Autumn leaf colour brightness is an important cultural ecosystem service. As its spatial patterns and ecophysiological mechanisms remain unclear, we analysed relationships among autumn leaf colour brightness, late summer chlorophyll content, and topographic position in both canopy-based micro-scale analysis and site-based macro-scale analysis. Multispectral drone observations were made in three Fagus crenata forests at elevations of 300, 600, and 900 m in Niigata Prefecture, Japan. In a drone-acquired digital surface model of canopy area distribution, we analysed the canopy-averaged autumn leaf colour brightness, four vegetation indices, topographic position index, and elevation. The macro-scale pattern showed brighter yellow leaves at lower elevations, suggesting the effect of temperature through productivity or evaporative stress. The micro-scale pattern showed brighter yellow leaves in canopies that had a low late summer chlorophyll content, or that grew on ridges. The negative correlation between autumn leaf colour and chlorophyll content suggests that low-chlorophyll trees experience stresses (e.g. irradiance, desiccation, senescence) that induce higher carotenoid content to defend against such stresses. Increased autumn leaf coloration on ridges is consistent with a stress gradient. Although further research is needed to reveal the underlying physiological and ecological mechanisms, autumn leaf colour brightness has different meanings at different scales and thus has potential not only as a cultural ecosystem service but also in forest management through early detection of stress.