Assessing the Feasibility of Global Long-Term Mitigation Scenarios

Abstract

This study explores the critical notion of how feasible it is to achieve long-term mitigation goals to limit global temperature change. It uses a model inter-comparison of three integrated assessment models (TIAM-Grantham, MESSAGE-GLOBIOM and WITCH) harmonized for socio-economic growth drivers using one of the new shared socio-economic pathways (SSP2), to analyse multiple mitigation scenarios aimed at different temperature changes in 2100, in order to assess the model outputs against a range of indicators developed so as to systematically compare the feasibility across scenarios. These indicators include mitigation costs and carbon prices, rates of emissions reductions and energy efficiency improvements, rates of deployment of key low-carbon technologies, reliance on negative emissions, and stranding of power generation assets. The results highlight how much more challenging the 2 °C goal is, when compared to the 2.5–4 °C goals, across virtually all measures of feasibility. Any delay in mitigation or limitation in technology options also renders the 2 °C goal much less feasible across the economic and technical dimensions explored. Finally, a sensitivity analysis indicates that aiming for less than 2 °C is even less plausible, with significantly higher mitigation costs and faster carbon price increases, significantly faster decarbonization and zero-carbon technology deployment rates, earlier occurrence of very significant carbon capture and earlier onset of global net negative emissions. Such a systematic analysis allows a more in-depth consideration of what realistic level of long-term temperature changes can be achieved and what adaptation strategies are therefore required.

1. Introduction

The Intergovernmental Panel on Climate Change (IPCC)’s 5th assessment report Working Group III [1] is based on hundreds of scenarios which assess the environmental, economic and energy technology consequences of reducing greenhouse gas (GHG) emissions in line with future long term climate goals. These scenarios have been produced using integrated assessment models (IAMs), which represent how future demands for energy, land use and other GHG-producing goods and services are linked to projections of population and economic growth, what technologies and energy sources are used to meet these future demands, and what GHG emissions result.

A detailed examination of the main implications of these scenarios [2] highlights that the 2 °C mitigation goal is still in reach at reasonable cost, although a substantial transformation of the global energy system is required throughout the 21st century, which means that any delays to action, any lack of ambition in energy efficiency improvements, and any absence of major technologies could result in significant additional costs and even jeopardise the achievability of this goal.

This study consists of a new, post-IPCC 5th assessment, set of scenarios designed to further explore the many dimensions of emissions reduction at a global level, with a particular focus on critically assessing the degree of feasibility and challenge associated with the most stringent mitigation scenarios. In constructing the scenarios, a number of novel aspects have been developed, compared to the hundreds of scenarios explored in the IPCC’s 5th assessment report:

• Constraints using newly-derived CO₂ budgets from Met Office Hadley Centre;
• Model inter-comparison using population and economic growth assumptions from one of the new shared socio-economic pathways (SSP2) [3];
• Production of a database of scenarios which allows key metrics (fossil share of primary energy, electricity share of final energy, mitigation costs, CO₂ sequestered) to be shown in a stepwise manner when moving between different temperature targets, different levels of delay (to 2020, to 2030) and different technology constraints. This goes further than what the IPCC 5th assessment database allows (as that focuses primarily on 2 and 2.5 °C scenarios, including a particular lack of sampling in the range 2.5–3.5 °C [4]);
• Some new technology constraint scenarios (carbon capture and storage (CCS) only available for deployment from 2050, as opposed to no CCS which has been widely explored in the IPCC’s 5th assessment, and constrained electrification of end-use sectors, which has not yet been explored).

The IPCC fifth assessment report, Working Group III (AR5 WGIII) [5] states that, “on the question of whether the [mitigation] pathways are feasible, integrated models can inform this question by providing relevant information such as rates of deployment of energy technologies, economic costs, finance transfers between regions and links to policy objectives (energy security, energy prices). However, these models cannot determine feasibility in an absolute sense. Scenario feasibility often arises from pushing models beyond the bounds they were designed to explore, but this doesn’t mean the scenario cannot be achieved—different models have different feasibility limits”. Riahi et al. [6] discuss such feasibility limits as being reached when a particular model cannot find a solution to a mitigation constraint, as a result of:

• Lack of mitigation options;
• Binding constraints for the diffusion of technologies;
• Extremely high price signals (such as rapid increases in carbon prices).

Riahi et al. [6] go on to caution that these feasibility limits concern technical and economic issues, and must be strictly differentiated from the feasibility of a low-carbon transformation in the real world, which also depends on a number of other factors such as political and social concerns.

Different indicators related to the degree of difficulty in meeting mitigation pathways have been discussed in the literature. These include:

