Coupling fine-grained nighttime Earth observation and optimal buffer for mobility-oriented measures of high-quality...

Earth observations can well facilitate the modeling of human living environments but still need to be coupled with proper geographic information system (GIS) techniques to fully unpack their power for human health services. In this study, we investigated the buffer zone with optimal size as the proper contextual unit for the high-quality measurements of people's exposure to outdoor artificial light at night (ALAN) using SDGSAT-1 glimmer data and the mobility-oriented research paradigm. A large sample set with comprehensive socio-demographic attributes and self-reported health statuses was collected from 940 participants in Hong Kong through a stratified scheme. Participants' trajectories were delineated from GPS-equipped smartphones, while Hong Kong's nightscape was delineated from SDGSAT-1 images at a spatial resolution of 10 m. We derived participants' exposures to outdoor ALAN along their nighttime trajectories using a range of buffer sizes from 10 m to 100 m. We employed binary logistic regression to evaluate the quality of exposure measurements by modeling people's overall health outcomes. Our results clearly showed the U-shaped p-value curves along the gradient of buffer size in two representative communities, which indicates the existence of an optimal buffer size at 70 m in pertinent geographic contexts. We also observed that improper settings of contextual units cannot adequately capture the high-quality environmental exposure, and the spatial non-stationary issues may further complicate the quality issues in exposure measures. Our results provide essential insights into a range of health geography, epidemiological, and public health studies that need Earth observation data for accurate environmental exposure measures.