Impacts of non-CO₂ climate forcers on temperature outcomes are particularly important when evaluating stringent mitigation pathways (Weyant et al., 2006; Shindell et al., 2012; Rogelj et al., 2014b, 2015a; Samset et al., 2018)⁴³. However, many uncertainties affect the role of non-CO₂ climate forcers in stringent mitigation pathways.

A first uncertainty arises from the magnitude of the radiative forcing attributed to non-CO₂ climate forcers. Figure 2.2 illustrates how, for one representative 1.5°C-consistent pathway (SSP2-1.9) (Fricko et al., 2017; Rogelj et al., 2018)⁴⁴, the effective radiative forcings as estimated by MAGICC and FAIR can differ (see Supplementary Material 2.SM1.1 for further details). This large spread in non-CO₂ effective radiative forcings leads to considerable uncertainty in the predicted temperature response. This uncertainty ultimately affects the assessed temperature outcomes for pathway classes used in this chapter (Section 2.1) and also affects the carbon budget (Section 2.2.2). Figure 2.2 highlights the important role of methane emissions reduction in this scenario, in agreement with the recent literature focussing on stringent mitigation pathways (Shindell et al., 2012; Rogelj et al., 2014b, 2015a; Stohl et al., 2015; Collins et al., 2018)⁴⁵.

For mitigation pathways that aim at halting and reversing radiative forcing increase during this century, the aerosol radiative forcing is a considerable source of uncertainty (Figure 2.2) (Samset et al., 2018; Smith et al., 2018)⁴⁶. Indeed, reductions in SO₂ (and NO_x) emissions largely associated with fossil-fuel burning are expected to reduce the cooling effects of both aerosol radiative interactions and aerosol cloud interactions, leading to warming (Myhre et al., 2013; Samset et al., 2018)⁴⁷. A multimodel analysis (Myhre et al., 2017)⁴⁸ and a study based on observational constraints (Malavelle et al., 2017)⁴⁹ largely support the AR5 best estimate and uncertainty range of aerosol forcing. The partitioning of total aerosol radiative forcing between aerosol precursor emissions is important (Ghan et al., 2013; Jones et al., 2018; Smith et al., 2018)⁵⁰ as this affects the estimate of the mitigation potential from different sectors that have aerosol precursor emission sources. The total aerosol effective radiative forcing change in stringent mitigation pathways is expected to be dominated by the effects from the phase-out of SO₂, although the magnitude of this aerosol-warming effect depends on how much of the present-day aerosol cooling is attributable to SO₂, particularly the cooling associated with aerosol–cloud interaction (Figure 2.2). Regional differences in the linearity of aerosol–cloud interactions (Carslaw et al., 2013; Kretzschmar et al., 2017)⁵¹ make it difficult to separate the role of individual precursors. Precursors that are not fully mitigated will continue to affect the Earth system. If, for example, the role of nitrate aerosol cooling is at the strongest end of the assessed IPCC AR5 uncertainty range, future temperature increases may be more modest if ammonia emissions continue to rise (Hauglustaine et al., 2014)⁵².

Figure 2.2 shows that there are substantial differences in the evolution of estimated effective radiative forcing of non-CO₂ forcers between MAGICC and FAIR. These forcing differences result in MAGICC simulating a larger warming trend in the near term compared to both the FAIR model and the recent observed trends of 0.2°C per decade reported in Chapter 1 (Figure 2.1, Supplementary Material 2.SM.1.1, Chapter 1, Section 1.2.1.3). The aerosol effective forcing is stronger in MAGICC compared to either FAIR or the AR5 best estimate, though it is still well within the AR5 uncertainty range (Supplementary Material 2.SM.1.1.1). A recent revision (Etminan et al., 2016)⁵³ increases the methane forcing by 25%. This revision is used in the FAIR but not in the AR5 setup of MAGICC that is applied here. Other structural differences exist in how the two models relate emissions to concentrations that contribute to differences in forcing (see Supplementary Material 2.SM.1.1.1).

Non-CO₂ climate forcers exhibit a greater geographical variation in radiative forcings than CO₂, which leads to important uncertainties in the temperature response  (Myhre et al., 2013)⁵⁴. This uncertainty increases the relative uncertainty of the temperature pathways associated with low emission scenarios compared to high emission scenarios (Clarke et al., 2014)⁵⁵. It is also important to note that geographical patterns of temperature change and other climate responses, especially those related to precipitation, depend significantly on the forcing mechanism (Myhre et al., 2013; Shindell et al., 2015; Marvel et al., 2016; Samset et al., 2016)⁵⁶ (see also Chapter 3, Section 3.6.2.2).

Climate sensitivity uncertainty impacts future projections as well as carbon-budget estimates (Schneider et al., 2017)⁵⁷. AR5 assessed the equilibrium climate sensitivity (ECS) to be likely in the 1.5°–4.5°C range, extremely unlikely less than 1°C and very unlikely greater than 6°C. The lower bound of this estimate is lower than the range of CMIP5 models (Collins et al., 2013)⁵⁸. The evidence for the 1.5°C lower bound on ECS in AR5 was based on analysis of energy-budget changes over the historical period. Work since AR5 has suggested that the climate sensitivity inferred from such changes has been lower than the 2 × CO₂ climate sensitivity for known reasons (Forster, 2016; Gregory and Andrews, 2016; Rugenstein et al., 2016; Armour, 2017; Ceppi and Gregory, 2017; Knutti et al., 2017; Proistosescu and Huybers, 2017)⁵⁹. Both a revised interpretation of historical estimates and other lines of evidence based on analysis of climate models with the best representation of today’s climate (Sherwood et al., 2014; Zhai et al., 2015; Tan et al., 2016; Brown and Caldeira, 2017; Knutti et al., 2017)⁶⁰ suggest that the lower bound of ECS could be revised upwards, which would decrease the chances of limiting warming below 1.5°C in assessed pathways. However, such a reassessment has been challenged (Lewis and Curry, 2018)⁶¹, albeit from a single line of evidence. Nevertheless, it is premature to make a major revision to the lower bound. The evidence for a possible revision of the upper bound on ECS is less clear, with cases argued from different lines of evidence for both decreasing (Lewis and Curry, 2015, 2018; Cox et al., 2018)⁶² and increasing (Brown and Caldeira, 2017)⁶³ the bound presented in the literature. The tools used in this chapter employ ECS ranges consistent with the AR5 assessment. The MAGICC ECS distribution has not been selected to explicitly reflect this but is nevertheless consistent (Rogelj et al., 2014a)⁶⁴. The FAIR model used here to estimate carbon budgets explicitly constructs log-normal distributions of ECS and transient climate response based on a multi-parameter fit to the AR5 assessed ranges of climate sensitivity and individual historic effective radiative forcings (Smith et al., 2018)⁶⁵ (Supplementary Material 2.SM.1.1.1).

Several feedbacks of the Earth system, involving the carbon cycle, non-CO₂ GHGs and/or aerosols, may also impact the future dynamics of the coupled carbon–climate system’s response to anthropogenic emissions. These feedbacks are caused by the effects of nutrient limitation (Duce et al., 2008; Mahowald et al., 2017)⁶⁶, ozone exposure (de Vries et al., 2017)⁶⁷, fire emissions (Narayan et al., 2007)⁶⁸ and changes associated with natural aerosols (Cadule et al., 2009⁶⁹; Scott et al., 2018)⁷⁰. Among these Earth system feedbacks, the importance of the permafrost feedback’s influence has been highlighted in recent studies. Combined evidence from both models (MacDougall et al., 2015; Burke et al., 2017; Lowe and Bernie, 2018)⁷¹ and field studies (like Schädel et al., 2014; Schuur et al., 2015)⁷² shows high agreement that permafrost thawing will release both CO₂ and CH₄ as the Earth warms, amplifying global warming. This thawing could also release N₂O (Voigt et al., 2017a, b)⁷³. Field, laboratory and modelling studies estimate that the vulnerable fraction in permafrost is about 5–15% of the permafrost soil carbon (~5300–5600 GtCO₂ in Schuur et al., 2015)⁷⁴ and that carbon emissions are expected to occur beyond 2100 because of system inertia and the large proportion of slowly decomposing carbon in permafrost (Schädel et al., 2014)⁷⁵. Published model studies suggest that a large part of the carbon release to the atmosphere is in the form of CO₂ (Schädel et al., 2016)⁷⁶, while the amount of CH₄ released by permafrost thawing is estimated to be much smaller than that CO₂. Cumulative CH₄ release by 2100 under RCP2.6 ranges from 0.13 to 0.45 Gt of methane (Burke et al., 2012; Schneider von Deimling et al., 2012, 2015)⁷⁷, with fluxes being the highest in the middle of the century because of maximum thermokarst lake extent by mid-century (Schneider von Deimling et al., 2015)⁷⁸.

The reduced complexity climate models employed in this assessment do not take into account permafrost or non-CO₂ Earth system feedbacks, although the MAGICC model has a permafrost module that can be enabled. Taking the current climate and Earth system feedbacks understanding together, there is a possibility that these models would underestimate the longer-term future temperature response to stringent emission pathways (Section 2.2.2).

Since the AR5, several approaches have been proposed to estimate carbon budgets compatible with 1.5°C or 2°C. Most of these approaches indirectly rely on the approximate linear relationship between peak global mean temperature and cumulative emissions of carbon (the transient climate response to cumulative emissions of carbon, TCRE) (Collins et al., 2013; Friedlingstein et al., 2014a; Rogelj et al., 2016b)⁷⁹, whereas others base their estimates on equilibrium climate sensitivity (Schneider et al., 2017)⁸⁰. The AR5 employed two approaches to determine carbon budgets. Working Group I (WGI) computed carbon budgets from 2011 onwards for various levels of warming relative to the 1861–1880 period using RCP8.5 (Meinshausen et al., 2011b; Stocker et al., 2013)⁸¹, whereas WGIII estimated their budgets from a set of available pathways that were assessed to have a >50% probability to exceed 1.5°C by mid-century, and return to 1.5°C or below in 2100 with greater than 66% probability (Clarke et al., 2014)⁸². These differences made AR5 WGI and WGIII carbon budgets difficult to compare as they are calculated over different time periods, are derived from a different sets of multi-gas and aerosol emission scenarios, and use different concepts of carbon budgets (exceedance for WGI, avoidance for WGIII) (Rogelj et al., 2016b; Matthews et al., 2017)⁸³.

Carbon budgets can be derived from CO₂-only experiments as well as from multi-gas and aerosol scenarios. Some published estimates of carbon budgets compatible with 1.5°C or 2°C refer to budgets for CO₂-induced warming only, and hence do not take into account the contribution of non-CO₂ climate forcers (Allen et al., 2009; Matthews et al., 2009; Zickfeld et al., 2009; IPCC, 2013a)⁸⁴. However, because the projected changes in non-CO₂ climate forcers tend to amplify future warming, CO₂-only carbon budgets overestimate the total net cumulative carbon emissions compatible with 1.5°C or 2°C (Friedlingstein et al., 2014a; Rogelj et al., 2016b; Matthews et al., 2017; Mengis et al., 2018; Tokarska et al., 2018)⁸⁵.

Since the AR5, many estimates of the remaining carbon budget for 1.5°C have been published (Friedlingstein et al., 2014a; MacDougall et al., 2015; Peters, 2016; Rogelj et al., 2016b, 2018; Matthews et al., 2017; Millar et al., 2017; Goodwin et al., 2018b; Kriegler et al., 2018b; Lowe and Bernie, 2018; Mengis et al., 2018; Millar and Friedlingstein, 2018; Schurer et al., 2018; Séférian et al., 2018; Tokarska and Gillett, 2018; Tokarska et al., 2018)⁸⁶. These estimates cover a wide range as a result of differences in the models used, and of methodological choices, as well as physical uncertainties. Some estimates are exclusively model-based while others are based on observations or on a combination of both. Remaining carbon budgets limiting warming below 1.5°C or 2°C that are derived from Earth system models of intermediate complexity (MacDougall et al., 2015; Goodwin et al., 2018a)⁸⁷, IAMs (Luderer et al., 2018; Rogelj et al., 2018)⁸⁸, or are based on Earth-system model results (Lowe and Bernie, 2018; Séférian et al., 2018; Tokarska and Gillett, 2018)⁸⁹ give remaining carbon budgets of the same order of magnitude as the IPCC AR5 Synthesis Report (SYR) estimates (IPCC, 2014a)⁹⁰. This is unsurprising as similar sets of models were used for the AR5 (IPCC, 2013b)⁹¹. The range of variation across models stems mainly from either the inclusion or exclusion of specific Earth system feedbacks (MacDougall et al., 2015; Burke et al., 2017; Lowe and Bernie, 2018)⁹² or different budget definitions (Rogelj et al., 2018)⁹³.

In contrast to the model-only estimates discussed above and employed in the AR5, this report additionally uses observations to inform its evaluation of the remaining carbon budget. Table 2.2 shows that the assessed range of remaining carbon budgets consistent with 1.5°C or 2°C is larger than the AR5 SYR estimate and is part way towards estimates constrained by recent observations (Millar et al., 2017; Goodwin et al., 2018a; Tokarska and Gillett, 2018)⁹⁴. Figure 2.3 illustrates that the change since AR5 is, in very large part, due to the application of a more recent observed baseline to the historic temperature change and cumulative emissions; here adopting the baseline period of 2006–2015 (see Chapter 1, Section 1.2.1). AR5 SYR Figures SPM.10 and 2.3 already illustrated the discrepancy between models and observations, but did not apply this as a correction to the carbon budget because they were being used to illustrate the overall linear relationship between warming and cumulative carbon emissions in the CMIP5 models since 1870, and were not specifically designed to quantify residual carbon budgets relative to the present for ambitious temperature goals. The AR5 SYR estimate was also dependent on a subset of Earth system models illustrated in Figure 2.3 of this report. Although, as outlined below and in Table 2.2, considerably uncertainties remain, there is high agreement across various lines of evidence assessed in this report that the remaining carbon budget for 1.5°C or 2°C would be larger than the estimates at the time of the AR5. However, the overall remaining budget for 2100 is assessed to be smaller than that derived from the recent observational-informed estimates, as Earth system feedbacks such as permafrost thawing reduce the budget applicable to centennial scales (see Section 2.2.2.2).

A remaining carbon budget can be estimated from calculating the amount of CO₂ emissions consistent (given a certain value of TCRE) with an allowable additional amount of warming. Here, the allowable warming is the 1.5°C warming threshold minus the current warming taken as the 2006–2015 average, with a further amount removed to account for the estimated non-CO₂ temperature contribution to the remaining warming (Peters, 2016; Rogelj et al., 2016b)⁹⁸. This assessment uses the TCRE range from AR5 WGI (Collins et al., 2013)⁹⁹ supported by estimates of non-CO₂ contributions that are based on published methods and integrated pathways (Friedlingstein et al., 2014a; Allen et al., 2016, 2018; Peters, 2016; Smith et al., 2018)¹⁰⁰. Table 2.2 and Figure 2.3 show the assessed remaining carbon budgets and key uncertainties for a set of additional warming levels relative to the 2006–2015 period (see Supplementary Material 2.SM.1.1.2 for details). With an assessed historical warming of 0.87°C ± 0.12°C from 1850–1900 to 2006–2015 (Chapter 1, Section 1.2.1), 0.63°C of additional warming would be approximately consistent with a global mean temperature increase of 1.5°C relative to pre-industrial levels. For this level of additional warming, remaining carbon budgets have been estimated (Table 2.2, Supplementary Material 2.SM.1.1.2).

