Forrest M. Hoffman (University of California-Irvine and Oak Ridge National Laboratory, USA) and James T. Randerson (University of California-Irvine, USA)
Rapidly increasing atmospheric carbon dioxide (CO2) concentrations are altering the Earth’s climate, and anthropogenic perturbation of the global carbon cycle is expected to induce feedbacks on both future CO2 concentrations and the climate system. Such feedbacks are highly uncertain and potentially large. The objectives of this project are to quantify climate-carbon cycle feedback responses in global models contributing to the Coupled Model Intercomparison Project Phase 5 (CMIP5) for the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5), and to reduce the range of uncertainty in climate predictions by constraining model representations of feedbacks through comparisons with contemporary observations. Friedlingstein et al. (2006) developed a feedback analysis methodology for separating the biosphere model responses to climate change, or rising temperature, from the effects of rising atmospheric CO2 concentrations. Their analysis of the terrestrial feedbacks from 11 global models in C4MIP demonstrated that the model climate sensitivities varied by a factor of nine while the model concentration sensitivities varied by a factor 14. However, this model intercomparison included no comparisons with observational data, the next crucial step required for reducing feedback uncertainties in Earth System Models (ESMs). To reduce uncertainties using contemporary observations, relationships must be found between contemporary variability and future trends within multiple models, and it must be possible to constrain contemporary variability using observations. For this analysis, a new conceptual framework for evaluating climate-carbon cycle feedbacks in the multi-model CMIP5 archive using contemporary observations will be presented. Biogeochemical feedbacks in both concentration- and emissions-forced historical and RCP simulations will be quantified and compared. Such model-data comparisons are being contributed to the International Land Model Benchmarking (ILAMB) Project (http://www.ilamb.org/), an effort to develop internationally accepted benchmarks for land model performance and to design a new, open source, benchmarking software system for use by the larger modeling community.