Projected Global Temperature Changes After Net Zero Are Small But Significant

As more countries make net zero greenhouse gas emissions pledges, it is crucial to understand the effects on global climate after achieving net zero emissions. The climate has been found to continue to evolve even after the abrupt cessation of CO2 emissions, with some models simulating a small warming and others simulating a small cooling. In this study, we analyze if the temperature and precipitation changes post abrupt cessation of CO2 emissions are significantly different compared to natural climate variations. We find that the temperature changes are outside of natural variability for most models, whilst the precipitation changes are mostly non‐significant. We also demonstrate that post‐net zero temperature changes have implications for the remaining carbon budget. The possibility of further global warming post‐net zero adds to the evidence supporting more rapid emissions reductions in the near‐term.


Introduction
Global greenhouse gas emissions must be reduced to near zero to prevent continued global warming (MacDougall et al., 2020;Matthews & Zickfeld, 2012).This is required if the target of " […] holding the increase in the global average temperature to well below 2˚C above preindustrial levels and pursuing efforts to limit the temperature increase to 1.5°C […]" (UNFCCC, 2015) outlined by the Paris Agreement is to be achieved.As such, many countries have made net zero pledges (Energy & Climate Intelligence Unit|Net Zero Scorecard, 2023).
Several experiments have been run that aim to quantify the response of the global and local climate after the cessation of emissions (Dvorak et al., 2022;Jones et al., 2019aJones et al., , 2019b;;Sherwood et al., 2022).One such experiment, the Zero Emissions Commitment Model Intercomparison Project (ZECMIP) (Jones et al., 2019a(Jones et al., , 2019b)), aims to understand the evolution of the climate after the abrupt cessation of CO 2 emissions at around 1.5˚C.In this scenario, Earth System Models (ESMSs) and ESMSs of Intermediate Complexity (EMICs) predict a post-net zero global average temperature change of 0.07˚C ( 0.36˚C to 0.29˚C between models) 50 years after emission cease -this is the Zero Emissions Commitment (ZEC) (MacDougall et al., 2020).Following net zero CO 2 emissions, the thermal inertia of the oceans will drive an increase in the global mean surface temperature, which is counteracted by the removal of carbon dioxide by the terrestrial biosphere and oceans.Ultimately, the trajectory of the climate after the cessation of emissions, and whether the climate will warm or cool, depends on the magnitude of these two effects, and results in a post-net zero change in temperature that is close to zero (MacDougall et al., 2020(MacDougall et al., , 2022)).
As the ZEC is small, it is important to know if this is robustly distinct from the background climate variations.Previous studies have averaged ZEC across a multi-model ensemble but not assessed the internal variability of each model which is known to cause apparent trends on short timescales.Here, we analyze whether global temperature changes after the abrupt cessation of CO 2 emissions are significantly different from natural variability for several ESMs compared to the natural variability of their climate.We also analyze if the global mean precipitation changes post-net zero are significant compared to their natural variability.
Any post-net zero changes will have implications on the remaining carbon budget (RCB).The RCB is the cumulative CO 2 that can be emitted while keeping the peak global average temperature rise below a global warming level (Dvorak et al., 2022).The RCB allows emissions reduction targets to be aligned with global warming levels.If annual emissions remain at 2022 levels (40.2 GtCO 2 ), nine years (380 GtCO 2 ) remain in the carbon budget from the beginning of 2023 for a 50% chance of avoiding exceeding 1.5˚C (Global Carbon Project (GCP), 2023).In previous work that used the ZEC when quantifying the RCB, the ZEC value is either ignored (Friedlingstein et al., 2020;Rypdal et al., 2021), considered zero (Rogelj et al., 2019), or has an uncertainty distributed around zero (IPCC, 2021;Lamboll et al., 2023;Matthews et al., 2021).Here we calculate the impact of the ZEC on the RCB.

Pre-Industrial Control Simulations
Pre-industrial control simulations are initialized with greenhouse gas levels from the reference year 1850 (Eyring et al., 2016).This year is chosen as it precedes the commencement of large-scale industrialization.Pre-industrial control simulations illustrate the climate's natural variability without human interference.The global mean temperature and global mean precipitation for all pre-industrial control simulations can be seen in Figures S3 and S4 of Supporting Information S1.In this study, we have used pre-industrial control simulations with constant atmospheric CO 2 concentrations.

ZEC Calculation
The temperature and precipitation changes after the cessation of emissions are compared with the 20-year average of the 1%CO2 run at the point at which the A1 experiment branches from this experiment.The ZEC 25 and ZEC 50 values are then calculated as the 20-year average centered on the years 25 and 50, respectively.To compare these values with the range in the pre-industrial control, we calculate the difference between two 20-year average periods with end and start points separated by five years (middle points separated by 25 years) corresponding to ZEC 25 , and 30 years (middle points separated by 50 years) corresponding to ZEC 50 .This calculation is done with maximally overlapping sets, generating the number of samples equal to the length of the piControl minus 44 years for ZEC25 and 69 for ZEC50 due to diminishing data availability at endpoints (the length of the pi-Control simulations can be found in Table S2 of Supporting Information S1).These values can then be used to create a distribution, as shown in Figure 1 Values are considered outside the range of natural variability if they are below the first percentile or above the 99th percentile of the pre-industrial control anomalies.The global mean temperature and precipitation calculation is weighted by grid-cell area and the precipitation is represented with units of millimeters per day (mm/day).

