Remote sensing is a powerful tool for characterizing ecosystems at large scales. However, the relative importance of leaf traits and canopy structure in characterizing the spatial distribution of functionally distinct tropical forests – the most diverse, structurally complex, and heterogeneous ecosystems on Earth – remains under-explored. Using satellite-resolution LiDAR and imaging spectroscopy metrics, we map spatial turnover in tropical forest function, examine the relative importance of leaf traits and canopy structure, and analyze differences in aboveground carbon and demography. We find that leaf phosphorus, LMA, and canopy height are key distinguishing properties of forest types, achieving accuracies of 85-96% and correspond to differences in community growth and mortality rates. Our remotely sensed forest types align with ground-based forest definitions but enable mapping of their entire extent. At 30 m resolution, our method can be used at large scales with spaceborne data to reveal important differences in structure and function across tropical forests.