Millennium time-scale experiments on climate-carbon cycle with doubled CO2 concentration

Earth system models (ESMs) are commonly used for simulating the climate–carbon (C) cycle and for projecting future global warming. While ESMs are most often applied to century-long climate simulations, millennium-long simulations, which have been conducted by other types of models but not by ESM because of the computational cost, can provide basic fundamental properties of climate–C cycle models and will be required for estimating the carbon dioxide (CO2) concentration and subsequent climate stabilization in the future. This study used two ESMs (the Model for Interdisciplinary Research on Climate, the Earth system model version (MIROC-ESM) and the MIROC Earth system version 2 for long-term simulation (MIROC-ES2L)) to investigate millennium-scale climate and C cycle adjustment to external forcing. The CO2 concentration was doubled abruptly at the beginning of the model simulations and kept at that level for the next 1000 or 2000 years; these model simulations were compared with transient simulations where the CO2 was increased at the rate of 1% year−1 for up to 140 years (1pctCO2). Model simulations to separate and evaluate the C cycle feedbacks were also performed. Unlike the 1pctCO2 experiment, the change in temperature–cumulative anthropogenic C emission (∆T–CE) relationship was non-linear over the millennium time-scales; there were differences in this nonlinearity between the two ESMs. The differences in ∆T–CE among existing models suggest large uncertainty in the ∆T and CE in the millennium-long climate-C simulations. Ocean C and heat transport were found to be disconnected over millennium time-scales, leading to longer time-scale of ocean C accumulation than heat uptake. Although the experimental design used here was highly idealized, this long-lasting C uptake by the ocean should be considered as part of the stabilization of CO2 concentration and global warming. Future studies should perform millennium time-scale simulations using a hierarchy of models to clarify climate-C cycle processes and to understand the long-term response of the Earth system to anthropogenic perturbations.