Novel integration of high resolution satellites with drone flights improves monitoring of tree-crown scale autumn leaf phenology

Autumn leaf phenology signals the end of leaf growing season and shows large inter-crown variability in response to global climate change, which strongly regulates the carbon, water, and nutrient cycles from individual tree-crowns up to ecosystems. However, critical challenges remain with the monitoring of tree-crown scale autumn leaf phenology over large spatial coverage due to the lack of spatially explicit information of individual tree-crowns and high-quality and high-resolution time-series observations. Traditional field and proximate remote sensing methods (e.g. field observations, phenocam measurements and UAV flights) are the commonly-used means to monitor crown-scale leaf phenology, but are constrained to a very limited footprint and time span. Satellite remote sensing might provide another alternative solution, but most satellites remain too coarse spatial resolution to resolve individual tree-crowns.

To address the above challenges, we integrated local drone surveys that enable segmentation of each individual tree-crown with high-resolution PlanetScope satellite measurements of 3-m resolution and near-daily revisit cycle that enable autumn leaf phenology monitoring. To test this integrated method, we used a temperate forest in Northeast China as an example, as all the relevant data are available there. Our results show that the proposed drone-PlanetScope integration enabled to capture large inter-crown variation in leaf autumn phenology (i.e. ~ 30 days difference in leaf fall date across different species). The crown-scale phenology derived from PlanetScope also agreed well with local phenocam measurements (R²=0.81). These findings demonstrate large spatial heterogeneity in crown-scale autumn leaf phenology within a temperate forest, suggesting the importance of using high-resolution satellites to advance crown-scale phenology studies over large geographical areas.