An assessment of land-based climate and carbon reversibility in the Australian Community Climate and Earth System Simulator

Future levels of climate change depend not only on carbon emissions but also on carbon uptake by the land and the ocean. Here we are using the Earth system model (ESM1) version of the Australian Community Climate and Earth System Simulator (ACCESS) to explore the potential and impact of removing carbon dioxide (CO₂) from the atmosphere through the climate and carbon cycle reversibility experiment. This experiment builds on the standard Coupled Model Intercomparison Project (CMIP) experiment, increasing CO₂ at 1% per year until 4xCO₂ is reached. The atmospheric CO₂ levels are then decreased at the same rate which brings the CO₂ back to pre-industrial levels. We then continue to run the model with constant CO₂ for another 350 years. Our analysis focuses on the response of the land carbon cycle. We find that carbon stores are largely reversible at the global scale over the timescale of changing CO₂. However, carbon stores continue to decrease after CO₂ returns to its initial value, and the land loses another 40 Pg of carbon (PgC) with the largest change in the tropics. It takes about 300 years beyond the period of changing CO₂ for the carbon stores to recover. Interestingly, we saw strong regional variations in the strength of the land response to changing CO₂. Australia showed the largest increase/decrease in biomass carbon (about 40%) and the largest variability in productivity, which was strongly correlated with rainfall. This highlights the importance of assessing the regional response to understanding the processes underlying the response and the sensitivity of these processes within each model. This understanding will benefit future multi-model analyses of this reversibility experiment. It also illustrates more generally the potential to use Earth system model experiments as part of the evaluation of proposed applications of carbon dioxide removal (CDR) technologies. As such, we recommend that these types of modelling experiments be included when mitigation policies are developed.