Trends In Secular Mean State And Variance Changes In Ocean Carbon Uptake In RCP8.5 Large Ensemble Experiments With Two Earth System Models
Abstract:It is well established that the ocean carbon cycle is changing. However, to date is has proven difficult to deconvolve changes associated with secular trends and natural variability. Ensemble simulations with Earth System Models provide a powerful tool for evaluating the secular trend (ensemble mean) and the intrinsic natural variability (ensemble spread). Here we consider the large initial-condition ensemble experiments of two Earth System Models, GFDL-ESM2M (Geophysical Fluid Dynamics Laboratory Earth System Model, 30 member) and CESM1-BGC (Community Earth System Model with Biogeochemistry, 24 member), run with historical and RCP8.5 boundary conditions. Our goal is to not only characterize the anthropogenic trend for the ensemble mean for each model, but also to characterize the changes in variance between the two models over the 21st century. Our analysis reveals that the GFDL and CESM have divergent trends in 21st century ocean carbon uptake variability, with the GFDL-ESM2M and CESM-BGC1 having decreased and increased variance, respectively. Preliminary work towards attribution indicates that these differences are driven by a tendency towards decreased ENSO variance for the case of GFDL-ESM2M and increased ENSO variance for the case of CESM1-BGC.
Additionally, we have used the ocean carbon cycle variance from the two models to estimate the degree to which internal variability may account for multi-model ensemble spread between the CMIP5 suite of Earth System Models. Preliminary results indicate that as much as 25% of the CMIP5 multi-model ensemble spread in carbon uptake may reflect internal variability in the climate system.