Skip to main content

CHanges Of CO2 Levels during pAst and fuTure intErglacials

Periodic Reporting for period 1 - CHOCOLATE (CHanges Of CO2 Levels during pAst and fuTure intErglacials)

Reporting period: 2015-10-15 to 2017-10-14

The concentration of carbon dioxide (CO2) in the atmosphere depends on carbon cycle processes, i.e. sources and sinks of carbon. The future evolution of the carbon sinks is not well known, which inhibits robust quantification of future atmospheric CO2 concentration and the resulting climate change. Understanding warm past periods is essential to constrain climate models and accurately predict future changes. During the last million years, warmer periods, called interglacials, happened every ~100,000 years. CO2 levels measured in interglacials before the mid-Bruhnes event (MBE), a large climate shift taking place ~430,000 years ago, are lower than the CO2 in interglacials after the MBE. The cause for this drastic evolution is still unexplained, resulting in uncertainty in the carbon cycle response to global warming, whose quantification is still subject to large unknowns.

The aim of this project is therefore to reconcile paleoclimate proxy data and model simulations and evaluate if new key processes are necessary to reproduce past climate variations and therefore coming ones. A specific focus will be dedicated to carbon cycle.

It has been suggested that a major mechanism that could explain MBE was a slower circulation during interglacials before the MBE, resulting in more ocean carbon storage and lower atmospheric CO2. Nevertheless, this hypothesis was based on a model that do not include a carbon cycle. We plan to evaluate this hypothesis by using a model including a carbon cycle.
Furthermore, we also propose an alternative hypothesis that is that sea-level changes may have played a considerable role by altering carbon sinks from land vegetation and shallowing ocean carbonate sedimentation, notably by coral reefs.

The objective of the project is to include these mechanisms in a state-of-the-art climate model applicable to long timescales, and compare its modified behavior with paleoclimate data and more complex models used for projections. This will provide a step change in our understanding of the impact of ocean circulation and sea-level changes on the carbon cycle.

It will benefit the European and international scientific community by shedding new light on these processes, and by setting the basis to include these new mechanisms in climate models used for projections, pivotal for adaptation policy to climate change.
The iLOVECLIM model has been set up in Bordeaux supercomputer, and available to a wider community of climate researcher there (now used by Thibaut Caley and Anne-Sophie Kremer for their own research). We have therefore disseminated this tool to a new community.

A set of 9 simulations one for each interglacials, with different reservoirs of the carbon cycle acitvated have been performed. One set with only the ocean, the other with ocean and vegetation and the last one including also the ice sheet variations. The total number of simulations is therefore of 27. Each simulation lasts for 5,000 year. None of the three sets compare correctly with atmospheric CO2 concentration from the observation, indicating that a key process is missing in the modelled carbon cycle, therefore invalidating the former hypothesis about ocean circulation changes.

Comparison with a few available observations in the ocean (temperature, δ13C as a tracer of circulation) and land (pollen data) have been performed through compilation of existing data. This comparison has confirmed that the changes in ocean circulation are unlikely to explain the Mid-Brunhes transition. Another hypothesis is now tested concerning the role played by coral reef in the storage of CO2. A coral reef module is under development in fortran and will be incorporated in the iLOVECLIM model.
We are also comparing the iLOVECLIM simulations with others from a more comprehensive climate models named IPSL-CM5A-LR. First results indicate that changes in circulation are likely largely in this model. Nevertheless, as stated before, the compilation of δ13C data indicate that changes in circulation are unlikely to explain observed atmospheric CO2 concentration, since no coherent pattern of changes emerge between after and before MBE.

We planned to produce publications of at least four scientific papers at the end of the two years. At the end of year 1, one publication is now submitted and two other in preparation: one concerning the δ13C new database and comparison with iLOVECLIM simulations, and the other one concerning the development of the coral module within iLOVECLIM.

A lot of climate variables from simulations performed with iLOVECLIM and IPSL-CM5A-LR have been stored and are available upon request notably through a thredds portal.

The new version of the iLOVECLIM model including a coral module and a parameterisation of the effect of sea level change sea-continent mask should be released soon.

The fellow has presented her results in a large number of international (2), national (3) and broad audience (>2) conferences.
The project has now clearly established that a former hypothesis, recently published in Nature, stating that ocean circulation changes as simulated by iLOVECLIM could explain the Mid-Bruhnes transition in terms of atmospheric CO2 concentration variations during interglacial (warm) periods is not validated. Another process need to be accounted for in the models, as it was intuited within the proposal of this Marie Curie fellowship. Another hypothesis based on the role played by coral reef in terms of carbon storage is now explored and could have serious consequences in terms of coming climate change. Thus, this project, by using modelling tool applied to past climate to evaluate key processes that may be missing to climate models, has shown here the potentially key role of coral reef in the carbon cycle response to a large warming.
Summary of the chocolate project.