Exploring population spatiotemporal structure of cities with cellular data: A case study of Beijing

Amid rapid urbanization, prior studies frequently overlook non-resident populations who also utilize urban functions. Different population groups change dynamically over observation time scales. However, due to the difficulty of traditional data acquisition, it has been challenging to observe and study the temporal dynamics over the long term. This study utilizes cellular data to conduct a six-month tracking of the population in Beijing, aiming to provide a more precise analysis of urban population composition and behavior. The key findings are as follows: 1) Based on date trajectory data and considering long-term stay patterns, this study proposes a novel framework for examining urban population structures by integrating clustering algorithms and complex network theory. 2) The observation time window (OTW) was found to be closely associated with the identified urban population structure. A multi-level and multi-perspective sensitivity analysis was conducted on population size and proportion across different OTWs. The classification results were validated against resident population data from comprehensive transportation surveys, resulting in the identification of the optimal OTW for capturing the urban population structure. 3) Based on complex network theory, urban travel networks (UTNs) were constructed. Results revealed pronounced spatial heterogeneity and multi-centric patterns in travel behaviors across different groups, as reflected in global characteristics, local characteristics, and community structure. These findings offer valuable insights for urban transportation planning and management, highlighting the need for differentiated strategies tailored to the characteristics of diverse population groups.