Although forest phenology arises from changes in leaves and trees, it controls processes including the carbon and water cycles, often observed at the ecosystem scale. Therefore it is important to characterize the relation between phenophase transition events at different spatial scales. We use an unmanned aerial vehicle (UAV) equipped with a digital camera to observe plant phenology over a large area (5.4 ha) of the Prospect Hill tract of Harvard Forest, with high spatial (10 m) and temporal (~5 day in spring, weekly in autumn) resolutions. We compare UAV-derived phenophase transition dates with in-situ observations, PhenoCams, and satellite remote sensing, and examine correlates of spatial variance in phenology using a detailed species inventory and remote sensing of land surface temperature. We find that species distribution explains most of the observed variance in phenology dates. We also explore statistical relations governing the scaling of phenology from an organismic scale (10 m) to forested landscapes (1 km) by incorporating Landsat and MODIS data, finding that spatial variance in transition dates decreases linearly with the logarithm of coarser spatial resolution. Our study is the first to observe ecosystem-scale phenology with fine enough spatial resolution to describe organism-level correlates and scaling phenomena.