Vacant Land

[Objective] In the new phase of industrialization and urbanization in China, resourcedepleted cities are facing various development challenges. Taking the characteristics of resourcedependent cities as a starting point, this study, using Hegang City as an example, proposed a method for identifying inefficient spaces to address typical spatial issues in resource-depleted cities. [Methods] Through a literature review, this study systematically identified a series of spatial issues faced by resource-depleted cities. Based on the actual situation in Hegang City, problems related to mining subsidence areas, urban vacant land, spatial disorder areas, and abandoned buildings were recognized. Building upon existing data, this study introduced innovative deep learning models for automatic detection that identify urban vacant land, spatial disorder areas, and abandoned buildings. [Results] This study employed the DeepLab V3 and SegNet models to generate a dataset of inefficient spaces in Hegang City. The identification results were refined through field surveys. The research visualized the distribution of mining subsidence areas, urban vacant land, spatial disorder areas, and abandoned buildings within the city. [Conclusion] The practical application in Hegang City demonstrated that the research methods are capable of efficiently, quickly, and accurately identifying inefficient spaces at the city scale. This provides an effective technical support for the identification of inefficient spaces in resource-depleted cities. However, there is still room for improvement in the definition of the objects being identified and in the technical details of the proposed research methods, necessitating further research for enhancement.

Urban vacant land (UVL) has been an important issue in the urbanization process, especially for shrinking cities. Identifying UVL and analyzing its spatiotemporal characteristics are the foundation for coping with this issue. This study identified UVL in 497 shrinking cities on the globe (10 % of shrinking cities in total) in 2016 and 2021 using manual labeling and deep learning to reflect the distribution patterns of UVL and its spatiotemporal changes. The results reveal that a global expansion of UVL from 2016 to 2021 in 497 shrinking cities, with diverse distribution patterns and varying changes across different regions. As for socioeconomic factors, UVL is related to population shrinkage, and the UVL ratio presents a phased change with the increase of the urbanization rate, revealing an inverted U-shaped relationship between the UVL ratio and the urbanization rate. The distribution patterns of UVL also vary globally in different urbanization phases. This study can provide theoretical and practical insights for improving urban planning and promoting sustainable urbanization. 

Urban vacant land is a growing issue worldwide. However, most of the existing research on urban vacant land has focused on small-scale city areas, while few studies have focused on large-scale national areas. Large-scale identification of urban vacant land is hindered by the disadvantage of high cost and high variability when using the conventional manual identification method. Criteria inconsistency in cross-domain identification is also a major challenge. To address these problems, we propose a large-scale automatic identification framework of urban vacant land based on semantic segmentation of high-resolution remote sensing images and select 36 major cities in China as study areas. The framework utilizes deep learning techniques to realize automatic identification and introduces the city stratification method to address the challenge of identification criteria inconsistency. The results of the case study on 36 major Chinese cities indicate two major conclusions. First, the proposed framework of vacant land identification can achieve over 90 percent accuracy of the level of professional auditors with much higher result stability and approximately 15 times higher efficiency compared to the manual identification method. Second, the framework has strong robustness and can maintain high performance in various cities. With the above advantages, the proposed framework provides a practical approach to large-scale vacant land identification in various countries and regions worldwide, which is of great significance for the academic development of urban vacant land and future urban development.