Global fossil fuel reduction pathways under different climate mitigation strategies and ambitions

The mitigation scenarios database of the Intergovernmental Panel on Climate Change’s Sixth Assessment Report is an important resource for informing policymaking on energy transitions. However, there is a large variety of models, scenario designs, and resulting outputs. Here we analyse the scenarios consistent with limiting warming to 2 °C or below regarding the speed, trajectory, and feasibility of different fossil fuel reduction pathways. In scenarios limiting warming to 1.5 °C with no or limited overshoot, global coal, oil, and natural gas supply (intended for all uses) decline on average by 95%, 62%, and 42%, respectively, from 2020 to 2050, but the long-term role of gas is highly variable. Higher-gas pathways are enabled by higher carbon capture and storage (CCS) and carbon dioxide removal (CDR), but are likely associated with inadequate model representation of regional CO2 storage capacity and technology adoption, diffusion, and path-dependencies. If CDR is constrained by limits derived from expert consensus, the respective modelled coal, oil, and gas reductions become 99%, 70%, and 84%. Our findings suggest the need to adopt unambiguous near- and long-term reduction benchmarks in coal, oil, and gas production and use alongside other climate mitigation targets.


SUPPLEMENTARY Discussion
Additional results and discussions for "Characteristics of scenarios with different roles for gas" section C1 scenarios Figure 4 shows that the "rebound" C1 scenarios are generally associated with one or more of the following features: (1) much higher fossil fuel use coupled to CCS, including for electricity generation from gas; (2) higher CDR via negative AFOLU emissions and DACCS; (3) lower carbon prices, especially after 2050; (4) higher gas demand in the transportation sector; (5) higher primary energy supply from nuclear; (6) lower capacity additions for electricity generation from solar and wind; and (7) lower capital costs for electricity generation from coal-fired power plants (without CCS) but higher from offshore wind.(For the last two features, we interpret the relevant variables with caution give the relatively limited reporting.)Conversely, the "fast decline" scenarios are typically associated with much less reliance on fossil CCS, higher carbon prices, and sustained increases in renewables integration from 2020 onwards.The "slow decline" scenarios share similar characteristics, but generally have higher CDR via BECCS and less extremely high carbon prices seen in some of the "fast decline" scenarios.The reduction pathways for gaspowered electricity generation without CCS are, however, similar between all three clusters, showing an almost complete phase-out by around 2040.
Compared to other scenario projects, the ENGAGE project 1 has the largest number of scenarios represented in each of the C1-C3 categories (Figure S3).Consequently, we pay special attention as to whether scenarios from this project all show relatively higher or lower gas supply values compared to scenarios from other projects within the AR6 scenario ensemble.In each of the C1-C3 categories, we find that scenarios from the ENGAGE project appear in all three gas clusters.The largest number of ENGAGE scenarios is found in the "rebound" cluster in C1, and in the "fast decline" or "decline" clusters in C2 and C3, respectively.

C2 scenarios
The C2 clusters display similar characteristics to those of the C1.Supplementary Figure 13 shows that the "rebound" C2 scenarios are generally associated with one or more of the following features: (1) much higher fossil fuel use coupled to CCS, especially for electricity generation from gas; (2) higher CDR via negative AFOLU emissions and DACCS; (3) lower carbon prices, especially after 2060; (4) higher hydrogen production from gas ; (5) lower capacity additions for electricity generation from solar and wind; and (g) higher capital costs for electricity generation from offshore wind.(For the last two features, we interpret the relevant variables with caution give the relatively limited reporting.)Conversely, the "fast decline" scenarios are typically associated with much less reliance on fossil CCS, higher carbon prices, and sustained increases in renewables integration from 2020 onwards.The "slow decline" scenarios share similar characteristics, but generally have higher CDR via BECCS and less extremely high carbon prices seen in some of the "fast decline" scenarios.The reduction pathways for gas-powered electricity generation without CCS are, however, similar between all three clusters, showing an almost complete phase-out by around 2050.
As shown in Supplementary Figures 15-16, all 10 C2 scenarios from the GEM-E3_V2021 model (associated with the ENGAGE project) are grouped into the "rebound" cluster.All C2 scenarios from the COFFEE 1.1 and IMAGE 3.2 models, and 25 out of 43 scenarios from the MESSAGE model family, are found in the "slow decline" cluster.All scenarios from the AIM/CGE 2.2 and POLES ENGAGE models and from the WITCH model family, as well as the majority of scenarios from the REMIND model family, are grouped into the "fast decline" cluster.

