Chapter 3 takes a long-term perspective on climate change mitigation pathways. Its focus is on the implications of long-term targets for the required short- and medium-term system changes and associated greenhouse gas (GHG) emissions. This focus dictates a more global view and on issues related to path-dependency and up-scaling of mitigation options necessary to achieve different emissions trajectories, including particularly deep mitigation pathways that require rapid and fundamental changes.

Stabilising global average-temperature change requires reducing CO2 emissions to net zero. Thus, a central cross-cutting topic within the chapter is the timing of reaching net zero CO2 emissions and how a ‘balance between anthropogenic emissions by sources and removals by sinks’ could be achieved across time and space. This includes particularly the increasing body of literature since the IPCC Special Report on Global Warming of 1.5°C (SR1.5) which focuses on net zero CO2 emissions pathways that avoid temperature overshoot and hence do not rely on net negative CO2 emissions. The chapter conducts a systematic assessment of the associated economic costs as well as the benefits of mitigation for other societal objectives, such as the Sustainable Development Goals (SDGs). In addition, the chapter builds on SR1.5 and introduces a new conceptual framing for the assessment of possible social, economic, technical, political, and geophysical ‘feasibility’ concerns of alternative pathways, including the enabling conditions that would need to fall into place so that stringent climate goals become attainable.

The structure of the chapter is as follows: Section 3.2 introduces different types of mitigation pathways as well as the available modelling. Section 3.3 explores different emissions trajectories given socio-economic uncertainties and consistent with different long-term climate outcomes. A central element in this section is the systematic categorisation of the scenario space according to key characteristics of the mitigation pathways (including e.g., global average-temperature change, socio-economic development, technology assumptions, etc.). In addition, the section introduces selected Illustrative Mitigation Pathways (IMPs) that are used across the whole report. Section 3.4 conducts a sectoral analysis of the mitigation pathways, assessing the pace and direction of systems changes across sectors. Among others, this section aims at the integration of the sectoral information across AR6 WGIII chapters through a comparative assessment of the sectoral dynamics in economy-wide systems models compared to the insights from bottom-up sectoral models (from Chapters 6 to 11). Section 3.5 focuses on the required timing of mitigation actions, and the implication of near-term choices for the attainability of a range of long-term climate goals. After having explored the underlying systems transitions and the required timing of the mitigation actions, Section 3.6 assesses the economic implications, mitigation costs and benefits; and Section 3.7 assesses related co-benefits, synergies, and possible trade-offs for sustainable development and other societal (non-climate) objectives. Section 3.8 assumes a central role in the chapter and introduces a multidimensional feasibility metric that permits the evaluation of mitigation pathways across a range of feasibility concerns. Finally, methods of the assessment and knowledge gaps are discussed in Section 3.9, followed by Frequently Asked Questions (FAQs).

Chapter 3 is linked to many other chapters in the report. The most important connections exist with Chapter 4 on mitigation and development pathways in the near to mid-term; with the sectoral chapters (Chapters 6–11); with the chapters dealing with cross-cutting issues (Chapters 12 and 17, e.g., feasibility); and finally also with AR6 WGI and WGII.

Within the overall framing of the AR6 report, Chapter 3 and Chapter 4 provide important complementary views of the required systems transitions across different temporal and spatial scales. While Chapter 3 focuses on the questions concerning the implications of the long-term objectives for the medium-to-near-term transformations, Chapter 4 comes from the other direction, and focuses on current near-term trends and policies (such as the Nationally Determined Contributions – NDCs) and their consequences with regards to GHG emissions. The latter chapter naturally focuses much more on the regional and national dimensions, and the heterogeneity of current and planned policies. Bringing together the information from these two chapters enables the assessment of whether current and planned actions are consistent with the required systems changes for the long-term objectives of the Paris Agreement.

Important other linkages comprise the collaboration with the ‘sectoral’ Chapters 6 to 11 to provide an integrated cross-sectoral perspective. This information (including information also from the sectoral chapters) is taken up ultimately also by Chapter 5 on demand/services and Chapter 12 for a further assessment of sectoral potential and costs.

Linkages to other chapters exist also on the topic of feasibility, which are informed by the policy, the sectoral and the demand chapters, the technology and finance chapters, as well as Chapter 4 on national circumstances.

Close collaboration with WGI permitted the use of AR6-calibrated emulators, which assure full consistency across the different working groups. Linkages to WGII concern the assessment of macroeconomic benefits of avoided impacts that are put into the context of mitigation costs as well as co-benefits and trade-offs for sustainable development.

The assessment of mitigation pathways explores a wide scenario space from the literature within which seven Illustrative Pathways (IPs) are explored. The overall process is indicated in Figure 3.5a.

For a comprehensive assessment, a large ensemble of scenarios is collected and made available through an interactive AR6 Scenarios Database 4 . The collected information is shared across the chapters of AR6 and includes more than 3000 different pathways from a diverse set of studies. After an initial screening and quality control, scenarios were further vetted to assess if they sufficiently represented historical trends (Annex III.II.3.1). Subsequently, the climate consequences of each scenario were assessed using the climate emulator (leading to further classification). The assessment in Chapter 3 is, however, not limited to the scenarios from the database, and wherever necessary other literature sources are also assessed in order to bring together multiple lines of evidence.

In parallel, based on the overall AR6 assessment, seven illustrative pathways (IP) were defined representing critical mitigation strategies discussed in the assessment. The seven pathways are composed of two sets: (i) one set of five Illustrative Mitigation Pathways (IMPs) and (ii) one set of two reference pathways illustrative for high emissions. The IMPs are on the one hand representative of the scenario spac but also help to communicate archetypes of distinctly different systems transformations and related policy choices. Subsequently, seven scenarios were selected from the full database that fitted these storylines of each IP best. For these scenarios more strict vetting criteria were applied. The selection was done by first applying specific filters based on the storyline followed by a final selection (Box 3.1 and Figure 3.5a).

Scenario and emission pathways are used to explore possible long-term trajectories, the effectiveness of possible mitigation strategies, and to help understand key uncertainties about the future. Ascenario is an integrated description of a possible future of the human–environment system (Clarke et al. 2014), and could be a qualitative narrative, quantitative projection, or both. Scenarios typically capture interactions and processes that change key driving forces such as population, GDP, technology, lifestyles, and policy, and the consequences on energy use, land use, and emissions. Scenarios are not predictions or forecasts. An emission pathway is a modelled trajectory of anthropogenic emissions (Rogelj et al. 2018 a) and, therefore, a part of a scenario.

There is no unique or preferred method to develop scenarios, and future pathways can be developed from diverse methods, depending on user needs and research questions (Turnheim et al. 2015; Trutnevyte et al. 2019a; Hirt et al. 2020). The most comprehensive scenarios in the literature are qualitative narratives that are translated into quantitative pathways using models (Clarke et al. 2014; Rogelj et al. 2018 a). Schematic or illustrative pathways can also be used to communicate specific features of more complex scenarios (Allen et al. 2018). Simplified models can be used to explain the mechanisms operating in more complex models (e.g., Emmerling et al. 2019). Ultimately, a diversity of scenario and modelling approaches can lead to more robust findings (Schinko et al. 2017; Gambhir et al. 2019).

Integrated Assessment Models (IAMs) are critical for understanding the implications of long-term climate objectives for the required near-term transition. For doing so, an integrated systems perspective including the representation of all sectors and GHGs is necessary. IAMs are used to explore the response of complex systems in a formal and consistent framework. They cover a broad range of modelling frameworks (Keppo et al. 2021). Given the complexity of the systems under investigation, IAMs necessarily make simplifying assumptions and therefore results need to be interpreted in the context of these assumptions. IAMs can range from economic models that consider only carbon dioxide emissions through to detailed process-based representations of the global energy system, covering separate regions and sectors (such as energy, transport, and land use), all GHG emissions and air pollutants, interactions with land and water, and a reduced representation of the climate system. IAMs are generally driven by economics and can have a variety of characteristics such as partial-, general- or non-equilibrium; myopic or perfect foresight; be based on optimisation or simulation; have exogenous or endogenous technological change amongst many other characteristics. IAMs take as input socio-economic and technical variables and parameters to represent various systems. There is no unique way to integrate this knowledge into a model, and due to their complexity, various simplifications and omissions are made for tractability. IAMs therefore have various advantages and disadvantages which need to be weighed up when interpreting IAM outcomes. Annex III.I contains an overview of the different types of models and their key characteristics.

Most IAMs are necessarily broad as they capture long-term dynamics. IAMs are strong in showing the key characteristics of emission pathways and are most suited to questions related to short- versus long-term trade-offs, key interactions with non-climate objectives, long-term energy and land-use characteristics, and implications of different overarching technological and policy choices (Clarke et al. 2014; Rogelj et al. 2018 a). While some IAMs have a high level of regional and sectoral detail, for questions that require higher levels of granularity (e.g., local policy implementation) specific region and sector models may be better suited. Utility of the IAM pathways increases when the quantitative results are contextualized through qualitative narratives or other additional types of knowledge to provide deeper insights (Geels et al. 2016a; Weyant 2017; Gambhir et al. 2019).