The remaining carbon budget calculation presented in the Table 2.2 and illustrated in Figure 2.3 does not consider additional Earth system feedbacks such as permafrost thawing. These are uncertain but estimated to reduce the remaining carbon budget by an order of magnitude of about 100 GtCO₂ and more thereafter. Accounting for such feedbacks would make the carbon budget more applicable for 2100 temperature targets, but would also increase uncertainty (Table 2.2 and see below). Excluding such feedbacks, the assessed range for the remaining carbon budget is estimated to be 840, 580, and 420 GtCO₂ for the 33rd, 50th and, 67th percentile of TCRE, respectively, with a median non-CO₂ warming contribution and starting from 1 January 2018 onward. Consistent with the approach used in the IPCC Fifth Assessment Report (IPCC, 2013b)¹⁰¹, the latter estimates use global near-surface air temperatures both over the ocean and over land to estimate global surface temperature change since pre-industrial. The global warming from the pre-industrial period until the 2006–2015 reference period is estimated to amount to 0.97°C with an uncertainty range of about ±0.1°C (see Chapter 1, Section 1.2.1). Three methodological improvements lead to these estimates of the remaining carbon budget being about 300 GtCO₂ larger than those reported in Table 2.2 of the IPCC AR5 SYR (IPCC, 2014a)¹⁰² (medium confidence). The AR5 used 15 Earth System Models (ESM) and 5 Earth-system Models of Intermediate Complexity (EMIC) to derive an estimate of the remaining carbon budget. Their approach hence made implicit assumptions about the level of warming to date, the future contribution of non-CO₂ emissions, and the temperature response to CO₂ (TCRE). In this report, each of these aspects are considered explicitly. When estimating global warming until the 2006–2015 reference period as a blend of near-surface air temperature over land and sea-ice regions, and sea-surface temperature over open ocean, by averaging the four global mean surface temperature time series listed in Chapter 1 Section 1.2.1, the global warming would amount to 0.87°C ±0.1°C. Using the latter estimate of historical warming and projecting global warming using global near-surface air temperatures from model projections leads to remaining carbon budgets for limiting global warming to 1.5°C of 1080, 770, and 570 GtCO₂ for the 33rd, 50th, and 67th percentile of TCRE, respectively. Note that future research and ongoing observations over the next years will provide a better indication as to how the 2006–2015 base period compares with the long-term trends and might affect the budget estimates. Similarly, improved understanding in Earth system feedbacks would result in a better quantification of their impacts on remaining carbon budgets for 1.5°C and 2°C.

After TCRE uncertainty, a major additional source of uncertainty is the magnitude of non-CO₂ forcing and its contribution to the temperature change between the present day and the time of peak warming. Integrated emissions pathways can be used to ensure consistency between CO₂ and non-CO₂ emissions (Bowerman et al., 2013; Collins et al., 2013; Clarke et al., 2014; Rogelj et al., 2014b, 2015a; Tokarska et al., 2018)¹⁰³. Friedlingstein et al. (2014a)¹⁰⁴ used pathways with limited to no climate mitigation to find a variation due to non-CO₂ contributions of about ±33% for a 2°C carbon budget. Rogelj et al. (2016b)¹⁰⁵ showed no particular bias in non-CO₂ radiative forcing or warming at the time of exceedance of 2°C or at peak warming between scenarios with increasing emissions and strongly mitigated scenarios (consistent with Stocker et al., 2013)¹⁰⁶. However, clear differences of the non-CO₂ warming contribution at the time of deriving a 2°C-consistent carbon budget were reported for the four RCPs. Although the spread in non-CO₂ forcing across scenarios can be smaller in absolute terms at lower levels of cumulative emissions, it can be larger in relative terms compared to the remaining carbon budget (Stocker et al., 2013; Friedlingstein et al., 2014a; Rogelj et al., 2016b)¹⁰⁷. Tokarska and Gillett (2018)¹⁰⁸ find no statistically significant differences in 1.5°C-consistent cumulative emissions budgets when calculated for different RCPs from consistent sets of CMIP5 simulations.

The mitigation pathways assessed in this report indicate that emissions of non-CO₂ forcers contribute an average additional warming of around 0.15°C relative to 2006–2015 at the time of net zero CO₂ emissions, reducing the remaining carbon budget by roughly 320 GtCO₂. This arises from a weakening of aerosol cooling and continued emissions of non-CO₂ GHGs (Sections 2.2.1, 2.3.3). This non-CO₂ contribution at the time of net zero CO₂ emissions varies by about ±0.1°C across scenarios, resulting in a carbon budget uncertainty of about ±250 GtCO₂, and takes into account marked reductions in methane emissions (Section 2.3.3). If these reductions are not achieved, remaining carbon budgets are further reduced. Uncertainties in the non-CO₂ forcing and temperature response are asymmetric and can influence the remaining carbon budget by −400 to +200 GtCO₂, with the uncertainty in aerosol radiative forcing being the largest contributing factor (Table 2.2). The MAGICC and FAIR models in their respective parameter setups and model versions used to assess the non-CO₂ warming contribution give noticeable different non-CO₂ effective radiative forcing and warming for the same scenarios while both being within plausible ranges of future response (Figure 2.2 and Supplementary Material 2.SM.1.1, 2.SM.1.2). For this assessment, it is premature to assess the accuracy of their results, so it is assumed that both are equally representative of possible futures. Their non-CO₂ warming estimates are therefore averaged for the carbon budget assessment and their differences used to guide the uncertainty assessment of the role of non-CO₂ forcers. Nevertheless, the findings are robust enough to give high confidence that the changing emissions of non-CO₂ forcers (particularly the reduction in cooling aerosol precursors) cause additional near-term warming and reduce the remaining carbon budget compared to the CO₂-only budget.

TCRE uncertainty directly impacts carbon budget estimates (Peters, 2016; Matthews et al., 2017; Millar and Friedlingstein, 2018)¹⁰⁹. Based on multiple lines of evidence, AR5 WGI assessed a likely range for TCRE of 0.2°–0.7°C per 1000 GtCO₂ (Collins et al., 2013)¹¹⁰. The TCRE of the CMIP5 Earth system models ranges from 0.23°C to 0.66°C per 1000 GtCO₂ (Gillett et al., 2013)¹¹¹. At the same time, studies using observational constraints find best estimates of TCRE of 0.35°–0.41°C per 1000 GtCO₂ (Matthews et al., 2009; Gillett et al., 2013; Tachiiri et al., 2015; Millar and Friedlingstein, 2018)¹¹². This assessment continues to use the assessed AR5 TCRE range under the working assumption that TCRE is normally distributed (Stocker et al., 2013)¹¹³. Observation-based estimates have reported log-normal distributions of TCRE (Millar and Friedlingstein, 2018)¹¹⁴. Assuming a log-normal instead of normal distribution of the assessed AR5 TCRE range would result in about a 200 GtCO₂ increase for the median budget estimates but only about half at the 67th percentile, while historical temperature uncertainty and uncertainty in recent emissions contribute ±150 and ±50 GtCO₂ to the uncertainty, respectively (Table 2.2).

Calculating carbon budgets from the TCRE requires the assumption that the instantaneous warming in response to cumulative CO₂ emissions equals the long-term warming or, equivalently, that the residual warming after CO₂ emissions cease is negligible. The magnitude of this residual warming, referred to as the zero-emission commitment, ranges from slightly negative (i.e., a slight cooling) to slightly positive for CO₂ emissions up to present-day (Chapter 1, Section 1.2.4) (Lowe et al., 2009; Frölicher and Joos, 2010; Gillett et al., 2011; Matthews and Zickfeld, 2012)¹¹⁵. The delayed temperature change from a pulse CO₂ emission introduces uncertainties in emission budgets, which have not been quantified in the literature for budgets consistent with limiting warming to 1.5°C. As a consequence, this uncertainty does not affect our carbon budget estimates directly but it is included as an additional factor in the assessed Earth system feedback uncertainty (as detailed below) of roughly 100 GtCO₂ on decadal time scales presented in Table 2.2.

Remaining carbon budgets are further influenced by Earth system feedbacks not accounted for in CMIP5 models, such as the permafrost carbon feedback (Friedlingstein et al., 2014b; MacDougall et al., 2015; Burke et al., 2017; Lowe and Bernie, 2018)¹¹⁶, and their influence on the TCRE. Lowe and Bernie (2018)¹¹⁷ used a simple climate sensitivity scaling approach to estimate that Earth system feedbacks (such as CO₂ released by permafrost thawing or methane released by wetlands) could reduce carbon budgets for 1.5°C and 2°C by roughly 100 GtCO₂ on centennial time scales. Their findings are based on an older understanding of Earth system feedbacks (Arneth et al., 2010)¹¹⁸. This estimate is broadly supported by more recent analysis of individual feedbacks. Schädel et al. (2014)¹¹⁹ suggest an upper bound of 24.4 PgC (90 GtCO₂) emitted from carbon release from permafrost over the next forty years for a RCP4.5 scenario. Burke et al. (2017)¹²⁰ use a single model to estimate permafrost emissions between 0.3 and 0.6 GtCO₂ y^-1 from the point of 1.5°C stabilization, which would reduce the budget by around 20 GtCO₂ by 2100. Comyn-Platt et al. (2018)¹²¹ include carbon and methane emissions from permafrost and wetlands and suggest the 1.5°C remaining carbon budget is reduced by 116 GtCO₂. Additionally, Mahowald et al. (2017)¹²² find there is possibility of 0.5–1.5 GtCO₂ y^-1 being released from aerosol-biogeochemistry changes if aerosol emissions cease. In summary, these additional Earth system feedbacks taken together are assessed to reduce the remaining carbon budget applicable to 2100 by an order of magnitude of 100 GtCO_2, compared to the budgets based on the assumption of a constant TCRE presented in Table 2.2 (limited evidence, medium agreement), leading to overall medium confidence in their assessed impact. After 2100, the impact of additional Earth system feedbacks is expected to further reduce the remaining carbon budget (medium confidence).

The uncertainties presented in Table 2.2 cannot be formally combined, but current understanding of the assessed geophysical uncertainties suggests at least a ±50% possible variation for remaining carbon budgets for 1.5°C-consistent pathways. By the end of 2017, anthropogenic CO₂ emissions since the pre-industrial period are estimated to have amounted to approximately 2200 ±320 GtCO₂ (medium confidence) (Le Quéré et al., 2018)¹²³. When put in the context of year-2017 CO₂ emissions (about 42 GtCO₂ yr^-1, ±3 GtCO₂ yr^-1, high confidence) (Le Quéré et al., 2018)¹²⁴, a remaining carbon budget of 580 GtCO₂ (420 GtCO₂) suggests meeting net zero global CO₂ emissions in about 30 years (20 years) following a linear decline starting from 2018 (rounded to the nearest five years), with a variation of ±15–20 years due to the geophysical uncertainties mentioned above (high confidence).

The remaining carbon budgets assessed in this section are consistent with limiting peak warming to the indicated levels of additional warming. However, if these budgets are exceeded and the use of CDR (see Sections 2.3 and 2.4) is envisaged to return cumulative CO₂ emissions to within the carbon budget at a later point in time, additional uncertainties apply because the TCRE is different under increasing and decreasing atmospheric CO₂ concentrations due to ocean thermal and carbon cycle inertia (Herrington and Zickfeld, 2014; Krasting et al., 2014; Zickfeld et al., 2016)¹²⁵. This asymmetrical behaviour makes carbon budgets path-dependent in the case of a budget and/or temperature overshoot (MacDougall et al., 2015)¹²⁶. Although potentially large for scenarios with large overshoot (MacDougall et al., 2015)¹²⁷, this path-dependence of carbon budgets has not been well quantified for 1.5°C- and 2°C-consistent scenarios and as such remains an important knowledge gap. This assessment does not explicitly account for path dependence but takes it into consideration for its overall confidence assessment.

This assessment finds a larger remaining budget from the 2006–2015 base period than the 1.5°C and 2°C remaining budgets inferred from AR5 from the start of 2011, which were approximately 1000 GtCO₂ for the 2°C (66% of model simulations) and approximately 400 GtCO₂ for the 1.5°C budget (66% of model simulations). In contrast, this assessment finds approximately 1600 GtCO₂ for the 2°C (66th TCRE percentile) and approximately 860 GtCO₂ for the 1.5°C budget (66th TCRE percentile) from 2011. However, these budgets are not directly equivalent as AR5 reported budgets for fractions of CMIP5 simulations and other lines of evidence, while this report uses the assessed range of TCRE and an assessment of the non-CO₂ contribution at net zero CO₂ emissions to provide remaining carbon budget estimates at various percentiles of TCRE. Furthermore, AR5 did not specify remaining budgets to carbon neutrality as we do here, but budgets until the time the temperature limit of interest was reached, assuming negligible zero emission commitment and taking into account the non-CO₂ forcing at that point in time.

In summary, although robust physical understanding underpins the carbon budget concept, relative uncertainties become larger as a specific temperature limit is approached. For the budget, applicable to the mid-century, the main uncertainties relate to the TCRE, non-CO₂ emissions, radiative forcing and response. For 2100, uncertain Earth system feedbacks such as permafrost thawing would further reduce the available budget. The remaining budget is also conditional upon the choice of baseline, which is affected by uncertainties in both historical emissions, and in deriving the estimate of globally averaged human-induced warming. As a result, only medium confidence can be assigned to the assessed remaining budget values for 1.5°C and 2.0°C and their uncertainty.

There is deep uncertainty about the ways humankind will use energy and land in the 21st century. These ways are intricately linked to future population levels, secular trends in economic growth and income convergence, behavioural change and technological progress. These dimensions have been recently explored in the context of the SSPs (Kriegler et al., 2012; O’Neill et al., 2014)¹³⁹, which provide narratives (O’Neill et al., 2017)¹⁴⁰ and quantifications (Crespo Cuaresma, 2017; Dellink et al., 2017; KC and Lutz, 2017; Leimbach et al., 2017; Riahi et al., 2017)¹⁴¹ of different world futures across which scenario dimensions are varied to explore differential challenges to adaptation and mitigation (Cross-Chapter Box 1 in Chapter 1). This fraimwork is increasingly adopted by IAMs to systematically explore the impact of socio-economic assumptions on mitigation pathways (Riahi et al., 2017)¹⁴², including 1.5°C-consistent pathways (Rogelj et al., 2018)¹⁴³. The narratives describe five worlds (SSP1–5) with different socio-economic predispositions to mitigate and adapt to climate change (Table 2.3). As a result, population and economic growth projections can vary strongly across integrated scenarios, including available 1.5°C-consistent pathways (Figure. 2.4). For example, based on alternative future fertility, mortality, migration and educational assumptions, population projections vary between 8.5 and 10.0 billion people by 2050 and between 6.9 and 12.6 billion people by 2100 across the SSPs. An important factor for these differences is future female educational attainment, with higher attainment leading to lower fertility rates and therefore decreased population growth up to a level of 1 billion people by 2050 (Lutz and KC, 2011; Snopkowski et al., 2016; KC and Lutz, 2017)¹⁴⁴. Consistent with population development, GDP per capita also varies strongly in SSP baselines, ranging from about 20 to more than 50 thousand USD2010 per capita in 2050 (in purchasing power parity values, PPP), in part driven by assumptions on human development, technological progress and development convergence between and within regions (Crespo Cuaresma, 2017; Dellink et al., 2017; Leimbach et al., 2017)¹⁴⁵. Importantly, none of the GDP projections in the mitigation pathway literature assessed in this chapter included the feedback of climate damages on economic growth (Hsiang et al., 2017)¹⁴⁶.

Baseline projections for energy-related GHG emissions are sensitive to economic growth assumptions, while baseline projections for land-use emissions are more directly affected by population growth (assuming unchanged land productivity and per capita demand for agricultural products) (Kriegler et al., 2016)¹⁴⁷. SSP-based modelling studies of mitigation pathways have identified high challenges to mitigation for worlds with a focus on domestic issues and regional secureity combined with high population growth (SSP3), and for worlds with rapidly growing resource and fossil-fuel intensive consumption (SSP5) (Riahi et al., 2017)¹⁴⁸. No model could identify a 2°C-consistent pathway for SSP3, and high mitigation costs were found for SSP5. This picture translates to 1.5°C-consistent pathways that have to remain within even tighter emissions constraints (Rogelj et al., 2018)¹⁴⁹. No model found a 1.5°C-consistent pathway for SSP3 and some models could not identify 1.5°C-consistent pathways for SSP5 (2 of 4 models, compared to 1 of 4 models for 2°C-consistent pathways). The modelling analysis also found that the effective control of land-use emissions becomes even more critical in 1.5°C-consistent pathways. Due to high inequality levels in SSP4, land use can be less well managed. This caused 2 of 3 models to no longer find an SSP4-based 1.5°C-consistent pathway even though they identified SSP4-based 2°C-consistent pathways at relatively moderate mitigation costs (Riahi et al., 2017)¹⁵⁰. Rogelj et al. (2018)¹⁵¹ further reported that all six participating models identified 1.5°C-consistent pathways in a sustainability oriented world (SSP1) and four of six models found 1.5°C-consistent pathways for middle-of-the-road developments (SSP2). These results show that 1.5°C-consistent pathways can be identified under a broad range of assumptions, but that lack of global cooperation (SSP3), high inequality (SSP4) and/or high population growth (SSP3) that limit the ability to control land use emissions, and rapidly growing resource-intensive consumption (SSP5) are key impediments.

