Disentangling relations between dynamic urban structure and its efficiency in 287 cities across China

Abstract

Conventionally urban structures are mainly inferred from the spatial arranges of related urban elements such as land use and population density. China's unprecedented urbanization process has significantly changed urban spatial structure and it is not clear how urban structure impacts the sustainability and livability of cities across a large geographic space. Furthermore, urban structure always shows dynamical variations during the day and over the night, which are a challenge to be captured by the conventional research methods and data. In this paper, we evaluated and compared the characteristics of the urban structure of the 287 cities in China by examining the dynamic changes of urban human activities distribution inferred from location requests collected by Tencent's platform. We developed two new indexes, the Dispersion Variation Index (DVI) and Spatial Dispersion Variation Index (SDVI), using the Lorenz curve, Gini coefficient, and adjusted Ripley's K function to describe the overall agglomeration intensity and the temporal variability of human activities. We also applied the urban scaling law to create two further indexes, the Scale-Adjusted Metropolitan Indicator DVI (SAMI_DVI) and the Scale-Adjusted Metropolitan Indicator SDVI (SAMI_SDVI), which enable us to evaluate the efficiency of urban structures. Our results show that the daily fluctuations of the urban structure are related to urban population, and also impacted by urban planning, topography, and other factors. The 1st-tier cities have the highest DVI (median: 0.020) and SDVI (0.052) values, which suggest the most volatile urban structure during the day. The 2nd-tier cities have the lowest urban structure efficiency, as suggested by higher SAMI_DVI (0.17) and SAMI_SDVI (0.28) values. Also, 69.2% and 76.9% of the 2nd-tier cities have DVI and SDVI values over the predicted values based on their population sizes. We also demonstrate that cities with high unstable structural characteristics tend to exhibit lower commuting efficiency and higher carbon emission intensity. Our framework outperforms previous metrics, is highly scalable, and can be implemented at little cost, allowing studies to be conducted even in small and medium-sized cities with little traditional data.