economic status, indicating that people who are economically disadvantaged continue to be the most at risk for fatal crashes.