DOE Ecological Research Facility Tests Ecosystem Models

Summary

A paper to be published in the August issue of Ecological Monographs, authored by nine DOE/Office of Science researchers and 10 of their collaborators from a total of 13 institutions, presents a comprehensive test of the ability of 13 ecosystem models to simulate exchanges of carbon and water between a forest and the atmosphere. The model testing used eight years (1993-2000) of data from DOE’s Throughfall Displacement Experiment at the Oak Ridge National Laboratory (an Office of Science ecosystem research facility) and other data collected on the DOE Oak Ridge National Environmental Research Park; this represents the longest experimental dataset used to test the largest number of ecosystem models ever. Models that included more detailed energy balance and carbon metabolism calculations provided consistently better predictions, indicating that the level of detail in the models is important. Most of the models were able to predict carbon and water exchanges relatively well when growing conditions were favorable, but many models failed during periods of drought, which occurred during several years of the study period. The loss of model accuracy with drought indicates that considerable uncertainty may exist with respect to present predictions of ecological effects of climatic change on forest ecosystems. Although a single model was not the best predictor of all important ecological variables, the mean of all model outputs was a robust predictor of the observations, even under drought. This result indicates that multiple models, rather than a single “best” model, may be needed to reliably predict effects of environmental changes on ecosystem carbon and water balances.