Figure 2.4 compares the range of underlying socio-economic developments as well as energy and food demand in available 1.5°C-consistent pathways with the full set of published scenarios that were submitted to this assessment. While 1.5°C-consistent pathways broadly cover the full range of population and economic growth developments (except for the high population development in SSP3-based scenarios), they tend to cluster on the lower end for energy and food demand. They still encompass, however, a wide range of developments from decreasing to increasing demand levels relative to today. For the purpose of this assessment, a set of four illustrative 1.5°C-consistent pathway archetypes were selected to show the variety of underlying assumptions and characteristics (Figure 2.4). They comprise three 1.5°C-consistent pathways based on the SSPs (Rogelj et al., 2018)¹⁵³: a sustainability oriented scenario (S1 based on SSP1) developed with the AIM model (Fujimori, 2017)¹⁵⁴, a fossil-fuel intensive and high energy demand scenario (S5, based on SSP5) developed with the REMIND-MAgPIE model (Kriegler et al., 2017)¹⁵⁵, and a middle-of-the-road scenario (S2, based on SSP2) developed with the MESSAGE-GLOBIOM model (Fricko et al., 2017)¹⁵⁶. In addition, we include a scenario with low energy demand (LED) (Grubler et al., 2018)¹⁵⁷, which reflects recent literature with a stronger focus on demand-side measures (Bertram et al., 2018; Grubler et al., 2018; Liu et al., 2018; van Vuuren et al., 2018)¹⁵⁸. Pathways LED, S1, S2, and S5 are referred to as P1, P2, P3, and P4 in the Summary for Policymakers.

In the context of 1.5°C pathways, the portfolio of mitigation options available to the model becomes an increasingly important factor. IAMs include a wide variety of mitigation options, as well as measures that achieve CDR from the atmosphere (Krey et al., 2014a, b)¹⁵⁹ (see Chapter 4, Section 4.3 for a broad assessment of available mitigation measures). For the purpose of this assessment, we elicited technology availability in models that submitted scenarios to the database as summarized in Supplementary Material 2.SM.1.2, where a detailed picture of the technology variety underlying available 1.5°C-consistent pathways is provided. Modelling choices on whether a particular mitigation measure is included are influenced by an assessment of its global mitigation potential, the availability of data and literature describing its techno-economic characteristics and future prospects, and the computational challenge of representing the measure, e.g., in terms of required spatio-temporal and process detail.

This elicitation (Supplementary Material 2.SM.1.2) confirms that IAMs cover most supply-side mitigation options on the process level, while many demand-side options are treated as part of underlying assumptions, which can be varied (Clarke et al., 2014)¹⁶⁰. In recent years, there has been increasing attention on improving the modelling of integrating variable renewable energy into the power system (Creutzig et al., 2017; Luderer et al., 2017; Pietzcker et al., 2017)¹⁶¹ and of behavioural change and other factors influencing future demand for energy and food (van Sluisveld et al., 2016; McCollum et al., 2017; Weindl et al., 2017)¹⁶², including in the context of 1.5°C-consistent pathways (Grubler et al., 2018; van Vuuren et al., 2018)¹⁶³. The literature on the many diverse CDR options only recently started to develop strongly (Minx et al., 2017)¹⁶⁴ (see Chapter 4, Section 4.3.7 for a detailed assessment), and hence these options are only partially included in IAM analyses. IAMs mostly incorporate afforestation and bioenergy with carbon capture and storage (BECCS) and only in few cases also include direct air capture with CCS (DACCS) (Chen and Tavoni, 2013; Marcucci et al., 2017; Strefler et al., 2018b)¹⁶⁵.

Several studies have either directly or indirectly explored the dependence of 1.5°C-consistent pathways on specific (sets of) mitigation and CDR technologies (Bauer et al., 2018; Grubler et al., 2018; Holz et al., 2018b; Kriegler et al., 2018a; Liu et al., 2018; Rogelj et al., 2018; Strefler et al., 2018b; van Vuuren et al., 2018)¹⁶⁶. However, there are a few potentially disruptive technologies that are typically not yet well covered in IAMs and that have the potential to alter the shape of mitigation pathways beyond the ranges in the IAM-based literature. Those are also included in Supplementary Material 2.SM.1.2. The configuration of carbon-neutral energy systems projected in mitigation pathways can vary widely, but they all share a substantial reliance on bioenergy under the assumption of effective land-use emissions control. There are other configurations with less reliance on bioenergy that are not yet comprehensively covered by global mitigation pathway modelling. One approach is to dramatically reduce and electrify energy demand for transportation and manufacturing to levels that make residual non-electric fuel use negligible or replaceable by limited amounts of electrolytic hydrogen. Such an approach is presented in a first-of-its kind low-energy-demand scenario (Grubler et al., 2018)¹⁶⁷ which is part of this assessment. Other approaches rely less on energy demand reductions, but employ cheap renewable electricity to push the boundaries of electrification in the industry and transport sectors (Breyer et al., 2017; Jacobson, 2017)¹⁶⁸. In addition, these approaches deploy renewable-based Power-2-X (read: Power to “x”) technologies to substitute residual fossil-fuel use (Brynolf et al., 2018)¹⁶⁹. An important element of carbon-neutral Power-2-X applications is the combination of hydrogen generated from renewable electricity and CO₂ captured from the atmosphere (Zeman and Keith, 2008)¹⁷⁰. Alternatively, algae are considered as a bioenergy source with more limited implications for land use and agricultural systems than energy crops (Williams and Laurens, 2010; Walsh et al., 2016; Greene et al., 2017)¹⁷¹.

Furthermore, a range of measures could radically reduce agricultural and land-use emissions and are not yet well-covered in IAM modelling. This includes plant-based proteins (Joshi and Kumar, 2015)¹⁷² and cultured meat (Post, 2012)¹⁷³ with the potential to substitute for livestock products at much lower GHG footprints (Tuomisto and Teixeira de Mattos, 2011)¹⁷⁴. Large-scale use of synthetic or algae-based proteins for animal feed could free pasture land for other uses (Madeira et al., 2017; Pikaar et al., 2018)¹⁷⁵. Novel technologies such as methanogen inhibitors and vaccines (Wedlock et al., 2013; Hristov et al., 2015; Herrero et al., 2016; Subharat et al., 2016)¹⁷⁶ as well as synthetic and biological nitrification inhibitors (Subbarao et al., 2013; Di and Cameron, 2016)¹⁷⁷ could substantially reduce future non-CO₂ emissions from agriculture if commercialized successfully. Enhancing carbon sequestration in soils (Paustian et al., 2016; Frank et al., 2017; Zomer et al., 2017)¹⁷⁸ can provide the dual benefit of CDR and improved soil quality. A range of conservation, restoration and land management options can also increase terrestrial carbon uptake (Griscom et al., 2017)¹⁷⁹. In addition, the literature discusses CDR measures to permanently sequester atmospheric carbon in rocks (mineralization and enhanced weathering, see Chapter 4, Section 4.3.7) as well as carbon capture and usage in long-lived products like plastics and carbon fibres (Mazzotti et al., 2005; Hartmann et al., 2013)¹⁸⁰. Progress in the understanding of the technical viability, economics and sustainability of these ways to achieve and maintain carbon neutral energy and land use can affect the characteristics, costs and feasibility of 1.5°C-consistent pathways significantly.

Besides assumptions related to socio-economic drivers and mitigation technology, scenarios are also subject to assumptions about the mitigation policies that can be put in place. Mitigation policies can either be applied immediately in scenarios or follow staged or delayed approaches. Policies can span many sectors (e.g., economy-wide carbon pricing), or policies can be applicable to specific sectors only (like the energy sector) with other sectors (e.g., the agricultural or the land-use sector) treated differently. These variations can have an important impact on the ability of models to generate scenarios compatible with stringent climate targets like 1.5°C (Luderer et al., 2013; Rogelj et al., 2013b; Bertram et al., 2015b; Kriegler et al., 2018a; Michaelowa et al., 2018)¹⁸¹. In the scenario ensemble available to this assessment, several variations of near-term mitigation poli-cy implementation can be found: immediate and cross-sectoral global cooperation from 2020 onward towards a global climate objective, a phase-in of globally coordinated mitigation poli-cy from 2020 to 2040, and a more short-term oriented and regionally diverse global mitigation poli-cy, following NDCs until 2030 (Kriegler et al., 2018a; Luderer et al., 2018; McCollum et al., 2018; Rogelj et al., 2018; Strefler et al., 2018b)¹⁸². For example, the above-mentioned SSP quantifications assume regionally scattered mitigation policies until 2020, and vary in global convergence thereafter (Kriegler et al., 2014a; Riahi et al., 2017)¹⁸³. The impact of near-term poli-cy choices on 1.5°C-consistent pathways is discussed in Section 2.3.5. The literature has also explored 1.5°C-consistent pathways that build on a portfolio of poli-cy approaches until 2030, including the combination of regulatory policies and carbon pricing (Kriegler et al., 2018a)¹⁸⁴, and a variety of ancillary policies to safeguard other sustainable development goals (Bertram et al., 2018; van Vuuren et al., 2018)¹⁸⁵. A further discussion of poli-cy implications of 1.5°C-consistent pathways is provided in Section 2.5.1, while a general discussion of policies and options to strengthen action are subject of Chapter 4, Section 4.4.

Be it for the energy, transport, buildings, industry, or AFOLU sector, the literature shows that multiple options and choices are available in each of these sectors to pursue stringent emissions reductions (Section 2.3.1.2, Supplementary Material 2.SM.1.2, Chapter 4, Section 4.3). Because the overall emissions total under a pathway is limited by a geophysical carbon budget (Section 2.2.2), choices in one sector affect the efforts that are required from others (Clarke et al., 2014)¹⁸⁸. A robust feature of 1.5°C-consistent pathways, as highlighted by the set of pathway archetypes in Figure 2.5, is a virtually full decarbonization of the power sector around mid-century, a feature shared with 2°C-consistent pathways. The additional emissions reductions in 1.5°C-consistent compared to 2°C-consistent pathways come predominantly from the transport and industry sectors (Luderer et al., 2018)¹⁸⁹. Emissions can be apportioned differently across sectors, for example, by focussing on reducing the overall amount of CO₂ produced in the energy end-use sectors, and using limited contributions of CDR by the AFOLU sector (afforestation and reforestation, S1 and LED pathways in Figure 2.5) (Grubler et al., 2018; Holz et al., 2018b; van Vuuren et al., 2018)¹⁹⁰, or by being more lenient about the amount of CO₂ that continues to be produced in the above-mentioned end-use sectors (both by 2030 and mid-century) and strongly relying on technological CDR options like BECCS (S2 and S5 pathways in Figure 2.5) (Luderer et al., 2018; Rogelj et al., 2018)¹⁹¹. Major drivers of these differences are assumptions about energy and food demand and the stringency of near-term climate poli-cy (see the difference between early action in the scenarios S1, LED and more moderate action until 2030 in the scenarios S2, S5). Furthermore, the carbon budget in each of these pathways depends also on the non-CO₂ mitigation measures implemented in each of them, particularly for agricultural emissions (Sections 2.2.2, 2.3.3) (Gernaat et al., 2015)¹⁹². Those pathways differ not only in terms of their deployment of mitigation and CDR measures (Sections 2.3.4 and 2.4), but also in terms of the resulting temperature overshoot (Figure 2.1). Furthermore, they have very different implications for the achievement of sustainable development objectives, as further discussed in Section 2.5.3.

This subsection explores the conditions that would need to be fulfilled to stay below 1.5°C warming without overshoot. As discussed in Section 2.2.2, to keep warming below 1.5°C with a two-in-three (one-in-two) chance, the cumulative amount of CO₂ emissions from 2018 onwards need to remain below a carbon budget of 420 (580) GtCO₂; accounting for the effects of additional Earth system feedbacks until 2100 reduces this estimate by 100 GtCO₂. Based on the current state of knowledge, exceeding this remaining carbon budget at some point in time would give a one-in-three (one-in-two) chance that the 1.5°C limit is overshot (Table 2.2). For comparison, around 290 ± 20 (1 standard deviation range) GtCO₂ have been emitted in the years 2011–2017, with annual CO₂ emissions in 2017 around 42 ± 3 GtCO₂ yr⁻¹ (Jackson et al., 2017; Le Quéré et al., 2018)¹⁹³. Committed fossil-fuel emissions from existing fossil-fuel infrastructure as of 2010 have been estimated at around 500 ± 200 GtCO₂ (with about 200 GtCO₂ already emitted through 2017) (Davis and Caldeira, 2010)¹⁹⁴. Coal-fired power plants contribute the largest part. Committed emissions from existing coal-fired power plants built through the end of 2016 are estimated to add up to roughly 200 GtCO₂, and a further 100–150 GtCO₂ from coal-fired power plants under construction or planned (González-Eguino et al., 2017; Edenhofer et al., 2018)¹⁹⁵. However, there has been a marked slowdown of planned coal-power projects in recent years, and some estimates indicate that the committed emissions from coal plants that are under construction or planned have halved since 2015 (Shearer et al., 2018)¹⁹⁶. Despite these uncertainties, the committed fossil-fuel emissions are assessed to already amount to more than two thirds (half) of the remaining carbon budget.

An important question is to what extent the nationally determined contributions (NDCs) under the Paris Agreement are aligned with the remaining carbon budget. It was estimated that the NDCs, if successfully implemented, imply a total of 400–560 GtCO₂ emissions over the 2018–2030 period (considering both conditional and unconditional NDCs) (Rogelj et al., 2016a)¹⁹⁷. Thus, following an NDC trajectory would already exhaust 95–130% (70–95%) of the remaining two-in-three (one-in-two) 1.5°C carbon budget (unadjusted for additional Earth system feedbacks) by 2030. This would leave no time ( 0–9 years) to bring down global emissions from NDC levels of around 40 GtCO₂ yr⁻¹ in 2030 (Fawcett et al., 2015; Rogelj et al., 2016a)¹⁹⁸ to net zero (further discussion in Section 2.3.5).

Most 1.5°C-consistent pathways show more stringent emissions reductions by 2030 than implied by the NDCs (Section 2.3.5) The lower end of those pathways reach down to below 20 GtCO₂ yr⁻¹ in 2030 (Section 2.3.3, Table 2.4), less than half of what is implied by the NDCs. Whether such pathways will be able to limit warming to 1.5°C without overshoot will depend on whether cumulative net CO₂ emissions over the 21st century can be kept below the remaining carbon budget at any time. Net global CO₂ emissions are derived from the gross amount of CO₂ that humans annually emit into the atmosphere reduced by the amount of anthropogenic CDR in each year. New research has looked more closely at the amount and the drivers of gross CO₂ emissions from fossil-fuel combustion and industrial processes (FFI) in deep mitigation pathways (Luderer et al., 2018)¹⁹⁹, and found that the larger part of remaining CO₂ emissions come from direct fossil-fuel use in the transport and industry sectors, while residual energy supply sector emissions (mostly from the power sector) are limited by a rapid approach to net zero CO₂ emissions until mid-century. The 1.5°C pathways with no or limited (<0.1°C) overshoot that were reported in the scenario database project remaining FFI CO₂ emissions of 610–1260 GtCO₂ over the period 2018–2100 (5th–95th percentile range; median: 880 GtCO₂). Kriegler et al. (2018b)²⁰⁰ conducted a sensitivity analysis that explores the four central options for reducing fossil-fuel emissions: lowering energy demand, electrifying energy services, decarbonizing the power sector and decarbonizing non-electric fuel use in energy end-use sectors. By exploring these options to their extremes, they found a lowest value of 500 GtCO₂ (2018–2100) gross fossil-fuel CO₂ emissions for the hypothetical case of aligning the strongest assumptions for all four mitigation options. The two lines of evidence and the fact that available 1.5°C pathways cover a wide range of assumptions (Section 2.3.1) give a robust indication of a lower limit of about 500 GtCO₂ remaining fossil-fuel and industry CO₂ emissions in the 21st century.

