COMMENTARY

Constraining future greenhouse gas emissions by a cumulative target

Matthew H. England1, Alexander Sen Gupta, and Andrew J. Pitman Climate Change Research Centre, University of New South Wales, Sydney, New South Wales 2052, Australia

y 1994 all major industrialized Emissions (PgC yr-1) acted as a tremendous buffer of climate

nations, including the United 16 6 change to date; these systems have ab- 1 States, had ratified the United 14 sorbed more than half of our industrial emissions (6). Yet the ability of these Nations Framework Convention 12 Bon Climate Change (UNFCCC), yet 15 2 stores to sequester is changing years later policymakers still debate how 10 over time. Fifty years ago natural car- best to formulate emissions legislation. 8 bon sinks removed Ϸ600 kg of every ton Article 2 of the UNFCCC calls for ‘‘sta- 6 of CO2 emitted to the atmosphere. To- bilization of greenhouse gas concentra- 4 day, these sinks remove only Ϸ550 kg 4 5 per ton emitted (6), and this amount is tions...atalevel that would prevent 2 3

dangerous anthropogenic interference 950 PgC expected to continue to fall (3, 4). with the climate system.’’ Emissions tar- 1950 2000 2050 2100 2150 There are also risks of abrupt change in Cummulative Emissions gets are commonly quoted as a percent- 6 1785 PgC the terrestrial ; for example, relative to 2000 (PgC) in 2200 5 700 PgC age reduction relative to a baseline year. 590 PgC climate change could trigger Amazon 600 1 3 66% probability forest die-back (7), and permafrost melt A different framework for emissions 2 of not exceeding 2o warming

targets is presented in a recent issue of 400 590 PgC could expose northern peatlands to PNAS (1), wherein the targets are set as large releases of carbon (8), both result- 200 170 PgC ing in strong positive feedbacks. In a cumulative emissions inventory, spell- 4 74 PgC 90% probability emitted in of not exceeding short, the carbon sinks that have served ing out to policymakers the net emis- 2o warming 0 last 8 years 200 PgC sions allowable to avoid the worst im- 170 PgC us so well are by no means stable: they 440 PgC pacts of climate change. Cummulative are changing and, unfortunately, chang- -200 emissions since Setting emissions targets around a net 1750 ing in the wrong sense. Add to this land cumulative quota is a familiar paradigm 1950 2000 2050 2100 2150 2200 clearing and the associated release of -220 PgC for policymakers. It is analogous to CO2 and loss of natural carbon storage planning for expenditure against a net Fig. 1. Historical and future scenarios of global (9), and the combined land–ocean sink carbon emissions (Upper) and their cumulative is showing diminishing returns in time. income or setting a catch quota to main- ϭ ϫ 15 growth (Lower) in PgC (1 PgC 1 10 gof Carbon feedbacks can thus greatly tain a sustainable fishery. In such cases, carbon). Pathways 1–3 show three scenarios where the available resource is fundamentally total post-2000 emissions accumulate to 590 PgC, reduce the allowable emissions to stabi- limited in a cumulative sense; harvest or giving a 66% chance that long-term warming will lize at some policy-prescribed target. spend too much and things become un- not exceed 2 °C (1). Curve 4 yields 170 PgC post- Yet these feedbacks are generally not sustainable. For the global harvesting of 2000 emissions and a 90% probability of avoiding incorporated into future climate projec- fossil fuels the message becomes clear: 2 °C warming. Curve 5 shows an example of rapid tions. By doing an inverse calculation burn beyond a cumulative cap and you decarbonization (25% reduction by 2020 and 80% within their model, Zickfeld et al. (1) reduction by 2100, relative to a 2007 baseline), but are able to progressively track an emis- commit the planet to a high risk of dan- with weaker mitigation into the future, resulting gerous anthropogenic climate change. sions pathway that leads to a given sta- in ongoing increasing cumulative emissions be- bilization target for global warming, The article by Zickfeld et al. (1) uses yond 2100. Error bars at right indicate the range in a coupled climate model to carefully post-2000 emissions for the 66% and 90% proba- while also incorporating carbon feed- diagnose, via inverse methods, the level bility cases of 2 °C warming, when a suite of climate backs. This approach differs from the of emissions allowable to track toward a sensitivities and carbon feedbacks are taken into conventional methodology in that they given stabilization target for global account. Scenario 6 shows business as usual for (i) use an inverse method working back warming. Normally, the problem is ad- several decades to come, with emissions peaking in from a given temperature target to 2050 before gradually declining over the ensuing quantify the allowable CO2 emissions; dressed in reverse: namely, for a given century (cumulative post-2000 emissions eventu- (ii) explicitly link CO2 emissions with future emission pathway (2), how will ally stabilize at Ϸ1,890 PgC). This net emission the climate system respond? Both ap- Ϸ CO2 concentrations, taking into account yields a 50–50 chance of exceeding 4 °C warming the coupled ; (iii) quantify proaches are valid, yet to make mean- using the median climate sensitivity and carbon ingful projections they each need to feedback range examined by Zickfeld et al. (1). the emissions in terms of a cumulative carefully incorporate coupled carbon target that turns out to be more robust feedbacks. than the time-dependent pathway, and Carbon feedbacks occur when there pacity of seawater to dissolve CO2, (iv) use a 3D coupled carbon cycle are climate-induced changes in the net which weakens the as model. This procedure allows an exami- the planet warms (3). Carbon feedbacks fluxes of carbon between the land/ocean nation of the evolution of the coupled are particularly uncertain for the terres- and the atmosphere. The strength of the trial biosphere (4, 5), with changes in feedback depends on the scale of physi- atmospheric quantities directly affecting Author contributions: M.H.E., A.S.G., and A.J.P. wrote the cal climate change and biophysical pro- vegetation type, properties, and physio- paper. cesses in the ocean and land systems. A logical function. These can then affect The authors declare no conflict of interest. simple example of an ocean carbon land carbon uptake via both positive See companion article on page 16129 in issue 38 of volume feedback is caused by the solubility of and negative feedbacks (4). 106. CO2, which varies inversely with temper- The anthropogenic carbon stored in 1To whom correspondence should be addressed. E-mail: ature. Ocean warming reduces the ca- the oceans and terrestrial biosphere has [email protected].

