The spatiotemporal dynamics of global urban expansion: Evidence from 2D urban area and 3D urban building volume

Accurate and comprehensively quantification of the dynamics of urban expansion is important for improving land use efficiency and sustainability in the context of rapidly urbanization in urban critical zones. However, existing studies still lack the understanding of the long-term dynamics of global urban expansion from both two-dimensional and three-dimensional expansions. In this study, we quantified the spatiotemporal dynamics of global urban expansion from 1990 to 2020, and used machine learning models and the SHAP method to explore the potential driving factors of urban expansion. The results show that the world as a whole has been expanding continuously over the past 30 years, with 5567 cities expanding to varying degrees, accounting for 74.3% of the total number of cities. Among them, the speed of urban expansion in South Asia is faster than that in other regions (2D UEI = 1.48, 3D UEI = 1.27). In addition, global urban expansion has shown an overall trend from a slow growth to a fast growth, and then a gradually decelerating growth. From the perspective of urban expansion type, the number of cities with vertical expansion is the largest, accounting for 32.8% of the total number of cities in the world, followed by cities with horizontal expansion and cities with unclear expansion. In addition, urban infrastructure construction and socioeconomic factors played important roles in urban expansion, among which population density can explain 55.1% of the variations of the two-dimensional urban expansion, and per capita urban building volume can explain 33.8% of the variations of the three-dimensional expansion. This study can provide a scientific basis for formulating urban planning according to local conditions and improving urban land use efficiency.