Characterizing annual dynamics of urban form at the horizontal and vertical dimensions using long-term Landsat time series data
Yixuan Wang,Xuecao Li,Peiyi Yin,Guojiang Yu,Wenting Cao,Jinxiu Liu,Lin Pei,Tengyun Hu,Yuyu Zhou,Xiaoping Liu,Jianxi Huang,Peng Gong
DOI: https://doi.org/10.1016/j.isprsjprs.2023.07.025
IF: 12.7
2023-01-01
ISPRS Journal of Photogrammetry and Remote Sensing
Abstract:The dynamics of built-up height are a crucial aspect of urban form, enabling the characterization of urban growth patterns and the attainment of sustainable development goals. While past studies have focused on urban extent mapping, little has been done to reveal changes in vertical structures in built-up areas. In this study, we reconstructed annual urban form dynamics in Beijing, China, from 1990 to 2020, utilizing Landsat time-series data. First, we identified urban growth and renewal years using a temporal segmentation approach. Then, we reconstructed the annual dynamics of built-up heights through logical reasoning, using the reference height in 2020 and the identified changing years. We evaluated the derived dataset and analyzed the urban growth patterns (i.e., upward and outward) over the past three decades. The detected change years are reliable, with an overall accuracy of 88% (±1 year). Meanwhile, the reconstructed built-up height data showed good agreement with surveyed results (R2: 0.92), with a root mean square error of 2.87 m and a mean absolute error of 2.07 m. Most expanded built-up areas during 1990–2020 were located at the fringe of the central city, and about 16% experiencing multiple changes. Overall, outward growth was prominent in Beijing during the first decade, while upward growth became dominant in most districts thereafter. The proposed framework provides a new way to reveal annual urban form dynamics through time series analysis, with great potential to support relevant studies such as urban climate modeling and carbon emissions studies.