IAMs have a long history in addressing environmental problems, particularly in the IPCC assessment process (van Beek et al. 2020). Many policy discussions have been guided by IAM-based quantifications, such as the required emission reduction rates, net zero years, or technology deployment rates required to meet certain climate outcomes. This has led to the discussion about whether IAM scenarios have become performative, meaning that they act upon, transform or bring into being the scenarios they describe (Beck and Mahony 2017, 2018). Transparency of underlying data and methods is critical for scenario users to understand what drives different scenario results (Robertson 2020). A number of community activities have thus focused on the provision of transparent and publicly accessible databases of both input and output data (Riahi et al. 2012; Huppmann et al. 2018; Krey et al. 2019; Daioglou et al. 2020), as well as the provision of open-source code, and increased documentation (Annex III.I.9). Transparency is needed to reveal conditionality of results on specific choices in terms of assumptions (e.g., discount rates) and model architecture. More detailed explanations of underlying model dynamics would be critical to increase the understanding of what drives results (Bistline et al. 2020; Butnar et al. 2020; Robertson 2020).

Mitigation scenarios developed for a long-term climate constraint typically focus on cost-effective mitigation action towards a long-term climate goal. Results from IAM as well as sectoral models depend on model structure (Mercure et al. 2019), economic assumptions (Emmerling et al. 2019), technology assumptions (Pye et al. 2018), climate/emissions target formulation (Johansson et al. 2020), and the extent to which pre-existing market distortions are considered (Guivarch et al. 2011). The vast majority of IAM pathways do not consider climate impacts (Schultes et al. 2021). Equity hinges upon ethical and normative choices. As most IAM pathways follow the cost-effectiveness approach, they do not make any additional equity assumptions. Notable exceptions include Tavoni et al. (2015), Pan et al. (2017), van den Berg et al. (2020), and Bauer et al. (2020). Regional IAM results therefore need to be assessed with care, considering that emissions reductions are happening where it is most cost-effective, which needs to be separated from who is ultimately paying for the mitigation costs. Cost-effective pathways can provide a useful benchmark, but may not reflect real-world developments (Calvin et al. 2014a; Trutnevyte 2016). Different modelling frameworks may lead to different outcomes (Mercure et al. 2019). Recent studies have shown that other desirable outcomes can evolve with only minor deviations from cost-effective pathways (Bauer et al. 2020; Neumann and Brown 2021). IAM and sectoral models represent social, political, and institutional factors only in a rudimentary way. This assessment is thus relying on new methods for the expost assessment of feasibility concerns (Jewell and Cherp 2020; Brutschin et al. 2021). A literature is emerging that recognises and reflects on the diversity and strengths/weaknesses of model-based scenario analysis (Keppo et al. 2021).

The climate constraint implementation can have a meaningful impact on model results. The literature so far includes many temperature overshoot scenarios with heavy reliance on long-term CDR and net negative CO2 emissions to bring back temperatures after the peak (Rogelj et al. 2019b; Johansson et al. 2020). New approaches have been developed to avoid temperature overshoot. The new generation of scenarios show that CDR is important beyond its ability to reduce temperature, but is essential also for offsetting residual emissions to reach net zero CO2 emissions (Rogelj et al. 2019b; Johansson et al. 2020; Riahi et al. 2021; Strefler et al. 2021b).

Many factors influence the deployment of technologies in the IAMs. Since AR5, there has been fervent debate on the large-scale deployment of bioenergy with carbon capture and storage (BECCS) in scenarios (Fuss et al. 2014; Geden 2015; Anderson and Peters 2016; Smith et al. 2016; van Vuuren et al. 2017; Galik 2020; Köberle 2019). Hence, many recent studies explore mitigation pathways with limited BECCS deployment (Grubler et al. 2018; van Vuuren et al. 2019; Riahi et al. 2021; Soergel et al. 2021a). While some have argued that technology diffusion in IAMs occurs too rapidly (Gambhir et al. 2019), others argued that most models prefer large-scale solutions resulting in a relatively slow phase-out of fossil fuels (Carton 2019). While IAMs are particularly strong on supply-side representation, demand-side measures still lag in detail of representation despite progress since AR5 (Grubler et al. 2018; Lovins et al. 2019; van den Berg et al. 2019; O’Neill et al. 2020b; Hickel et al. 2021; Keyßer and Lenzen 2021). The discount rate has a significant impact on the balance between near-term and long-term mitigation. Lower discount rates <4% (than used in IAMs) may lead to more near-term emissions reductions – depending on the stringency of the target (Emmerling et al. 2019; Riahi et al. 2021). Models often use simplified policy assumptions (O’Neill et al. 2020b) which can affect the deployment of technologies (Sognnaes et al. 2021). Uncertainty in technologies can lead to more or less short-term mitigation (Grant et al. 2021; Bednar et al. 2021). There is also a recognition to put more emphasis on what drives the results of different IAMs (Gambhir et al. 2019) and suggestions to focus more on what is driving differences in result across IAMs (Nikas et al. 2021). As noted by Weyant (2017, p. 131), ‘IAms can provide very useful information, but this information needs to be carefully interpreted and integrated with other quantitative and qualitative inputs in the decision-making process.’

IPCC reports have often used voluntary submissions to a scenario database in its assessments. The database is an ensemble of opportunity, as there is not a well-designed statistical sampling of the hypothetical model or scenario space: the literature is unlikely to cover all possible models and scenarios, and not all scenarios in the literature are submitted to the database. Model intercomparisons are often the core of scenario databases assessed by the IPCC (Cointe et al. 2019; Nikas et al. 2021). Single-model studies may allow more detailed sensitivity analyses or address specific research questions. The scenarios that are organised within the scientific community are more likely to enter the assessment process via the scenario database (Cointe et al. 2019), while scenarios from different communities, in the emerging literature, or not structurally consistent with the database may be overlooked. Scenarios in the grey literature may not be assessed even though they may have greater weight in a policy context.

One notable development since AR5 is the Shared Socio-economic Pathways (SSPs), conceptually outlined in Moss et al. (2010) and subsequently developed to support integrated climate research across the IPCC Working Groups (O’Neill et al. 2014). Initially, a set of SSP narratives were developed, describing worlds with different challenges to mitigation and adaptation (O’Neill et al. 2017a): SSP1 (sustainability), SSP2 (middle of the road), SSP3 (regional rivalry), SSP4 (inequality) and SSP5 (rapid growth). The SSPs have now been quantified in terms of energy, land-use, and emission pathways (Riahi et al. 2017), for both no-climate-policy reference scenarios and mitigation scenarios that follow similar radiative-forcing pathways as the Representative Concentration Pathways (RCPs) assessed in AR5 WGI. Since then the SSPs have been successfully applied in thousands of studies (O’Neill et al. 2020b) including some critiques on the use and application of the SSP framework (Pielke and Ritchie 2021; Rosen 2021). A selection of the quantified SSPs are used prominently in AR6 WGI as they were the basis for most climate modelling since AR5 (O’Neill et al. 2016). Since 2014, when the first set of SSP data was made available, there has been a divergence between scenario and historic trends (Burgess et al. 2020). As a result, the SSPs require updating (O’Neill et al. 2020b). Most of the scenarios in the AR6 database are SSP-based and consider various updates compared to the first release (Riahi et al. 2017).

To facilitate this assessment, a large ensemble of scenarios has been collected and made available through an interactive AR6 WGIII scenario database. The collection of the scenario outputs is coordinated by Chapter 3 and expands upon the IPCC SR1.5 scenario explorer (Huppmann et al. 2018; Rogelj et al. 2018 a). A complementary database for national pathways has been established by Chapter 4. Annex III.II.3 contains full details on how the scenario database was compiled.

The AR6 scenario database contains 3131 scenarios (Figure 3.5a). After an initial screening and quality control, scenarios were further vetted to assess if they sufficiently represented historical trends (Annex III.II.3.1). Of the initial 2266 scenarios with global scope, 1686 scenarios passed the vetting process and are assessed in this chapter. The scenarios that did not pass the vetting are still available in the database. The vetted scenarios were from over 50 different model families, or over 100 when considering all versions of the same family (Figure 3.1). The scenarios originated from over 15 different model intercomparison projects, with around one-fifth originating from individual studies (Figure 3.2). Because of the uneven distribution of scenarios from different models and projects, uncorrected statistics from the database can be misleading.

