Relationships between canopy surface temperature measured from drones and below-canopy forest microclimate in a tree diversity experiment

Abstract

Understanding forest microclimate is critical for predicting forest ecosystem responses to climate change. Yet its fine-scale variability–driven by tree diversity and vegetation structure–remains challenging to quantify and predict. Uncrewed Aerial Vehicles (UAVs) offer a promising tool for capturing high-resolution and spatially continuous thermal data to link canopy characteristics to microclimate dynamics. Despite growing UAV use, it remains unclear whether (above) canopy surface temperature (Tcan) via thermal imaging can predict below-canopy (understory) air temperature Tmicro, and how this relationship is influenced by canopy characteristics, tree species richness and wind speed. In a unique forest tree diversity experiment with varying tree richness levels (1, 2, and 4 tree species), we assessed the correlation between Tcan, measured via UAVs, and forest Tmicro, measured by understory loggers across two growing seasons. We found a significantly positive correlation between Tcan and Tmicro and between their offsets from the macroclimate temperature throughout the entire growing season. However, for single flights on individual days or at specific hours, this correlation was weaker. Incorporating vegetation indices related to leaf chlorophyll or biomass derived from UAV imagery as covariates sometimes, but not systematically, improved the prediction of Tmicro based on Tcan. Tree diversity effects were not significant. Overall, UAV-based predictions of microclimate were most accurate when informed by multiple flights across the growing season. Our study advances the integration of remote sensing and forest microclimate ecology for microclimate prediction across tree diversity gradients.