Simulations of the flux of atmospheric carbon dioxide into the ocean show that changes in flux associated with human activities are currently masked by natural climate variations, but will be evident in the near future. See Letter p.469
The world ocean has absorbed about one-third of the carbon released by humans1, and therefore has a key role in moderating climate change. Observations2 of the ocean interior confirm that increases in carbon dioxide emissions from fossil-fuel burning are accompanied by an increase in carbon content in the upper ocean. But, surprising as it may seem, McKinley et al.3 report on page 469 that, in many ocean regions, changes in the uptake of CO2 induced by human activities are currently indistinguishable from changes driven by natural climate variations. So, are anthropogenic trends in the ocean carbon sink concealed by Earth's own variability?
As atmospheric CO2 levels increase, the ocean takes up this gas at a rate proportional to the difference of the partial pressure of CO2 (a measure of CO2 concentration in a mixture of gases) between the air and sea2. The strength of the ocean carbon sink is determined by chemical reactions in seawater, biological processes such as photosynthesis and respiration, and physical processes, including ocean circulation and vertical mixing4. But even though these key mechanisms are known, there are considerable uncertainties regarding their year-to-year (interannual) and decadal variations5. These variations are tightly linked to modes of internal variability in the climate system — such as the El Niño–Southern Oscillation (ENSO) — that have regional to worldwide effects on weather and climate, and thereby modulate air–sea CO2 fluxes and ocean biogeochemical cycles.
Advances in observations and models have shed light on how internal climate variability controls the ocean carbon sink. Modern Earth-system models (ESMs) that were analysed in the fifth Coupled Model Intercomparison Project (CMIP5, which compared the output of models in which components of Earth's system are coupled) consistently predict that ocean carbon uptake increases in line with rises in fossil-fuel carbon emissions. But these ESMs poorly capture natural climate variability. Furthermore, CMIP5 projections generate a large spread in estimates of the ocean carbon sink because the models have different numerical schemes, process descriptions and spatial resolutions6.
Further complications arise because there is a large spread in observed variations of CO2 fluxes, due to both the use of different mapping procedures and gaps in the observational record7. This spread of variation from observational data is at least as large as the spread from the CMIP5 ensemble of models (Fig. 1). Estimates of the internal variability of the ocean carbon sink therefore remain unconstrained, impeding the detection and attribution of changes in air–sea CO2 fluxes.
McKinley and co-workers report a crucial step towards detecting changes in the ocean carbon sink — they have generated a large ensemble of simulations based on an ESM. The authors repeated their model runs 32 times over the same time interval, spanning the years 1920 to 2100. The model follows historical climate evolution until the end of 2005, and then a climate-change scenario (known as RCP8.5) that projects high levels of atmospheric CO2. Few modelling centres are able to perform so many runs because doing so is computationally expensive, but the value of using ensembles with a high number of simulations is increasingly being recognized.
External forcing factors such as the concentrations of atmospheric greenhouse gases and aerosols, volcanic eruptions and solar variability were identical across all simulations. The only difference between the ensemble members was their climate state at initialization: each simulation started with a different state generated by a small perturbation of the air temperature. As a result, individual ensemble members were not identical to each other, even though the model variables for each member followed the same general trajectory.
The authors considered two outcomes from the simulations: the forced trend, which is the average trend in model variables across all ensemble members produced under specified external forcing, and the internal trend, which is the difference between each member's trend and the forced trend, caused by the model's internal variability. They observed that the forced trends in ocean carbon uptake are indistinguishable from internal model variability in vast ocean regions between 1990 and 1999. When the period between 1990 and 2019 is considered, the forced trends become statistically significant in many more areas than during 1990 to 1999, and emerge almost everywhere in the ocean. This suggests that the predicted increase in oceanic carbon uptake is attributable to anthropogenic forcing. These trends intensify as atmospheric CO2 levels increase, and so become detectable ocean-wide when the period from 1990 to 2089 is considered.
The researchers confirm that the spatial pattern in which trends emerge seems to be closely linked to the internal variability of the climate system. For instance, the largest internal variations are in the equatorial Pacific Ocean, a region known to be affected by ENSO. In regions such as this that exhibit strong seasonal and interannual variability, it is difficult to detect anthropogenic changes in CO2 uptake. By contrast, anthropogenic trends emerge early in the subpolar North Atlantic and equatorial Atlantic oceans and in some regions of the Southern Ocean. A similar pattern has been reported8 for several ocean biogeochemical parameters modelled in the CMIP5 ensemble.
McKinley and colleagues conclude that forced trends should be detectable in observations once they have emerged and become statistically significant. However, although internal model variability can give an indication of the chaotic behaviour of some natural processes, it is not equivalent to Earth's natural climate variability. Furthermore, predictions of the time of emergence are model-dependent8,9, and may change if a different model or ensemble size is considered (Fig. 1).
Nevertheless, the current study makes a valuable contribution to the quantification of internal variability in the rates of change of the ocean carbon sink. Changes driven by human activities are undoubtedly there, but may be concealed by natural variations in many ocean regions because of the slow timescales on which ocean processes occur. Future work based on coordinated observational frameworks and large ensemble simulations using ESMs should enable the natural variability reported in this study to be verified.Footnote 1
Le Quéré, C. et al. Earth Syst. Sci. Data 7, 349–396 (2015).
IPCC. Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker, T. F. et al.) (Cambridge Univ. Press, 2013).
McKinley, G. A. et al. Nature 530, 469–472 (2016).
Heinze, C. et al. Earth Syst. Dynam. 6, 327–358 (2015).
Marotzke, J. & Forster, P. M. Nature 517, 565–570 (2015).
Bopp, L. et al. Biogeosciences 10, 6225–6245 (2013).
Rödenbeck, C. et al. Biogeosciences 12, 7251–7278 (2015).
Keller, K. M., Joos, F. & Raible, C. C. Biogeosciences 11, 3647–3659 (2014).
Dobrynin, M., Murawski, J., Baehr, J. & Ilyina, T. J. Clim. 28, 1578–1591 (2015).
Related links in Nature Research
About this article
Spatial and Temporal Variations of Particulate Organic Carbon Sinking Flux in Global Ocean from 2003 to 2018
Remote Sensing (2019)
Seasonal Asymmetry in the Evolution of Surface OceanpCO2and pH Thermodynamic Drivers and the Influence on Sea-Air CO2Flux
Global Biogeochemical Cycles (2018)
Current and Future Decadal Trends in the Oceanic Carbon Uptake Are Dominated by Internal Variability
Geophysical Research Letters (2018)