Using Initial Condition Large Ensemble Experiments to Interpret Observed Trends and 21st Century Projections of Ocean Carbon Uptake

We find data-based products for carbon fluxes over the 1990s and 2000s show a pronounced trend towards stronger outgassing of CO2 over the Tropical Pacific. This is consistent with the dynamics of a persistent La Nina-like state in which enhanced upwelling in the East and Equatorial Pacific brings carbon rich waters to the surface, increasing pCO2 and thus Sea-to-Air carbon fluxes. Results from multi-model and initial-condition ensemble simulations do not capture the observed trends in the Equatorial Pacific outgassing of carbon. We interpret this model-observation disagreement to reflect missing processes in the current generation of Earth System Models, namely, model biases in Equatorial Pacific zonal wind stress trends identified by England et al., (2014) and McGregor et al., (2015).

More generally, we have evaluated variability and trends of the ocean carbon cycle with two initial condition large ensemble (ICLE) experiments run with Earth System Models (ESMs) to interpret historical observations and future projections of ocean carbon uptake. We analyze (1) ICLE’s run with the GFDL-ESM2M and CESM1-BGC using historical to RCP8.5 boundary conditions, (2) four observational data-based products spanning the period 1990-2009 of ocean carbon fluxes from the Surface Ocean pCO2 Mapping intercomparison (SOCOM) initiative and (3) historical to RCP8.5 simulations with the multi-model CMIP5 suite of ESM’s. We find that internal variability contributes significantly to annual but not cumulative global ocean carbon uptake over the 20th and 21st centuries. Internal variability is of first-order importance for detecting trends in Air-to-Sea carbon fluxes over the observational period (1990-2009), where the models show no locally emergent trends. Some regional trends are emergent, with the degree of emergence and regions of emergence being model dependent.