Ensemble Ecosystem Model Experiment and Intercomparison using the Terrestrial Observation and Prediction System (TOPS)

TOPS is a common modeling framework that seamlessly integrates observational datasets, meteorological records and numerical models to simulate key ecosystem processes. In support of North American Carbon Program (NACP), we are using TOPS to conduct a VEMAP-like modeling exercise, focusing on characterizing sources of uncertainty in carbon flux estimates from forward-modeling approaches. The current experiment uses public-domain versions of Biome-BGC, CASA, LPJ, SIMCYCLE, and BEAMS, driven by a consistent set of inputs for the period 1982-2006. We developed a novel approach to analyze model outputs by generating complete distributions of simulated fluxes (e.g., GPP) in the climatological temperature-precipitation domain (that is, we bin simulation results of grid cells according to their "climate coordinates" instead of their geographic locations). These 2-D distributions concisely characterize the dynamics of different ecosystem models and thus facilitate comprehensive model intercomparisons as well as model-observation comparisons. Applying this method to analyze model-simulated carbon fluxes (GPP, NPP, and NEE) and flux- net measurements indicates general agreement but also remarkable differences, arising from the diverse approaches to characterize and parameterize ecosystem processes. These results serve to highlight the problem inherent in relying on only one modeling approach to map surface carbon fluxes. They also emphasize the pressing necessity of expanded and enhanced monitoring systems to narrow critical structural and parametrical uncertainties among ecosystem models.