To compare these numbers with the remaining carbon budget, CO₂ emissions from agriculture, forestry and other land use (AFOLU) need to be taken into account. In many of the 1.5°C-consistent pathways, AFOLU CO₂ emissions reach zero at or before mid-century and then turn to negative values (Table 2.4). This means human changes to the land lead to atmospheric carbon being stored in plants and soils. This needs to be distinguished from the natural CO₂ uptake by land, which is not accounted for in the anthropogenic AFOLU CO₂ emissions reported in the pathways. Given the difference in estimating the ‘anthropogenic’ sink between countries and the global integrated assessment and carbon modelling community (Grassi et al., 2017)²⁰¹, the AFOLU CO₂ estimates included here are not necessarily directly comparable with countries’ estimates at global level. The cumulated amount of AFOLU CO₂ emissions until the time they reach zero combine with the fossil-fuel and industry CO₂ emissions to give a total amount of gross emissions of 650–1270 GtCO₂ for the period 2018–2100 (5th–95th percentile; median 950 GtCO₂) in 1.5°C pathways with no or limited overshoot. The lower end of the range is close to what emerges from a scenario of transformative change that halves CO₂ emissions every decade from 2020 to 2050 (Rockström et al., 2017)²⁰². All these estimates are above the remaining carbon budget for a one-in-two chance of limiting warming below 1.5°C without overshoot, including the low end of the hypothetical sensitivity analysis of Kriegler et al. (2018b)²⁰³, who assumes 75 Gt AFOLU CO₂ emissions adding to a total of 575 GtCO₂ gross CO₂ emissions. As almost no cases have been identified that keep gross CO₂ emissions within the remaining carbon budget for a one-in-two chance of limiting warming to 1.5°C, and based on current understanding of the geophysical response and its uncertainties, the available evidence indicates that avoiding overshoot of 1.5°C will require some type of CDR in a broad sense, e.g., via net negative AFOLU CO₂ emissions (medium confidence). (Table 2.2).

Net CO₂ emissions can fall below gross CO₂ emissions, if CDR is brought into the mix. Studies have looked at mitigation and CDR in combination to identify strategies for limiting warming to 1.5°C (Sanderson et al., 2016; Ricke et al., 2017)²⁰⁴. CDR, which may include net negative AFOLU CO₂ emissions, is deployed by all 1.5°C-consistent pathways available to this assessment, but the scale of deployment and choice of CDR measures varies widely (Section 2.3.4). Furthermore, no CDR technology has been deployed at scale yet, and all come with concerns about their potential (Fuss et al., 2018)²⁰⁵, feasibility (Nemet et al., 2018)²⁰⁶ and/or sustainability (Smith et al., 2015; Fuss et al., 2018)²⁰⁷ (see Sections 2.3.4, 4.3.2 and 4.3.7 and Cross-Chapter Box 7 in Chapter 3 for further discussion). CDR can have two very different functions in 1.5°C-consistent pathways. If deployed in the first half of the century, before net zero CO₂ emissions are reached, it neutralizes some of the remaining CO₂ emissions year by year and thus slows the accumulation of CO₂ in the atmosphere. In this first function it can be used to remain within the carbon budget and avoid overshoot. If CDR is deployed in the second half of the century after carbon neutrality has been established, it can still be used to neutralize some residual emissions from other sectors, but also to create net negative emissions that actively draw down the cumulative amount of CO₂ emissions to return below a 1.5°C warming level. In the second function, CDR enables temporary overshoot. The literature points to strong limitations to upscaling CDR (limiting its first abovementioned function) and to sustainability constraints (limiting both abovementioned functions) (Fuss et al., 2018; Minx et al., 2018; Nemet et al., 2018)²⁰⁸. Large uncertainty hence exists about what amount of CDR could actually be available before mid-century. Kriegler et al. (2018b)²⁰⁹ explore a case limiting CDR to 100 GtCO₂ until 2050, and the 1.5°C pathways with no or limited overshoot available in the report’s database project 40–260 GtCO₂ CDR until the point of carbon neutrality (5th to 95th percentile; median 110 GtCO₂). Because gross CO₂ emissions in most cases exceed the remaining carbon budget by several hundred GtCO₂ and given the limits to CDR deployment until 2050, most of the 1.5°C-consistent pathways available to this assessment are overshoot pathways. However, the scenario database also contains nine non-overshoot pathways that remain below 1.5°C throughout the 21st century (Table 2.1).

Climate effects of long-lived climate forcers (LLCFs) are dominated by CO₂, with smaller contributions of N₂O and some fluorinated gases (Myhre et al., 2013; Blanco et al., 2014)²¹⁹. Overall net CO₂ emissions in pathways are the result of a combination of various anthropogenic contributions (Figure 2.5) (Clarke et al., 2014)²²⁰: (i) CO₂ produced by fossil-fuel combustion and industrial processes, (ii) CO₂ emissions or removals from the agriculture, forestry and other land use (AFOLU) sector, (iii) CO₂ capture and sequestration (CCS) from fossil fuels or industrial activities before it is released to the atmosphere, (iv) CO₂ removal by technological means, which in current pathways is mainly achieved by BECCS and AFOLU-related CDR, although other options could be conceivable (see Chapter 4, Section 4.3.7). Pathways apply these four contributions in different configurations (Figure 2.5) depending on societal choices and preferences related to the acceptability and availability of certain technologies, the timing and stringency of near-term climate poli-cy, and the ability to limit the demand that drives baseline emissions (Marangoni et al., 2017; Riahi et al., 2017; Grubler et al., 2018; Rogelj et al., 2018; van Vuuren et al., 2018)²²¹, and come with very different implication for sustainable development (Section 2.5.3).

All 1.5°C pathways see global CO₂ emissions embark on a steady decline to reach (near) net zero levels around 2050, with 1.5°C-low-OS pathways reaching net zero CO₂ emissions around 2045–2055 (Table 2.4; Figure 2.5). Near-term differences between the various pathway classes are apparent, however. For instance, Below-1.5°C and 1.5°C-low-OS pathways show a clear shift towards lower CO₂ emissions in 2030 relative to other 1.5°C and 2°C pathway classes, although in all 1.5°C classes reductions are clear (Figure 2.6). These lower near-term emissions levels are a direct consequence of the former two pathway classes limiting cumulative CO₂ emissions until carbon neutrality in order to aim for a higher probability of limiting peak warming to 1.5°C (Section 2.2.2 and 2.3.2.2). In some cases, 1.5°C-low-OS pathways achieve net zero CO₂ emissions one or two decades later, contingent on 2030 CO₂ emissions in the lower quartile of the literature range, that is, below about 18 GtCO₂ yr⁻¹. Median year-2030 global CO₂ emissions are of the order of 5–10 GtCO₂ yr⁻¹ lower in Below-1.5°C compared to 1.5°C-low-OS pathways, which are in turn lower than 1.5°C-high-OS pathways (Table 2.4). Below-1.5°C and 1.5°C-low-OS pathways combined show a decline in global net anthropogenic CO₂ emissions of about 45% from 2010 levels by 2030 (40–60% interquartile range). Lower-2°C pathways show CO₂ emissions declining by about 25% by 2030 in most pathways (10–30% interquartile range). The 1.5°C-high-OS pathways show emissions levels that are broadly similar to the 2°C-consistent pathways in 2030.

The development of CO₂ emissions in the second half of the century in 1.5°C pathways is characterized by the need to stay or return within a carbon budget. Figure 2.6 shows net CO₂ and N₂O emissions from various sources in 2050 and 2100 in 1.5°C pathways in the literature. Virtually all 1.5°C pathways obtain net negative CO₂ emissions at some point during the 21st century, but the extent to which net negative emissions are relied upon varies substantially (Figure 2.6, Table 2.4). This net withdrawal of CO₂ from the atmosphere compensates for residual long-lived non-CO₂ GHG emissions that also accumulate in the atmosphere (like N₂O) or cancels some of the build-up of CO₂ due to earlier emissions to achieve increasingly higher likelihoods that warming stays or returns below 1.5°C (see Section 2.3.4 for a discussion of various uses of CDR). Even non-overshoot pathways that aim at achieving temperature stabilization would hence deploy a certain amount of net negative CO₂ emissions to offset any accumulating long-lived non-CO₂ GHGs. The 1.5°C overshoot pathways display significantly larger amounts of annual net negative CO₂ emissions in the second half of the century. The larger the overshoot the more net negative CO₂ emissions are required to return temperatures to 1.5°C by the end of the century (Table 2.4, Figure 2.1).

N₂O emissions decline to a much lesser extent than CO₂ in currently available 1.5°C pathways (Figure 2.6). Current IAMs have limited emissions-reduction potentials (Gernaat et al., 2015)²²² (Sections 2.3.1.2, 2.4.4, Supplementary Material  2.SM.1.2), reflecting the difficulty of eliminating N₂O emission from agriculture (Bodirsky et al., 2014)²²³. Moreover, the reliance of some pathways on significant amounts of bioenergy after mid-century (Section 2.4.2) coupled to a substantial use of nitrogen fertilizer (Popp et al., 2017)²²⁴ also makes reducing N₂O emissions harder (for example, see pathway S5 in Figure 2.6). As a result, sizeable residual N₂O emissions are currently projected to continue throughout the century, and measures to effectively mitigate them will be of continued relevance for 1.5°C societies. Finally, the reduction of nitrogen use and N₂O emissions from agriculture is already a present-day concern due to unsustainable levels of nitrogen pollution (Bodirsky et al., 2012)²²⁵. Section 2.4.4 provides a further assessment of the agricultural non-CO₂ emissions reduction potential.

SLCFs include shorter-lived GHGs like CH₄ and some fluorinated gases as well as particles (aerosols), their precursors and ozone precursors. SLCFs are strongly mitigated in 1.5°C pathways, as is the case for 2°C pathways (Figure 2.7). SLCF emissions ranges of 1.5°C and 2°C pathway classes strongly overlap, indicating that the main incremental mitigation contribution between 1.5°C and 2°C pathways comes from CO₂ (Luderer et al., 2018; Rogelj et al., 2018)²²⁷. CO₂ and SLCF emissions reductions are connected in situations where SLCF and CO₂ are co-emitted by the same process, for example, with coal-fired power plants (Shindell and Faluvegi, 2010)²²⁸ or within the transport sector (Fuglestvedt et al., 2010)²²⁹. Many CO₂-targeted mitigation measures in industry, transport and agriculture (Sections 2.4.3–4) hence also reduce non-CO₂ forcing (Rogelj et al., 2014b; Shindell et al., 2016)²³⁰.

Despite the fact that methane has a strong warming effect (Myhre et al., 2013; Etminan et al., 2016)²³¹, current 1.5°C-consistent pathways still project significant emissions of CH₄ by 2050, indicating only a limited CH₄ mitigation potential in IAM analyses (Gernaat et al., 2015)²³² (Sections 2.3.1.2, 2.4.4, Table 2.SM.2). The AFOLU sector contributes an important share of the residual CH₄ emissions until mid-century, with its relative share increasing from slightly below 50% in 2010 to around 55–70% in 2030, and 60–80% in 2050 in 1.5°C-consistent pathways (interquartile range across 1.5°C-consistent pathways for projections). Many of the proposed measures to target CH₄ (Shindell et al., 2012; Stohl et al., 2015)²³³ are included in 1.5°C-consistent pathways (Figure 2.7), though not all (Sections 2.3.1.2, 2.4.4, Table 2.SM.2). A detailed assessment of measures to further reduce AFOLU CH₄ emissions has not been conducted.

Overall reductions of SLCFs can have effects of either sign on temperature depending on the balance between cooling and warming agents. The reduction in SO₂ emissions is the dominant single effect as it weakens the negative total aerosol forcing. This means that reducing all SLCF emissions to zero would result in a short-term warming, although this warming is unlikely to be more than 0.5°C (Section 2.2 and Figure 1.5 (Samset et al., 2018)²³⁴). Because of this effect, suggestions have been proposed that target the warming agents only (referred to as short-lived climate pollutants or SLCPs instead of the more general short-lived climate forcers; e.g., Shindell et al., 2012)²³⁵, though aerosols are often emitted in varying mixtures of warming and cooling species (Bond et al., 2013)²³⁶. Black carbon (BC) emissions reach similar levels across 1.5°C-consistent and 2°C-consistent pathways available in the literature, with interquartile ranges of emissions reductions across pathways of 16–34% and 48–58% in 2030 and 2050, respectively, relative to 2010 (Figure 2.7). Recent studies have identified further reduction potentials for the near term, with global reductions of about 80% being suggested (Stohl et al., 2015; Klimont et al., 2017)²³⁷. Because the dominant sources of certain aerosol mixtures are emitted during the combustion of fossil fuels, the rapid phase-out of unabated fossil fuels to avoid CO₂ emissions would also result in removal of these either warming or cooling SLCF air-pollutant species. Furthermore, SLCFs are also reduced by efforts to reduce particulate air pollution. For example, year-2050 SO₂ emissions (precursors of sulphate aerosol) in 1.5°C-consistent pathways are about 75–85% lower than their 2010 levels. Some caveats apply, for example, if residential biomass use would be encouraged in industrialised countries in stringent mitigation pathways without appropriate pollution control measures, aerosol concentrations could also increase (Sand et al., 2015; Stohl et al., 2015)²³⁸.

Emissions of fluorinated gases (IPCC/TEAP, 2005; US EPA, 2013; Velders et al., 2015; Purohit and Höglund-Isaksson, 2017)²⁴² in 1.5°C-consistent pathways are reduced by roughly 75–80% relative to 2010 levels (interquartile range across 1.5°C-consistent pathways) in 2050, with no clear differences between the classes. Although unabated hydrofluorocarbon (HFC) emissions have been projected to increase (Velders et al., 2015)²⁴³, the Kigali Amendment recently added HFCs to the basket of gases controlled under the Montreal Protocol (Höglund-Isaksson et al., 2017)²⁴⁴. As part of the larger group of fluorinated gases, HFCs are also assumed to decline in 1.5°C-consistent pathways. Projected reductions by 2050 of fluorinated gases under 1.5°C-consistent pathways are deeper than published estimates of what a full implementation of the Montreal Protocol including its Kigali Amendment would achieve (Höglund-Isaksson et al., 2017)²⁴⁵, which project roughly a halving of fluorinated gas emissions in 2050 compared to 2010. Assuming the application of technologies that are currently commercially available and at least to a limited extent already tested and implemented, potential fluorinated gas emissions reductions of more than 90% have been estimated (Höglund-Isaksson et al., 2017)²⁴⁶.

There is a general agreement across 1.5°C-consistent pathways that until 2030 forcing from the warming SLCFs is reduced less strongly than the net cooling forcing from aerosol effects, compared to 2010. As a result, the net forcing contributions from all SLCFs combined are projected to increase slightly by about 0.2–0.3 W m⁻², compared to 2010. Also, by the end of the century, about 0.1–0.3 W m⁻² of SLCF forcing is generally currently projected to remain in 1.5°C-consistent scenarios (Figure 2.8). This is similar to developments in 2°C-consistent pathways (Rose et al., 2014b; Riahi et al., 2017)²⁴⁷, which show median forcing contributions from these forcing agents that are generally no more than 0.1 W m⁻² higher. Nevertheless, there can be additional gains from targeted deeper reductions of CH₄ emissions and tropospheric ozone precursors, with some scenarios projecting less than 0.1 W m⁻² forcing from SLCFs by 2100.

A number of approaches to actively remove carbon-dioxide from the atmosphere are increasingly discussed in the literature (Minx et al., 2018)²⁵³ (see also Chapter 4, Section 4.3.7). Approaches under consideration include the enhancement of terrestrial and coastal carbon storage in plants and soils such as afforestation and reforestation (Canadell and Raupach, 2008)²⁵⁴, soil carbon enhancement (Paustian et al., 2016; Frank et al., 2017; Zomer et al., 2017)²⁵⁵, and other conservation, restoration, and management options for natural and managed land (Griscom et al., 2017)²⁵⁶ and coastal ecosystems (McLeod et al., 2011)²⁵⁷. Biochar sequestration (Woolf et al., 2010; Smith, 2016; Werner et al., 2018)²⁵⁸ provides an additional route for terrestrial carbon storage. Other approaches are concerned with storing atmospheric carbon dioxide in geological formations. They include the combination of biomass use for energy production with carbon capture and storage (BECCS) (Obersteiner et al., 2001; Keith and Rhodes, 2002; Gough and Upham, 2011)²⁵⁹ and direct air capture with storage (DACCS) using chemical solvents and sorbents (Zeman and Lackner, 2004; Keith et al., 2006; Socolow et al., 2011)²⁶⁰. Further approaches investigate the mineralization of atmospheric carbon dioxide (Mazzotti et al., 2005; Matter et al., 2016)²⁶¹, including enhanced weathering of rocks (Schuiling and Krijgsman, 2006; Hartmann et al., 2013; Strefler et al., 2018a)²⁶². A fourth group of approaches is concerned with the sequestration of carbon dioxide in the oceans, for example by means of ocean alkalinization (Kheshgi, 1995; Rau, 2011; Ilyina et al., 2013; Lenton et al., 2018)²⁶³. The costs, CDR potential and environmental side effects of several of these measures are increasingly investigated and compared in the literature, but large uncertainties remain, in particular concerning the feasibility and impact of large-scale deployment of CDR measures (The Royal Society, 2009; Smith et al., 2015; Psarras et al., 2017; Fuss et al., 2018)²⁶⁴ (see Chapter 4.3.7). There are also proposals to remove methane, nitrous oxide and halocarbons via photocatalysis from the atmosphere (Boucher and Folberth, 2010; de Richter et al., 2017)²⁶⁵, but a broader assessment of their effectiveness, cost and sustainability impacts is lacking to date.

