Articleshttps://doi.org/10.1038/s41558-018-0097-x

1Center for Global Environmental Research, National Institute for Environmental Studies (NIES), Tsukuba, Japan. 2Climate and Global Dynamics Laboratory, National Center for Atmospheric Research (NCAR), Boulder, CO, USA. *e-mail: tanaka.katsumasa@nies.go.jp

The Paris Agreement adopted in December 2015 aims to limit ‘the increase in the global average temperature to well below 2 °C above pre-industrial levels and to pursue efforts to limit

the temperature increase to 1.5 °C above pre-industrial levels’ (Article 2.1). It also sets an emissions reduction goal ‘to achieve a balance between anthropogenic emissions by sources and remov-als by sinks of greenhouse gases in the second half of this century’ (Article 4.1). Although the temperature targets have received the most attention in political and scientific discussion, the net zero greenhouse gas emissions goal may be more operationally impor-tant as ‘an actionable target’ for policy1. Understanding its consis-tency with ultimate temperature goals is therefore essential.

Although substantial analysis of the 2 °C and 1.5 °C targets has been carried out2–9, no study has modelled the relationship between the Paris Agreement net zero GHG emissions and tem-perature targets with the primary aim of assessing their consistency. Instead, most studies draw conclusions about this relationship based on an analysis of existing scenarios, mostly those compiled in the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5) Scenario Database, which are heteroge-neous collections of integrated assessment model (IAM) outputs designed for other purposes. The current literature draws two con-clusions: (1) net GHG emissions need to be close to or below zero by the end of this century to avoid warming above 2 °C (refs 4–6), and (2) achieving net zero GHG emissions by 2100 does not necessarily ensure limiting warming to the target levels5,7. These claims imply that net zero GHG emissions are necessary, but not sufficient, for achieving the Paris temperature targets. However, it is unclear under what conditions these statements hold and to what degree the rela-tionship between the temperature and emissions targets depends on pathway features (for example, overshoot), uncertain parameters (for example, climate sensitivity) and key assumptions (for example, socioeconomic inertia). Analyses with internally consistent model-ling approaches are required to explore the target consistency in a sys-tematic yet flexible manner. One recent study10 addresses the target

consistency issue by carrying out a multi-gas study that focuses on the consequence for CO2 emissions on centennial timescales. However, the consistency between net zero GHG emissions and the temperature targets during this century, which is arguably more rel-evant to the Agreement itself, remains unaddressed.

We address three questions. First, do the Paris temperature tar-gets imply that net GHG emissions need to fall to zero? The answer to this question may depend on allowing for overshoot of the tem-perature target before returning to it, a possibility about which the Paris Agreement text is not explicit. Second—the converse—does achieving the net zero GHG emissions goal imply temperature out-comes consistent with the Paris temperature targets? For example, it is not clear yet whether the emissions target would pave the way to achieving the temperature target, especially at the 1.5 °C level, and whether this relationship would depend on the timing of achieving zero net emissions within the 50-year period referred to in the Agreement or the metric used to calculate total GHG emis-sions. Third, would both temperature and emissions targets applied jointly lead to substantially different consequences than either type applied alone?

To answer these questions, we calculate least-cost multi-gas emis-sions pathways that would stabilize at 1.5 °C or 2 °C (also allowing for overshoot) and identify when net emissions fall to zero. We then reverse the roles of the targets and impose the net zero-emissions target alone, within the timeframe stipulated in the Agreement, and examine the implications for temperature change. Finally, we apply both types of target simultaneously and compare outcomes to the cases in which each is applied individually (Table 1). Because we focus on the long-term implications of the temperature and emis-sions goals, we do not directly include near-term nationally deter-mined contributions in the analysis. Our analysis assumes that the Agreement text requires the temperature not to exceed (or to be equilibrated at) a desired level, but does not call for further decline, although this is open to interpretation. Furthermore, although real emissions reductions that emerge from international negotiations

The Paris Agreement zero-emissions goal is not always consistent with the 1.5 °C and 2 °C temperature targetsKatsumasa Tanaka   1* and Brian C. O’Neill2

The Paris Agreement stipulates that global warming be stabilized at well below 2 °C above pre-industrial levels, with aims to further constrain this warming to 1.5 °C. However, it also calls for reducing net anthropogenic greenhouse gas (GHG) emissions to zero during the second half of this century. Here, we use a reduced-form integrated assessment model to examine the consis-tency between temperature- and emission-based targets. We find that net zero GHG emissions are not necessarily required to remain below 1.5 °C or 2 °C, assuming either target can be achieved without overshoot. With overshoot, however, the emissions goal is consistent with the temperature targets, and substantial negative emissions are associated with reducing warming after it peaks. Temperature targets are put at risk by late achievement of emissions goals and the use of some GHG emission metrics. Refinement of Paris Agreement emissions goals should include a focus on net zero CO2—not GHG—emissions, achieved early in the second half of the century.

