show that CO2 levels have fluctuated between around 180 parts per million (ppm) during ice ages and 280 ppm during warm periods. Only once did they briefly increase slightly above 300 ppm. At the time of writing, the last estimate of global mean CO2 stood at 415 ppm, with an annual growth rate of almost 3 ppm over the last five years (NOAA 2020b). The last time atmospheric CO2 was about as high as now was around three million years ago, during the Middle Pliocene (Lunt et al. 2008). Even that was only a temporary excursion, and we have to go back a staggering 25 million years to find values exceeding 500 ppm (Pagani et al. 2005), a value that at current trends we will reach in about 30 years.
When a kettle is switched on, the water in it does not heat up instantaneously, but there is a delay. The same happens with the earth’s climate: according to the International Panel on Climate Change (IPCC 2019a), 90 per cent of the heat from the enhanced greenhouse effect is absorbed by the oceans’ water. This means that the warming we have seen so far lags behind the steep rise in atmospheric carbon dioxide levels, and that even if we suddenly managed to stop the rise, warming will continue (Huntingford, Williamson and Nijsse 2020). Earth’s surface is now over 1°C warmer than during the late nineteenth century (NOAA 2020a), which is comparable to or slightly warmer than during the last geological warm period between 10,000 and 5,000 years ago (Marcott et al. 2013). But during that time the warming was caused by changes in the way the earth circles and wobbles around the sun, with the result that tropical lands were slightly cooler than today. Nowadays, the (over)heating effect is seen universally across the globe, and as such differs decidedly from any climate fluctuations since the end of the last ice age more than 10,000 years ago (Barbuzano 2019; Neukom et al. 2019).
Currently the enhanced greenhouse effect is created only to about two-thirds by CO2 and one third by other gases, of which about half by methane.1 Countering this warming to a certain extent are several other human-caused effects, notably from atmospheric pollution through aerosols, which cause some cooling (Myhre et al. 2013). These cooling effects probably approximately cancel out the warming effect of the non-CO2 greenhouse gases. CO2, however, is by far the most important greenhouse gas because its lifetime vastly exceeds almost all of the other greenhouse gases or aerosols. A question of substantial theoretical value is what would happen if we not only prevented a further rise of CO2 but stopped all emissions of CO2 and other greenhouse gases tomorrow. Because the oceans and land tend to take up more than half of human-made CO2 emissions (Global Carbon Project 2020), such an instantaneous net-zero balance of man-made carbon fluxes would lead to some drawdown of CO2 by natural sinks, and a lowering of atmospheric CO2 levels. The extent of the drawdown, however, is by no means understood and could easily be overestimated. Forests are responsible for a large part of that sink (Global Carbon Project 2020), but are increasingly being fragmented and damaged (Grantham et al. 2020). Some recent research also suggests that we already observe a declining efficiency of the sinks (Wang et al. 2020). Furthermore, even if such lowering occurs, it may not lead to a cessation of heating: because of the possible acceleration of overheating once the ‘global dimming effect’ (from aerosols) is reduced (Xu, Ramanathan and Victor 2018); and because of vicious climate feedbacks that may already be underway (Lenton et al. 2019).
How much more overheat will result if CO2 emissions alone stopped depends on both how rapidly the gas is removed from the atmosphere, and on the long-term climate response to the remaining CO2 level. Model results (Matthews and Zickfeld 2012) indicate that in such a scenario, two-thirds of the human-caused excess CO2 – i.e. above the pre-industrial level of 280 ppm – would still remain in the atmosphere after 190 years. For the case of stopping emissions in 2020 at a level of 415 ppm, this translates to more than 370 ppm by the year 2200. The earth during that time would continue to warm, but probably only by a few tenths of a degree. If all other greenhouse gas and aerosol emissions also stopped, the result might well be similar. However, the assumption that it would be excludes at least two further possibilities – possible stronger than expected carbon cycle feedback that we cannot reliably quantify, leading to higher than expected CO2 levels (Lenton et al. 2019); and the possibility of a higher long-term sensitivity of the earth’s temperature to CO2 (Bjordal et al. 2020).
Because of the limitations of models, a more prudent approach is to derive climate sensitivity from past climates. Most commonly, this is based on the temperature and CO2 changes during ice-age/warm-period fluctuations (Hansen et al. 2013). Results based on this approach generally support the model results (Sherwood et al. 2020). The problem, however, is that the earth is already in a different state from any time during those glacial cycles. Climate sensitivity could be higher in a warmer state due to positive climate system feedbacks, or tipping mechanisms, not yet quantified, which is the basis of the deeply alarming ‘Hothouse Earth’ hypothesis (Steffen et al. 2018). Support for this hypothesis comes from estimates for the Pliocene warm period, when CO2 was between 365 and 415 ppm and temperatures about 3°C warmer than during the pre-industrial era (Pagani et al. 2010; Sherwood et al. 2020). According to those data – from the last time earth was in a similar climate state to now – an immediate stop of CO2 emissions would still lead to substantial warming after today: about another 1°C for the higher end of the CO2 estimate for the Pliocene, and more than 2°C at the lower end.2
The mechanisms that may have led to such a high climate sensitivity are unknown, but there is some evidence that Arctic sea-ice feedback could have contributed. It is possible that even if we stopped emitting CO2 now, we could still experience an ice-free Arctic in the near future that could lock in significant warming for decades to come because of additional energy absorbed by the ice-free ocean in the long Arctic summer days. In the latest round of climate model simulations, those models that correctly simulate past sea-ice loss tend to have a higher climate sensitivity than usually assumed. Remarkably, even models driven by an extremely low-emissions scenario, approaching a stopnow scenario, still show an ice-free Arctic before 2050 (SIMIP Community 2020). The principal mechanism here is that even at declining CO2 concentrations, excess heat stored in the oceans will only decline very slowly (Solomon et al. 2010).
It is important to stress that the scenario just discussed is largely speculative and only serves to illustrate how far we have already proceeded on a route to irreversibly altering our planet’s climate state. Computer simulations of possible future climate states using certain scenarios of greenhouse gas emissions can be used to gain a general impression of how this trend might continue – as there is still no evidence of a lowered CO2 level due to climate policy (Knorr 2019; Le Quéré et al. 2020).
A high-profile publication by a group of US scientists (Burke et al. 2018) confirms that we are indeed in the process of driving our climate system well into uncharted territory. Different to the approach followed by the IPCC (Hoegh-Guldberg et al. 2018), the group did not try to assess the impacts of projected changes directly by assessing impacts of past changes or using computer models. Instead, they compared expected climate warming patterns derived from model simulations with what we know from the geological past. They concluded that, at even ‘moderate’ degrees of warming, the climate in large parts of the planet will not resemble anything seen anywhere on earth since at least the onset of agricultural civilization. Instead, the combination of extreme heat and humidity due to be encountered in large parts of the world will have their closest analogue in deep time. In the case of a rapid and unprecedented decarbonization of the world economy, climate is expected to eventually stabilize at a state most closely resembling the already-discussed Mid-Pliocene warm period, some 3–5 million years ago. In a much more likely higher-emissions scenario, however, large parts of the earth will revert to a climate state last seen ‘just’ – in geological terms – after the demise of the dinosaurs: the early Eocene, some 50 million years ago.
This scary scenario is not all, because it only considers the start and end point of warming, but not the path on which we get there. If, within a few generations, we turn back the earth’s geological CO2 levels by tens of millions of years, then the rapidity of this change must surely have an impact on the way climate heating will play out. Unfortunately, this rate exceeds anything we know of from the deep geological past (Zeebe,