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0908197106 PNAS ͉ September 29, 2009 ͉ vol. 106 ͉ no. 39 ͉ 16539–16540 Downloaded by guest on September 24, 2021 climate-carbon cycle toward temperature sions. Among the pathways shown are Zickfeld et al. (1); two examples are stabilization, consistently deriving cumu- three scenarios (labeled 1–3) that are shown on the right side of Fig. 1. These lative emissions along the way. Zickfeld likely (P ϭ 0.66) to yield a long-term indicate the range of cumulative emis- et al. (1) present their results in a prob- global warming of Ͻ2 °C. Also shown is sions allowable for the 66% and 90% abilistic framework, incorporating en- an emissions pathway (trajectory 4) that likelihoods of avoiding 2 °C warming. sembles of simulations adopting a wide gives a very likely chance (P ϭ 0.9) of So, at the high end of carbon feedback range of climate sensitivity levels, and a stabilizing warming at or below 2 °C, an and climate sensitivity, being 90% sure suite of model estimates of the carbon ambitious yet uncapped trajectory that of avoiding the 2 °C threshold would feedback magnitude. sees cumulative emissions still growing have required that emissions and land- Zickfeld et al. (1) show that to have a by 2200, and a ‘‘business-as-usual’’ sce- clearing ceased in the middle of the last decent chance of stabilizing warming to nario that sees emissions not peaking Ϫ Ͻ century ( 220 PgC). Today, this would 2 °C above preindustrial, net carbon until 2050 before gradually declining require active removal of greenhouse emissions (accumulated from the year over the ensuing century. Curves 1–4 in gasses from the atmosphere. At the 2000) must not exceed Ϸ590 PgC (1 Fig. 1 are based on the median values other extreme, taking the questionable PgC ϭ 1 ϫ 1015 g of carbon, or equiva- obtained by Zickfeld et al. (1), spanning risk of assuming very low climate sensi- lently 3.67 Pg of CO2). This value is the a range of climate sensitivities and car- median across a range of climate sensi- bon feedback rates. tivity and carbon feedback, we still need tivities and carbon feedback rates (see Apparent in scenarios 1–3 in Fig. 1 is to cap post-2000 emissions at only 700 Fig. 1). Here, a ‘‘decent’’ chance is de- the strong tradeoff between the date of PgC. This requires a peak in emissions fined as P Ͼ 0.66, equivalent to ‘‘likely’’ peak emissions and the subsequent re- within the next few decades followed by in Intergovernmental Panel on Climate quired rate of decarbonization of the rapid decarbonization. Change (IPCC) lexicon. Alarmingly, in economy. Put simply, if policymakers A global mean warming of 2 °C could the first 10 years since 2000, net carbon stall significant action for too long, then still have devastating impacts on climate emissions will reach Ϸ100 PgC, approxi- the required rate of decommissioning (13), ecosystems, human health, and in- mately a sixth of the cumulative emis- fossil fuel technologies shoots up. Ap- frastructure. This level of warming, for sions budgeted for above. If policymak- parent in scenario 5 is the need for ur- example, is likely to significantly reduce ers seek even greater certainty to avoid gent and sustained reduction strategies: food productivity over the tropics, sub- crossing the 2 °C threshold, say moving in that case deep emissions reductions stantially increase the risk of extinction into the ‘‘very likely’’ range in IPCC lex- are achieved in the near term, but the for 20–30% of species worldwide, bleach icon (P ϭ 0.9), then it is estimated we phasing out of carbon-intensive technol- most of the world’s coral reefs, and in- need to cap post-2000 emissions at only ogies then progresses slowly, leading to crease the likelihood of severe weather 170 PgC. Assuming unmitigated green- the continued growth of net emissions and extreme climate events (14). Global house gas emissions over the next few beyond the end of this century. This sce- warming to Ϸ2.7 °C could additionally years, we will exceed this cumulative nario is unlikely to stabilize global trigger a gradual but irreversible disinte- Ͻ quota as soon as 2017, which is a sober- warming 2 °C. gration of the West Antarctic Ice Sheet, ing metric in the lead-up to Copenhagen Perhaps the most troubling of the sce- with still lower thresholds thought to this year. narios is trajectory 4, which sees emis- apply to the Greenland Ice Sheet (15), Zickfeld et al. (1) further show that sions plunge dramatically from today, ultimately raising sea level by Ͼ10 m the allowable cumulative emissions for a capping at a mere 170 PgC of post-2000 and displacing hundreds of millions of given stabilization target are robust to emissions. According to the median esti- the chosen pathway toward stabilization. mates of Zickfeld et al. (1), this cumula- people worldwide. This finding agrees with three recent tive emission yields a 90% chance of Two degrees Celsius warming should studies (10–12), each exhibiting remark- avoiding a 2 °C warming. Although this thus not be seen as a mere aspirational ably consistent cumulative emissions tar- task is daunting, indeed all but impossi- target: it surely has to be the maximum gets. What this result implies is that ble, Article 2 of the UNFCCC suggests stabilization target for global warming, aiming for percentage reductions rela- this goal should be firmly in the minds with recognition that even this carries tive to a baseline year is not enough: of policymakers. Unfortunately, the past significant global-scale risks. Worryingly, cumulative emissions must also be 15 years of inaction has very real policy this once-modest target is becoming in- capped. This is demonstrated in Fig. 1, implications today. creasingly difficult to achieve, requiring which shows six different emissions tra- There are, of course, uncertainties in deep emissions cuts over a confronting jectories out to the year 2200, along the magnitude of carbon feedback and time frame, something that must be se- with the corresponding cumulative emis- climate sensitivity, as also described by cured in Copenhagen this year.