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATure ClimATe ChANge | VOL 8 | APRIL 2018 | 319–324 | www.nature.com/natureclimatechange 319

Articles Nature Climate ChaNge

may not resemble least-cost scenarios, these scenarios do provide an easily interpretable benchmark representing efficient emissions pathways towards targets.

We use a reduced-form version of an IAM—Aggregated Carbon Cycle, Atmospheric Chemistry, and Climate model (ACC2) version 4.2 (see Methods)—which generates emissions pathways under a cost-effectiveness approach using marginal abatement cost (MAC) functions for CO2, CH4 and N2O (ref. 11). ACC2 accounts for the diverse nature of GHGs (CO2, CH4, N2O, SF6, 29 species of halo-carbons, tropospheric and stratospheric O3, and stratospheric water vapour), pollutants (NOx, CO and VOC) and aerosols (direct effect of sulfate aerosols, direct effect of black carbon and organic car-bon, and indirect effect of all aerosols) in the Earth system, such as different atmospheric residence times and radiative properties12. Emissions pathways for CO2, CH4 and N2O, the gases representing a large fraction of total radiative forcing in baseline emissions scenar-ios13, are generated endogenously as the outcome of a cost minimiza-tion, subject to meeting a specified temperature or emissions target. Emissions for all other GHGs and aerosols are assumed to follow a fixed mitigation scenario consistent with approximate stabilization at 2 °C, and variations from this pathway would have only a second-order effect on the results (Supplementary Fig. 10). Unless noted otherwise, results are based on a default set of assumptions, with sensitivity analyses provided where appropriate (Supplementary

Figs. 2 to 11). These assumptions include a 3 °C equilibrium climate sensitivity and the use of global warming potential with a 100-year time horizon (GWP100)14 as weighting factors to calculate net GHG emissions as the sum of Kyoto gas emissions. In addition, we con-strain allowable rates of change in mitigation effort based on the results of a more complex IAM in order to exclude highly implau-sible mitigation pathways2. It should be emphasized that, although we generate pathways based on cost minimization under socioeco-nomic constraints unless noted otherwise, our main interest is to clarify geophysical relationships between the emissions goal and the temperature targets, offering policy-related insights into the Paris Agreement. Our global MAC function approach is suitable for representing first-order socioeconomic considerations influencing mitigation pathways, but is a simplification of more complex miti-gation modelling, which should be relied on for questions focused primarily on cost outcomes.

results and discussionEmission implications of temperature targets. We find that remaining below either 1.5 °C or 2 °C does not require GHG emis-sions to be reduced to a net zero level (Cases II and IV) (Fig. 1). These results are based on default model assumptions, except that in order to make achieving the 1.5 °C target feasible (Case II), we had to allow for extremely rapid mitigation by relaxing the abate-ment rate constraints (that is, first derivative constraint changed from 4%/year to 8%/year and second derivative constraint removed, see Methods). In the least-cost emissions pathway, net GHG emis-sions fall to approximately 10 GtCO2eq/year in about 2033 in the 1.5 °C case and to 16 GtCO2eq/year in about 2060 in the 2 °C case, and then remain nearly constant after that. Sustained positive net emissions after these reductions are consistent with a stable global average temperature because they comprise a mix of gases. CO2 emissions decline sharply to low levels and then decline slowly after that (becoming slightly negative by about 2050 in the 1.5 °C case), while emissions of CH4 and N2O are first reduced and then rise slowly over time (Fig. 2). Thus, slowly declining CO2 concentrations are mostly offset by gradually increasing CH4 and N2O concentra-tions, producing slowly declining forcing and stable temperature.

The conclusion that remaining below the Paris temperature tar-gets does not require net zero GHG emissions is robust to assump-tions about climate sensitivity (Supplementary Fig. 2). However, remaining below these temperature targets without temporarily exceeding them (overshooting) is only possible under limited con-ditions. If climate sensitivity is above 3.4 °C (the 2 °C target case) or 2.2 °C (the 1.5 °C target case), the targets are not achievable (with all other model assumptions intact) because emissions cannot be reduced fast enough to sufficiently limit rising temperatures. Given the likely range for climate sensitivity of 1.5 °C to 4.5 °C in IPCC AR514, this rules out achieving the targets for a considerable fraction of the climate sensitivity range in the 2 °C case and nearly the whole range in the 1.5 °C case. Even if the emissions abatement constraints are relaxed and higher climate sensitivities are assumed, net zero GHG emissions are still not required to meet the targets (fifth and sixth cases in Supplementary Fig. 2).

