Using Point of Interest Data and Satellite Observation for Urban Functional Zone Mapping

Abstract

Today, China's development mode has shifted from extensive management to meticulous management, so it is inevitable that data will be used to conduct quantitative analysis. In this study, we synthesized multi-source data and the term frequency-inverse document frequency (TF-IDF) algorithm to determine the dominant function of land within the Fifth Ring Road of Beijing, China. Additionally, we explored how point of interest (POI) distributions affect nighttime artificial light radiation using the Spearman correlation coefficient. The overall accuracy of land functional classification in the central urban area of Beijing was 88.07%, and the kappa coefficient was 85.28%. The core area of Beijing is dominated by commercial and business facilities, accounting for 29.37% of the total area, followed by residential (22.02%), mixed-use (16.51%), green space (10.62%), administration and public service (9.66%), transportation (9.31%), and industrial (2.52%) areas. The number of commercial and business facilities, administration and public services, and residential areas correlate positively with the vegetation adjusted nighttime light (NTL) urban index (VANUI) value. In contrast, the number of industrial facilities yielded a negative correlation with VANUI. This study provides support for the optimization of the functional layout and the distribution of public resources in the central urban area of Beijing.