Objectives  Based on taxi movement trajectories, urban traffic operating conditions can be analyzed. At present, the relevant research mainly focuses on the investigation of traffic origin and destination points and the extraction of traffic hotspots, but seldom analyzes the interaction between traffic hotspots. We focus on exploring the temporal and spatial patterns of urban traffic hotspots, with the purpose of providing a scientific basis for traffic management departments for traffic management in different periods and sections.

  Methods  We take the main urban city of Nanjing City as the research area, and use taxi GPS data in 2015, 2016, and 2017 for experiments. Firstly, the effective trajectory flow is extracted from the taxi GPS data through the local anomaly test method. Secondly, based on the GeoJSON geocoding to realize map matching, the Pathlet method is used to process the trajectory flow into a sequence which can be optimally solved by dynamic programming. Finally, based on the standard for evaluation of urban traffic operation status, the macroscopic fundamental diagram model is constructed according to the relationship between traffic flow, speed and density in the road network, and the evaluation index is established by combining travel time ratio and the delay time ratio.

  Results  (1) Judging from the average of the traffic operation indicators, the main urban area of Nanjing is the most congested in the morning and evening rush hours, and the traffic is smooth in the early morning hours, and the traffic hotspots are more active on working days than on rest days. (2) Judging from the traffic operation indicators of the nine different nodes, the traffic congestion near the parking lot and office building nodes is relatively increased on weekdays, and the traffic congestion near the nodes of shopping malls and residential quarters is relatively increased on the rest days.(3) Judging from the perspective of spatial distribution of traffic hotspots in five representative periods, traffic hotspots mostly appear near intersections and ramps, and there are more interactions between main roads and express roads, but between townships and village roads and county roads. There is less interaction. (4) Judging from the perspective of the temporal and spatial patterns of traffic hotspots near the school during the morning rush hour, the temporal and spatial interaction characteristics of traffic hotspots are significant in areas with high node coverage, while the spatial distribution characteristics of traffic hotspots change unclearly with time in areas with low node coverage.

  Conclusions  Urban traffic hotspots show significant differences in time and space distribution characteristics, and their spatiotemporal interactions become stronger with time and space aggregation, which has a significant positive correlation.