Given the extremely rapid abatement implied by the 1.5 °C stabilization scenario, we investigate pathways that allow for tem-porary overshoot of temperature targets (Cases I and III), requir-ing warming to not exceed 1.5 °C or 2 °C beginning in 2100. The least-cost overshoot pathways show a temperature peak at 1.95 °C in 2058 (1.5 °C target case) and 2.30 °C in 2070 (2 °C target case) (Supplementary Table 1). In the 1.5 °C target case, net GHG emis-sions fall to zero in 2070 and then stay negative for the rest of the cen-tury. The net CO2 emissions path reaches zero in 2058 and remains negative at about − 17 GtCO2/year for the rest of the century. This is in line with the previous finding that net zero CO2 emissions occur earlier than net zero GHG emissions4. Once climate stabilization is

Table 1 | Descriptions of the ten representative cases analyzed in this study

Cases experimental Designs

Temperature target

emissions goal

emission metric

emissions path

I 1.5 °C (overshoot)

– – –

II 1.5 °C – – Relaxed abatement constraints

III 2 °C (overshoot)

– – –

IV 2 °C – – –

V 2 °C Zero GHGs by 2060

GWP100 2030 peak

VI 2 °C Zero CO2 by 2060

— 2030 peak

VII – Zero GHGs by 2060

GWP100 2030 peak

VIII – Zero GHGs by 2100

GWP100 2030 peak and linear decline

IX – Zero GHGs by 2100

GWP100 2030 peak

X – Zero GHGs by 2060

GTP100 2030 peak

Targets and constraint(s) applied to each case are summarized. Cases I and III are cost-effective 1.5 and 2 °C stabilization pathways (respectively) including overshoot allowed before 2100. Case II is a 1.5 °C stabilization pathway without overshoot, a theoretical case assuming extremely rapid mitigation (see main text for details). Case IV is a cost-effective 2 °C stabilization pathway without overshoot. In Case V, the 2 °C target and net zero GHG emissions target to be met by 2060 are jointly applied. Case VI is the same as Case V, except that the zero-emissions requirement applies only to CO2. Case VII is a cost-effective pathway meeting the net zero GHG emissions target by 2060 without a temperature target. In Cases VIII and IX, the net zero GHG emissions target is achieved by 2100 through linearly increasing abatement from 2035 to 2095 (Case VIII) or cost-effectively (Case IX). Case X is equivalent to Case VII except for the use of GTP100 to weight the emissions of different GHGs to derive total GHG emissions. Unless noted otherwise, the conventional GWP100 is used for such weighting. All cases using the emissions target require GHG emissions to peak by 2030. GTP100, 100-year global temperature change potential; GWP100, 100-year global warming potential.

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATure ClimATe ChANge | VOL 8 | APRIL 2018 | 319–324 | www.nature.com/natureclimatechange320

ArticlesNature Climate ChaNge

achieved in 2100, the net GHG emissions become positive again, with positive CH4 and N2O emissions more than offsetting slightly negative CO2 emissions. A similar pattern of changes occurs in the 2 °C target case, with shorter periods of net negative CO2 and GHG emissions. As shown also by previous studies, net zero emissions leading to the 1.5 °C target occur one to two decade(s) earlier than those leading to the 2 °C target2–4,8. Furthermore, sensitivity analyses generally support that net CO2 and GHG emissions become nega-tive in overshoot cases (Supplementary Figs. 3–5 and 10). It was recently shown that zero CO2 emissions (and implicitly net GHG emissions) are not required to meet the temperature targets based on scenarios involving overshoots lasting for a few centuries (that is, until around 2300)10. Although we do not consider such long over-shoot in our analysis, different carbon cycle and climate dynamics may matter on centennial timescales, requiring further analysis.

The difference between the GHG emissions results for the over-shoot and non-overshoot cases in our analysis can be explained, at least in part, by the necessity of decreasing the global average temperature and GHG concentration. In the overshoot case, the least-cost pathway shows a peak and then decline in global average temperature, which requires decreasing CO2 concentrations and therefore net negative CO2 emissions that are large enough in

magnitude to imply net negative GHG emissions. In the non-overshoot case, no decline in temperature is required, and there-fore little decline in CO2 concentration is needed, meaning that negative CO2 emissions are not large enough to imply net negative (or zero) GHG emissions.

Temperature implications of emissions target. When the net zero GHG emissions goal is imposed alone, we find that it pro-duces temperature pathways that are consistent with the 1.5 and 2 °C targets only under limited conditions. Given the uncertainty in the required timing of achieving net zero emissions, we test two illustrative timings: achieving the goal early (2060) or late (2100). Meeting the target in 2050 is infeasible under our default assump-tions (the earliest possible year to make the target feasible is 2056). Furthermore, we assume that GHG emissions peak by 2030, given the statement in Article 4.1: ‘… Parties aim to reach global peaking of greenhouse gas emissions as soon as possible… ’ The late achieve-ment case has two further illustrative pathways after 2030: one in which net GHG emissions are linearly reduced to zero in 2100 (Case VIII) and another in which GHG emissions decline from the peak level only when economically motivated to do so (that is, a cost-effective pathway) (Case IX).

