Equifinality in parameterization of process-based biogeochemistry models: A significant uncertainty source to the estimation of regional carbon dynamics

Numerical biogeochemistry models suffer from equifinality problem in their parameterizations using eddy flux tower data, which can contribute to diverged estimates of regional carbon dynamics. To date, the uncertainty in regional estimates propagated from the site-level parameterization equifinality has not been well characterized. Here, we use a process-based biogeochemistry model, the Terrestrial Ecosystem Model (TEM), and a Bayesian inference framework to quantify the influence of parameterization equifinality on the estimates of carbon dynamics in boreal forest ecosystems during the 20th century. By conducting three groups of ensemble regional simulations, we find that, given a certain climate data set being used, (1) in comparison to the effects of random noises in climate forcing, the regional uncertainty due to parameterization equifinality is remarkably greater, (2) the parameterization equifinality results in drastically different decadal variations in the estimation of carbon storage during the 20th century, and (3) the uncertainties associated with parameterization equifinality and random noises in climate forcing vary both spatially and seasonally. We conclude that the equifinality from site-level parameterizations in biogeochemistry models is an important uncertainty source in estimating regional carbon dynamics. Simply extrapolating the site-level parameterization to large spatial and temporal scales could bias the regional estimates irrespective of regional climate data sets used in our analysis. Ensemble process-based biogeochemistry model simulations conditioned on observed ecosystem fluxes with Bayesian inference techniques could provide more serious estimates of regional carbon dynamics and their associated uncertainties.