Large spread in interannual variance of atmospheric CO2 concentration across CMIP6 Earth System Models

Abstract

Abstract Numerical Earth System Models (ESMs) are our best tool to predict the evolution of atmospheric CO2 concentration and its effect on Global temperature. However, large uncertainties exist among ESMs in the variance of the year-to-year changes of atmospheric CO2 concentration. This prevents us from precisely understanding its past evolution and from accurately estimating its future evolution. Here we analyze various ESMs simulations from the 6th Coupled Model Intercomparison Projects (CMIP6) to understand the origins of the inter-model uncertainty in the interannual variability of the atmospheric CO2 concentration. Considering the observed period 1986-2013, we show that most of this uncertainty is coming from the simulation of the land CO2 flux internal variability. Although models agree that those variations are driven by El Niño Southern Oscillation (ENSO), similar ENSO-related surface temperature and precipitation teleconnections across models drive different land CO2 fluxes, pointing to the land vegetation models as the dominant source of the inter-model uncertainty.