–20

–10

0

10

20

30

40

50

60

2000 2020 2040 2060 2080 2100 2120 2140

a

0.5

1.0

1.5

2.0

2.5

3.0

3.5

2000 2020 2040 2060 2080 2100 2120 2140

g

–10

0

10

20

30

40

50

60

70

2000 2020 2040 2060 2080 2100 2120 2140

d

–20

–10

0

10

20

30

40

50

60

2000 2020 2040 2060 2080 2100 2120 2140

b

–10

0

10

20

30

40

50

60

70

2000 2020 2040 2060 2080 2100 2120 2140

e

0.5

1.0

1.5

2.0

2.5

3.0

3.5

2000 2020 2040 2060 2080 2100 2120 2140

h

–20

–10

0

10

20

30

40

50

60

2000 2020 2040 2060 2080 2100 2120 2140

c

–10

0

10

20

30

40

50

60

70

2000 2020 2040 2060 2080 2100 2120 2140

f

0.5

1.0

1.5

2.0

2.5

3.0

3.5

2000 2020 2040 2060 2080 2100 2120 2140

i

CO

2 em

issi

ons

(GtC

O2

yr–1

)

CO

2 em

issi

ons

(GtC

O2

yr–1

)

CO

2 em

issi

ons

(GtC

O2

yr–1

)

GH

G e

mis

sion

s (G

tCO

2e y

r–1)

GH

G e

mis

sion

s (G

tCO

2e y

r–1)

GH

G e

mis

sion

s (G

tCO

2e y

r–1)

War

min

g si

nce

pre-

indu

stria

l tim

es (

°C)

War

min

g si

nce

pre-

indu

stria

l tim

es (

°C)

War

min

g si

nce

pre-

indu

stria

l tim

es (

°C)

(I) 1.5 °C (OS) (II) 1.5 °C* (III) 2 °C (OS) (IV) 2 °C (V) 2 °C, ZE (2060) (VI) 2 °C, ZE (2060, CO2)

(VII) ZE (2060) (VIII) ZE (2100, LN) (IX) ZE (2100) (X) ZE (2060, GTP100)

Fig. 1 | global emissions and warming for the ten representative cases. a–f, Global annual anthropogenic CO2 (a–c) and GHG (d–f) emissions from 2000 to 2140. g–i, Global average warming since pre-industrial time. OS, ZE and LN: overshoot, zero emissions and linearly increasing abatement, respectively. See Table 1 for case descriptions. A complete reporting of the results is available in Supplementary Fig. 1. *This case deliberately relaxes constraints for abatement rates so that the 1.5 °C target can be achieved without overshoot (see text for details), unlike all the other cases in this figure using default model assumptions.

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATure ClimATe ChANge | VOL 8 | APRIL 2018 | 319–324 | www.nature.com/natureclimatechange 321

Articles Nature Climate ChaNge

The three emissions target cases produce temperature pathways that peak above 2 °C and then decline. In the 2060 emissions tar-get case (Case VII) warming peaks at 2.1 °C in 2052 and eventually decreases to 1.5 °C in 2123. In the 2100 emissions target case, warm-ing peaks at 2.5 °C in 2083 (Case VIII) or 3.1 °C in 2093 (Case IX), and does not decline to the target levels until the 23rd century (Supplementary Table 1). Considering a range of climate sensi-tivities (Supplementary Fig. 6) and the possibility of achieving the emissions goal in 2050 (Supplementary Fig. 7, with relaxed mitiga-tion constraints) shows that preventing warming from rising signif-icantly above 1.5 °C is only possible if climate sensitivity is assumed to be low (2.0 °C) and if net zero GHG emissions is achieved by

2060, and limiting warming to 2 °C only if climate sensitivity is low or net zero GHG emissions is achieved by 2050. Consistent with previous studies5,7, achieving net zero GHG emissions by 2100 can produce peak warming substantially above 2 °C, and warming is quite sensitive to the time at which net zero emissions is achieved.

Significant impact of emission metrics. Any calculation of net GHG emissions depends on the metric used to aggregate gases15, and therefore an analysis of a GHG emissions target may also be sensitive to this metric. We test the effect of replacing the default GWP100 with two alternatives assessed by the IPCC AR514: the 20-year GWP (GWP20) and the 100-year global temperature