Each scenario with sufficient data is given a temperature classification using climate model emulators. Three emulators were used in the assessment: FAIR (Smith et al. 2018), CICERO-SCM (Skeie et al. 2021), MAGICC (Meinshausen et al. 2020). Only the

Table 3.1 | Classification of emissions scenarios into warming levels using MAGICC

aThe Illustrative Mitigation Pathway ‘Neg’ has extensive use of carbon dioxide removal (CDR) in the AFOLU, energy and the industry sectors to achieve net negative emissions. Warming peaks around 2060 and declines to below 1.5°C (50% likelihood) shortly after 2100. Whilst technically classified as C3, it strongly exhibits the characteristics of C2 high-overshoot pathways, hence it has been placed in the C2 category. See Box SPM.1 for an introduction of the IPs and IMPs.

results of MAGICC are shown in this chapter as it adequately covers the range of outcomes. The emulators are calibrated against the behaviour of complex climate models and observation data, consistent with the outcomes of AR6 WGI (Cross-Chapter Box 7.1). The climate assessment is a three-step process of harmonisation, infilling and a probabilistic climate model emulator run (Annex III.II.2.5). Warming projections until the year 2100 were derived for 1574 scenarios, of which 1202 passed vetting, with the remaining scenarios having insufficient information (Figure 3.3 and Table 3.1). For scenarios that limit warming to 2°C or lower, the SR1.5 classification was adopted in AR6, with more disaggregation provided for higher warming levels (Table 3.1). These choices can be compared with the selection of common global warming levels (GWLs) of 1.5°C, 2°C, 3°C and 4°C to classify climate change impacts in the WGII assessment.

In addition to the temperature classification, each scenario is assigned to one of the following policy categories: (P0) diagnostic scenarios – 99 of 1686 vetted scenarios; (P1) scenarios with no globally coordinated policy (500) and (P1a) no climate mitigation efforts – 124, (P1b) current national mitigation efforts – 59, (P1c) Nationally Determined Contributions (NDCs) – 160, or (P1d) other non-standard assumptions – 153; (P2) globally coordinated climate policies with immediate action (634) and (P2a) without any transfer of emission permits – 435, (P2b) with transfers – 70; or (P2c) with additional policy assumptions – 55; (P3) globally coordinated climate policies with delayed (i.e., from 2030 onwards or after 2030) action (451), preceded by (P3a) no mitigation commitment or current national policies – 7, (P3b) NDCs – 426, (P3c) NDCs and additional policies – 18; (P4) cost-benefit analysis (CBA) – 2. The policy categories were identified using text pattern matching on the scenario metadata and calibrated on the best-known scenarios from model intercomparisons, with further validation against the related literature, reported emission and carbon price trajectories, and exchanges with modellers. If the information available is enough to qualify a policy category number but not sufficient for a subcategory, then only the number is retained (e.g., P2 instead of P2a/b/c). A suffix added after P0 further qualifies a diagnostic scenario as one of the other policy categories. To demonstrate the diversity of the scenarios, the vetted scenarios were classified into different categories along the dimensions of population, GDP, energy, and cumulative emissions (Figure 3.4). The number of scenarios in each category provides some insight into the current literature, but this does not indicate a higher probability of that category occurring in reality. For population, the majority of scenarios are consistent with the SSP2 ‘middle of the road’ category, with very few scenarios exploring the outer extremes. GDP has a slightly larger variation, but overall most scenarios are around the SSP2 socio-economic assumptions. The level of CCS and CDR is expected to change depending on the extent of mitigation, but there remains extensive use of both CDR and CCS in scenarios. CDR is dominated by bioenergy with CCS (BECCS) and sequestration on land, with relatively few scenarios using direct air capture with carbon storage (DACCS) and even less with enhanced weathering (EW) and other technologies (not shown). In terms of energy consumption, final energy has a much smaller range than primary energy as conversion losses are not included in final energy. Both mitigation and reference scenarios are shown, so there is a broad spread in different energy carriers represented in the database. Bioenergy has a number of scenarios at around 100 EJ, representing a constraint used in many model intercomparisons.

Successive IPCC Assessment Reports (ARs) have used scenarios to illustrate key characteristics of possible climate (policy) futures. In AR5 four RCPs made the basis of climate modelling in WGI and WGII, with WGIII assessing over 1000 scenarios spanning those RCPs (Clarke et al. 2014). Of the over 400 scenarios assessed in SR1.5, four scenarios were selected to highlight the trade-off between short-term emission reductions and long-term deployment of BECCS (Rogelj et al. 2018 a), referred to as ‘Illustrative Pathways’ (IPs). AR6 WGI and WGII rely on the scenarios selected for CMIP6, called ScenarioMIP (O’Neill et al. 2016), to assess warming levels. In addition to the full set of scenarios, AR6 WGIII also uses selected Illustrative Mitigation Pathways (IMPs).

In WGIII, IMPs were selected to denote the implications of different societal choices for the development of future emissions and associated transformations of main GHG-emitting sectors (Figure 3.5a and Box 3.1). The most important function of the IMPs is to illustrate key themes that form a common thread in the report, both with a storyline and a quantitative illustration. The storyline describes the key characteristics that define an IMP. The quantitative versions of the IMPs provide numerical values that are internally consistent and comparable across chapters of the report. The quantitative IMPs have been selected from the AR6 scenario database. No assessment of the likelihood of each IMP has been made.

The selected scenarios (IPs) are divided into two sets (Figures 3.5 and 3.6): two reference pathways illustrative of high emissions and five Illustrative Mitigation Pathways (IMPs). The narratives are explained in full in Annex III.II.2.4. The two reference pathways explore the consequences of current policies and pledges: Current Policies (CurPol) and Moderate Action (ModAct ). The CurPol pathway explores the consequences of continuing along the path of implemented climate policies in 2020 and only a gradual strengthening after that. The scenario illustrates the outcomes of many scenarios in the literature that project the trend from implemented policies until the end of 2020. The ModAct pathway explores the impact of implementing the Nationally Determined Contributions (NDCs) as formulated in 2020 and some further strengthening after that. In line with current literature, these two reference pathways lead to an increase in global mean temperature of more than 2°C (Section 3.3).

The Illustrative Mitigation Pathways (IMPs) properly explore different pathways consistent with meeting the long-term temperature goals of the Paris Agreement. They represent five different pathways that emerge from the overall assessment. The IMPs differ in terms of their focus, for example, placing greater emphasis on renewables (IMP-Ren), deployment of carbon dioxide removal that results in net negative global GHG emissions (IMP-Neg), and efficient resource use and shifts in consumption patterns, leading to low demand for resources, while ensuring a high level of services (IMP-LD). Other IMPs illustrate the implications of a less rapid introduction of mitigation measures followed by a subsequent gradual strengthening (IMP-GS), and how shifting global pathways towards sustainable development, including by reducing inequality, can lead to mitigation (IMP-SP) In the IMP framework, IMP-GS is consistent with limiting warming to 2°C (>67%) (C3), IMP-Neg shows a strategy that also limits warming to 2°C (>67%) but returns to nearly 1.5°C (>50%) by the end of the century (hence indicated as C2*). The other variants that can limit warming to 1.5°C (>50%) (C1) were selected. In addition to these IMPs, sensitivity cases that explore alternative warming levels (C3) for IMP-Neg and IMP-Ren are assessed (IMP-Neg-2.0 and IMP-Ren-2.0).

The IMPs are selected to have different mitigation strategies, which can be illustrated looking at the energy system and emission pathways (Figure 3.7 and Figure 3.8). The mitigation strategies show the different options in emission reduction (Figure 3.7). Each panel shows the key characteristics leading to total GHG emissions, consisting of residual (gross) emissions (fossil CO2 emissions, CO2 emissions from industrial processes, and non-CO2 emissions) and removals (net land-use change, bioenergy with carbon capture and storage – BECCS, and direct air carbon capture and storage – DACCS), in addition to avoided emissions through the use of carbon capture and storage on fossil fuels. The IMP-Neg and IMP-GS scenarios were shown to illustrate scenarios with a significant role of CDR. The energy supply (Figure 3.8) shows the phase-out of fossil fuels in the IMP-LD, IMP-Ren and IMP-SP cases, but a less substantial decrease in the IMP-Neg case. The IMP-GS case needs to make up its slow start by (i) rapid reductions mid-century and (ii) massive reliance on net negative emissions by the end of the century. The CurPol and ModAct cases both result in relatively high emissions, showing a slight increase and stabilisation compared to current emissions, respectively.

Greenhouse gas (GHG) emissions mainly originate from the use and transformation of energy, agriculture, land use (change) and industrial activities. The future development of these sources is influenced by trends in socio-economic development, including population, economic activity, technology, politics, lifestyles, and climate policy. Trends for these factors are not independent, and scenarios provide a consistent outlook for these factors together (Section 3.2). Marangoni et al. (2017) show that in projections, assumptions influencing energy intensity (e.g., structural change, lifestyle and efficiency) and economic growth are the most important determinants of future CO2 emissions from energy combustion. Other critical factors include technology assumptions, preferences, resource assumptions and policy (van Vuuren et al. 2008). As many of the factors are represented differently in specific models, the model itself is also an important factor – providing a reason for the importance of model diversity (Sognnaes et al. 2021). For land use, Stehfest et al. (2019) show that assumptions on population growth are more dominant given that variations in per capita consumption of food are smaller than for energy. Here, we only provide a brief overview of some key drivers. We focus first on so-called reference scenarios (without stringent climate policy) and look at mitigation scenarios in detail later. We use the SSPs to discuss trends in more detail. The SSPs were published in 2017, and by now, some elements will have to be updated (O’Neill et al. 2020b). Still, the ranges represent the full literature relatively well.

