Estimating building-scale population using multi-source spatial data. (April 2021)

• Record Type: Journal Article
• Authors: Shang, Shuoshuo; Du, Shihong; Du, Shouji; Zhu, Shoujie
• Is Part Of: Cities. Volume 111(2021)
• Publication Date: 2021-04
• Subjects: Population distribution -- Building classification -- Urban functional zones -- Multi-source geographic data -- Urban studies; City planning -- Periodicals; Urban policy -- Periodicals; 711.4
• DOI: 10.1016/j.cities.2020.103002
• Languages: English
• ISSNs: 0264-2751

Abstract

Abstract: Fine-scale population distribution data is essential for social and geographical studies, such as social resource analysis, emergency evacuation, business decision-making and urban planning. At present, the estimated population distribution of most studies at fine scales are in the form of grid. For studies at building scale, however, the spatial heterogeneity is hardly considered. This study develops a novel method to estimate population at building scale by considering the spatially heterogeneous of population distribution through fusing urban functional zones (UFZs) data and multi-source geospatial data. First, residential buildings are classified into different categories based on UFZ data. Second, multiple residential indexes are defined to describe residential space by using multi-source geospatial data. Finally, a random forests model is established to estimate population at building scale. A study area of Ningbo, China, is employed to evaluate the proposed method. The R 2 between predicted population with statistical population is 94% at community level, and the MAPE (Mean Absolute Percentage Error) is 19%, which are better than the results of the state-of-the-art methods, illustrating the effectiveness of the proposed method. The estimated building-scale population data in this study can contribute to scientific management of urban modernization and optimal allocation of resources. Highlights: This study develops a novel approach to estimate population at building scale by using multi-source spatial data. This model considers the heterogeneity of the population distribution on the level of buildings. We consider socio-economic factors affecting population distribution characterized by POIs. The population estimation model in this study has high accuracy.