0

20

40

60

80

100

120

2000 2020 2040 2060 2080 2100 2120 2140

a

0

10

20

30

40

50

60

70

80

2000 2020 2040 2060 2080 2100 2120 2140

b

0

10

20

30

40

50

60

2000 2020 2040 2060 2080 2100 2120 2140

c

0

500

1,000

1,500

2,000

2000 2020 2040 2060 2080 2100 2120 2140

Tot

al a

bate

men

t cos

ts(U

S$

billi

on)

d

150

200

250

300

350

400

450

500

550

2000 2020 2040 2060 2080 2100 2120 2140

CH

4 em

issi

ons

(MtC

H4

yr–1

)

e

8

10

12

14

16

18

20

2000 2020 2040 2060 2080 2100 2120 2140

N2O

em

issi

ons

(MtN

2O y

r–1)

f

300

350

400

450

500

550

600

2000 2020 2040 2060 2080 2100 2120 2140

g

1,500

2,000

2,500

3,000

3,500

2000 2020 2040 2060 2080 2100 2120 2140

h

300

320

340

360

380

400

420

440

2000 2020 2040 2060 2080 2100 2120 2140

i

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

2000 2020 2040 2060 2080 2100 2120 2140

CO

2 fo

rcin

g (W

m–2

)

j

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

2000 2020 2040 2060 2080 2100 2120 2140

Non

-CO

2 fo

rcin

g (W

m–2

)

k

1

2

3

4

5

6

2000 2020 2040 2060 2080 2100 2120 2140

Tot

al r

adia

tive

forc

ing

(W m

–2)l

CO

2 ab

atem

ent l

evel

(%

)

CH

4 ab

atem

ent l

evel

(%

)

N2O

aba

tem

ent l

evel

(%

)

CO

2 co

ncen

trat

ion

(ppm

)

CH

4 co

ncen

trat

ion

(ppb

)

N2O

con

cent

ratio

n (p

pb)

(I) 1.5 °C (OS) (II) 1.5 °C* (III) 2 °C (OS) (IV) 2 °C (V) 2 °C, ZE (2060) (VI) 2 °C, ZE (2060, CO2)

(VII) ZE (2060) (VIII) ZE (2100, LN) (IX) ZE (2100) (X) ZE (2060, GTP100)

Fig. 2 | gas-specific abatement and selected earth system outcomes for the ten representative cases. a–c, Abatement levels of CO2 (a), CH4 (b) and N2O (c). d, Global annual abatement costs for all three endogenous gases. e–f, Global annual anthropogenic emissions of CH4 (e) and N2O (f). g–i, Atmospheric CO2 (g), CH4 (h) and N2O (i) concentrations. j, CO2 forcing. k, Non-CO2 forcing. l, Total radiative forcing. Global annual abatement costs are shown in terms of the net present value (that is, year 2010) in billions of US dollars. See Table 1 for case descriptions. *This case deliberately relaxes constraints for abatement rates so that the 1.5 °C target can be achieved without overshoot. As a result, no abatement costs are shown for this case (d).

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATure ClimATe ChANge | VOL 8 | APRIL 2018 | 319–324 | www.nature.com/natureclimatechange322

ArticlesNature Climate ChaNge

change potential (GTP100). GWP20 emphasizes shorter-term climate impacts than GWP100. GTP100 uses the temperature change 100 years after emissions occur as a basis for the metric, an approach argued to be more consistent with long-term climate goals16,17. Results show that the mix of GHGs constituting the least-cost emissions pathway is sensitive to the metric. When GTP100 is adopted, the negative CO2 emission level is not as deep as in the standard GWP100 case (Cases VII and X), because both CH4 and N2O emissions are valued less (but especially CH4) when converted to CO2. This results in a slightly higher peak temperature of 2.2 °C (as opposed to 2.1 °C). Because of the less negative CO2 emissions, the rate of temperature decline following the peak is slower, leading to a delay of 54 years in reducing the warming below 2 °C compared to the GWP100 case (Supplementary Table 1). When GWP20 is used as the metric instead of GWP100, the net zero GHG emissions target is infeasible, primarily because CH4 is valued so highly rela-tive to CO2 that negative CO2 emissions cannot adequately offset the residual positive CH4 emissions. Under different timings and pathways to meet the emissions target (Supplementary Fig. 9), the abovementioned findings generally hold, supporting the fact that the use of GTP100 leads to a higher peak temperature and a lower rate of temperature decrease after the peak. Although the influence of the metric on temperature stabilization pathways was shown to be small globally18,19, we found that its impact on zero-emissions pathways is substantial.

Joint temperature and emissions targets. When the temperature and emissions targets are applied jointly, we gain insight into how each target affects outcomes. For example, when adding the 2060 GHG emissions target to the 2 °C stabilization constraint (Cases IV and V), the peak warming is unchanged but a decline in tempera-ture is introduced after 2 °C is reached. This result emphasizes that a key effect of the net zero GHG emissions goal is to produce a decline in temperature after it peaks. In contrast, adding a net zero CO2 emissions target in 2060, rather than the net zero GHG emis-sions target, to the 2 °C stabilization goal (Cases IV and VI) does not substantially change the nature of the emissions or temperature pathways, reflecting the fact that the CO2 emissions target is more consistent with the 2 °C stabilization target. Similar conclusions can be drawn for the 1.5 °C target case with a climate sensitivity of 2 °C (Supplementary Fig. 11).