Historically, population and GDP have been growing over time. Scenario studies agree that further global population growth is likely up to 2050, leading to a range of possible outcomes of around 8.5–11 billion people (Figure 3.9a). After 2050, projections show a much wider range. If fertility drops below replacement levels, a decline in the global population is possible (as illustrated by SSP1 and SSP5). This typically includes scenarios with rapid development and investment in education. However, median projections mostly show a stabilisation of the world population (e.g., SSP2), while high-end projections show a continued growth (e.g., SSP3). The UN Population Prospects include considerably higher values for both the medium projection and the high end of the range than the SSP scenarios (KC and Lutz 2017; UN 2019). The most recent median UN projection reaches almost 11 billion people in 2100. The key differences are in Africa and China: here, the population projections are strongly influenced by the rate of fertility change (faster drop in SSPs). Underlying these differences, the UN approach is more based on current demographic trends while the SSPs assume a broader range of factors (including education) driving future fertility.

Economic growth is even more uncertain than the population projections (Figure 3.9c). The average growth rate of GDP was about 2.8% per year (constant USD) in the 1990–2019 period (The World Bank 2021). In 2020, the COVID-19 crisis resulted in a considerable drop in GDP (estimated around 4–5%) (IMF 2021). After a recovery period, most economic projections assume growth rates to converge back to previous projections, although at a lower level (IMF 2021; OECD 2021) (see also Box 3.2). In the long term, assumptions on future growth relate to political stability, the role of the progress of the technology frontier and the degree to which countries can catch up (Johansson et al. 2013). The SSP scenarios cover an extensive range, with low per-capita growth in SSP3 and SSP4 (mostly in developing countries) and rapid growth in SSP1 and SSP5. At the same, however, also scenarios outside the range have some plausibility – including the option of economic decline (Kallis et al. 2012) or much faster economic development (Christensen et al. 2018). The OECD long-term projection is at the global level reasonably consistent with SSP2. Equally important economic parameters include income distribution (inequity) and the type of growth (structural change, i.e., services vs manufacturing industries). Some projections (like SSP1) show a considerable convergence of income levels within and across countries, while in other projections, this does not occur (e.g., SSP3). Most scenarios reflect the suggested inverse relationship between the assumed growth rate for income and population growth (Figure 3.9e). SSP1 and SSP5 represent examples of scenarios with relatively low population increase and relatively high-income increase over the century. SSP3 represents an example of the opposite – while SSP2 and SSP4 are placed more in the middle. Nearly all scenarios assessed here do not account for climate impacts on growth (mostly for methodological reasons). As discussed in Section 3.5 these impacts can be considerable. An emerging area of literature emphasises the possibility of stabilisation (or even decline) of income levels in developed countries, arguing that such a trend would be preferred or even needed for environmental reasons (Anderson and Larkin 2013; Hickel and Kallis 2020; Kallis et al. 2020; Hickel et al. 2021; Keyßer and Lenzen 2021) (see also Chapter 5). Such scenarios are not common among IAM outcomes, that are more commonly based on the idea that decarbonisation can be combined with economic growth by a combination of technology, lifestyle and structural economic changes. Still, such scenarios could result in a dramatic reduction of energy and resource consumption.

Scenarios show a range of possible energy projections. In the absence of climate policy, most scenarios project the final energy demand to continue to grow to around 650–800 EJ yr –1 in 2100 (based on the AR6 Scenarios Database, Figure 3.9b). Some projections show a very high energy demand up to 1000 EJ yr –1 (comparable to SSP5). The scenario of the IEA lies within the SSP range but near the SSP1 projection. However, it should be noted that the IEA scenario includes current policies (most reference scenarios do not) and many scenarios published before 2021 did not account for the COVID-19 crisis. Several researchers discuss the possibility of decoupling material and energy demand from economic growth in the literature, mainly in developed countries (Kemp-Benedict 2018) (decoupling here refers to either a much slower increase in demand or even a decrease). In the scenario literature, this is reflected by scenarios with very low demand for final energy based on increased energy efficiency and less energy-intensive lifestyles (e.g., SSP1 and the LED scenario) (Grubler et al. 2018; van Vuuren et al. 2018). While these studies show the feasibility of such pathways, their energy efficiency improvement rates are considerably above the historic range of around 2% (Gütschow et al. 2018; Jeffery et al. 2018; Vrontisi et al. 2018; Haberl et al. 2020; Roelfsema et al. 2020; Giarola et al. 2021; Höhne et al. 2021; IEA 2021 a; Höhne et al. 2021; Sognnaes et al. 2021). These scenarios also show clear differences in food consumption and the amount of land used for agriculture. Food demand in terms of per-capita caloric intake is projected to increase in most scenarios (Figure 3.9d). However, it should be noted that there are large differences in dietary composition across the scenarios (from more meat-intensive in scenarios such as SSP5 to a decrease in meat consumptions in other scenarios such as SSP1). Land-use projections also depend on assumed changes in yield and the population scenarios (Figure 3.9f). Typically, changes in land use are less drastic than some other parameters (in fact, the 5–95th percentile database range is almost stable). Agriculture land is projected to increase in SSP3, SSP2, and SSP4 – it is more-or-less stable in SSP5 and is projected to decline in SSP1.

At the moment, climate change impact on mitigation potential is hardly considered in model-based scenarios. While a detailed overview of climate impacts is provided in IPCC AR6 WGII and Section 3.6 discusses the economic consequences, here we concentrate on the implications for mitigation potential. Climate change directly impacts the carbon budget via all kinds of feedbacks – which is included in the ranges provided for the carbon budget (e.g., 300–900 GtCO2 for 17th–83rd percentile for not exceeding 1.5°C; see AR6 WGI Chapter 5, 2021). Climate change, however, alters the production and consumption of energy (Section 6.5). An overview of the literature is provided by Yalew et al. (2020). In terms of supply, impacts could influence the cooling capacity of thermal plants, the potential and predictability of renewable energy, and energy infrastructure (van Vliet et al. 2016; Turner et al. 2017; Cronin et al. 2018a; Lucena et al. 2018; Yalew et al. 2020; Gernaat et al. 2021). Although the outcomes of these studies differ, they seem to suggest that although impacts might be relatively small at the global scale, they could be substantial at the regional scale (increasing or decreasing potential). Climate change can also impact energy demand, with rising temperatures resulting in decreases in heating demand and increases in cooling demand (Isaac and van Vuuren 2009; Zhou et al. 2014; Labriet et al. 2015; McFarland et al. 2015; Auffhammer et al. 2017; Clarke et al. 2018; van Ruijven et al. 2019; Yalew et al. 2020). As expected, the increase in cooling demand dominates the impact in warm regions and decreases in heating demand in cold regions (Isaac and van Vuuren 2009; Zhou et al. 2014; Clarke et al. 2018). Globally, most studies show a net increase in energy demand at the end of the century due to climate impacts (Isaac and van Vuuren 2009; Clarke et al. 2018; van Ruijven et al. 2019); however, one study shows a net decrease (Labriet et al. 2015). Only a few studies quantify the combined impacts of climate change on energy supply and energy demand (McFarland et al. 2015; Mima and Criqui 2015; Emodi et al. 2019; Steinberg et al. 2020). These studies show increases in electricity generation in the USA (McFarland et al. 2015; Steinberg et al. 2020) and increases in CO2 emissions in Australia (Emodi et al. 2019) or the USA (McFarland et al. 2015).

Climate change can impact the potential for AFOLU mitigation action by altering terrestrial carbon uptake, crop yields and bioenergy potential (Chapter 7). Carbon sequestration in forests may be positively or adversely affected by climate change and CO2 fertilisation. On the one hand, elevated CO2 levels and higher temperatures could enhance tree growth rates, carbon sequestration, and timber and biomass production (Beach et al. 2015; Kim et al. 2017; Anderegg et al. 2020). On the other hand, climate change could lead to greater frequency and intensity of disturbance events in forests, such as fires, prolonged droughts, storms, pests and diseases (Kim et al. 2017; Anderegg et al. 2020). The impact of climate change on crop yields could also indirectly impact the availability of land for mitigation and AFOLU emissions (Calvin et al. 2013; Bajželj and Richards 2014; Kyle et al. 2014; Beach et al. 2015; Meijl et al. 2018). The impact is, however, uncertain, as discussed in AR6 WGII Chapter 5. A few studies estimate the effect of climate impacts on AFOLU on mitigation, finding increases in carbon prices or mitigation costs by 1–6% in most scenarios (Calvin et al. 2013; Kyle et al. 2014).

In summary, a limited number of studies quantify the impact of climate on emissions pathways. The most important impact in energy systems might be through the impact on demand, although climate change could also impact renewable mitigation potential – certainly at the local and regional scale. Climate change might be more important for land-use related mitigation measures, including afforestation, bioenergy and nature-based solutions. The net effect of changes in climate and CO2 fertilisation are uncertain but could be substantial (Chapter 7).

