Congestion Detection and Distribution Pattern Analysis Based on Spatiotemporal Density Clustering

Abstract

Urban congestion has multiple hazards to city transportation, safety and environment. Researches on urban congestion are conducive to prompting traffic management, assisting in urban planning, and ensuring the harmonious development of cities. This study proposed an improved spatiotemporal DBSCAN approach aiming to investigate the spatiotemporal distribution and variation pattern of traffic congestion from GNSS taxi trajectory data and applied on Wuhan, China. Firstly, low-speed trajectory sequences are extracted from taxi trajectories. Secondly, resorting to the idea of similarity and dissimilarity, we propose a new method of measuring the time distance and spatial distance between trajectories to extend traditional DBSCAN algorithm to spatiotemporal DBSCAN algorithm. Afterwards, congestion-prone areas in Wuhan are detected by the proposed method and DBSCAN method respectively. Finally, through the analysis and contrast of the congestion distribution on holiday, weekend, and weekday in multi-scale (time-series scale and date scale), we obtain the potential spatiotemporal distribution pattern of urban congestion in Wuhan.