The insensitivity of the temperature stabilization pathways to the additional zero CO2 emission requirement can be inferred from the well-established rule of thumb: temperature stabilization eventually requires zero emissions of CO2 (ref. 20). Also recognized is that, in addition to zero CO2 emissions, temperature stabilization requires emissions of short-lived components to stop growing (that is, con-stant emissions)17,21,22; however, this is not very evident in our CH4 emissions pathways, which reflect cost minimization, not just the geophysical requirement.

The difference between combining the 2 °C stabilization target with either the net zero CO2 or net zero GHG emissions target is substantial, a finding worth highlighting because the distinction is sometimes blurred in debates on net zero emissions and carbon budgets. Not only does the GHG emissions target substantially change the temperature outcome, it also leads to a different mix of emissions. CO2 mitigation is much deeper, caused by the need to offset residual CH4 and N2O emissions, reflecting unabatable CH4 emissions from livestock and N2O emissions from agriculture.

Need for net zero GHG emissions. Our results provide insight into and illuminate limitations to the previous claim that net GHG emis-sions need to be close to or below zero to keep the warming below 2 °C (refs 4–6). First, it is important to recognize that this claim is based on scenario evidence in which there are many scenarios that achieve the temperature target without requiring net zero GHG

emissions. About a third of scenarios with a likely chance of achiev-ing the target, and about a half with a medium chance, do so without net zero emissions4. However, this does not fully explain the differ-ence with our results, because we find no cases in which net zero GHG emissions are required unless the temperature overshoots the target and then declines. To further understand this difference, we also performed an analysis using the AR5 Scenario Database, but separately for peak and decline (PD) versus all other scenarios (Supplementary Fig. 17 and Supplementary Table 2).

First, we find that all scenarios that reach net negative GHG emis-sions follow PD temperature pathways. Second, no scenario with a rising temperature pathway has net negative emissions. Third, 34 of 61 scenarios with PD temperature paths have net negative GHG emissions, which increases to 19 of 20 for scenarios with a decline of at least 0.1 °C.

Given these findings, we believe that the previous claim about the need for net zero GHG emissions is strongly influenced by the declining temperature trends in PD scenarios. In other words, the shape of the temperature pathways in the existing scenarios data-base, rather than the temperature target itself, has probably led to an overemphasis of net zero GHG emissions to achieve the 1.5 °C or 2 °C target.

ConclusionsThis study aims to illuminate the relationships between the Paris emissions goal and temperature targets, demonstrating that achieving the zero GHG emissions goal is not a requirement for meeting the 1.5 °C or 2 °C targets, assuming these targets are to be met without overshoot. Achieving net zero GHG emissions is associated with substantially negative CO2 emissions and a decline in global temperature, whereas achieving net zero CO2 emissions is associated more closely with stabilized temperature. Thus a zero-emissions target for CO2, rather than GHGs, is more likely to be consistent with the Paris temperature targets, unless either (i) overshoot of the temperature targets were allowed, or (ii) the intent of the Paris Agreement is not only to keep warming from exceeding a given level but to then reduce it substantially thereafter.

A net zero CO2 emissions goal would be easier to achieve than the net zero GHG emissions goal, which requires substantially nega-tive CO2 emissions, a point of recent concern23. However, achieving net zero CO2 emissions would still require rapid decarbonization, and separate limits on non-CO2 emissions may be needed as well. If overshoot of the temperature goal occurred, as seems likely in the case of the 1.5 °C target, further action to create negative CO2 emis-sions would be necessary to lower temperatures.

methodsMethods, including statements of data availability and any asso-ciated accession codes and references, are available at https://doi.org/10.1038/s41558-018-0097-x.

Received: 27 July 2017; Accepted: 2 February 2018; Published online: 26 March 2018

references 1. Geden, O. An actionable climate target. Nat. Geosci. 9, 340–342 (2016). 2. Azar, C., Johansson, D. J. A. & Mattsson, N. Meeting global temperature

targets—the role of bioenergy with carbon capture and storage. Environ. Res. Lett. 8, 034004 (2013).

3. Rogelj, J. et al. Energy system transformations for limiting end-of-century warming to below 1.5 °C. Nat. Clim. Change 5, 519–527 (2015).

4. Rogelj, J. et al. Zero emission targets as long-term global goals for climate protection. Environ. Res. Lett. 10, 105007 (2015).

5. Hallegatte, S. et al. Mapping the climate change challenge. Nat. Clim. Change 6, 663–668 (2016).

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATure ClimATe ChANge | VOL 8 | APRIL 2018 | 319–324 | www.nature.com/natureclimatechange 323

Articles Nature Climate ChaNge

6. Sanderson, B. M., O’NeillB. C. & Tebaldi, C. What would it take to achieve the Paris temperature targets? Geophys. Res. Lett. 43, 7133–7142 (2016).