Without mitigation, energy consumption and supply emissions continue to rise ( high confidence) (Kriegler et al. 2016; Bauer et al. 2017; Riahi et al. 2017; Mcjeon et al. 2021) (Section 6.7). While the share of renewable energy continues to grow in reference scenarios, fossil fuel accounts for the largest share of primary energy (Bauer et al. 2017; Price and Keppo 2017; Riahi et al. 2017). In scenarios that limit warming to 2°C or lower, transition of the energy-supply sector to a low- or no-carbon system is rapid (Rogelj et al. 2016, 2018b; Grubler et al. 2018; Luderer et al. 2018; van Vuuren et al. 2018). CO2 emissions from energy supply reach net zero around 2041 (2033–2057) in pathways limiting warming to 1.5°C (>50%) with no or limited overshoot and around 2053 (2040–2066) in pathways that limit warming to 2°C (>67%). Emissions reductions continue, with emissions reaching –7.1 GtCO2 yr –1 (–15 to –2.3 GtCO2 yr –1) in 2100 in all pathways that limit warming to 2°C (>67%) or lower.

All pathways that limit warming to 2°C (>67%) or lower show substantial reductions in fossil fuel consumption and a near elimination of the use of coal without CCS ( high confidence) (Bauer et al. 2017; van Vuuren et al. 2018; Grubler et al. 2018; Luderer et al. 2018; Rogelj et al. 2018 a,b; Azevedo et al. 2021; Mcjeon et al. 2021; Welsby et al. 2021) (Figure 3.22). In these pathways, the use of coal, gas and oil is reduced by 90%, 25%, and 41%, respectively, between 2019 and 2050 and 91%, 39%, and 78% between 2019 and 2100; coal without CCS is further reduced to 99% below its 2019 levels in 2100. These pathways show an increase in low-carbon energy, with 88% (69–97%) of primary energy from low-carbon sources in 2100, with different combinations of low-carbon fuels (e.g., non-biomass renewables, biomass, nuclear, and CCS) (Rogelj et al. 2018 a,b; van Vuuren et al. 2018) (Sections 3.4.1 and 6.7). Across all pathways that limit warming to 2°C and below, non-biomass renewables account for 52% (24–77%) of primary energy in 2100 (Creutzig et al. 2017; Pietzcker et al. 2017; Rogelj et al. 2018 b) (Chapter 6 and Figure 3.22). There are some studies analysing the potential for 100% renewable energy systems (Hansen et al. 2019); however, there are a range of issues around such systems (Box 6.6).

Stringent emissions reductions at the level required to limit warming to 2°C (>67%) or 1.5°C are achieved through increased electrification of end use, resulting in increased electricity generation in all pathways ( high confidence) (Rogelj et al. 2018 a; Azevedo et al. 2021) (Figure 3.23). Nearly all electricity in pathways likely to limit warming to 2°C and below is from low- or no-carbon fuels (Rogelj et al. 2018 a; Azevedo et al. 2021), with different shares of nuclear, biomass, non-biomass renewables, and fossil CCS across pathways. Low-emissions scenarios also show increases in hydrogen use (Figure 3.23).

Global final energy use inthe building sector increases in all pathways as a result of population growth and increasing affluence (Figure 3.24). There is very little difference in final energy intensity for the buildings sector across scenarios. Direct CO2 emissions from the buildings sector vary more widely across temperature stabilisation levels than energy consumption. In 2100, scenarios above 3°C [C7–C8] still show an increase of CO2 emissions from buildings around 29% above 2019, while all scenarios likely to limit warming to 2°C and below have emission reductions of around 85% (8–100%). Carbon intensity declines in all scenarios, but much more sharply as the warming level is reduced.

In all scenarios, the share of electricity in final energy use increases, a trend that is accelerated by 2050 for the scenarios likely to limit warming to 2°C and below (Figure 3.23). By 2100, the low-warming scenarios show large shares of electricity in final energy consumption for buildings. The opposite is observed for gases.

While several global IAM models have developed their buildings modules considerably over the past decade (Daioglou et al. 2012; Knobloch et al. 2017; Clarke et al. 2018; Edelenbosch et al. 2021; Mastrucci et al. 2021), the extremely limited availability of key sectoral variables in the AR6 scenarios database (such as floor space and energy use for individual services) prohibit a detailed analysis of sectoral dynamics. Individual studies in the literature often focus on single aspects of the buildings sector, though collectively providing a more comprehensive overview (Edelenbosch et al. 2020; Ürge-Vorsatz et al. 2020). For example, energy demand is driven by economic development that fulfills basic needs (Mastrucci et al. 2019; Rao et al. 2019a), but also drives up floor space in general (Daioglou et al. 2012; Levesque et al. 2018; Mastrucci et al. 2021) and ownership of energy-intensive appliances such as air conditioners (Isaac and van Vuuren 2009; Colelli and Cian 2020; Poblete-Cazenave et al. 2021). These dynamics are heterogeneous and lead to differences in energy demand and emission mitigation potential across urban/rural buildings and income levels (Krey et al. 2012; Poblete-Cazenave et al. 2021). Mitigation scenarios rely on fuel switching and technology (Knobloch et al. 2017; Dagnachew et al. 2020), efficiency improvement in building envelopes (Levesque et al. 2018; Edelenbosch et al. 2021) and behavioural changes (van Sluisveld et al. 2016; Niamir et al. 2018, 2020). The in-depth dynamics of mitigation in the building sector are explored in Chapter 9. 13

Reference scenarios show growth in transport demand, particularly in aviation and freight (Yeh et al. 2017; Sharmina et al. 2020; Müller-Casseres et al. 2021b). Energy consumption continues to be dominated by fossil fuels in reference scenarios, with some increases in electrification (Yeh et al. 2017; Edelenbosch et al. 2020; Yeh et al. 2017). CO2 emissions from transport increase for most models in reference scenarios (Yeh et al. 2017; Edelenbosch et al. 2020).

The relative contribution of demand-side reduction, energy- efficiency improvements, fuel switching, and decarbonisation of fuels, varyies by model, level of mitigation, mitigation options available, and underlying socio-economic pathway (Longden 2014; Wise et al. 2017; Yeh et al. 2017; Luderer et al. 2018; Yeh et al. 2017; Edelenbosch et al. 2020; Müller-Casseres et al. 2021a,b). IAMs typically rely on technology-focused measures like energy- efficiency improvements and fuel switching to reduce carbon emissions (Pietzcker et al. 2014; Edelenbosch et al. 2017a; Yeh et al. 2017; Zhang et al. 2018a,b; Rogelj et al. 2018 b; Zhang et al. 2018a,b; Sharmina et al. 2020). Many mitigation pathways show electrification of the transport system (Luderer et al. 2018; Pietzcker et al. 2014; Longden 2014; Luderer et al. 2018; Zhang et al. 2018a); however, without decarboniszation of the electricity system, transport electrification can increase total energy system emissions (Zhang and Fujimori 2020). A small number of pathways include demand-side mitigation measures in the transport sector; these studies show reduced carbon prices and reduced dependence on CDR (Grubler et al. 2018; Méjean et al. 2019; van de Ven et al. 2018; Zhang et al. 2018c; Méjean et al. 2019) (Section 3.4.1). 14

Across all IAM scenarios assessed, final energy demand for transport continues to grow, including in many stringent mitigation pathways (Figure 3.25). The carbon intensity of energy declines substantially by 2100 in likely 2°C (>67%) and below scenarios, leading to substantial declines in transport sector CO2 emissions with increased electrification of the transport system (Figure 3.23).

The transport sector has more detail than other sectors in many IAMs (Edelenbosch et al. 2020); however, there is considerable variation across models. Some models (e.g., GCAM, IMAGE, MESSAGE-GLOBIOM) represent different transport modes with endogenous shifts across modes as a function of income, price, and modal speed (Edelenbosch et al. 2020). 15 However, IAMs, including those with detailed transport, exclude several supply-side (e.g., synthetic fuels) and demand-side (e.g., behaviour change, reduced shipping, telework and automation) mitigation options (Pietzcker et al. 2014; Creutzig et al. 2016; Mittal et al. 2017; Davis et al. 2018; Köhler et al. 2020; Mittal et al. 2017; Gota et al. 2019; Wilson et al. 2019; Creutzig et al. 2016; Köhler et al. 2020; Sharmina et al. 2020; Pietzcker et al. 2014; Lefèvre et al. 2021; Müller-Casseres et al. 2021a,b).

As a result of these missing options and differences in how mitigation is implemented, IAMs tend to show less mitigation than the potential from national transport/energy models (Wachsmuth and Duscha 2019; Gota et al. 2019; Yeh et al. 2017; Gota et al. 2019; Wachsmuth and Duscha 2019; Edelenbosch et al. 2020). For the transport sector as a whole, studies suggest a mitigation potential of 4–-5 GtCO2 per year in 2030 (Edelenbosch et al. 2020) with complete decarbonization decarbonisation possible by 2050 (Gota et al. 2019; Wachsmuth and Duscha 2019). However, in the scenarios assessed in this chapter that limit warming to below 1.5°C (>50%) with no or limited overshoot, transport sector CO2 emissions are reduced by only 59% (28–% to 81%) in 2050 compared to 2015. IAM pathways also show less electrification than the potential from other studies; pathways that limit warming to 1.5°C with no or limited overshoot show a median of 25% (7– to 43%) of final energy from electricity in 2050, while the IEA NZE scenario includes 45% (IEA 2021 a).

