An equation has been derived that allows the timing of the onset of glaciations to be predicted. This confirms that Earth has just missed entering a new glacial period, and is unlikely to enter one for another 50,000 years. See Letter p.200
A subject of much debate is whether atmospheric levels of carbon dioxide were already significantly altered by emissions associated with human activities before the Industrial Revolution in the eighteenth century. One estimate suggests that the atmospheric concentration of CO2 would have been only 240 parts per million (p.p.m.) in an agriculture-free world, rather than 280 p.p.m., as was measured just before the Industrial Revolution1. On page 200 of this issue, Ganopolski et al.2 report modelling studies confirming that we would now be entering an ice age if the concentration had remained at 240 p.p.m. By contrast, they report that glacial inception — the onset of an ice age — could not have occurred at CO2 concentrations that were typical of the eighteenth century.
The Quaternary period has conventionally been divided into two epochs: the Pleistocene, which lasted from about 2.59 million to 12,000 years ago, and the Holocene, which followed the Pleistocene and continues to the present day. The Pleistocene was a time of great, successive glaciations interspersed with interglacial periods, during which environmental conditions were similar to those occurring today. During the Holocene — the latest interglacial period — humans invented agriculture, and their impact on the environment increased at an exponential rate. One of the signatures of this impact is the rising concentration of CO2 in the atmosphere. But at what point does this impact become sufficiently large to affect climate and glacial inception?
A modelling study3 in 2000 established that pre-industrial levels of CO2 were high enough to guarantee a period of interglacial conditions for at least 50,000 years (Fig. 1). Consistent with Ganopolski and colleagues' findings, this earlier study also predicted that the next glacial inception (which would have led to a glaciation that reached an ice maximum 60,000 years from now), could not now occur owing to the warming effect of anthropogenic emissions. Moreover, probabilistic assessments4,5 of the timing of the next glacial inception have been provided by using simple dynamical systems for climate prediction, calibrated using data about past CO2 levels and ice volumes. All of these studies used different models and assumptions, but they broadly agree on the potential timing for a glacial inception because their forecasts are determined by predictable drops in incoming solar radiation (insolation) in the Northern Hemisphere caused by changes in Earth's orbit.
Ganopolski and co-workers' study is an advance on previous work because it provides a simple equation for predicting when glacial inception will occur. The researchers observed that, in the Earth-system model they used for their study (CLIMBER-2), ice begins to form when insolation in the Northern Hemisphere at the summer solstice falls below a certain value that depends logarithmically on the concentration of atmospheric CO2. They were thus able to work out an equation that describes this behaviour.
To calibrate the equation, the authors performed several simulations that differed by the value of a parameter that controls cloud height in their model. This sampling process effectively generates a family of model versions, which the authors tested to see which ones predicted past glacial inceptions. Past glaciations and interglacials have been identified on the basis of isotopic data from marine sediments, and they follow a numbering scheme in which isotope 'stages' with odd numbers roughly correspond to interglacials. The authors paid special attention to the glacial inceptions after marine isotope stages 19 and 11, and to the period after marine isotope stage 1 (that is, the Holocene), because insolation evolved in a similar way at those times but led to different outcomes (stage 1 did not produce a glacial inception). Only the parameter values that yielded correct simulations of all past glacial inceptions were used to establish the equation.
The authors were thus able to confirm that Earth had a narrow escape from glacial inception during the Holocene: the increase in atmospheric CO2 levels during this period was sufficient to prevent the planet from entering a glacial period. The authors also report that an interglacial climate would have continued for at least 20,000 years, and more plausibly for 50,000 years, if CO2 concentrations had been sustained at levels typical of the eighteenth century. However, almost 500 gigatonnes of carbon (GTC; 1 GTC is equivalent to 3.6 gigatonnes of CO2) have been released into the atmosphere since the Industrial Revolution. Ganopolski et al. show that this means that we will probably skip the next glacial inception too: emissions of 1,000 GTC (a scenario that is quite likely) will almost guarantee 100,000 years without any glaciation.
Such long-term consequences may seem surprising, given that the emissions will occur over a few centuries at most and that anthropogenic CO2 will eventually be absorbed by the oceans. But for this absorption to occur, carbonate minerals in the ocean will need to be dissolved, to counteract the increase in ocean acidity that occurs when CO2 is absorbed, and which limits the amount of CO2 that can be dissolved. This takes time. In fact, the mean half-life of CO2 in the atmosphere is of the order of 35,000 years6. Consequently, anthropogenic CO2 will still be in the atmosphere in 50,000 years' time, and even 100,000 years, which is enough to prevent any glaciation.
The method used by Ganopolski et al. is known as 'perturbed physics' sampling. This means that the different scenarios for future climate were sampled by modifying one parameter, which controls one of the physical effects described by the model. But no model is perfect, and all the possible errors associated with the model cannot be entirely compensated for by adjusting this parameter. To provide better predictions, we need to pay special attention to climate processes that are currently not well quantified.
Among them, the causes of CO2 changes during past interglacial periods and during the early stages of glaciation remain a matter of controversy. For example, we are uncertain about the amplitude and dynamics of carbon sequestered in peatlands1,7. More fundamentally, we do not yet know whether natural CO2 dynamics have an active role in causing glacial inception, or whether they passively amplify the effects of accumulating ice at northern high latitudes. In spite of these uncertainties, Ganopolski and colleagues' main conclusion is likely to stand. It reinforces previous assessments asserting that humanity's collective footprint on Earth already extends beyond any imaginable future of our society.