Dynamic Global Vegetation Models (DGVMs) are the components of Earth System Models that predict vegetation distribution, composition, response to climate and disturbance, and fluxes to and from the atmosphere. These models are critical for understanding the global consequences of ecosystem processes. Due to the complexity of ecosystem processes, DGVMs greatly simplify ecosystem structure and dynamics for the sake of computational efficiency. As a result, they have not captured important heterogeneity within ecosystems and differential sensitivities to global change, such as the enhanced vulnerability of large trees to drought or the role of nutrients in modulating vegetation responses. In recent years, more complex processes have been integrated into a new generation of DGVMs. Detailed information on the demography, functional biology, and environmental conditions of tree species is needed to help improve and evaluate these new models. Tropical ecologists have collected such field data over decades, and advances in remote sensing have made information on vegetation structure, dynamics, and composition, over large areas, available. However, there remains a lack of integration between field and remote-sensing data, and DGVMs. This symposium aims to integrate empirical field ecology, remote sensing, and modeling to achieve novel inferences about tropical forest structure and function, and to improve predictions of the future of these critical ecosystems and their benefits to society. Part of the divide between modeling and ecological disciplines is related to scale: while the modeling community looks for broad, global-scale patterns, field-based ecological studies are often site-based, highly detailed, process-level studies. Bridging this gap is necessary to move tropical forest science and ecology forward. The talks in this symposium will describe the latest representations of tropical forests in DGVMs. They will describe some of the current limitations of the models and how field and remote-sensing data can be used to both improve the parameterization of models and provide much more robust testbeds for model evaluation. The symposium will greatly expand inter-disciplinary collaboration. Given the broad sweep of the topic, we propose two linked sessions. One session will introduce DGVMs and will investigate the linkages between DGVMs and empirical field ecology. The second session will explore how remote sensing can help link site-based field data with large-scale models. We further propose that this symposium be coupled with an informal workshop, after the symposium, where modelers, remote sensing scientists, and ecologists can engage, answer questions, and plan collaborations.