Reference scenarios show declines in energy intensity, but increases in final energy use in the industrial sector (Edelenbosch et al. 2017b). These scenarios show increases in CO2 emissions both for the total industrial sector (Edelenbosch et al. 2017b, 2020; Luderer et al. 2018) and individual subsectors such as cement and iron and steel (van Ruijven et al. 2016; van Sluisveld et al. 2021) or chemicals (Daioglou et al. 2014; van Sluisveld et al. 2021).

In mitigation pathways, CO2 emissions reductions are achieved through a combination of energy savings (via energy-efficiency improvements and energy conservation), structural change, fuel switching, and decarbonisation of fuels (Edelenbosch et al. 2017b, 2020; Grubler et al. 2018; Luderer et al. 2018). Mitigation pathways show reductions in final energy for industry compared to the baseline (Edelenbosch et al. 2017b; Luderer et al. 2018; Edelenbosch et al. 2020) and reductions in the carbon intensity of the industrial sector through both fuel switching and the use of CCS (van Ruijven et al. 2016; Edelenbosch et al. 2017b, 2020; Luderer et al. 2018; Paltsev et al. 2021; van Sluisveld et al. 2021). The mitigation potential differs depending on the industrial subsector and the availability of CCS, with larger potential reductions in the steel sector (van Ruijven et al. 2016) and cement industry (Sanjuán et al. 2020) than in the chemicals sector (Daioglou et al. 2014). Many scenarios, including stringent mitigation scenarios, show continued growth in final energy; however, the carbon intensity of energy declines in all mitigation scenarios (Figure 3.26).

The representation of the industry sector is very aggregated in most IAMs, with only a small subset of models disaggregating key sectors such as cement, fertiliser, chemicals, and iron and steel (Daioglou et al. 2014; Edelenbosch et al. 2017b; Pauliuk et al. 2017; Napp et al. 2019; van Sluisveld et al. 2021). IAMs often account for both energy combustion and feedstocks (Edelenbosch et al. 2017b), but IAMs typically ignore material flows and miss linkages between sectors (Pauliuk et al. 2017; Kermeli et al. 2019). By excluding these processes, IAMs misrepresent the mitigation potential of the industry sector, for example by overlooking mitigation from material efficiency and circular economies (Sharmina et al. 2020), which can have substantial mitigation potential (Sections 5.3.4 and 11.3).

Sectoral studies indicate a large mitigation potential in the industrial sector by 2050, including the potential for net zero CO2 emissions for steel, plastics, ammonia, and cement (Section 11.4.1). Detailed industry sector pathways show emissions reductions between 39% and 94% by mid-century compared to the present day 17 (Section 11.4.2) and a substantial increase in direct electrification (IEA 2021 a). IAMs show comparable mitigation potential to sectoral studies with median reductions in CO2 emissions between 2019 and 2050 of 70% in scenarios likely to limit warming to 2°C (>67%) and below and a maximum reduction of 96% (Figure 3.26). Some differences between IAMs and sectoral models can be attributed to differences in technology availability, with IAMs sometimes including more technologies (van Ruijven et al. 2016) and sometimes less (Sharmina et al. 2020).

Mitigation pathways show substantial reductions in CO2 emissions, but more modest reductions in AFOLU CH4 and N2O emissions ( high confidence) (Popp et al. 2017; Roe et al. 2019; Reisinger et al. 2021) (Figure 3.27). Pathways limiting warming to likely 2°C or lower are projected to reach net zero CO2 emissions in the AFOLU sector around 2033 (2024–2060); however, AFOLU CH4 and N2O emissions remain positive in all pathways (Figure 3.27). While IAMs include many land-based mitigation options, these models exclude several options with large mitigation potential, such as biochar, agroforestry, restoration/avoided conversion of coastal wetlands, and restoration/avoided conversion of peatland (IPCC 2019a; Smith et al. 2019) (Chapter 7 and Section 3.4). Sectoral studies show higher mitigation potential than IAM pathways, as these studies include more mitigation options than IAMs (medium confidence) (Chapter 7).

Limiting warming to likely 2°C (>67%) or lower can result in large-scale transformation of the land surface ( high confidence) (Popp et al. 2017; Rogelj et al. 2018 a,b; Brown et al. 2019; Roe et al. 2019). The scale of land transformation depends, inter alia, on the temperature goal and the mitigation options included (Popp et al. 2017; Rogelj et al. 2018 a; IPCC 2019a). Pathways with more demand-side mitigation options show less land transformation than those with more limited options (Grubler et al. 2018; van Vuuren et al. 2018; IPCC 2019a). Most of these pathways show increases in forest cover, with an increase of 322 million ha (–67 to 890 million ha) in 2050 in pathways that limit warming to 1.5°C (>50%) with no or limited overshoot, whereas bottom-up models portray an economic potential of 300–500 million ha of additional forest (Chapter 7). Many IAM pathways also include large amounts of energy cropland area, to supply biomass for bioenergy and BECCS, with 199 (56–482) million ha in 2050 in pathways that limit warming to 1.5°C (>50%) with no or limited overshoot. Large land transformations, such as afforestation/reforestation and widespread planting of energy crops, can have implications for biodiversity and sustainable development (Sections 3.7, 7.7.4 and 12.5).

Delayed mitigation has implications for land-use transitions (Hasegawa et al. 2021a). Delaying mitigation action can result in a temporary overshoot of temperature and large-scale deployment of CDR in the second half of the century to reduce temperatures from their peak to a given level (Smith et al. 2019; Hasegawa et al. 2021a). IAM pathways rely on afforestation and BECCS as CDR measures, so delayed mitigation action results in substantial land-use change in the second half of the century with implications for sustainable development (Hasegawa et al. 2021a) (Section 3.7). Shifting to earlier mitigation action reduces the amount of land required for this, though at the cost of larger land-use transitions earlier in the century (Hasegawa et al. 2021a). Earlier action could also reduce climate impacts on agriculture and land-based mitigation options (Smith et al. 2019).

Some AFOLU mitigation options can enhance vegetation and soil carbon stocks such as reforestation, restoration of degraded ecosystems, protection of ecosystems with high carbon stocks and changes to agricultural land management to increase soil carbon ( high confidence) (Griscom et al. 2017; de Coninck et al. 2018; Fuss et al. 2018; Smith et al. 2019) (AR6 WGIII Chapter 7). The time scales associated with these options indicate that carbon sinks in terrestrial vegetation and soil systems can be maintained or enhanced so as to contribute towards long-term mitigation ( high confidence); however, many AFOLU mitigation options do not continue to sequester carbon indefinitely (Fuss et al. 2018; de Coninck et al. 2018; IPCC 2019a) (AR6 WGIII Chapter 7). In the very long term (the latter part of the century and beyond), it will become more challenging to continue to enhance vegetation and soil carbon stocks, so that the associated carbon sinks could diminish or even become sources ( high confidence) (de Coninck et al. 2018; IPCC 2019a) (AR6 WGI Chapter 5). Sustainable forest management, including harvest and forest regeneration, can help to remediate and slow any decline in the forest carbon sink, for example by restoring degraded forest areas, and so go some way towards addressing the issue of sink saturation (IPCC 2019) (AR6 WGI Chapter 5; and Chapter 7 in this report). The accumulated carbon resulting from mitigation options that enhance carbon sequestration (e.g., reforestation, soil carbon sequestration) is also at risk of future loss due to disturbances (e.g., fire, pests) (Boysen et al. 2017; de Coninck et al. 2018; Fuss et al. 2018; Smith et al. 2019; IPCC 2019a; Anderegg et al. 2020) (AR6 WGI Chapter 5). Maintaining the resultant high vegetation and soil carbon stocks could limit future land-use options, as maintaining these carbon stocks would require retaining the land use and land-cover configuration implemented to achieve the increased stocks.

Anthropogenic land CO2 emissions and removals in IAM pathways cannot be directly compared with those reported in national GHG inventories ( high confidence) (Grassi et al. 2018, 2021) (Section 7.2). Due to differences in definitions for the area of managed forests and which emissions and removals are considered anthropogenic, the reported anthropogenic land CO2 emissions and removals differ by about 5.5 GtCO2 yr –1 between IAMs, which rely on bookkeeping approaches (e.g., Houghton and Nassikas 2017), and national GHG inventories (Grassi et al. 2021). Such differences in definitions can alter the reported time at which anthropogenic net zero CO2 emissions are reached for a given emission scenario. Using national inventories would lead to an earlier reported time of net zero (van Soest et al. 2021b) or to lower calculated cumulative emissions until the time of net zero (Grassi et al. 2021) as compared to IAM pathways. The numerical differences are purely due to differences in the conventions applied for reporting the anthropogenic emissions and do not have any implications for the underlying land-use changes or mitigation measures in the pathways. Grassi et al. (Grassi et al. 2021) offer a methodology for adjusting to reconcile these differences and enable a more accurate assessment of the collective progress achieved under the Paris Agreement (Chapter 7 and Cross-Chapter Box 6 in Chapter 7).