C3 scenarios
Supplementary Figure 14 shows that the "decline" C3 scenarios are generally associated with one or more of the following features: (1) lower reliance on fossil CCS; (2) lower CDR via AFOLU and DACCS; (3) higher carbon prices after around 2060; (4) higher primary energy supply from renewables, including sustained increases in renewable electricity capacity additions; and (5) higher capital costs for electricity generation from coal-CCS power plants but lower from offshore wind.(For the last two features, we interpret the relevant variables with caution give the relatively limited reporting.)Compared to the "decline" cluster, the "rebound" scenarios rely on higher fossil CCS and CDR via AFOLU and DACCS, have higher gas demand for buildings, and lower capacity additions from solar (but not wind) for electricity generation.Meanwhile, the "increase" cluster is associated with higher levels of fossil CCS, gas demand for transportation, higher CDR via AFOLU and DACCS, and lower capacity additions from renewables.Both the "rebound" and "increase" scenarios generally see carbon prices plateauing after around 2070.As in the C1 and C2 scenarios, the reduction pathways for gas-powered electricity generation without CCS are similar between all three C3 clusters, showing an almost complete phase-out by around 2050.
As shown in Supplementary Figures 15-16, around half of the scenarios in the "rebound" cluster are from the WITCH model family (19 out of 39).All C3 scenarios from the GEM-E3 and IMAGE 3.2 models, all but one scenario from the GCAM 5.3 model, and the majority of scenarios from the MESSAGE model family are grouped into the "increase" cluster.The majority of scenarios from the REMIND, TIAM, POLES, AIM, and COFFEE model families are grouped into the "decline" cluster, with 77 out of 164 scenarios (47%) in this cluster coming from the REMIND model family.

Additional results and discussion for the "Sensitivity of fossil fuel reduction pathways to different mitigation strategies and uncertainties in CDR potential" section
Here we describe the fossil fuel reduction pathways shown in Figure 5 for the five illustrative mitigation pathways (IMPs) of the AR6 in detail.
The IPCC AR6 assessment of all submitted mitigation scenarios identified five different "illustrative mitigation pathways" (IMPs) that reflect different prominent mitigation strategies for reaching a given temperature outcome.There are three IMPs in the C1 category: "IMP-LD" -strong emphasis on energy demand reductions; "IMP-Ren" -heavy reliance on renewables; "IMP-SP" -mitigation in the context of broader sustainable development."IMP-Neg", which relies on extensive CDR in the energy and the industry sectors, is a C2 scenario."IMP-GS", which entails a less rapid and gradual strengthening of near-term mitigation actions, is a C3 scenario.
In the C1 category, the IMP-LD scenario meets the 1.5°C threshold with no or limited overshoot through much faster reductions in oil and gasand much higher carbon prices -than almost all other scenarios (Supplementary Fig. 20).For oil, this is accomplished through rapid and dramatic transitions to shared vehicle fleets, flexible transit systems, and rapid vehicle electrification; as well as reductions in freight volumes due to longer-lasting and more material-efficient goods.For gas, this is accomplished through extensive end-use efficiency improvements (e.g., through building retrofits), plus preferring renewable power to gas with CCS.In fact, the IMP-LD scenario deploys no CCS or BECCS for normative reasons, informed by concern over innovation failure, investment risks, and public opposition (though it does rely extensively on CDR via the AFOLU sector) 5 .
By contrast, the IMP-Ren scenario lacks as much focus on shared and flexible transit or material and end-use efficiency, but goes even harder on electrification, as powered by renewable electricity 6 .Lastly, IMP-SP phases out all fossil fuels almost completely by the end of the century and with relatively limited CDR and CCS reliance, driven in part by the "substantially reduce[d] detrimental effects of outdoor air pollution on public health", as well as by additional ecosystem and human health constraints that limit certain mitigation options related to bioenergy and land availability 7 .Still, IMP-SP's focus on availability of and access to modern energy services (e.g., electric and LPG cook stoves) in developing regions causes it to reduce oil and gas demand slightly slower than do scenarios without this explicit focus on improving livelihoods and limiting cost impacts on low-income households.
In the C2 category, the IMP-Neg scenario allows dramatically more room for coal than most scenarios but, over time, less room for gas (Figure 5).The continued use of coal in this scenario is driven by ongoing use in industry (and to a lesser extent for coal-powered electricity generation coupled to CCS), which is enabled by extensive CDR reliance: up to 8 GtCO2 of BECCS, 1-2 GtCO2 of sequestration through the AFOLU sector, and up to 6 GtCO2 by other methods annually in the latter half of the century (Supplementary Fig. 21).This extensive reliance on long-term CDR is typical of C2 scenarios, in which net negative CO2 emissions are needed to compensate for emissions in the first half of the century and bring down temperatures after the peak 8 .
In the C3 category that limits warming to below 2°C, the IMP-GS scenario is generally well within the interquartile range of scenario outcomes, with the exception that it reduces gas by much more in mid-century than most other scenarios (Figure 5).This is largely because this scenario builds much more renewable power (solar and wind) and electricity storage than do most C3 scenarios (Supplementary Fig. 22), which enables it to phase down gas power more rapidly in the near-term.(Over time, gas supply rebounds slightly due to demand in the buildings and industry sectors, with overall emissions primarily being offset by DACCS.) For our CDR sensitivity analysis, we find that 4 out of 26 C1, 0 out of 15 C2, and 6 out of 62 C3 scenarios that report the relevant variables do not exceed the limits based on Grant et al.The subsets of "model"--"scenario" (and their project study) that meet these criteria are as follows:

SUPPLEMENTARY TABLES
Supplementary Table 1.The years during which global coal, oil, and gas supply peak* and the percentage changes in supply relative to 2020 in the C1-C3 pathways shown in Figure 5. Percent changes shown are rounded to the nearest integer.(*Note: Data are available at 5-year intervals; all C2 scenarios exceed the CDR limits of our sensitivity analysis.)-14.To allow easy visualization, the projects with a relatively small number of scenarios (Supplementary Fig. 3) have been grouped into the "Others" category here.C1: "fast decline" cluster C1: "slow decline" cluster C1: "rebound" cluster  C3: "decline" cluster C3: "rebound" cluster C3: "increase" cluster Supplementary Fig. 23.Non-energy use of coal, oil, and gas ("Final Energy|Non-Energy Use|xx", in exajoules, EJ) as modelled by the IPCC AR6 mitigation scenarios consistent with limiting warming to 1.5°C or 2°C.In subplots a-i, the 2010-2100 annual timeseries of the following are plotted at each 5-year interval within a given temperature category (C1-C3): individual pathways (light lines); median values (dark lines); and the 25 th -75 th percentiles (shaded ranges).Subplots j-l show the boxplot distributions of the 2020-2100 cumulative values across the scenario ensemble within each temperature category.The horizontal center line depicts the median, the box spans the interquartile range (IQR) between the 25 th percentile (Q1) and 75 th percentile (Q3), the lower whisker represents the minimum value or Q1 -1.5 x IQR (whichever is larger), and the upper whisker represents the maximum value or Q3 + 1.5 x IQR (whichever is smaller).Supplementary Fig. 24.Non-energy industry use of fossil fuel-and bio-derived solids, liquids, and gases ("Final Energy|Industry|xx", in exajoules, EJ) as modelled by the IPCC AR6 mitigation scenarios consistent with limiting warming to 1.5°C or 2°C.In subplots a-i, the 2010-2100 annual timeseries of the following are plotted at each 5-year interval within a given temperature category (C1-C3): individual pathways (light lines); median values (dark lines); and the 25 th -75 th percentiles (shaded ranges).Subplots j-l show the boxplot distributions of the 2020-2100 cumulative values across the scenario ensemble within each temperature category.The horizontal center line depicts the median, the box spans the interquartile range (IQR) between the 25 th percentile (Q1) and 75 th percentile (Q3), the lower whisker represents the minimum value or Q1 -1.5 x IQR (whichever is larger), and the upper whisker represents the maximum value or Q3 + 1.5 x IQR (whichever is smaller).

. 15 .
Number of scenarios from different model families in each of the gas clusters in Figures 4 and S13-S14.
Number of scenarios from different major scenario projects in each of the gas clusters in Figure4and Supplementary Figures 13

Table 2 .
As in Supplementary Table1but for primary energy supply from coal, oil, gas without CCS.

Table 4 .
Gas-cluster classification and CCS-related assumptions of the model families underlying the C1 scenarios.