Extracting physical urban areas of 81 major Chinese cities from high-resolution land uses

Abstract

Past two decades have witnessed a rapid urbanization process in China, with the urbanization ratio suddenly increasing from 30.9 % to 63.9 %. Physical urban areas (PUA) are fundamental indicators to monitoring and evaluating urbanization, which differ from administrative urban areas and are much complicated to identify, as PUA contain heterogeneous land uses which are shaped by variant physical structures and diverse socioeconomic activities. Previous studies extracted PUA by densely populated, night-lighted, built-up, or artificial impervious surfaces, which consider either physical or socioeconomic aspect of PUA, but cannot measure both. Accordingly, this study firstly integrates physical and socioeconomic features derived from high-resolution (HR) satellite images and points of interests (POI) to extract HR land uses; then, a knowledge-based morphological aggregation method is proposed to aggregate different land uses and generate PUA based on spatial land-use structures. As the result, 450 PUA in 81 major Chinese cities are extracted and a China PUA dataset (namely CPUA) is generated. The CPUA is evaluated by reference to a widely-used global urban boundary dataset. The evaluation shows an accuracy of 92.5 %, demonstrating the effectiveness of the proposed method and the reliability of generated dataset. The evaluation also indicates that the generated CPUA outperforms the reference dataset in identifying urban parks and eliminating rural homesteads. Furthermore, the CPUA can be employed as fundamental data to monitor urbanization process and its spatial patterns, and thus plays an important role in evaluating sustainable city development. The CPUA is freely available on http://geoscape.pku.edu.cn/otherdata_en.html.