7. Schleussner, C.-F. et al. Science and policy characteristics of the Paris Agreement temperature goal. Nat. Clim. Change 6, 827–835 (2016).

8. Robiou du Pont, Y. et al. Equitable mitigation to achieve the Paris Agreement goals. Nat. Clim. Change 7, 38–43 (2017).

9. Su, X. et al. Emission pathways to achieve 2.0 and 1.5 °C climate targets. Earth’s Future 5, 592–604 (2017).

10. Wigley, T. M. L. The Paris warming targets: emissions requirements and sea level consequences. Climatic Change https://doi.org/10.1007/s10584-017-2119-5 (2017).

11. Tanaka, K., Johansson, D. J. A., O’Neill, B. C. & Fuglestvedt, J. S. Emission metrics under the 2 °C climate stabilization target. Climatic Change 117, 933–941 (2013).

12. Tanaka, K. et al. Aggregated Carbon Cycle, Atmospheric Chemistry, and Climate Model (ACC2) – Description of the Forward and Inverse Modes 188 (Max Planck Institute for Meteorology, 2007).

13. Riahi, K., Grübler, A. & Nakicenovic, N. Scenarios of long-term socio-economic and environmental development under climate stabilization. Technol. Forecast. Soc. Change 74, 887–935 (2007).

14. IPCC Climate Change 2013: The Physical Science Basis (eds Stocker, T. F. et al.) (Cambridge Univ. Press: 2013).

15. Tanaka, K., Peters, G. P. & Fuglestvedt, J. S. Policy update: multicomponent climate policy: why do emission metrics matter? Carbon Manag. 1, 191–197 (2010).

16. Tol, R. S. J., Berntsen, T. K., O’Neill, B. C., Fuglestvedt, J. S. & Shine, K. P. A unifying framework for metrics for aggregating the climate effect of different emissions. Environ. Res. Lett. 7, 044006 (2012).

17. Allen, M. R. et al. New use of global warming potentials to compare cumulative and short-lived climate pollutants. Nat. Clim. Change 6, 773–776 (2016).

18. O’Neill, B. C. Economics, natural science, and the costs of global warming potentials. Climatic Change 58, 251–260 (2003).

19. Johansson, D., Persson, U. & Azar, C. The cost of using global warming potentials: analysing the trade off between CO2, CH4 and N2O. Climatic Change 77, 291–309 (2006).

20. Matthews, H. D. & Caldeira, K. Stabilizing climate requires near-zero emissions. Geophys. Res. Lett. 35, L04705 (2008).

21. Rogelj, J., Meinshausen, M., Schaeffer, M., Knutti, R. & Riahi, K. Impact of short-lived non-CO2 mitigation on carbon budgets for stabilizing global warming. Environ. Res. Lett. 10, 075001 (2015).

22. Pierrehumbert, R. T. Short-lived climate pollution. Annu. Rev. Earth Planet. Sci. 42, 341–379 (2014).

23. Field, C. B. & Mach, K. J. Rightsizing carbon dioxide removal. Science 356, 706–707 (2017).

AcknowledgementsThe authors thank D. Johansson for providing data associated with the maximum abatement cost curves used in this study. The authors are grateful for comments from B. Sanderson, X. Su, C. Tebaldi and T. Wigley, which were useful to improve this study. This research was partially supported by the Environment Research and Technology Development Fund (2-1702) of the Environmental Restoration and Conservation Agency, Japan.

Author contributionsB.C.O. conceived of the research. K.T. led the study. K.T. and B.C.O designed the experiment. K.T. performed the model simulations and generated all the figures and tables. K.T. and B.C.O. analysed the results. K.T. and B.C.O. drafted the manuscript.

Competing interestsThe authors declare no competing interests.

Additional informationSupplementary information is available for this paper at https://doi.org/10.1038/s41558-018-0097-x.

Reprints and permissions information is available at www.nature.com/reprints.

Correspondence and requests for materials should be addressed to K.T.

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATure ClimATe ChANge | VOL 8 | APRIL 2018 | 319–324 | www.nature.com/natureclimatechange324

ArticlesNature Climate ChaNge

model and 450 ppm CO2eq scenarios from other IAMs in the AR5 Scenario Database.

A note of caution in comparing our results with others is that, while our analysis is based on a deterministic approach, most other recent analyses adopt probabilistic approaches. In the mitigation literature, two different probability criteria exist to determine whether a target can be achieved or not: (1) ‘likely chance (> 66%)’ criteria, used predominantly by default, and (2) ‘medium chance (50–66%)’ criteria, less common but more comparable to our threshold of judgement (used, for example, in ref. 4 for the 1.5 °C target). Previous studies also report results with the medium chance criteria, but often not equally highlighted. This indicates that our model may show less stringent results than other IAMs using the likely chance criteria.

