Outdoor artificial light at night (ALAN) may impact human health through disturbing human circadian rhythms. Laboratory experiments have shown that spectral differences in ALAN may lead to disparities in disturbing human circadian rhythms, but such spectral differences are overlooked in past urban studies due to the difficulties in measuring exposure. This study employed SDGSAT-1 data with multispectral nighttime light and the most advanced mobility-oriented research paradigm to investigate the different effects of outdoor ALAN exposures associated with different spectral bands. Based on surveys in Hong Kong, 502 participants provided their socio-demographic attributes, sleep quality, and GPS-derived trajectories. Outdoor ALAN exposures were derived at 10-m or 40-m spatial resolution using a 60-m buffer at each visited place and then spatiotemporally accumulated. The residence-based, mobility-based, and non-residential outdoor ALAN exposures were measured using red, green, blue, and panchromatic bands, respectively. Linear mixed-effects models were employed to estimate the associations between outdoor ALAN exposures in different spectral bands and the sleep quality indicators. The results indicated that panchromatic outdoor ALAN exposure measurements are the most accurate and comprehensive exposure measurements that minimize spectral configuration errors for sleep quality modeling. Along with the decreasing wavelength (from red to blue), the magnitude of estimated health associations and the confidence of estimation both decreased. This phenomenon can be well explained by Rayleigh scattering and spectral configuration errors. Consequently, this study provided essential implications for a range of studies in urban health and sustainability that need accurate outdoor ALAN exposures and improved outdoor ALAN settings.