Carbon Budget Calculation
The carbon budget is defined as the allowable remaining emissions for keeping the peak global average temperature below a certain global warming level.The allowable emission can be derived using a known linear relationship between warming and cumulative CO 2 emissions (Matthews et al., 2009).This relationship is commonly estimated using the transient response to cumulative emissions (TCRE).The transient response to cumulative emissions is the change in the global mean temperature (ΔT ) to increasing cumulative emissions (E).

TCRE = ΔT E
In this study, we have used a common method of estimating this parameter from the 1% CO 2 simulations by taking the 20-year average once 1,000 PgC has been emitted.The 20-year average can then be converted to an anomaly by subtracting the average of the pre-industrial control simulation (all years for each model).This gives the sensitivity of the climate to increasing cumulative emissions (˚C/GtCO 2 ), which can then be used to calculate the remaining emissions for a given peak global warming.In this study, we use this relationship to infer the associated emissions (Budget C ) with a given ZEC 50 where, 3,670, is the conversion factor from 1,000 PgC to GtCO 2 .

Results
The temperature changes after cessation of emissions are outside the range of natural variability in most models (Figure 1 left column) (see Table S1 in Supporting Information S1 for a summary of how the ZEC values compare with natural variability).Only ACCESS-ESM1-5 and CanESM5 predict little global average temperature changes, which are non-significant.Two models simulate warming after net zero (ACCESS-ESM1-5 and UKESM1-0-LL); with, the warming only significant compared to natural variability in the UKESM1-0-LL model for both ZEC timeframes.The remainder of the models cool after emission cessation.The cooling is only significant after 25 years in two models (MPI-ESM1-2-LR and NorEMS2-LR).After 50 years, the cooling is significant in six of the nine models (CESM2, GFDL-ESM4, GISS-E2-1-G-CC and MIROC-E2SL now also have cooling significant relative to natural variability).As the cooling is not significant after 25 years in several models, but is after 50 years, this suggests that the post-net zero changes occur on slow multi-decadal timescales, and have not yet emerged compared to a natural unforced climate in several of the models after 25 years (CESM2, GFDL-ESM4, GISS-E2-1-G-CC, MIROC-E2SL).Thus, the response most commonly simulated after the cessation of emissions is a cooling that is significant compared to natural variability, based on the model simulations examined here.
We next analyze the post-abrupt CO 2 emissions cessations effect on precipitation (Figure 1 right column).The simulations show significant increases in precipitation in only a select few models after 25 years (ACCESS, two CanESM5 simulations, and UKESM).By the year 50, precipitation changes are still positive for all models, excluding MPI-ESM1-2-LR; however, these changes are not significant for most models.As atmospheric CO 2 concentrations decline, the reverse of what is described in Andrews et al. (2009) may occur.The decreasing CO 2 will decrease the positive radiative component at the top of the atmosphere greater than the surface, increasing latent heat flux, resulting in increased evaporation, and more rainfall.This effect may counteract or add to the Clausius Clapeyron effect, which results in air temperature increases as surface temperatures increase.The only two models that see significant increases in precipitation after 50 years (ACCESS-ESM1-5 and UKESM1-0-LL), are both models that see increased global mean surface temperature (although ACCESS-ESM1-5 warming is nonsignificant).The decreased CO 2 and increased surface temperature, thus both act to increase precipitation.For all other models, the precipitation change is significant at year 25 compared to natural variability likely due to the inertia of global average precipitation that persists even after the global temperature is no longer increasing (Mitchell et al., 2016).However, by the year 50, as CO 2 concentrations are still decreasing in models, and surface temperature has decreased, these effects are likely to have counteracted, resulting in non-significant precipitation changes.Additionally, the variability of precipitation is larger compared to temperature, and thus, the forced changes in precipitation must be greater in order to be significant relative to natural variability (Milinski et al., 2020).Our findings show that whilst the temperature changes after zero CO 2 emissions have a discernible influence on short-term precipitation patterns, longer-term changes in precipitation do not exceed the bounds of natural variability.These post-net zero temperature changes may have implications for the RCB.The multi-model average RCB for 1.5˚C, based upon current warming of 1.2˚C (globalwarmingindex.org --Tracking progress to a safe climate, 2023) is 535 GtCO 2 (see Table S2 in Supporting Information S1 for details on each model).This value is larger than other reported values (Friedlingstein et al., 2022;GCP, 2023;IPCC, 2021;Matthews et al., 2021) however, is still within the range of possible values considering the large spread in the RCB and the limited number of models available for this analysis.Should the climate exhibit a significant warming post-cessation of emissions, a scenario identified here solely in the UKESM1-0-LL model, this results in a reduction of the carbon budget by 518 GtCO 2 .Consequently, with the RCB estimate being 525 GtCO 2 from this subset of models, this suggests that even if CO 2 emissions ceased, the climate could still approach the 1.5˚C warming level.In models that predict a significant cooling after zero emissions, an average temperature reduction of 0.21˚C is simulated.This cooling could potentially increase the RCB for limiting global warming to 1.5˚C by an additional 513 GtCO 2 , representing a substantial 95% boost to the available carbon budget.However, this extra carbon budget is only applicable under certain conditions.If CO 2 emissions were to be abruptly halted, the cooling effect would have no impact on the RCB.This is because the RCB by definition relates emissions remaining to the peak global mean temperature, and a negative ZEC would not reduce the peak temperature reached in the ZECMIP esm-1pct-brch-1000PgC experiment (see Figure S1 in Supporting Information S1).Rather, the negative ZEC will influence the climate evolution after the global mean temperature has peaked, returning the global mean temperature to stabilization at a lower level.This will still potentially reduce some of the negative impacts from climate change, as the negative ZEC would cause the global mean temperature to stabilize at a lower level under a continued net zero emissions pathway.In reality, the reduction in emissions is likely to be a gradual and phased process.Some of the ZEC may be realized before reaching net-zero emissions (Koven et al., 2023), impacting the RCB.It is important to emphasize that the 513 GtCO 2 represents a maximum potential increase for the RCB for 1.5˚C, and the actual impact of ZEC on the RCB can vary depending on the emissions reduction trajectory and the cumulative emissions (Allen et al., 2022).There is currently a lack of available simulations to precisely determine how much of the ZEC effect will be realised with different pathways to net-zero emissions.

