A 3-dimension urban growth analysis using local climate zone mapping

Abstract

To comprehend urban evolution, it is necessary to investigate changes in urban land in megacities. However, volumetric expansion has not been well studied. This study proposes an approach to characterize a city's volumetric expansion for exploring urban land changes based on Local Climate Zone (LCZ) mapping by using a new convolutional neural network (CNN) model termed as DenseNetLCZ. The approach identifies three processes of urban growth: new urbanization, intensified compactness, and intensified height. The method was applied to four megacities, Beijing, Moscow, Paris, and Houston, from 2000 to 2020. The results showed satisfactory overall accuracy, ranging between 80% and 90%. The expansion of new urbanization was found to be consistently faster than intensified compactness and intensified height in all cities. Analysis of Beijing revealed that during this period, new urbanization increased by 52.8%, while intensified compactness decreased by 25%, and intensified height increased by 87.5%. However, due to the initially small base area of intensified height, this growth was less significant in terms of overall land coverage compared to new urbanization. Additionally, the diffusion to coalescence pattern was found to be beneficial for urban intensification. Our research forecasts urban expansion and intensification in 2100 under different Shared Socioeconomic Pathways (SSPs), indicating that stricter sustainability policies may promote concentrated, vertical urban growth, while looser ones may lead to more dispersed expansion, underscoring the crucial role of these policies in shaping future urban development strategies.