This subsection includes other CDR options not discussed in the previous subsections, including direct air carbon capture and storage (DACCS), enhanced weathering (EW), and ocean-based approaches, focusing on the role of these options in long-term mitigation pathways, using both IAMs (Chen and Tavoni 2013; Marcucci et al. 2017; Rickels et al. 2018; Fuhrman et al. 2019, 2020, 2021; Realmonte et al. 2019; Akimoto et al. 2021; Strefler et al. 2021a) and non-IAMs (Fuss et al. 2013; González and Ilyina 2016; Bednar et al. 2021; Shayegh et al. 2021). There are other options discussed in the literature, such as methane capture (Jackson et al. 2019), however, the role of these options in long-term mitigation pathways has not been quantified and is thus excluded here. Chapter 12 includes a more detailed description of the individual technologies, including their costs, potentials, financing, risks, impacts, maturity and upscaling.

Very few studies and pathways include other CDR options (Table 3.5). Pathways with DACCS include potentially large removal from DACCS (up to 37 GtCO2 yr –1 in 2100) in the second half of the century (Chen and Tavoni 2013; Marcucci et al. 2017; Realmonte et al. 2019; Fuhrman et al. 2020, 2021; Shayegh et al. 2021; Akimoto et al. 2021) and reduced cost of mitigation (Bistline and Blanford 2021; Strefler et al. 2021a). At large scales, the use of DACCS has substantial implications for energy use, emissions, land, and water; substituting DACCS for BECCS results in increased energy usage, but reduced land-use change and water withdrawals (Fuhrman et al., 2020, 2021) (Chapter 12.3.2; AR6 WGI Chapter 5). The level of deployment of DACCS is sensitive to the rate at which it can be scaled up, the climate goal or carbon budget, the underlying socio-economic scenario, the availability of other decarbonisation options, the cost of DACCS and other mitigation options, and the strength of carbon-cycle feedbacks (Chen and Tavoni 2013; Fuss et al. 2013; Honegger and Reiner 2018; Realmonte et al. 2019; Fuhrman et al. 2020; Bistline and Blanford 2021; Fuhrman et al. 2021; Strefler et al. 2021a) (AR6 WGI Chapter 5). Since DACCS consumes energy, its effectiveness depends on the type of energy used; the use of fossil fuels would reduce its sequestration efficiency (Creutzig et al. 2019; NASEM 2019; Babacan et al. 2020). Studies with additional CDR options in addition to DACCS (e.g., enhanced weathering, BECCS, afforestation, biochar, and soil carbon sequestration) find that CO2 removal is spread across available options (Holz et al. 2018; Strefler et al. 2021a). Similar to DACCS, the deployment of deep-ocean storage depends on cost and the strength of carbon-cycle feedbacks (Rickels et al. 2018).

Table 3.5 |Carbon dioxide removal in assessed pathways. Scenarios are grouped by temperature categories, as defined in Section 3.2.4. Quantity indicates the median and 5–95th percentile range of cumulative sequestration from 2020 to 2100 in GtCO2. Count indicates the number of scenarios with positive values for that option. Source: AR6 Scenarios Database.

1Cumulative CDR from AFOLU cannot be quantified precisely because models use different reporting methodologies that in some cases combine gross emissions and removals, and use different baselines.

The close link between cumulative CO2 emissions and warming has strong implications for the relationship between near-, medium-, and long-term climate action to limit global warming. The AR6 WGI Assessment has estimated a remaining carbon budget of 500 (400) GtCO2 from the beginning of 2020 onwards for staying below 1.5°C with 50% (67%) likelihood, subject to additional uncertainties about historic warming and the climate response, and variations in warming from non-CO2 climate forcers (Canadell and Monteiro 2019) (AR6 WGI Chapter 5, Section 5.5). For comparison, if current CO2 emissions of more than 40 GtCO2 are keeping up until 2030, more than 400 GtCO2 will be emitted during 2021–2030, already exhausting the remaining carbon budget for 1.5°C by 2030.

The relationship between warming limits and near-term action is illustrated in Figure 3.29, using a set of 1.5°C–2°C scenarios with different levels of near-term action, overshoot and non-CO2 warming contribution from a recent study (Riahi et al. 2021). In general, the more CO2 is emitted until 2030, the less CO2 can be emitted thereafter to stay within a remaining carbon budget and below a warming limit. Scenarios with immediate action to observe the warming limit give the longest time to exhaust the associated remaining carbon budget and reach net zero CO2 emissions (see light blue lines in Figure 3.29 and Cross-Chapter Box 3 in this chapter). In comparison, following projected NDC emissions until 2030 would imply a more pronounced drop in emissions from 2030 levels to net zero to make up for the additional near-term emissions (see orange lines in Figure 3.29). If such a drop does not occur, the remaining carbon budget is exceeded and net negative CO2 emissions are required to return global mean temperature below the warming limit (see black lines in Figure 3.29) (Clarke et al. 2014; Fuss et al. 2014; Rogelj et al. 2018 a).

The relationship between warming limits and near-term action is also affected by the warming contribution of non-CO2 greenhouse gases and other short-lived climate forcers (Section 3.3; AR6 WGI Section 6.7). The estimated budget values for limiting warming to 1.5°C–2°C already assume stringent reductions in non-CO2 greenhouse gases and non-CO2 climate forcing as found in 1.5°C–2°C pathways (Section 3.3 and Cross-Working Group Box 1 in this chapter; AR6 WGI Section 5.5 and Box 5.2 in Chapter 5). Further variations in non-CO2 warming observed across 1.5°C–2°C pathways can vary the median estimate for the remaining carbon budget by 220 GtCO2 (AR6 WGI Section 5.5). In 1.5°C–2°C pathways, the non-CO2 warming contribution differs strongly between the near, medium and long term. Changes to the atmospheric composition of short-lived climate forcers (SLCFs) dominate the warming response in the near term (AR6 WGI Section 6.7). CO2 reductions are combined with strong reductions in air pollutant emissions due to rapid reduction in fossil fuel combustion and in some cases the assumption of stringent air quality policies (Rao et al. 2017b; Smith et al. 2020c). As air pollutants exert a net-cooling effect, their reduction drives up non-CO2 warming in the near term, which can be attenuated by the simultaneous reduction of methane and black carbon (Shindell and Smith 2019; Smith et al. 2020b) (AR6 WGI Section 6.7). After 2030, the reduction in methane concentrations and associated reductions in tropospheric ozone levels tend to dominate so that a peak and decline in non-CO2 forcing and non-CO2-induced warming can occur before net zero CO2 is reached (Figure 3.29) (Rogelj et al. 2018 a). The more stringent the reductions in methane and other short-lived warming agents such as black carbon, the lower this peak and the earlier the decline of non-CO2 warming, leading to a reduction of warming rates and overall warming in the near to medium term (Harmsen et al. 2020; Smith et al. 2020b). This is important for keeping warming below a tight warming limit that is already reached around mid-century as is the case in 1.5°C pathways (Xu and Ramanathan 2017). Early and deep reductions of methane emissions, and other short-lived warming agents such as black carbon, provide space for residual CO2-induced warming until the point of net zero CO2 emissions is reached (see purple lines in Figure 3.29). Such emissions reductions have also been advocated due to co-benefits for, for example, reducing air pollution (Rao et al. 2016; Shindell et al. 2017a, 2018; Shindell and Smith 2019; Rauner et al. 2020a; Vandyck et al. 2020).

The relationship between long-term climate goals and near-term action is further constrained by social, technological, economic and political factors (Cherp et al. 2018; van Sluisveld et al. 2018b; Aghion et al. 2019; Mercure et al. 2019; Trutnevyte et al. 2019b; Jewell and Cherp 2020). These factors influence path dependency and transition speed (Pahle et al. 2018; Vogt-Schilb et al. 2018). While detailed integrated assessment modelling of global mitigation pathways accounts for technology inertia (Bertram et al. 2015a; Mercure et al. 2018) and technology innovation and diffusion (Wilson et al. 2013; van Sluisveld et al. 2018a; Luderer et al. 2021), there are limitations in capturing socio-technical and political drivers of innovation, diffusion and transition processes (Gambhir et al. 2019; Köhler et al. 2019; Hirt et al. 2020; Keppo et al. 2021). Mitigation pathways show a wide range of transition speeds that have been interrogated in the context of socio-technical inertia (Gambhir et al. 2017; Kefford et al. 2018; Kriegler et al. 2018a; Brutschin et al. 2021) vs accelerating technological change and self-enforcing socio-economic developments (Creutzig et al. 2017; Zenghelis 2019) (Section 3.8). Diagnostic analysis of detailed IAMs found a lag of 8–20 years between the convergence of emissions pricing and the convergence of emissions response after a period of differentiated emission prices (Harmsen et al. 2021). This provides a measure of the inertia to changing policy signals in the model response. It is about half the time scale of 20–40 years observed for major energy transitions (Grubb et al. 2021). Hence, the mitigation pathways assessed here capture socio-technical inertia in reducing emissions, but the limited modelling of socio-political factors may alter the extent and persistence of this inertia.