1. Zickfeld K, Eby M, Matthews HD, Weaver A (2009) 5. Heimann M, Reichstein M (2008) Terrestrial ecosystem 11. Meinshausen M, et al. (2009) Greenhouse-gas emission Setting cumulative emissions targets to reduce the risk carbon dynamics and climate feedbacks. Nature targets for limiting global warming to 2 °C. Nature of dangerous climate change. Proc Natl Acad Sci USA 451:289–292. 458:1158–1162. 106:16129–16134. 6. Canadell JG, et al. (2007) Contributions to accelerating 12. Allen MR, et al. (2009) Warming caused by cumulative 2. Meehl GA, et al. (2007) Global climate projections. atmospheric CO2 growth from economic activity, car- carbon emissions toward the trillionth tonne. Nature Climate Change 2007: The Physical Science Basis. Con- bon intensity, and efficiency of natural sinks. Proc Natl 458:1163–1166. Acad Sci USA 104:18866–18870. 13. Lenton TM, et al. (2008) Tipping elements in the Earth’s tribution of Working Group I to the Fourth Assessment 7. Cox PM, et al. (2000) Acceleration of global warming climate system. Proc Natl Acad Sci USA 105:1786–1793. Report of the Intergovernmental Panel on Climate due to carbon cycle feedbacks in a coupled climate 14. Parry ML, et al. (2007) Technical summary. Climate Change, eds Solomon S, et al. (Cambridge Univ Press, model. Nature 408:184–187. Change 2007: Impacts, Adaptation and Vulnerability. Cambridge, UK), pp 747–845. 8. Dorrepaal E, et al. (2009) Carbon respiration from sub- Contribution of Working Group II to the Fourth As- 3. Sarmiento JL, et al. (1998) Simulated response of the surface peat accelerated by climate warming in the sessment Report of the Intergovernmental Panel on ocean carbon cycle to anthropogenic climate warming. subarctic. Nature 460:616–619. Climate Change, eds Parry ML, et al. (Cambridge Univ Nature 393:245–249. 9. Houghton RA (2007) Balancing the global carbon bud- Press, Cambridge, UK), pp 23–78. 4. Friedlingstein P, et al. (2006) Climate–carbon cycle get. Annu Rev Earth Plan Sci 35:313–347. 15. Oppenheimer M, Alley RB (2005) Ice sheets, global feedback analysis: Results from the C4MIP model inter- 10. Weaver AJ (2008) Keeping Our Cool: Canada in a warming, and Article 2 of the UNFCCC. Clim Change comparison. J Clim 19:3337–3353. Warming World (Penguin, Toronto). 68:257–267.

16540 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0908197106 England et al. Downloaded by guest on September 24, 2021