With more than three trillion individual trees determining the composition, structure and function of ecosystems around the world, the challenge of understanding their integrated role in the Earth system is complex. Forest and savanna systems are shaped by a variety of intrinsic and extrinsic processes related to climate, soils, and both natural and anthropogenic disturbances and influence biodiversity, biogeochemistry and biophysical feedbacks between the land and atmosphere. However, Earth system modeling approaches, where the land surface is represented by dynamic global vegetation models, or DGVMs, have struggled in their representation of forest dynamics because theoretical advances have not been computationally tractable and because observational data have been unavailable at global scales.
Three advances in representing forest demography in Earth system models are discussed with a focus on scaling principles from self-thinning theory and data integration with forest inventory and remote sensing. First, using long-term forest inventory data, we confirm that self-thinning drives population dynamics in undisturbed tropical forests, and that the relationship between mean-biomass and density can be used to scale individual-demographics. Second, Earth system models have incorporated mainly land cover data to represent deforestation and regrowth, and thus a new global dataset on forest age, derived from inventory, is presented as an alternative for initializing and benchmarking forest simulations. Lastly, new global remote sensing data are improving, by coupling Lidar methods with spaceborne observations, or radar methods to forest structural properties, which can be ingested within ecosystem models. Advancing how forest demographic processes are represented within Earth system models will help forecast how ecosystems respond to anthropogenic climate change in the 21st century.