On land, the most important regions for sucking up carbon from the atmosphere are the tropical rainforests of Amazonia, the Congo basin and Southeastern Asia, and the boreal forests and Arctic tundra.2 Collectively known as the ‘green lungs’ of the planet, these regions have vast quantities of carbon locked up in vegetation and soil.3 Sizeable fractions of boreal forest and tundra regions have an added store of carbon in their underlying permafrost layer. In a warming climate, thawing of permafrost could thus release large amounts of carbon as CO2 or, in swamps and bogs, as CH4, which would further amplify climate change4.
In the oceans, two important carbon ‘hot-spots’ exist in the North Atlantic Ocean and the Southern Ocean around Antarctica. Here, excess carbon moves from the surface into deep waters where it is stored over timescales of centuries to millennia. Changes in the oceanic circulation in these areas, which might happen as temperatures rise, could decrease the oceans’ capacity to store carbon. Initial studies suggest that this is already happening in the Southern Ocean, which calls into question whether, in the future, carbon sinks will continue to operate or will saturate and perhaps even become carbon sources5.
The challenges for research are clear, starting with understanding the carbon cycle as an integral component of the global climate system. For example, further studies are needed to elucidate the key processes that transform carbon in terrestrial and marine ecosystems, and to understand how the carbon cycle is coupled both to the cycles of nutrients, such as nitrogen and phosphorus, and to the hydrological cycle.
Also warranting further attention is the multitude of climate system–carbon cycle feedbacks that operate on timescales ranging from days to to geological epochs. This cannot be achieved without improving our modelling tools. Here, the international research community has a long-term commitment to the improvement of Earth-system models — in other words, global climate models with a closed carbon cycle3,6.
Complex models must be constrained with real-world observations. Long-term observations of key carbon hot-spots are thus imperative. One operational example is the Zotino Tall Tower Observatory facility in the Siberian taiga forest, which includes a 300-m-tall mast for measuring regional atmospheric greenhouse gases, reactive chemistry, aerosols and meteorology.7 A similar observatory will be established in the Amazon forest in the short term. These ground-based measurements should be complemented with repeated air- and space-borne remote-sensing systems.