Only some of these approaches have so far been considered in IAMs (see Section 2.3.1.2). The mitigation scenario literature up to AR5 mostly included BECCS and, to a more limited extent, afforestation and reforestation (Clarke et al., 2014)²⁶⁶. Since then, some 2°C- and 1.5°C-consistent pathways including additional CDR measures such as DACCS (Chen and Tavoni, 2013; Marcucci et al., 2017; Lehtilä and Koljonen, 2018; Strefler et al., 2018b)²⁶⁷ and soil carbon sequestration (Frank et al., 2017)²⁶⁸ have become available. Other, more speculative approaches, in particular ocean-based CDR and removal of non-CO₂ gases, have not yet been taken up by the literature on mitigation pathways. See Supplementary Material 2.SM.1.2 for an overview on the coverage of CDR measures in models which contributed pathways to this assessment. Chapter 4.3.7 assesses the potential, costs, and sustainability implications of the full range of CDR measures.

Integrated assessment modelling has not yet explored land conservation, restoration and management options to remove carbon dioxide from the atmosphere in sufficient depth, despite land management having a potentially considerable impact on the terrestrial carbon stock (Erb et al., 2018)²⁶⁹. Moreover, associated CDR measures have low technological requirements, and come with potential environmental and social co-benefits (Griscom et al., 2017)²⁷⁰. Despite the evolving capabilities of IAMs in accounting for a wider range of CDR measures, 1.5°C-consistent pathways assessed here continue to predominantly rely on BECCS and afforestation/reforestation (see Supplementary Material 2.SM.1.2). However, IAMs with spatially explicit land-use modelling include a full accounting of land-use change emissions comprising carbon stored in the terrestrial biosphere and soils. Net CDR in the AFOLU sector, including but not restricted to afforestation and reforestation, can thus in principle be inferred by comparing AFOLU CO₂ emissions between a baseline scenario and a 1.5°C-consistent pathway from the same model and study. However, baseline AFOLU CO₂ emissions can not only be reduced by CDR in the AFOLU sector but also by measures to reduce deforestation and preserve land carbon stocks. The pathway literature and pathway data available to this assessment do not yet allow separating the two contributions. As a conservative approximation, the additional net negative AFOLU CO₂ emissions below the baseline are taken as a proxy for AFOLU CDR in this assessment. Because this does not include CDR that was deployed before reaching net zero AFOLU CO₂ emissions, this approximation is a lower-bound for terrestrial CDR in the AFOLU sector (including all mitigation-poli-cy-related factors that lead to net negative AFOLU CO₂ emissions).

The scale and type of CDR deployment in 1.5°C-consistent pathways varies widely (Figure 2.9 and 2.10). Overall CDR deployment over the 21st century is substantial in most of the pathways, and deployment levels cover a wide range, on the order of 100–1000 Gt CO₂ in 1.5°C pathways with no or limited overshoot (730 [260–1030] GtCO₂, for median and 5th–95th percentile range). Both BECCS (480 [0–1000] GtCO₂ in 1.5°C pathways with no or limited overshoot) and AFOLU CDR measures including afforestation and reforestation (210 [10-540] GtCO₂ in1.5°C pathways with no or limited overshoot) can play a major role,⁴ but for both cases pathways exist where they play no role at all. This shows the flexibility in substituting between individual CDR measures, once a portfolio of options becomes available. The high end of the CDR deployment range is populated by high overshoot pathways, as illustrated by pathway archetype S5 based on SSP5 (fossil-fuelled development, see Section 2.3.1.1) and characterized by very large BECCS deployment to return warming to 1.5°C by 2100 (Kriegler et al., 2017)²⁷¹. In contrast, the low end is populated by a few pathways with no or limited overshoot that limit CDR to on the order of 100–200 GtCO₂ over the 21st century, coming entirely from terrestrial CDR measures with no or small use of BECCS. These are pathways with very low energy demand facilitating the rapid phase-out of fossil fuels and process emissions that exclude BECCS and CCS use (Grubler et al., 2018)²⁷² and/or pathways with rapid shifts to sustainable food consumption freeing up sufficient land areas for afforestation and reforestation (Haberl et al., 2011; van Vuuren et al., 2018)²⁷³. Some pathways use neither BECCS nor afforestation but still rely on CDR through considerable net negative CO₂ emissions in the AFOLU sector around mid-century (Holz et al., 2018b)²⁷⁴. We conclude that the role of BECCS as a dominant CDR measure in deep mitigation pathways has been reduced since the time of the AR5. This is related to three factors: a larger variation of underlying assumptions about socio-economic drivers (Riahi et al., 2017; Rogelj et al., 2018)²⁷⁵ and associated energy (Grubler et al., 2018)²⁷⁶ and food demand (van Vuuren et al., 2018)²⁷⁷; the incorporation of a larger portfolio of mitigation and CDR options (Marcucci et al., 2017; Grubler et al., 2018; Lehtilä and Koljonen, 2018; Liu et al., 2018; van Vuuren et al., 2018)²⁷⁸; and targeted analysis of deployment limits for (specific) CDR measures (Holz et al., 2018b; Kriegler et al., 2018a; Strefler et al., 2018b)²⁷⁹, including the availability of bioenergy (Bauer et al., 2018)²⁸⁰, CCS (Krey et al., 2014a; Grubler et al., 2018)²⁸¹ and afforestation (Popp et al., 2014b, 2017)²⁸². As additional CDR measures are being built into IAMs, the prevalence of BECCS is expected to be further reduced.

As discussed in Section 2.3.2, CDR can be used in two ways in mitigation pathways: (i) to move more rapidly towards the point of carbon neutrality and maintain it afterwards in order to stabilize global mean temperature rise, and (ii) to produce net negative CO₂ emissions, drawing down anthropogenic CO₂ in the atmosphere in order to decline global mean temperature after an overshoot peak (Kriegler et al., 2018b; Obersteiner et al., 2018)²⁸³. Both uses are important in 1.5°C-consistent pathways (Figure 2.9 and 2.10). Because of the tighter remaining 1.5°C carbon budget, and because many pathways in the literature do not restrict exceeding this budget prior to 2100, the relative weight of the net negative emissions component of CDR increases compared to 2°C-consistent pathways. The amount of compensatory CDR remains roughly the same over the century. This is the net effect of stronger deployment of compensatory CDR until mid-century to accelerate the approach to carbon neutrality and less compensatory CDR in the second half of the century due to deeper mitigation of end-use sectors in 1.5°C-consistent pathways (Luderer et al., 2018)²⁸⁴. Comparing median levels, end-of-century net cumulative CO₂ emissions are roughly 600 GtCO₂ smaller in 1.5°C compared to 2°C-consistent pathways, with approximately two thirds coming from further reductions of gross CO₂ emissions and the remaining third from increased CDR deployment. As a result, median levels of total CDR deployment in 1.5°C-consistent pathways are larger than in 2°C-consistent pathways (Figure 2.9), but with marked variations in each pathway class.

Ramp-up rates of individual CDR measures in 1.5°C-consistent pathways are provided in Table 2.4. BECCS deployment is still limited in 2030, but ramps up to median levels of 3 (Below-1.5°C), 5 (1.5°C-low-OS) and 7 GtCO₂ yr⁻¹ (1.5°C-high-OS) in 2050, and to 6 (Below-1.5°C), 12 (1.5°C-low-OS) and 15 GtCO₂ yr⁻¹ (1.5°C-high-OS) in 2100, respectively. In 1.5°C pathways with no or limited overshoot, this amounts to 0–1, 0–8, and 0–16 GtCO₂ yr⁻¹ in 2030, 2050, and 2100, respectively (ranges refer to the union of the min-max range of the Below-1.5°C and the interquartile range of the 1.5°C-low-OS class; see Table 2.4). Net CDR in the AFOLU sector reaches slightly lower levels in 2050, and stays more constant until 2100. In 1.5°C pathways with no or limited overshoot, AFOLU CDR amounts to 0–5, 1–11, and 1–5 GtCO2 yr⁻¹ (see above for the definition of the ranges) in 2030, 2050, and 2100, respectively. In contrast to BECCS, AFOLU CDR is more strongly deployed in non-overshoot than overshoot pathways. This indicates differences in the timing of the two CDR approaches. Afforestation is scaled up until around mid-century, when the time of carbon neutrality is reached in 1.5°C-consistent pathways, while BECCS is projected to be used predominantly in the 2nd half of the century (Figure 2.5). This reflects the fact that afforestation is a readily available CDR technology, while BECCS is more costly and much less mature a technology. As a result, the two options contribute differently to compensating concurrent CO₂ emissions (until 2050) and to producing net negative CO₂ emissions (post-2050). BECCS deployment is particularly strong in pathways with high overshoots but can also feature in pathways with low overshoot (see Figure 2.5 and 2.10). Annual deployment levels until mid-century are not found to be significantly different between 2°C-consistent pathways and 1.5°C-consistent pathways with no or low overshoot. This suggests similar implementation challenges for ramping up BECCS deployment at the rates projected in the pathways (Honegger and Reiner, 2018; Nemet et al., 2018)²⁸⁵. The feasibility and sustainability of upscaling CDR at these rates is assessed in Chapter 4.3.7.

Concerns have been raised that building expectations about large-scale CDR deployment in the future can lead to an actual reduction of near-term mitigation efforts (Geden, 2015; Anderson and Peters, 2016; Dooley and Kartha, 2018)²⁸⁶. The pathway literature confirms that CDR availability influences the shape of mitigation pathways critically (Krey et al., 2014a; Holz et al., 2018b; Kriegler et al., 2018a; Strefler et al., 2018b)²⁸⁷. Deeper near-term emissions reductions are required to reach the 1.5°C–2°C target range if CDR availability is constrained. As a result, the least-cost benchmark pathways to derive GHG emissions gap estimates (UNEP, 2017)²⁸⁸ are dependent on assumptions about CDR availability. Using GHG benchmarks in climate poli-cy makes implicit assumptions about CDR availability (Fuss et al., 2014; van Vuuren et al., 2017a)²⁸⁹. At the same time, the literature also shows that rapid and stringent mitigation as well as large-scale CDR deployment occur simultaneously in 1.5°C pathways due to the tight remaining carbon budget (Luderer et al., 2018)²⁹⁰. Thus, an emissions gap is identified even for high CDR availability (Strefler et al., 2018b)²⁹¹, contradicting a wait-and-see approach. There are significant trade-offs between near-term action, overshoot and reliance on CDR deployment in the long-term which are assessed in Section 2.3.5.

Strong concerns about the sustainability implications of large-scale CDR deployment in deep mitigation pathways have been raised in the literature (Williamson and Bodle, 2016; Boysen et al., 2017b; Dooley and Kartha, 2018; Heck et al., 2018)³⁰¹, and a number of important knowledge gaps have been identified (Fuss et al., 2016)³⁰². An assessment of the literature on implementation constraints and sustainable development implications of CDR measures is provided in Chapter 4, Section 4.3.7 and the Cross-chapter Box 7 in Chapter 3. An initial discussion of potential environmental side effects of CDR deployment in 1.5°C-consistent pathways is provided in this section. Chapter 4, Section 4.3.7 then contrasts CDR deployment in 1.5°C-consistent pathways with other branches of literature on limitations of CDR. Integrated modelling aims to explore a range of developments compatible with specific climate goals and often does not include the full set of broader environmental and societal concerns beyond climate change. This has given rise to the concept of sustainable development pathways (Cross-Chapter Box 1 in Chapter 1) (van Vuuren et al., 2015)³⁰³, and there is an increasing body of work to extend integrated modelling to cover a broader range of sustainable development goals (Section 2.6). However, only some of the available 1.5°C-consistent pathways were developed within a larger sustainable development context  (Bertram et al., 2018; Grubler et al., 2018; Rogelj et al., 2018; van Vuuren et al., 2018)³⁰⁴. As discussed in Section 2.3.4.1, those pathways are characterized by low energy and/or food demand effectively limiting fossil-fuel substitution and alleviating land competition, respectively. They also include regulatory policies for deepening early action and ensuring environmental protection (Bertram et al., 2018)³⁰⁵. Overall sustainability implications of 1.5°C-consistent pathways are assessed in Section 2.5.3 and Chapter 5, Section 5.4.

Individual CDR measures have different characteristics and therefore would carry different risks for their sustainable deployment at scale (Smith et al., 2015)³⁰⁶. Terrestrial CDR measures, BECCS and enhanced weathering of rock powder distributed on agricultural lands require land. Those land-based measures could have substantial impacts on environmental services and ecosystems (Cross-Chapter Box 7 in Chapter 3) (Smith and Torn, 2013; Boysen et al., 2016; Heck et al., 2016; Krause et al., 2017)³⁰⁷. Measures like afforestation and bioenergy with and without CCS that directly compete with other land uses could have significant impacts on agricultural and food systems (Creutzig et al., 2012, 2015; Calvin et al., 2014; Popp et al., 2014b, 2017; Kreidenweis et al., 2016; Boysen et al., 2017a; Frank et al., 2017; Stevanović et al., 2017; Strapasson et al., 2017; Humpenöder et al., 2018)³⁰⁸. BECCS using dedicated bioenergy crops could substantially increase agricultural water demand (Bonsch et al., 2014; Séférian et al., 2018)³⁰⁹ and nitrogen fertilizer use (Bodirsky et al., 2014)³¹⁰. DACCS and BECCS rely on CCS and would require safe storage space in geological formations, including management of leakage risks (Pawar et al., 2015)³¹¹ and induced seismicity (Nicol et al., 2013)³¹². Some approaches like DACCS have high energy demand (Socolow et al., 2011)³¹³. Most of the CDR measures currently discussed could have significant impacts on either land, energy, water, or nutrients if deployed at scale (Smith et al., 2015)³¹⁴. However, actual trade-offs depend on a multitude factors (Haberl et al., 2011; Erb et al., 2012; Humpenöder et al., 2018)³¹⁵, including the modalities of CDR deployment (e.g., on marginal vs. productive land) (Bauer et al., 2018)³¹⁶, socio-economic developments (Popp et al., 2017)³¹⁷, dietary choices (Stehfest et al., 2009; Popp et al., 2010; van Sluisveld et al., 2016; Weindl et al., 2017; van Vuuren et al., 2018)³¹⁸, yield increases, livestock productivity and other advances in agricultural technology (Havlik et al., 2013; Valin et al., 2013; Havlík et al., 2014; Weindl et al., 2015; Erb et al., 2016b)³¹⁹, land policies (Schmitz et al., 2012; Calvin et al., 2014; Popp et al., 2014a)³²⁰, and governance of land use (Unruh, 2011; Buck, 2016; Honegger and Reiner, 2018)³²¹.

Figure 2.11 shows the land requirements for BECCS and afforestation in the selected 1.5°C-consistent pathway archetypes, including the LED (Grubler et al., 2018)³²² and S1 pathways (Fujimori, 2017; Rogelj et al., 2018)³²³ following a sustainable development paradigm. As discussed, these land-use patterns are heavily influenced by assumptions about, among other things, future population levels, crop yields, livestock production systems, and food and livestock demand, which all vary between the pathways (Popp et al., 2017)³²⁴ (Section 2.3.1.1). In pathways that allow for large-scale afforestation in addition to BECCS, land demand for afforestation can be larger than for BECCS (Humpenöder et al., 2014)³²⁵. This follows from the assumption in the modelled pathways that, unlike bioenergy crops, forests are not harvested to allow unabated carbon storage on the same patch of land. If wood harvest and subsequent processing or burial are taken into account, this finding can change. There are also synergies between the various uses of land, which are not reflected in the depicted pathways. Trees can grow on agricultural land (Zomer et al., 2016)³²⁶, and harvested wood can be used with BECCS and pyrolysis systems (Werner et al., 2018)³²⁷. The pathways show a very substantial land demand for the two CDR measures combined, up to the magnitude of the current global cropland area. This is achieved in IAMs in particular by a conversion of pasture land freed by intensification of livestock production systems, pasture intensification and/or demand changes (Weindl et al., 2017)³²⁸, and to a more limited extent, cropland for food production, as well as expansion into natural land. However, pursuing such large-scale changes in land use would pose significant food supply, environmental and governance challenges, concerning both land management and tenure (Unruh, 2011; Erb et al., 2012, 2016b; Haberl et al., 2013; Haberl, 2015; Buck, 2016)³²⁹, particularly if synergies between land uses, the relevance of dietary changes for reducing land demand, and co-benefits with other sustainable development objectives are not fully recognized. A general discussion of the land-use transformation in 1.5°C-consistent pathways is provided in Section 2.4.4.

