Traceable components of modeled carbon storage capacity in terrestrial ecosystem

Background/Question/Methods: Data-model and model-model intercomparisons have become a popular method to examine differences between data and models or among models.  Most of the intercomparison studies have shown that modeling results differ substantially among models, leading to great uncertainties in model projections. Although many of the intercomparison studies have generally showed that the mean of the model projections fits with data better than any individual models, there is no approach available to explain variations of predictions among models. In this study, we developed a analytical framework for dissecting modeled carbon storage capacity into a few tractable components. These components include climate forcings, environmental scalars, model parameters for carbon processes, carbon residence times, and ecosystem carbon input (i.e., net primary productivity). This framework has been used to analyze how model structure determined the differences in carbon storage capacity among biomes in the Australian Community Atmosphere Biosphere Land Exchange (CABLE) model.

Results/Conclusions: Take evgreen forests as examples, the complementary combination of NPP (0.39 and 1.20 kg C m^-2 yr^-1) and  (86.44 and 26.30 years) resulted in comparable carbon storage capacities between evergreen needle- (34.12 kg C m^-2) and broadleaf (31.53 kg C m^-2) forests. The longer  in evergreen needleleaf forest was ascribed to its longer actual carbon residence time (43.59 years) and smaller environmental scalar (0.14 on litter/soil carbon decay rates) than those in evergreen broadleaf forest (18.24 years and 0.40). The larger environmental scalar in evergreen broadleaf forest was resulted from its warmer and wetter climate than evergreen needleleaf forest. These results illustrate that decomposing current complex models into several traceable components is a efficient way to explain model uncertainties. The framework described here could help us reduce uncertainties in future projections of terrestrial carbon cycle by facilitating several model analyses, including assessments of new biogeochemical modules, model intercomparison, benchmarking analysis and data assimilation.