Main content area

Modelling spatial distribution of fine-scale populations based on residential properties

Han, Dongrui, Yang, Xiaohuan, Cai, Hongyan, Xu, Xinliang, Qiao, Zhi, Cheng, Chuanzhou, Dong, Nan, Huang, Dong, Liu, Andi
International journal of remote sensing 2019 v.40 no.14 pp. 5287-5300
data collection, decision making, models, planning, prediction, remote sensing, residential areas, resource allocation, traffic, China
Fine-scale population gridded datasets are of great significance in emergency response, resource allocation, and traffic planning. Many studies have developed fine-scale population spatialization models based on building patch area (BPA) and building floor (BF). However, little attention has been given to house occupancy rate (HOR). Based on BPA, BF and HOR, this study proposed a novel fine-scale population spatialization method, taking the six districts of Beijing as the study area. The results showed that the HOR in central Beijing was higher than that of the surrounding area. The model with consideration of HOR was more accurate than that without it. In addition, the fine-scale population gridded map generated by this novel method was more accurate (mean prediction error = 8.47%). For all the testing samples, the relative errors of the population gridded data were between −13.54% and 16.04%. Thus, this study suggested that HOR could be a key indicator for fine-scale population modelling. Furthermore, the proposed method could be employed in generating fine-scale population gridded maps and could provide credible and fundamental data for rapid response and decision-making.