Measuring Traffic Congestion with Novel Metrics: A Case Study of Six U.S. Metropolitan Areas

Abstract

Quantifying traffic congestion is a critical task for transportation planning and research. Numerous metrics have been developed, mainly focusing on changes in vehicle speeds, their extents, and travel time. In this study, new metrics are presented using the Hägerstrand’s space-time cube that has been studied from time geography perspectives since the 1960s. Particularly, the product of distance and time, i.e., distanceTime, is proposed as a base metric to measure traffic congestion amounts. Using the base metric such as mileHours, metrics of weighted congestion and normalized congestion amounts were also developed. New metrics were applied to six metropolitan areas and their vicinities in the United States (Atlanta, Chicago, Washington, D.C. and Baltimore, Dallas and Fort Worth, Los Angeles, and New York), and congestion amounts were calculated and compared. The Google Traffic Layer API was used to obtain traffic congestion datasets for six months (April–September 2022), and GIS (geographic information systems) was used for delineating road features and traffic intensity levels. Among the six areas, New York and its vicinity showed the largest congestion when only heavy congestion was used. Los Angeles and its vicinity showed the largest congestion when all congestion levels were considered. This study shows that the proposed metrics are very effective in summarizing traffic amounts and broadly applicable for further analyses of traffic congestion phenomena by associating various other factors, such as weekdays, months, or gas prices. The new metrics developed in this research may help transportation researchers and practitioners by providing them with a set of metrics applicable to summarizing congestion amounts by synthesizing congestion intensity, extent, and duration.

Method for the Identification and Classification of Zones with Vehicular Congestion

“…In parallel, traffic congestion assessment has established itself as a critical component in the effective management of road networks. Numerous investigations have devised methodologies and approaches for accurately pinpointing congested zones, employing a diverse array of criteria, spanning intricate elements like delay constraints as well as more straightforward factors such as traffic speeds [23,27,[29][30][31]. In addition, machine learning has proven effective in making use of both historical and real-time data to detect recurring congestion and anticipate congestion conditions in real time [32,33].…”

Methodology for the Identification of Vehicle Congestion Based on Dynamic Clustering

“…Several studies have developed methodologies and techniques for identifying congested areas accurately, using a variety of traffic and environmental characteristics [30][31][32][33].…”