The implications of near-term climate action for long-term climate outcomes can be explored by comparing mitigation pathways with different near-term emissions developments aiming for the same climate target (Riahi et al. 2015; Vrontisi et al. 2018; Roelfsema et al. 2020). A particular example is the comparison of cost-effective pathways with immediate action to limit warming to 1.5°C–2°C with mitigation pathways pursuing more moderate mitigation action until 2030. After the adoption of the Paris Agreement, near-term action was often modelled to reflect conditional and unconditional elements of originally submitted NDCs (2015–2019) (Fawcett et al. 2015; Fujimori et al. 2016a; Kriegler et al. 2018a; Vrontisi et al. 2018; Roelfsema et al. 2020). The most recent modelling studies also include submission of updated NDCs or announcements of planned updates in the first half of 2021 (Network for Greening the Financial System 2021; Riahi et al. 2021). Emissions levels under NDCs announced prior to COP26 are assessed to range between 47–57 GtCO2-eq in 2030 (Section 4.2.2). This assessed range corresponds well to 2030 emissions levels in 2°C mitigation pathways in the literature that are designed to follow the original or updated NDCs until 2030. 20 For the 139 scenarios of this kind that are collected in the AR6 scenario database and that still limit warming to 2°C (>67%), the 2030 emissions range is 53 (45–58) GtCO2-eq (based on native model reporting) and 52.5 (47–56.5) GtCO2-eq, respectively (based on harmonised emissions data for climate assessment (Annex III.2.5.1); median and 5–95th percentile). This close match allows a robust assessment of the implications of implementing NDCs announced prior to COP26 for post-2030 mitigation efforts and warming outcomes based on the literature and the AR6 scenarios database.

Without a strengthening of policies beyond those that are implemented by the end of 2020, GHG emissions are projected to rise beyond 2025, leading to a median global warming of 3.2 [2.2 to 3.5] °C by 2100. Modelled pathways that are consistent with NDCs announced prior to COP26 until 2030 and assume no increase in ambition thereafter have lower emissions, leading to a median global warming of 2.8°C [2.1–3.4°C] by 2100.

The assessed emission ranges from implementing the unconditional (unconditional and conditional) elements of NDCs announced prior to COP26 implies an emissions gap to cost-effective mitigation pathways of 19–26 (16–23) GtCO2-eq in 2030 for limiting warming to 1.5°C (>50%) with no or limited overshoot and 10–16 (6–14) GtCO2-eq in 2030 for limiting warming to 2°C (>67%) (Cross-Chapter Box 4 in Chapter 4). The emissions gap gives rise to a number of mitigation challenges (Kriegler et al. 2013a, 2018a,b; Luderer et al. 2013, 2018; Rogelj et al. 2013a; Fawcett et al. 2015; Riahi et al. 2015; Fujimori et al. 2016b; Strefler et al. 2018; Winning et al. 2019; SEI et al. 2020; UNEP 2020): (i) larger transitional challenges post-2030 to still remain under the warming limit, in particular higher CO2 emissions reduction rates and technology transition rates required during 2030–2050; (ii) larger lock-in into carbon-intensive infrastructure and increased risk of stranded fossil fuel assets (Section 3.5.2.2); and (iii) larger reliance on CDR to reach net zero CO2 more rapidly and compensate excess emissions in the second half of the century (Section 3.5.2.1). All these factors exacerbate the socio-economic strain of implementing the transition, leading to an increased risk of overshooting the warming and a higher risk of climate change impacts (Drouet et al. 2021).

The challenges are illustrated in Table 3.6 and Figure 3.30, surveying global mitigation pathways in the literature that were collected in the AR6 scenarios database. There is a clear trend of increasing peak warming with increasing 2030 GHG emission levels (Figure 3.30a,b). In particular, there is no mitigation pathway designed to follow the NDCs until 2030 that can limit warming to 1.5°C (>50%) with no or limited overshoot. Our assessment confirms the finding of the IPCCSpecial Report on Global Warming of 1.5°C (Rogelj et al. 2018) for the case of NDCs announced prior to COP26 that pathways following the NDCs until 2030 ‘would not limit global warming to 1.5°C, even if supplemented by very challenging increases in the scale and ambition of emissions reductions after 2030’ (SR1.5 SPM). This assessment is now more robust than in SR1.5 as it is based on a larger set of 1.5°C–2°C pathways with better representation of current trends and plans covering a wider range of post-2030 emissions developments. In particular, a recent multi-model study limiting peak cumulative CO2 emissions for a wide range of carbon budgets and immediate vs NDC-type action until 2030 established a feasibility frontier for the existence of such pathways across participating models (Riahi et al. 2021).

Table 3.6 | Comparison of key scenario characteristics for five scenario classes (see Table 3. 2): (i) immediate action to limit warming to 1.5°C (>50%) with no or limited overshoot, (ii) near team action following the NDCs until 2030 and returning warming to below 1.5°C (>50%) by 2100 after a high overshoot, (iii) immediate action to limit warming to 2°C (>67%), (iv) near term action following the NDCs until 2030 followed by post-2030 action to limit warming to 2°C (>67%). Also shown are the characteristics for (v) the combined class of all scenarios that limit warming to 2°C (>67%). The classes (ii) and (iv) comprise the large majority of scenarios indicated by red dots, and the classes (i) and (iii) comprise the scenarios depicted by blue dots in Figure 3.30. Shown are median and interquartile ranges (in brackets) for selected global indicators. Emissions ranges are based on harmonized emissions data for the climate assessment with the exception of land use CO2 emissions for which uncertainty in historic estimates is large. Numbers are rounded to the nearest 5, with the exception of cumulative CCS, BECCS, and net negative CO2 emissions rounded to the nearest 10.

The 2030 emissions levels in the NDCS announced prior to COP26 also tighten the remaining space to limit warming to 2°C (>67%). As shown in Figure 3.30b, the 67th percentile of peak warming reaches values above 1.7°C in pathways with 2030 emissions levels in this range. To still limit warming to 2°C (>67%), the global post-2030 GHG emission reduction rates would need to be abruptly raised in 2030 from 0–0.7 GtCO2-eq yr –1 to an average of 1.4–2.0 GtCO2-eq yr –1 during the period 2030–2050 (Figure 3.30c), around 70% higher than in immediate mitigation pathways confirming findings in the literature (Winning et al. 2019). Their average reduction rate of 0.6–1.4 GtCO2 yr –1 would already be unprecedented at the global scale and, with a few exceptions, national scale for an extended period of time (Riahi et al. 2015). For comparison, the impact of COVID-19 on the global economy is projected to have lead to a decline of around 2.5–3 GtCO2 of global CO2 emissions from fossil fuels and industry in 2020 (Friedlingstein et al. 2020) (Section 2.2).

The increased post-2030 transition challenge in mitigation pathways with moderate near-term action is also reflected in the timing of reaching net zero CO2 emissions (Figure 3.30d and Cross-Chapter Box 3 in this chapter). As 2030 emission levels and the cumulated CO2 emissions until 2030 increase, the remaining time for dropping to net zero CO2 and staying within the remaining carbon budget shortens (Figure 3.29). This gives rise to an inverted v-shape of the lower bound on the year of reaching net zero as a function of 2030 emissions levels. Reaching low emissions in 2030 facilitates reaching net zero early (left leg of the inverted v), but staying high until 2030 also requires reaching net zero CO2 faster to compensate for higher emissions early on (right leg of the inverted v). Overall, there is a considerable spread of the timing of net zero CO2 for any 2030 emissions level due to variation in the timing of spending the remaining carbon budget and the non-CO2 warming contribution (Cross-Chapter Box 3 in this chapter).

There is also a profound impact on the underlying transition of energy and land use (Figure 3.30f–h and Table 3.6). Scenarios following NDCs until 2030 show a much smaller reduction in fossil fuel use, a slower growth in renewable energy use, and a smaller reduction in CO2 and CH4 land-use emissions in 2030 compared to immediate action scenarios. This is then followed by a much faster reduction of land-use emissions and fossil fuels, and a larger increase of nuclear energy, bioenergy and non-biomass renewable energy during the medium term in order to get close to the levels of the immediate action pathways in 2050. This is combined with a larger amount of net negative CO2 emissions that are used to compensate the additional emissions before 2030. The faster transition during 2030–2050 is taking place from a greater investment in fossil fuel infrastructure and lower deployment of low-carbon alternatives in 2030, adding to the socio-economic challenges to realise the higher transition rates (Section 3.5.2.2). Therefore, these pathways also show higher mitigation costs, particularly during the period 2030–2050, than immediate action scenarios (Section 3.6.1 and Figure 3.34d) (Liu et al. 2016; Kriegler et al. 2018a; Vrontisi et al. 2018). Given these circumstances and the fact the modelling of socio-political and institutional constraints is limited in Integrated Assessment Models (IAMs) (Gambhir et al. 2019; Köhler et al. 2019; Hirt et al. 2020; Keppo et al. 2021), the feasibility of realising these scenarios is assessed to be lower (Gambhir et al. 2017; Napp et al. 2017; Brutschin et al. 2021) (cf. Section 3.8), increasing the risk of an overshoot of climate goals.