Finally, net zero emissions calculated in our analysis are only with respect to anthropogenic fluxes, in contrast to a different interpretation including natural fluxes31. It is not explicit in the English text whether ‘removals’ in Article 4.1 of the Paris Agreement concern only anthropogenic fluxes without accounting for natural sinks; however, it can be interpreted so from the original French text.

Data availability. The data that support the findings of this study are available from the corresponding author upon request.

references 24. Hooss, G., Voss, R., Hasselmann, K., Maier-Reimer, E. & Joos, F. A nonlinear

impulse response model of the coupled carbon cycle-climate system (NICCS). Clim. Dynam. 18, 189–202 (2001).

25. Bruckner, T., Hooss, G., Füssel, H.-M. & Hasselmann, K. Climate system modeling in the framework of the tolerable windows approach: the ICLIPS climate model. Climatic Change 56, 119–137 (2003).

26. Tanaka, K., Raddatz, T., O’Neill, B. C. & Reick, C. H. Insufficient forcing uncertainty underestimates the risk of high climate sensitivity. Geophys. Res. Lett. 36, L16709 (2009).

27. Tanaka, K., Raddatz, T. Correlation between climate sensitivity and aerosol forcing and its implication for the ‘climate trap’. Climatic Change 109, 815–825 (2011).

28. Joos, F. et al. Carbon dioxide and climate impulse response functions for the computation of greenhouse gas metrics: a multi-model analysis. Atmos. Chem. Phys. 13, 2793–2825 (2013).

29. Johansson, D. Temperature stabilization, ocean heat uptake and radiative forcing overshoot profiles. Climatic Change 108, 107–134 (2011).

30. van Vuuren, D. P., Weyant, J. & de la Chesnaye, F. Multi-gas scenarios to stabilize radiative forcing. Energy Econ. 28, 102–120 (2006).

31. Walsh, B. et al. Pathways for balancing CO2 emissions and sinks. Nat. Commun. 8, 14856 (2017).

methodsThe ACC2 model (version 4.2) is a globally aggregated IAM comprising the following four components: carbon cycle, atmospheric chemistry, climate and economy11,12. ACC2 was developed from earlier simple climate models24,25. The main difference from the previous version 4.011 is revision of the CO2 MAC function to account for negative CO2 emissions through technologies such as bioenergy combined with carbon capture and storage (bioCCS)2. The carbon cycle component is a four-box model for the ocean–atmosphere coupled system, linked to another four-box model for the terrestrial biosphere. The saturation of ocean CO2 uptake under rising atmospheric CO2 concentration is described by thermodynamic equilibrium associated with carbonate species in the ocean. No climate–carbon cycle feedbacks are assumed; in other words, we treat the carbon cycle as insensitive to the change in climate. The atmospheric chemistry component consists of parameterizations obtained from sensitivity runs of more detailed chemistry transport models. The radiative forcings of CO2, CH4 and N2O are calculated based on parameterizations. The climate component is an energy balance model coupled with a diffusion model. The climate sensitivity in the default set-up is assumed to be 3 °C, a conventional best estimate within the latest IPCC likely range of 1.5–4.5 °C (ref. 14). Other uncertain parameters, including those determining aerosol forcing, are adjusted based on a Bayesian approach using historical data26,27. Note that we obtain best estimates of parameters without probabilistic interpretations. Responses of the carbon cycle and climate components are within the respective ranges from multi-model comparisons28.

The economy component contains the MAC functions for CO2, CH4 and N2O (refs 2,29). The abatement level for each gas is defined relative to the respective baseline level, here taken to be the A2r scenario13, a representative baseline. Note that the CO2 MAC function applies to the emissions of fossil fuel origin. We treat the abatement of each gas independently without accounting for co-reductions from common sources30. Abatements are allowed to start in year 2020. The maximum abatement levels for CO2, CH4 and N2O are assumed to be 112, 70 and 50% from the baseline, respectively2,29. Socioeconomic inertia is modelled through constraints imposed on the first and second derivatives of abatement levels for endogenous gases (CO2, CH4 and N2O) (4%/year and 0.4%/year2, respectively)2. The second derivative constraint means that it takes 10 years for the annual growth of abatement to increase by four percentage points. The first and second derivatives of abatement levels in our model runs generally span larger ranges than those of the 450 ppm CO2eq scenarios in the AR5 Scenario Database (Supplementary Figs. 12 to 14). Emissions pathways of other components (including land-use CO2 emissions) follow GGI A2r 480 ppm CO2eq (ref. 13) until 2100 and remain at the respective 2100 levels thereafter. The default discount rate is 4%. Supplementary Figs. 12 to 16 show comparisons between CO2, CH4, N2O and GHG emissions and temperature projections from our

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.

NATure ClimATe ChANge | www.nature.com/natureclimatechange