Diagnosing the uncertainty of Earth system models using a traceability framework
A huge uncertainty exists between the predictions of Earth system models in the terrestrial carbon cycle and its feedback to climate change. Therefore, it is critical to understand the underlying reasons for the uncertainty. However, the Earth system models have become increasingly complex, making it difficult to diagnose. In this study, we have developed a traceability framework to trace the reasons step by step. We first simulated the outputs from Coupled Model Intercomparison Project Phase 5 (CMIP5) using a simplified carbon cycle model, and derived the necessary components for the traceability analysis using a data assimilation technique. Then, we decomposed the carbon cycle analytically to trace the reasons for the uncertainty between Earth system models in CMIP5.
The traceability framework decomposed the terrestrial carbon into two parts, carbon in the steady state (which shows the capacity of the terrestrial carbon storage under given climate conditions) and the remaining potential (which is the difference between the capacity and the actual carbon storage). The results showed that the capacity of the terrestrial carbon storage is different dramatically among models. The difference could be attributed to the residence time and net primary production (NPP). Specifically, the residence times of all root, wood, litter and soil were significantly different. The difference of the residence times could further be attributed to the decay rate, transfer coefficients between pools and allocations of NPP.