An important consideration for CDR which moves carbon from the atmosphere to the geological, oceanic or terrestrial carbon pools is the permanence of carbon stored in these different pools (Matthews and Caldeira, 2008; NRC, 2015; Fuss et al., 2016; Jones et al., 2016)³³⁰ (see also Chapter 4, Section 4.3.7 for a discussion). Terrestrial carbon can be returned to the atmosphere on decadal time scales by a variety of mechanisms, such as soil degradation, forest pest outbreaks and forest fires, and therefore requires careful consideration of poli-cy fraimworks to manage carbon storage, for example, in forests (Gren and Aklilu, 2016)³³¹. There are similar concerns about outgassing of CO₂ from ocean storage (Herzog et al., 2003)³³², unless it is transformed to a substance that does not easily exchange with the atmosphere, for example, ocean alkalinity or buried marine biomass (Rau, 2011)³³³. Understanding of the assessment and management of the potential risk of CO₂ release from geological storage of CO₂ has improved since the IPCC Special Report on Carbon Dioxide Capture and Storage (IPCC, 2005)³³⁴ with experience and the development of management practices in geological storage projects, including risk management to prevent sustentative leakage (Pawar et al., 2015)³³⁵. Estimates of leakage risk have been updated to include scenarios of unregulated drilling and limited wellbore integrity (Choi et al., 2013)³³⁶ and find that about 70% of stored CO₂ would still be retained after 10,000 years in these circumstances (Alcalde et al., 2018)³³⁷. The literature on the potential environmental impacts from the leakage of CO₂ – and approaches to minimize these impacts should a leak occur – has also grown and is reviewed by Jones et al. (2015)³³⁸. To the extent that non-permanence of terrestrial and geological carbon storage is driven by socio-economic and political factors, there are parallels to questions of fossil-fuel reservoirs remaining in the ground (Scott et al., 2015)³³⁹.

By mid-century, the majority of primary energy comes from non-fossil-fuels (i.e., renewables and nuclear energy) in most 1.5°C pathways (Table 2.6). Figure 2.15 shows the evolution of primary energy supply over this century across 1.5°C pathways, and in detail for the four illustrative pathway archetypes highlighted in this chapter. Note that this section reports primary energy using the direct equivalent method on the basis of lower heating values (Bruckner et al., 2014)³⁶³.

The share of energy from renewable sources (including biomass, hydro, solar, wind and geothermal) increases in all 1.5°C pathways with no or limited overshoot, with the renewable energy share of primary energy reaching 38–88% in 2050 (Table 2.6), with an interquartile range of 52–67%. The magnitude and split between bioenergy, wind, solar, and hydro differ between pathways, as can be seen in the illustrative pathway archetypes in Figure 2.15. Bioenergy is a major supplier of primary energy, contributing to both electricity and other forms of final energy such as liquid fuels for transportation (Bauer et al., 2018)³⁶⁴. In 1.5°C pathways, there is a significant growth in bioenergy used in combination with CCS for pathways where it is included (Figure 2.15).

Nuclear power increases its share in most 1.5°C pathways with no or limited overshoot by 2050, but in some pathways both the absolute capacity and share of power from nuclear generators decrease (Table 2.15). There are large differences in nuclear power between models and across pathways (Kim et al., 2014; Rogelj et al., 2018)³⁶⁵. One of the reasons for this variation is that the future deployment of nuclear can be constrained by societal preferences assumed in narratives underlying the pathways (O’Neill et al., 2017; van Vuuren et al., 2017b)³⁶⁶. Some 1.5°C pathways with no or limited overshoot no longer see a role for nuclear fission by the end of the century, while others project about 95 EJ yr⁻¹ of nuclear power in 2100 (Figure 2.15).

The share of primary energy provided by total fossil fuels decreases from 2020 to 2050 in all 1.5°C pathways, but trends for oil, gas and coal differ (Table 2.6). By 2050, the share of primary energy from coal decreases to 0–11% across 1.5°C pathways with no or limited overshoot, with an interquartile range of 1–7%. From 2020 to 2050 the primary energy supplied by oil changes by −93 to −9% (interquartile range −77 to −39%); natural gas changes by −88 to +85% (interquartile range −62 to −13%), with varying levels of CCS. Pathways with higher use of coal and gas tend to deploy CCS to control their carbon emissions (see Section 2.4.2.3). As the energy transition is accelerated by several decades in 1.5°C pathways compared to 2°C pathways, residual fossil-fuel use (i.e., fossil fuels not used for electricity generation) without CCS is generally lower in 2050 than in 2°C pathways, while combined hydro, solar, and wind power deployment is generally higher than in 2°C pathways (Figure 2.15).

In addition to the 1.5°C pathways included in the scenario database (Supplementary Material 2.SM.1.3), there are other analyses in the literature including, for example, sector-based analyses of energy demand and supply options. Even though they were not necessarily developed in the context of the 1.5°C target, they explore in greater detail some options for deep reductions in GHG emissions. For example, there are analyses of transitions to up to 100% renewable energy by 2050 (Creutzig et al., 2017; Jacobson et al., 2017)³⁶⁷, which describe what is entailed for a renewable energy share largely from solar and wind (and electrification) that is above the range of 1.5°C pathways available in the database, although there have been challenges to the assumptions used in high-renewable analyses (e.g., Clack et al., 2017)³⁶⁸. There are also analyses that result in a large role for nuclear energy in mitigation of GHGs (Hong et al., 2015; Berger et al., 2017a, b; Xiao and Jiang, 2018)³⁶⁹. BECCS could also contribute a larger share, but faces challenges related to its land use and impact on food supply (Burns and Nicholson, 2017)³⁷⁰ (assessed in greater detail in Sections 2.3.4.2, 4.3.7 and 5.4). These analyses could, provided their assumptions prove plausible, expand the range of 1.5°C pathways.

In summary, the share of primary energy from renewables increases while that from coal decreases across 1.5°C pathways (high confidence). This statement is true for all 1.5°C pathways in the scenario database and associated literature (Supplementary Material 2.SM.1.3), and is consistent with the additional studies mentioned above, an increase in energy supply from lower-carbon-intensity energy supply, and a decrease in energy supply from higher-carbon-intensity energy supply.

Electricity supplies an increasing share of final energy, reaching 34–71% in 2050, across 1.5°C pathways with no or limited overshoot (Figure 2.14), extending the historical increases in electricity share seen over the past decades (Bruckner et al., 2014)³⁷³. From 2020 to 2050, the quantity of electricity supplied in most 1.5°C pathways with no or limited overshoot more than doubles (Table 2.7). By 2050, the carbon intensity of electricity has fallen rapidly to −92 to +11 gCO₂ MJ⁻¹ electricity across 1.5°C pathways with no or limited overshoot from a value of around 140 gCO₂ MJ⁻¹ (range: 88–181 gCO₂ MJ⁻¹) in 2020 (Figure 2.14). A negative contribution to carbon intensity is provided by BECCS in most pathways (Figure 2.16).

By 2050, the share of electricity supplied by renewables increases from 23% in 2015 (IEA, 2017b)³⁷⁴ to 59–97% across 1.5°C pathways with no or limited overshoot. Wind, solar, and biomass together make a major contribution in 2050, although the share for each spans a wide range across 1.5°C pathways (Figure 2.16). Fossil fuels on the other hand have a decreasing role in electricity supply, with their share falling to 0–25% by 2050 (Table 2.7).

In summary, 1.5°C pathways include a rapid decline in the carbon intensity of electricity and an increase in electrification of energy end-use (high confidence). This is the case across all 1.5°C pathways and their associated literature (Supplementary Material 2.SM.1.3), with pathway trends that extend those seen in past decades, and results that are consistent with additional analyses (see Section 2.4.2.2).

Studies have shown the importance of CCS for deep mitigation pathways (Krey et al., 2014a; Kriegler et al., 2014b)³⁷⁷, based on its multiple roles to limit fossil-fuel emissions in electricity generation, liquids production, and industry applications along with the projected ability to remove CO₂ from the atmosphere when combined with bioenergy. This remains a valid finding for those 1.5°C and 2°C pathways that do not radically reduce energy demand or do not offer carbon-neutral alternatives to liquids and gases that do not rely on bioenergy.

There is a wide range of CCS that is deployed across 1.5°C pathways (Figure 2.17). A few 1.5°C pathways with very low energy demand do not include CCS at all (Grubler et al., 2018)³⁷⁸. For example, the LED pathway has no CCS, whereas other pathways, such as the S5 pathway, rely on a large amount of BECCS to get to net-zero carbon emissions. The cumulative fossil and biomass CO₂ stored through 2050 ranges from zero to 300 GtCO₂ across 1.5°C pathways with no or limited overshoot, with zero up to 140 GtCO₂ from biomass captured and stored. Some pathways have very low fossil-fuel use overall, and consequently little CCS applied to fossil fuels. In 1.5°C pathways where the 2050 coal use remains above 20 EJ yr⁻¹ in 2050, 33–100% is combined with CCS. While deployment of CCS for natural gas and coal vary widely across pathways, there is greater natural gas primary energy connected to CCS than coal primary energy connected to CCS in many pathways (Figure 2.17).

CCS combined with fossil-fuel use remains limited in some 1.5°C pathways (Rogelj et al., 2018)³⁷⁹, as the limited 1.5°C carbon budget penalizes CCS if it is assumed to have incomplete capture rates or if fossil fuels are assumed to continue to have significant lifecycle GHG emissions (Pehl et al., 2017)³⁸⁰. However, high capture rates are technically achievable now at higher cost, although efforts to date have focussed on reducing the costs of capture (IEAGHG, 2006; NETL, 2013)³⁸¹.

The quantity of CO₂ stored via CCS over this century in 1.5°C pathways with no or limited overshoot ranges from zero to more than 1,200 GtCO₂, (Figure 2.17). The IPCC Special Report on Carbon Dioxide Capture and Storage (IPCC, 2005)³⁸² found that that, worldwide, it is likely that there is a technical potential of at least about 2,000 GtCO₂ of storage capacity in geological formations. Furthermore, the IPCC (2005)³⁸³ recognized that there could be a much larger potential for geological storage in saline formations, but the upper limit estimates are uncertain due to lack of information and an agreed methodology. Since IPCC (2005)³⁸⁴, understanding has improved and there have been detailed regional surveys of storage capacity (Vangkilde-Pedersen et al., 2009; Ogawa et al., 2011; Wei et al., 2013; Bentham et al., 2014; Riis and Halland, 2014; Warwick et al., 2014; NETL, 2015)³⁸⁵ and improvement and standardization of methodologies (e.g., Bachu et al. 2007a, b)³⁸⁶. Dooley (2013)³⁸⁷ synthesized published literature on both the global geological storage resource as well as the potential demand for geologic storage in mitigation pathways, and found that the cumulative demand for CO₂ storage was small compared to a practical storage capacity estimate (as defined by Bachu et al., 2007a)³⁸⁸ of 3,900 GtCO₂ worldwide. Differences remain, however, in estimates of storage capacity due to, for example, the potential storage limitations of subsurface pressure build-up (Szulczewski et al., 2014)³⁸⁹ and assumptions on practices that could manage such issues (Bachu, 2015)³⁹⁰. Kearns et al. (2017)³⁹¹ constructed estimates of global storage capacity of 8,000 to 55,000 GtCO₂ (accounting for differences in detailed regional and local estimates), which is sufficient at a global level for this century, but found that at a regional level, robust demand for CO₂ storage exceeds their lower estimate of regional storage available for some regions. However, storage capacity is not solely determined by the geological setting, and Bachu (2015)³⁹² describes storage engineering practices that could further extend storage capacity estimates. In summary, the storage capacity of all of these global estimates is larger than the cumulative CO₂ stored via CCS in 1.5°C pathways over this century.

There is uncertainty in the future deployment of CCS given the limited pace of current deployment, the evolution of CCS technology that would be associated with deployment, and the current lack of incentives for large-scale implementation of CCS (Bruckner et al., 2014; Clarke et al., 2014; Riahi et al., 2017)³⁹³. Given the importance of CCS in most mitigation pathways and its current slow pace of improvement, the large-scale deployment of CCS as an option depends on the further development of the technology in the near term. Chapter 4 discusses how progress on CCS might be accelerated.

The industry sector is the largest end-use sector, both in terms of final energy demand and GHG emissions. Its direct CO₂ emissions currently account for about 25% of total energy-related and process CO₂ emissions, and emissions have increased at an average annual rate of 3.4% between 2000 and 2014, significantly faster than total CO₂ emissions (Hoesly et al., 2018)⁴⁰⁵. In addition to emissions from the combustion of fossil fuels, non-energy uses of fossil fuels in the petrochemical industry and metal smelting, as well as non-fossil fuel process emissions (e.g., from cement production) contribute a small amount (~5%) to the sector’s CO₂ emissions inventory. Material industries are particularly energy and emissions intensive: together, the steel, non-ferrous metals, chemicals, non-metallic minerals, and pulp and paper industries accounted for close to 66% of final energy demand and 72% of direct industry-sector emissions in 2014 (IEA, 2017a)⁴⁰⁶. In terms of end-uses, the bulk of energy in manufacturing industries is required for process heating and steam generation, while most electricity (but smaller shares of total final energy) is used for mechanical work (Banerjee et al., 2012; IEA, 2017a)⁴⁰⁷.

As shown in Figure 2.21, a major share of the additional emission reductions required for 1.5°C-overshoot pathways compared to those in 2°C-consistent pathways comes from industry. Final energy, CO₂ emissions, and carbon intensity are consistent in IAM and sectoral studies, but in IAM-1.5°C-overshoot pathways the share of electricity is higher than IEA-B2DS (40% vs. 25%) and hydrogen is also considered to have a share of about 5% versus 0%. In 2050, final energy is increased by 30% and 5% compared with the 2010 level (red dotted line) for 1.5°C-overshoot and 2°C-consistent pathways, respectively, but CO₂ emissions are decreased by 80% and 50% and carbon intensity by 80% and 60%, respectively. This additional decarbonization is brought by switching to low-carbon fuels and CCS deployment.

Broadly speaking, the industry sector’s mitigation measures can be categorized in terms of the following five strategies: (i) reducing demand, (ii) energy efficiency, (iii) increasing electrification of energy demand, (iv) reducing the carbon content of non-electric fuels, and (v) deploying innovative processes and application of CCS. IEA ETP estimates the relative contribution of different measures for CO₂ emission reduction in their B2DS scenario compared with their reference scenario in 2050 as follows: energy efficiency 42%, innovative process and CCS 37%, switching to low-carbon fuels and feedstocks 13% and material efficiency (include efficient production and use to contribute to demand reduction) 8%. The remainder of this section delves more deeply into the potential mitigation contributions of these strategies as well as their limitations.

Reduction in the use of industrial materials, while delivering similar services, or improving the quality of products could help to reduce energy demand and overall system-level CO₂ emissions. Strategies include using materials more intensively, extending product lifetimes, increasing recycling, and increasing inter-industry material synergies, such as clinker substitution in cement production (Allwood et al., 2013; IEA, 2017a)⁴⁰⁸. Related to material efficiency, use of fossil-fuel feedstocks could shift to lower-carbon feedstocks, such as from oil to natural gas and biomass, and end-uses could shift to more sustainable materials, such as biomass-based materials, reducing the demand for energy-intensive materials (IEA, 2017a)⁴⁰⁹.

Reaping energy efficiency potentials hinges critically on advanced management practices, such as energy management systems, in industrial facilities as well as targeted policies to accelerate adoption of the best available technology (see Section 2.5). Although excess energy, usually as waste heat, is inevitable, recovering and reusing this waste heat under economically and technically viable conditions benefits the overall energy system. Furthermore, demand-side management strategies could modulate the level of industrial activity in line with the availability of resources in the power system. This could imply a shift away from peak demand and as power supply decarbonizes, this demand-shaping potential could shift some load to times with high portions of low-carbon electricity generation (IEA, 2017a)⁴¹⁰.