Discussion
Previous modeling studies have found that the changes in global average temperature after the immediate cessation of CO 2 emissions are small (global average range of 0.36˚C to 0.29˚C) (MacDougall et al., 2020), but it has not been investigated if this is only due to internal variability.The IPCC assessed that the changes after zero emissions are "[…] small compared with natural variability in Global Surface Air Temperature" (IPCC, 2021) but this was after averaging across the ensemble before making that comparison.Here, by comparing each model with its own natural variability we find that for most models in the ZECMIP esm-1pctbrch-1000PgC (A1) experimental ensemble, the changes in global average temperature are significant.The response is varied between models; however, the most commonly simulated response is a significant cooling compared to natural variability-six out of nine models cool significantly 50 years after the cessation of CO 2 emissions.However, the cooling after emissions are abruptly ceased does not affect the RCB for 1.5˚C, but may return global average temperatures to a lower global warming level, reducing impacts from climate change.Warming is less likely after the cessation of emissions, with only one model (UKESM1-0-LL) simulating significant warming compared to natural variability.This low-likelihood but high-impact outcome means we must still plan ambitious emissions reductions to avoid the possibility of exceeding the Paris Agreement warming levels following emissions cessation.Our study underlines the need for improved understanding and constraints on TCRE and ZEC, and we recommend larger ensembles of simulations are needed to enable more robust quantification of the magnitude of ZEC in the face of climate variability, as well as less idealized experiments to explore implications of net-zero.
The temperature and precipitation changes after net zero across climate models were found to be diverse, but the small ensemble of model simulations prevents robust probabilistic analysis of different outcomes under net-zero simulations.Given humanity's goal of reaching net zero emissions to prevent further global warming, it is imperative that the climate science community makes rapid advances in understanding the committed climate changes following net zero.Decision-makers need more information about the response of the climate to net zero emissions to plan accordingly.In the meantime, it may be prudent to account for the possibility of post-net zero emissions global warming and to take more rapid action to reduce greenhouse gas emissions as a result.AB, AK, JRB, and LC were supported by the Australian Research Council Grant CE170100023.ADK, JRB, TZ, and MM were supported by the Australian Government's National Environment Science Program.The research was completed with the assistance of resources from the National Computational Infrastructure, which is supported by the Australian Government.

•
Global mean surface temperature changes after abrupt CO 2 emissions cessation are significant compared to natural variability in most models • Global mean precipitation changes after 50 yrs are only significant compared to natural variability in models that warm • The uncertain temperature changes post-net zero have implications for the remaining carbon budget Supporting Information: Supporting Information may be found in the online version of this article.

Figure 1 .
Figure 1.The distribution of differences in 20-year average global mean temperatures (left column) and global mean precipitation (right column) separated by 5 years (blue) and 30 years (purple) for each model (rows), compared with the ZEC 25 (blue line(s)) and ZEC 50 (purple lines(s)).Precipitation is not available for GFDL-ESM4 and NorESM2-LM models.
Open access publishing facilitated by The University of Melbourne, as part of the Wiley-The University of Melbourne agreement via the Council of Australian University Librarians.CDJ was supported by the Joint UK BEIS/Defra Met Office Hadley Centre Climate Programme (GA01101) and the European Union's Horizon 2020 research and innovation programme under Grant Agreement No 101003536 (ESM2025-ESMSs for the Future).Open access publishing facilitated by The University of Melbourne, as part of the Wiley -The University of Melbourne agreement via the Council of Australian University Librarians.