The synergistic use of models and observations: understanding the mechanisms behind observed biomass dynamics at 14 Amazonian field sites and the implications for future biomass change

The Amazonian rainforests play a vital role in global water, energy and carbon cycling. The sensitivity of this system to natural and anthropogenic disturbances therefore has important implications for the global climate. Some global models have predicted large-scale forest dieback and the savannization of Amazonia over the next century [Meehl et al., 2007]. While several studies have demonstrated the sensitivity of dynamic global vegetation models to changes in temperature, precipitation, and dry season length [e.g. Galbraith et al., 2010; Good et al., 2011], the ability of these models to accurately reproduce ecosystem dynamics of present-day transitional or low biomass tropical forests has not been demonstrated. A model-data intercomparison was conducted with four state-of-the-art terrestrial ecosystem models to evaluate the ability of these models to accurately represent structure, function, and long-term biomass dynamics over a range of Amazonian ecosystems. Each modeling group conducted a series of simulations for 14 sites including mature forest, transitional forest, savannah, and agricultural/pasture sites. All models were run using standard physical parameters and the same initialization procedure. Model results were compared against forest inventory and dendrometer data in addition to flux tower measurements. While the models compared well against field observations for the mature forest sites, significant differences were observed between predicted and measured ecosystem structure and dynamics for the transitional forest and savannah sites. The length of the dry season and soil sand content were good predictors of model performance. In addition, for the big leaf models, model performance was highest for sites dominated by late successional trees and lowest for sites with predominantly early and mid-successional trees. This study provides insight into tropical forest function and sensitivity to environmental conditions that will aid in predictions of the response of the Amazonian rainforest to future anthropogenically induced changes.