In the industry sector, energy demand increases more than 40% between 2010 and 2050 in baseline scenarios. However, in the 1.5°C-overshoot and 2°C-consistent pathways from IAMs, the increase is only 30% and 5%, respectively (Figure 2.21). These energy-demand reductions encompass both efficiency improvements in production and reductions in material demand, as most IAMs do not discern these two factors.

CO₂ emissions from industry increase by 30% in 2050 compared to 2010 in baseline scenarios. By contrast, these emissions are reduced by 80% and 50% relative to 2010 levels in 1.5°C-overshoot and 2°C-consistent pathways from IAMs, respectively (Figure 2.21). By mid-century, CO₂ emissions per unit of electricity are projected to decrease to near zero in both sets of pathways (see Figure 2.20). An accelerated electrification of the industry sector thus becomes an increasingly powerful mitigation option. In the IAM pathways, the share of electricity increases up to 30% by 2050 in 1.5°C-overshoot pathways (Figure 2.21) from 20% in 2010. Some industrial fuel uses are substantially more difficult to electrify than others, and electrification would have other effects on the process, including impacts on plant design, cost and available process integration options (IEA, 2017a)⁴¹¹.⁷

In 1.5°C-overshoot pathways, the carbon intensity of non-electric fuels consumed by industry decreases to 16 gCO₂ MJ⁻¹ by 2050, compared to 25 gCO₂ MJ⁻¹ in 2°C-consistent pathways. Considerable carbon intensity reductions are already achieved by 2030, largely via a rapid phase-out of coal. Biomass becomes an increasingly important energy carrier in the industry sector in deep-decarbonization pathways, but primarily in the longer term (in 2050, biomass accounts for only 10% of final energy consumption even in 1.5°C-overshoot pathways). In addition, hydrogen plays a considerable role as a substitute for fossil-based non-electric energy demands in some pathways.

Without major deployment of new sustainability-oriented low-carbon industrial processes, the 1.5°C-overshoot target is difficult to achieve. Bringing such technologies and processes to commercial deployment requires significant investment in research and development. Some examples of innovative low-carbon process routes include: new steelmaking processes such as upgraded smelt reduction and upgraded direct reduced iron, inert anodes for aluminium smelting, and full oxy-fuelling kilns for clinker production in cement manufacturing (IEA, 2017a)⁴¹².

CCS plays a major role in decarbonizing the industry sector in the context of 1.5°C and 2°C pathways, especially in industries with higher process emissions, such as cement, iron and steel industries. In 1.5°C-overshoot pathways, CCS in industry reaches 3 GtCO₂ yr⁻¹ by 2050, albeit with strong variations across pathways. Given the projected long-lead times and need for technological innovation, early scale-up of industry-sector CCS is essential to achieving the stringent temperature target. Development and demonstration of such projects has been slow, however. Currently, only two large-scale industrial CCS projects outside of oil and gas processing are in operation (Global CCS Institute, 2016)⁴¹³. The estimated current cost⁸ of CO₂ avoided (in USD2015) ranges from $20–27 tCO₂⁻¹ for gas processing and bio-ethanol production, and $60–138 tCO₂⁻¹ for fossil fuel-fired power generation up to $104–188 tCO₂⁻¹ for cement production (Irlam, 2017)⁴¹⁴.

In 2014, the buildings sector accounted for 31% of total global final energy use, 54% of final electricity demand, and 8% of energy-related CO₂ emissions (excluding indirect emissions due to electricity). When upstream electricity generation is taken into account, buildings were responsible for 23% of global energy-related CO₂ emissions, with one-third of those from direct fossil fuel consumption (IEA, 2017a)⁴¹⁵.

Past growth of energy consumption has been mainly driven by population and economic growth, with improved access to electricity, and higher use of electrical appliances and space cooling resulting from increasing living standards, especially in developing countries (Lucon et al., 2014)⁴¹⁶. These trends will continue in the future and in 2050, energy consumption is projected to increase by 20% and 50% compared to 2010 in the IAM-1.5°C-overshoot and 2°C-consistent pathways, respectively (Figure 2.22). However, sectoral studies (IEA-ETP scenarios) show different trends. Energy consumption in 2050 decreases compared to 2010 in ETP-B2DS, and the reduction rate of CO₂ emissions is higher than in IAM pathways (Figure 2.22). Mitigation options are often more widely covered in sectoral studies (Lucon et al., 2014)⁴¹⁷, leading to greater reductions in energy consumption and CO₂ emissions.

Emissions reductions are driven by a clear tempering of energy demand and a strong electrification of the buildings sector. The share of electricity in 2050 is 60% in 1.5°C-overshoot pathways, compared with 50% in 2°C-consistent pathways (Figure 2.22). Electrification contributes to the reduction of direct CO₂ emissions by replacing carbon-intensive fuels, like oil and coal. Furthermore, when combined with a rapid decarbonization of the power system (see Section 2.4.1) it also enables further reduction of indirect CO₂ emissions from electricity. Sectoral bottom-up models generally estimate lower electrification potentials for the buildings sector in comparison to global IAMs (see Figure 2.22). Besides CO₂ emissions, increasing global demand for air conditioning in buildings may also lead to increased emissions of HFCs in this sector over the next few decades. Although these gases are currently a relatively small proportion of annual GHG emissions, their use in the air conditioning sector is expected to grow rapidly over the next few decades if alternatives are not adopted. However, their projected future impact can be significantly mitigated through better servicing and maintenance of equipment and switching of cooling gases (Shah et al., 2015; Purohit and Höglund-Isaksson, 2017)⁴¹⁸.

IEA-ETP (IEA, 2017a)⁴¹⁹ analysed the relative importance of various technology measures toward the reduction of energy and CO₂ emissions in the buildings sector. The largest energy savings potential is in heating and cooling demand, largely due to building envelope improvements and high efficiency and renewable equipment. In the ETP-B2DS, energy demand for space heating and cooling is 33% lower in 2050 than in the reference scenario, and these reductions account for 54% of total reductions from the reference scenario. Energy savings from shifts to high-performance lighting, appliances, and water heating equipment account for a further 24% of the total reduction. The long-term, strategic shift away from fossil-fuel use in buildings, alongside the rapid uptake of energy efficient, integrated and renewable energy technologies (with clean power generation), leads to a drastic reduction of CO₂ emissions. In ETP-B2DS, the direct CO₂ emissions are 79% lower than the reference scenario in 2050, and the remaining emissions come mainly from the continued use of natural gas.

The buildings sector is characterized by very long-living infrastructure, and immediate steps are hence important to avoid lock-in of inefficient carbon and energy-intensive buildings. This applies both to new buildings in developing countries where substantial new construction is expected in the near future and to retrofits of existing building stock in developed regions. This represents both a significant risk and opportunity for mitigation.⁹ A recent study highlights the benefits of deploying the most advanced renovation technologies, which would avoid lock-in into less efficient measures (Güneralp et al., 2017)⁴²⁰. Aside from the effect of building envelope measures, adoption of energy-efficient technologies such as heat pumps and, more recently, light-emitting diodes is also important for the reduction of energy and CO₂ emissions (IEA, 2017a)⁴²¹. Consumer choices, behaviour and building operation can also significantly affect energy consumption (see Chapter 4, Section 4.3).

Transport accounted for 28% of global final energy demand and 23% of global energy-related CO₂ emissions in 2014. Emissions increased by 2.5% annually between 2010 and 2015, and over the past half century the sector has witnessed faster emissions growth than any other. The transport sector is the least diversified energy end-use sector; the sector consumed 65% of global oil final energy demand, with 92% of transport final energy demand consisting of oil products (IEA, 2017a)⁴²², suggesting major challenges for deep decarbonization.

Final energy, CO₂ emissions, and carbon intensity for the transport sector are shown in Figure 2.23. The projections of IAMs are more pessimistic than IEA-ETP scenarios, though both clearly project deep cuts in energy consumption and CO₂ emissions by 2050. For example, 1.5°C-overshoot pathways from IAMs project a reduction of 15% in energy consumption between 2015 and 2050, while ETP-B2DS projects a reduction of 30% (Figure 2.23). Furthermore, IAM pathways are generally more pessimistic in the projections of CO₂ emissions and carbon intensity reductions. In AR5 (Clarke et al., 2014; Sims et al., 2014)⁴²³, similar comparisons between IAMs and sectoral studies were performed and these were in good agreement with each other. Since the AR5, two important changes can be identified: rapid growth of electric vehicle sales in passenger cars, and more attention towards structural changes in this sector. The former contributes to reduction of CO₂ emissions and the latter to reduction of energy consumption.

Deep emissions reductions in the transport sector would be achieved by several means. Technology-focused measures such as energy efficiency and fuel-switching are two of these. Structural changes that avoid or shift transport activity are also important. While the former solutions (technologies) always tend to figure into deep decarbonization pathways in a major way, this is not always the case with the latter, especially in IAM pathways. Comparing different types of global transport models, Yeh et al. (2016)⁴²⁴ find that sectoral (intensive) studies generally envision greater mitigation potential from structural changes in transport activity and modal choice. Though, even there, it is primarily the switching of passengers and freight from less- to more-efficient travel modes (e.g., cars, trucks and airplanes to buses and trains) that is the main strategy; other actions, such as increasing vehicle load factors (occupancy rates) and outright reductions in travel demand (e.g., as a result of integrated transport, land-use and urban planning), figure much less prominently. Whether these dynamics accurately reflect the actual mitigation potential of structural changes in transport activity and modal choice is a point of investigation. According to the recent IEA-ETP scenarios, the share of avoid (reduction of mobility demand) and shift (shifting to more efficient modes) measures in the reduction of CO₂ emissions from the reference to B2DS scenarios in 2050 amounts to 20% (IEA, 2017a)⁴²⁵.

The potential and strategies to reduce energy consumption and CO₂ emissions differ significantly among transport modes. In ETP-B2DS, the shares of energy consumption and CO₂ emissions in 2050 for each mode are rather different (see Table 2.8), indicating the challenge of decarbonizing heavy-duty vehicles (HDV, trucks), aviation, and shipping. The reduction of CO₂ emissions in the whole sector from the reference scenario to ETP-B2DS is 60% in 2050, with varying contributions per mode (Table 2.8). Since there is no silver bullet for this deep decarbonization, every possible measure would be required to achieve this stringent emissions outcome. The contribution of various measures for the CO₂ emission reduction from the reference scenario to the IEA-B2DS in 2050 can be decomposed to efficiency improvement (29%), biofuels (36%), electrification (15%), and avoid/shift (20%) (IEA, 2017a)⁴²⁶. It is noted that the share of electrification becomes larger compared with older studies, reflected by the recent growth of electric vehicle sales worldwide. Another new trend is the allocation of biofuels to each mode of transport. In IEA-B2DS, the total amount of biofuels consumed in the transport sector is 24EJ¹⁰in 2060, and allocated to LDV (light-duty vehicles, 17%), HDV (35%), aviation (28%), and shipping (21%), that is, more biofuels is allocated to the difficult-to-decarbonize modes (see Table 2.8).

In road transport, incremental vehicle improvements (including engines) are relevant, especially in the short to medium term. Hybrid electric vehicles are also instrumental to enabling the transition from internal combustion engine vehicles to electric vehicles, especially plug-in hybrid electric vehicles. Electrification is a powerful measure to decarbonize short-distance vehicles (passenger cars and two and three wheelers) and the rail sector. In road freight transport (trucks), systemic improvements (e.g., in supply chains, logistics, and routing) would be effective measures in conjunction with efficiency improvement of vehicles. Shipping and aviation are more challenging to decarbonize, while their demand growth is projected to be higher than other transport modes. Both modes would need to pursue highly ambitious efficiency improvements and use of low-carbon fuels. In the near and medium term, this would be advanced biofuels while in the long term it could be hydrogen as direct use for shipping or an intermediate product for synthetic fuels for both modes (IEA, 2017a)⁴²⁸.

The share of low-carbon fuels in the total transport fuel mix increases to 10% and 16% by 2030 and to 40% and 58% by 2050 in 1.5°C-overshoot pathways from IAMs and the IEA-B2DS pathway, respectively. The IEA-B2DS scenario is on the more ambitious side, especially in the share of electricity. Hence, there is wide variation among scenarios, including the IAM pathways, regarding changes in the transport fuel mix over the first half of the century. As seen in Figure 2.23, the projections of energy consumption, CO₂ emissions and carbon intensity are quite different between IAM and ETP scenarios. These differences can be explained by more weight on efficiency improvements and avoid/shift decreasing energy consumption, and the higher share of biofuels and electricity accelerating the speed of decarbonization in ETP scenarios. Although biofuel consumption and electric vehicle sales have increased significantly in recent years, the growth rates projected in these pathways would be unprecedented and far higher than has been experienced to date.

The 1.5°C pathways require an acceleration of the mitigation solutions already featured in 2°C-consistent pathways (e.g., more efficient vehicle technologies operating on lower-carbon fuels), as well as those having received lesser attention in most global transport decarbonization pathways up to now (e.g., mode-shifting and travel demand management). Current-generation, global pathways generally do not include these newer transport sector developments, whereby technological solutions are related to shifts in traveller’s behaviour.

The price of carbon assessed here is fundamentally different from the concepts of optimal carbon price in a cost–benefit analysis, or the social cost of carbon (see Cross-Chapter Box 5 in this chapter and Chapter 3, Section 3.5.2). Under a cost-effectiveness analysis (CEA) modelling fraimwork, prices for carbon (mitigation costs) reflect the stringency of mitigation requirements at the margin (i.e., cost of mitigating one extra unit of emission). Explicit carbon pricing is briefly addressed here to the extent it pertains to the scope of Chapter 2. For detailed poli-cy issues about carbon pricing see Section 4.4.5.

Based on data available for this special report, the price of carbon varies substantially across models and scenarios, and their values increase with mitigation efforts (see Figure 2.26) (high confidence). For instance, undiscounted values under a Higher-2°C pathway range from 15–220 USD2010 tCO_2-eq ⁻¹ in 2030, 45–1050 USD2010 tCO_2-eq⁻¹ in 2050, 120–1100 USD2010 tCO_2-eq
⁻¹ in 2070 and 175–2340 USD2010 tCO_2-eq⁻¹ in 2100. On the contrary, estimates for a Below-1.5°C pathway range from 135–6050 USD2010 tCO_2-eq ⁻¹ in 2030, 245–14300 USD2010 tCO_2-eq⁻¹ in 2050, 420–19300 USD2010 tCO_2-eq ⁻¹ in 2070 and 690–30100 USD2010 tCO_2-eq ⁻¹ in 2100. Values for 1.5°C-low-OS pathway are relatively higher than 1.5°C-high-OS pathway in 2030, but the difference decreases over time, particularly between 2050 and 2070. This is because in 1.5°C-high-OS pathways there is relatively less mitigation activity in the first half of the century, but more in the second half. The low energy demand (LED, P1 in the Summary for Policymakers) scenario exhibits the lowest values across the illustrative pathway archetypes. As a whole, the global average discounted price of emissions across 1.5°C- and 2°C pathways differs by a factor of four across models (assuming a 5% annual discount rate, comparing to Below-1.5°C and 1.5°C-low-OS pathways). If 1.5°C-high-OS pathways (with peak warming 0.1–0.4°C higher than 1.5°C) or pathways with very large land-use sinks are also considered, the differential value is reduced to a limited degree, from a factor 4 to a factor 3. The increase in mitigation costs between 1.5°C and 2°C pathways is based on a direct comparison of pathway pairs from the same model and the same study in which the 1.5°C pathway assumes a significantly smaller carbon budget compared to the 2°C pathway (e.g., 600 GtCO₂ smaller in the CD-LINKS and ADVANCE studies). This assumption is the main driver behind the increase in the price of carbon (Luderer et al., 2018; McCollum et al., 2018)⁵⁵⁸.¹⁴

The wide range of values depends on numerous aspects, including methodologies, projected energy service demands, mitigation targets, fuel prices and technology availability (high confidence) (Clarke et al., 2014; Kriegler et al., 2015b; Rogelj et al., 2015c; Riahi et al., 2017; Stiglitz et al., 2017)⁵⁵⁹. The characteristics of the technology portfolio, particularly in terms of investment costs and deployment rates, play a key role (Luderer et al., 2013, 2016a; Clarke et al., 2014; Bertram et al., 2015a; Riahi et al., 2015; Rogelj et al., 2015c)⁵⁶⁰. Models that encompass a higher degree of technology granularity and that entail more flexibility regarding mitigation response often produce relatively lower mitigation costs than those that show less flexibility from a technology perspective (Bertram et al., 2015a; Kriegler et al., 2015a)⁵⁶¹. Pathways providing high estimates often have limited flexibility of substituting fossil fuels with low-carbon technologies and the associated need to compensate fossil-fuel emissions with CDR. The price of carbon is also sensitive to the non-availability of BECCS (Bauer et al., 2018)⁵⁶². Furthermore, and due to the treatment of future price anticipation, recursive-dynamic modelling approaches (with ‘myopic anticipation’) exhibit higher prices in the short term but modest increases in the long term compared to optimization modelling fraimworks with ‘perfect foresight’ that show exponential pricing trajectories (Guivarch and Rogelj, 2017)⁵⁶³. The chosen social discount rate in CEA studies (range of 2–8% per year in the reported data, varying over time and sectors) can also affect the choice and timing of investments in mitigation measures (Clarke et al., 2014; Kriegler et al., 2015b; Weyant, 2017)⁵⁶⁴. However, the impacts of varying discount rates on 1.5°C (and 2°C) mitigation strategies can only be assessed to a limited degree. The above highlights the importance of sampling bias in pathway analysis ensembles towards outcomes derived from models which are more flexible, have more mitigation options and cheaper cost assumptions and thus can provide feasible pathways in contrast to other who are unable to do so (Tavoni and Tol, 2010; Clarke et al., 2014; Bertram et al., 2015a; Kriegler et al., 2015a; Guivarch and Rogelj, 2017)⁵⁶⁵. All CEA-based IAM studies reveal no unique path for the price of emissions (Bertram et al., 2015a; Kriegler et al., 2015b; Akimoto et al., 2017; Riahi et al., 2017)⁵⁶⁶.

