Xiangzhong Guo, Tao Lin, Yicheng Zheng, Matthew Ng, Junmao Zhang, Hongkai Geng, Zixu Jia, Jing Lin, Yuan Chen, Hoon Han, Christopher Pettit
Understanding the spatial distribution and temporal dynamics of population activities is critical for advancing interdisciplinary research in economic development, climate change, environmental management, and urban sustainability. Although nighttime light remote sensing data has been widely used to represent human activity, limited research has examined their spatiotemporal relationship between static nighttime light (NTL) and dynamic population activity intensity (PAI). This study integrates high-resolution SDGSAT-1 NTL imagery with hourly Baidu User Density data to investigate the multiscale spatiotemporal relationship between NTL and PAI in Shanghai, China. Using bivariate spatial autocorrelation and a multiscale analysis framework, we assess spatial association across four different time segments: satellite transit, nighttime, daytime, and whole day. At the 200m scale, the spatial association between NTL and PAI is both positively correlated and temporally stable, as evidenced by global Moran's I value (p < 0.001) ranging from 0.38 to 0.43 and a clustering consistency index of 0.88 across different time segments. Extending the analysis to ten different spatial scales (200m–2000m) reveals scale effects in the spatiotemporal relationship. While clustering patterns between NTL and PAI vary with spatial scale, temporal stability remains consistently high within each scale, with clustering consistency indices ranging from 0.88 to 0.96 across different time segments. A semi-variogram-based method identifies 1000-1400m as optimal spatial scale range at which NTL can best represent PAI, with 1200m being the optimal scale suitable for this study. These findings promote the use of NTL data in capturing the spatiotemporal dynamics of human activity, underscore the temporal stability of the relationship between NTL and PAI, and highlight the critical role of spatial scale selection in NTL-based population research.