Variability in Ecosystem Carbon Predictions in the CCSM4.0 Ensemble

Ecosystems have the capability to moderate climate change over the twenty-first century by removing carbon dioxide from the atmosphere and storing it in vegetation and soils. The responses of terrestrial carbon pools and fluxes are dependent on environmental variables in the physical climate, but model simulations of the physical climate can be highly uncertain due to unforced variability in the climate system. In this study, we evaluate the magnitude and time dependence of variability in terrestrial carbon pools and fluxes in North America using a 6-member ensemble of Community Climate System Model (CCSM) version 4.0 simulations for Representative Concentration Pathway 8.5 (RCP 8.5). Results indicate large ensemble variability in simulated carbon fluxes, and variability in instantaneous carbon fluxes is larger than in integrated carbon pools. The change in carbon pools in response to climate change ('climate signal') is smaller than the unforced variability 10 years into the future, however, the climate signal is larger than the unforced variability throughout North America by 50 years into the future. These results demonstrate that unforced variability needs to be considered when monitoring and predicting decadal-scale ecosystem responses to climate.