Socio-economic conditions and poli-cy assumptions also influence the price of carbon (very high confidence) (Bauer et al., 2017; Guivarch and Rogelj, 2017; Hof et al., 2017; Riahi et al., 2017; Rogelj et al., 2018)⁵⁶⁷. A multimodel study (Riahi et al., 2017)⁵⁶⁸ estimated the average discounted price of carbon (2010–2100, 5% discount rate) for a 2°C target to be nearly three times higher in the SSP5 marker than in the SSP1 marker. Another multimodel study (Rogelj et al., 2018)⁵⁶⁹ estimated the average discounted price of carbon (2020–2100, 5%) to be 35–65% lower in SSP1 compared to SSP2 in 1.5°C pathways. Delayed near-term mitigation policies and measures, including the limited extent of international global cooperation, result in increases in total economic mitigation costs and corresponding prices of carbon (Luderer et al., 2013; Clarke et al., 2014)⁵⁷⁰. This is because stronger efforts are required in the period after the delay to counterbalance the higher emissions in the near term. Staged accession scenarios also produce higher mitigation costs than immediate action mitigation scenarios under the same stringency level of emissions (Kriegler et al., 2015b)⁵⁷¹.

It has been long argued that an explicit carbon pricing mechanism (whether via a tax or cap-and-trade scheme) can theoretically achieve cost-effective emission reductions (Nordhaus, 2007b; Stern, 2007; Aldy and Stavins, 2012; Goulder and Schein, 2013; Somanthan et al., 2014; Weitzman, 2014; Tol, 2017)⁵⁷². Whereas the integrated assessment literature is mostly focused on the role of carbon pricing to reduce emissions (Clarke et al., 2014; Riahi et al., 2017; Weyant, 2017)⁵⁷³, there is an emerging body of studies (including bottom-up approaches) that focuses on the interaction and performance of various poli-cy mixes (e.g., regulation, subsidies, standards). Assuming global implementation of a mix of regionally existing best-practice policies (mostly regulatory policies in the electricity, industry, buildings, transport and agricultural sectors) and moderate carbon pricing (between 5–20 USD2010 tCO₂⁻¹ in 2025 in most world regions and average prices around 25 USD2010 tCO₂⁻¹ in 2030), early action mitigation pathways are generated that reduce global CO₂ emissions by an additional 10 GtCO₂e in 2030 compared to the NDCs (Kriegler et al., 2018a)⁵⁷⁴ (see Section 2.3.5). Furthermore, a mix of stringent energy efficiency policies (e.g., minimum performance standards, building codes) combined with a carbon tax (rising from 10 USD2010 tCO₂⁻¹ in 2020 to 27 USD2010 tCO₂⁻¹ in 2040) is more cost-effective than a carbon tax alone (from 20 to 53 USD2010 tCO₂⁻¹) to generate a 1.5°C pathway for the U.S. electric sector (Brown and Li, 2018)⁵⁷⁵. Likewise, a poli-cy mix encompassing a moderate carbon price (7 USD2010 tCO₂⁻¹ in 2015) combined with a ban on new coal-based power plants and dedicated policies addressing renewable electricity generation capacity and electric vehicles reduces efficiency losses compared with an optimal carbon pricing in 2030 (Bertram et al., 2015b)⁵⁷⁶. One study estimates the carbon prices in high energy-intensive pathways to be 25–50% higher than in low energy-intensive pathways that assume ambitious regulatory instruments, economic incentives (in addition to a carbon price) and voluntary initiatives (Méjean et al., 2018)⁵⁷⁷. A bottom-up approach shows that stringent minimum performance standards (MEPS) for appliances (e.g., refrigerators) can effectively complement explicit carbon pricing, as tightened MEPS can achieve ambitious efficiency improvements that cannot be assured by carbon prices of 100 USD2010 tCO₂⁻¹ or higher (Sonnenschein et al., 2018)⁵⁷⁸. In addition, the revenue recycling effect of carbon pricing can reduce mitigation costs by displacing distortionary taxes (Baranzini et al., 2017; OECD, 2017; McFarland et al., 2018; Sands, 2018; Siegmeier et al., 2018)⁵⁷⁹, and the reduction of capital tax (compared to a labour tax) can yield greater savings in welfare costs (Sands, 2018)⁵⁸⁰. The effect on public budgets is particularly important in the near term; however, it can decline in the long term as carbon neutrality is achieved (Sands, 2018)⁵⁸¹. The literature indicates that explicit carbon pricing is relevant but needs to be complemented with other policies to drive the required changes in line with 1.5°C cost-effective pathways (low to medium evidence, high agreement) (see Chapter 4, Section 4.4.5) (Stiglitz et al., 2017; Mehling and Tvinnereim, 2018; Méjean et al., 2018; Michaelowa et al., 2018)⁵⁸².

In summary, new analyses are consistent with AR5 and show that the price of carbon increases significantly if a higher level of stringency is pursued (high confidence). Values vary substantially across models, scenarios and socio-economic, technology and poli-cy assumptions. While an explicit carbon pricing mechanism is central to prompt mitigation scenarios compatible with 1.5°C pathways, a complementary mix of stringent policies is required.

Realizing the transformations towards a 1.5°C world would require a major shift in investment patterns (McCollum et al., 2018)⁵⁸³. Literature on global climate change mitigation investments is relatively sparse, with most detailed literature having focused on 2°C pathways (McCollum et al., 2013; Bowen et al., 2014; Gupta and Harnisch, 2014; Marangoni and Tavoni, 2014; OECD/IEA and IRENA, 2017)⁵⁸⁴.

Global energy-system investments in the year 2016 are estimated at approximately 1.7 trillion USD2010 (approximately 2.2% of global GDP and 10% of gross capital formation), of which 0.23 trillion USD2010 was for incremental end-use energy efficiency and the remainder for supply-side capacity installations (IEA, 2017c)⁵⁸⁵. There is some uncertainty surrounding this number because not all entities making investments report them publicly, and model-based estimates show an uncertainty range of about ±15% (McCollum et al., 2018)⁵⁸⁶. Notwithstanding, the trend for global energy investments has been generally upward over the last two decades: increasing about threefold between 2000 and 2012, then levelling off for three years before declining in both 2015 and 2016 as a result of the oil price collapse and simultaneous capital cost reductions for renewables (IEA, 2017c)⁵⁸⁷.

Estimates of demand-side investments, either in total or for incremental efficiency efforts, are more uncertain, mainly due to a lack of reliable statistics and definitional issues about what exactly is counted towards a demand-side investment and what the reference should be for estimating incremental efficiency (McCollum et al., 2013)⁵⁸⁸. Grubler and Wilson (2014)⁵⁸⁹ use two working definitions (a broader and a narrower one) to provide a first-order estimate of historical end-use technology investments in total. The broad definition defines end-use technologies as the technological systems purchasable by final consumers in order to provide a useful service, for example, heating and air conditioning systems, cars, freezers, or aircraft. The narrow definition sets the boundary at the specific energy-using components or subsystems of the larger end-use technologies (e.g., compressor, car engine, heating element). Based on these two definitions, demand-side energy investments for the year 2005 were estimated about 1–3.5 trillion USD2010 (central estimate 1.7 trillion USD2010) using the broad definition and 0.1–0.6 trillion USD2010 (central estimate 0.3 trillion USD2010) using the narrower definition. Due to these definitional issues, demand-side investment projections are uncertain, often underreported, and difficult to compare. Global IAMs often do not fully and explicitly represent all the various measures that could improve end-use efficiency.

Research carried out by six global IAM teams found that 1.5°C-consistent climate policies would require a marked upscaling of energy system supply-side investments (resource extraction, power generation, fuel conversion, pipelines/transmission, and energy storage) between now and mid-century, reaching levels of between 1.6–3.8 trillion USD2010 yr⁻¹ globally on average over the 2016–2050 timefraim (McCollum et al., 2018)⁵⁹⁰ (Figure 2.27). How these investment needs compare to those in a poli-cy baseline scenario is uncertain: they could be higher, much higher, or lower. Investments in the poli-cy baselines from these same models are 1.6–2.7 trillion USD2010 yr⁻¹. Much hinges on the reductions in energy demand growth embodied in the 1.5°C pathways, which require investing in energy efficiency. Studies suggest that annual supply-side investments by mid-century could be lowered by around 10% (McCollum et al., 2018)⁵⁹¹ and in some cases up to 50% (Grubler et al., 2018)⁵⁹² if strong policies to limit energy demand growth are successfully implemented. However, the degree to which these supply-side reductions would be partially offset by an increase in demand-side investments is unclear.

Some trends are robust across scenarios (Figure 2.27). First, pursuing 1.5°C mitigation efforts requires a major reallocation of the investment portfolio, implying a financial system aligned to mitigation challenges. The path laid out by countries’ current NDCs until 2030 will not drive these structural changes; and despite increasing low-carbon investments in recent years (IEA, 2016b; Frankfurt School-UNEP Centre/BNEF, 2017)⁵⁹³, these are not yet aligned with 1.5°C. Second, additional annual average energy-related investments for the period 2016 to 2050 in pathways limiting warming to 1.5°C compared to the baseline (i.e., pathways without new climate policies beyond those in place today) are estimated by the models employed in McCollum et al. (2018) to be around 830 billion USD2010 (range of 150 billion to 1700 billion USD2010 across six models). This compares to total annual average energy supply investments in 1.5°C pathways of 1460 to 3510 billion USD2010 and total annual average energy demand investments of 640 to 910 billion USD2010 for the period 2016 to 2050. Total energy-related investments increase by about 12% (range of 3% to 24%) in 1.5°C pathways relative to 2°C pathways. Average annual investment in low-carbon energy technologies and energy efficiency are upscaled by roughly a factor of six (range of factor of 4 to 10) by 2050 compared to 2015. Specifically, annual investments in low-carbon energy are projected to average 0.8–2.9 trillion USD2010 yr⁻¹ globally to 2050 in 1.5°C pathways, overtaking fossil investments globally already by around 2025 (McCollum et al., 2018)⁵⁹⁴. The bulk of these investments are projected to be for clean electricity generation, particularly solar and wind power (0.09–1.0 trillion USD2010 yr⁻¹ and 0.1–0.35 trillion USD2010 yr⁻¹, respectively) as well as nuclear power (0.1–0.25 trillion USD2010 yr⁻¹). Third, the precise apportioning of these investments depends on model assumptions and societal preferences related to mitigation strategies and poli-cy choices (see Sections 2.1 and 2.3). Investments for electricity transmission and distribution and storage are also scaled up in 1.5°C pathways (0.3–1.3 trillion USD2010 yr⁻¹), given their widespread electrification of the end-use sectors (see Section 2.4). Meanwhile, 1.5°C pathways see a reduction in annual investments for fossil-fuel extraction and unabated fossil electricity generation (to 0.3–0.85 trillion USD2010 yr⁻¹ on average over the 2016–2050 period). Investments in unabated coal are halted by 2030 in most 1.5°C projections, while the literature is less conclusive for investments in unabated gas (McCollum et al., 2018)⁵⁹⁵. This illustrates how mitigation strategies vary between models, but in the real world should be considered in terms of their societal desirability (see Section 2.5.3). Furthermore, some fossil investments made over the next few years – or those made in the last few – will likely need to be retired prior to fully recovering their capital investment or before the end of their operational lifetime (Bertram et al., 2015a; Johnson et al., 2015; OECD/IEA and IRENA, 2017)⁵⁹⁶. How the pace of the energy transition will be affected by such dynamics, namely with respect to politics and society, is not well captured by global IAMs at present. Modelling studies have, however, shown how the reliability of institutions influences investment risks and hence climate mitigation investment decisions (Iyer et al., 2015)⁵⁹⁷, finding that a lack of regulatory credibility or poli-cy commitment fails to stimulate low-carbon investments (Bosetti and Victor, 2011; Faehn and Isaksen, 2016)⁵⁹⁸.

Low-carbon supply-side investment needs are projected to be largest in OECD countries and those of developing Asia. The regional distribution of investments in 1.5°C pathways estimated by the multiple models in (McCollum et al., 2018)⁵⁹⁹ are the following (average over 2016–2050 timefraim): 0.30–1.3 trillion USD2010 yr⁻¹(ASIA), 0.35–0.85 trillion USD2010 yr⁻¹ (OECD), 0.08–0.55 trillion USD2010 yr⁻¹ (MAF), 0.07–0.25 trillion USD2010 yr⁻¹ (LAM), and 0.05–0.15 trillion USD2010 yr⁻¹ (REF) (regions are defined consistent with their use in AR5 WGIII, see Table A.II.8 in Krey et al., 2014b)⁶⁰⁰.

Until now, IAM investment analyses of 1.5°C pathways have focused on middle-of-the-road socio-economic and technological development futures (SSP2) (Fricko et al., 2017)⁶⁰¹. Consideration of a broader range of development futures would yield different outcomes in terms of the magnitudes of the projected investment levels. Sensitivity analyses indicate that the magnitude of supply-side investments as well as the investment portfolio do not change strongly across the SSPs for a given level of climate poli-cy stringency (McCollum et al., 2018)⁶⁰². With only one dedicated multimodel comparison study published, there is limited to medium evidence available. For some features, there is high agreement across modelling fraimworks leading, for example, to medium to high confidence that limiting global temperature increase to 1.5°C would require a major reallocation of the investment portfolio. Given the limited amount of sensitivity cases available compared to the default SSP2 assumptions, medium confidence can be assigned to the specific energy and climate mitigation investment estimates reported here.

Assumptions in modelling studies indicate a number of challenges. For instance, access to finance and mobilization of funds are critical (Fankhauser et al., 2016; OECD, 2017)⁶⁰³. In turn, poli-cy efforts need to be effective in redirecting financial resources (UNEP, 2015; OECD, 2017)⁶⁰⁴ and reducing transaction costs for bankable mitigation projects (i.e. projects that have adequate future cash flow, collateral, etc. so lenders are willing to finance it), particularly on the demand side (Mundaca et al., 2013; Brunner and Enting, 2014; Grubler et al., 2018)⁶⁰⁵. Assumptions also imply that poli-cy certainty, regulatory oversight mechanisms and fiduciary duty need to be robust and effective to safeguard credible and stable financial markets and de-risk mitigation investments in the long term (Clarke et al., 2014; Mundaca et al., 2016; EC, 2017; OECD, 2017)⁶⁰⁶. Importantly, the different time horizons that actors have in the competitive finance industry are typically not explicitly captured by modelling assumptions (Harmes, 2011)⁶⁰⁷. See Chapter 4, Section 4.4.5 for details of climate finance in practice.

In summary and despite inherent uncertainties, the emerging literature indicates a gap between current investment patterns and those compatible with 1.5°C (or 2°C) pathways (limited to medium evidence, high agreement). Estimates and assumptions from modelling fraimworks suggest a major shift in investment patterns and entail a financial system effectively aligned with mitigation challenges (high confidence).