• Mitigation costs: The latest IPCC assessment report (WGIII) [5] has costs of mitigation for “idealised implementation” scenarios (achieving a range of atmospheric GHG concentrations of between 430 and 480 ppm CO₂e of 1.5%–15% of Gross Domestic Product (GDP) (median = 3%, interquartile range 2%–6%) over the period 2015–2100 (Net Present Value, discounted at 5%).
• Carbon prices: For idealised implementation scenarios, carbon prices in the 430–480 ppm scenarios rise to between $100/tCO₂e and $6000/tCO₂e (median = $1500tCO₂e, interquartile range $1000–2000/tCO₂e) by 2100 [5].
• Model solution: As noted by the IPCC 5th assessment report [5], reported ranges may contain a downward bias towards costs of mitigation and carbon prices, since they only represent results for models that solve. Model solution has been discussed as a key facet of assessing the feasibility of low-carbon pathways [6,7], although as noted in Kriegler et al. [7], feasibility is subject to different interpretations around model solution, political actions or availability of any set of technologies or actions that could meet a target.
• Implications for idled high-carbon assets: International Energy Agency (IEA) [8] estimates that a 450 ppm scenario would result in $300 billion of stranded fossil fuel assets, and more if policy lacks clarity. Johnson et al. [9] show that, in a mitigation scenario aimed at achieving a 450 ppm GHG concentration following weak policy action to 2030, there would be on average 350 GW of stranded conventional coal plants over the period 2030–2050.
• Technology deployment rates: As demonstrated by van der Zwaan et al. [10], technology deployment rates between scenarios can highlight the degree of challenge of different scenario sets, with many hundreds of GW of key supply-side technologies such as nuclear, solar PV and wind deployed in least-cost low-carbon pathways—in many cases several multiples of historical deployment rates of these technologies.
• The degree of reliance on negative emissions and other specific technologies like CCS: Numerous studies have highlighted the degree of dependence of the cost-effectiveness of low-carbon pathways on the availability of CCS [6,11], with negative emissions (combining bio-energy with CCS) a key facet of achieving low-carbon pathways [12]
• Rates of decarbonisation and energy efficiency improvements: Rates of decarbonisation in low-carbon scenarios have been used to understand the degree of challenge associated with these scenarios, with high rates of decarbonisation (beyond 3.5% per year) having been asserted as “extreme” in Den Elzen et al.’s 2010 analysis [13], but far higher rates (beyond 10% per year) included in models deemed feasible in more recent analysis by Riahi et al. [6]. Economy-wide and sector-specific energy efficiency improvements have also been analysed in a range of low-carbon scenarios [5,14].

All of these aspects, or combinations of some of these aspects, have been drawn out of previous modelling exercises to assess the degree of difficulty or challenge in meeting low-carbon scenarios with either delayed action, technology limitations, or different temperature goals (see in particular Luderer et al. [15,16] and von Stechow et al. [17]). However, a multi-factor scenario comparison framework regarding mitigation feasibility has yet to be presented in a holistic and systematic way which allows direct comparison of the degree of challenge of different mitigation scenarios, as presented here.

It should also be noted that feasibility analysis is increasingly using historical energy transitions experience to understand how challenging future transitions might be, in light of relevant metrics which relate to past energy transitions [1,18,19,20,21,22]. This paper does not focus on an assessment of feasibility in light of such historical benchmarks, but rather on relative challenges of future scenarios. As is elaborated in the rest of this paper, such a systematic assessment makes clear the degree of challenge associated with achieving goals of below 2 °C, particularly with any delays to international mitigation action or technology limitations.

The rest of this paper is structured as follows. the full description of scenarios, and methods used to assess feasibility within them, is given in Section 2. Section 3 discusses the scenario results, with analysis of several different aspects of the most stringent mitigation scenarios in order to explore the range of implications associated with this degree of mitigation, and the reasons the models’ results differ, before presenting a comparison of the scenarios using the metrics presented in Section 2. This enables an assessment of the relative degree of challenge associated with each mitigation scenario. Section 4 presents a discussion of the implications of this systematic comparison, particularly from the perspective of the degree of challenge associated with achieving the 2 °C goal.

2. Materials and Methods

Table 1. Mitigation scenarios explored in this study.

Median Temperature Change/°C by 2100 (Relative to Pre-Industrial)	Cumulative (2000–2100) CO2 Emissions from Fossil Fuel Combustion and Industry (GtCO2)	Scenario Variants
2	1340	Immediate action from model base year 1
		Action from 2020, following moderate action
		Action from 2020, following moderate action, with the introduction of CCS delayed until 2050
		Action from 2020, following moderate action, with limited potential for electricity in end-use sectors
		Action from 2030, following moderate action
2.5	2260	Immediate action from model base year
		Action from 2020, following moderate action
		Action from 2030, following moderate action
3	3560	Immediate action from model base year
		Action from 2020, following moderate action
		Action from 2030, following moderate action
4	5280	Immediate action from model base year
		Action from 2020, following moderate action
		Action from 2030, following moderate action
4.6 2	6000	None

Notes: ¹ Model base years are shown in Table 2; ² Reference associated temperature change calculated for the TIAM (TIMES Integrated Assessment Model)-Grantham run only. WITCH (World Induced Technical Change Hybrid) and MESSAGE (Model for Energy Supply Strategy Alternatives and their General Environmental Impact)-GLOBIOM (Global Biosphere Management Model) reference runs have cumulative CO₂ levels of 5850 GtCO₂ and 5650 GtCO₂ respectively, so would have lower associated temperature changes in 2100.

describes the full scenario set used in this study. The scenario design has been focused on adding additional insight to those scenarios explored in studies included in the IPCC’s 5th assessment report, and to reflect some of the emerging policy-relevant challenges of decarbonisation. In particular, widespread commercial deployment of CCS continues to prove elusive, demanding an analysis of the implications of delays in CCS deployment. Furthermore, the importance of electrification in end-use sectors suggests analysing the implications of limited electrification is also important. Finally, a stepwise increase in long-term temperature goals (LTTGs) allows a systematic comparison of the implications of costs and rates of decarbonisation associated with more or less ambitious goals.

In the scenarios described in Table 1, “moderate” action refers to a level of emissions reductions (to 2020 or 2030, respectively) in line with the less stringent end of countries’ Cancun pledges (where these have been quantified) and reference or unmitigated emissions where these have not been quantified, with full details given in Appendix A. The 2020 and 2030 global CO₂ figures, at 39 GtCO₂ and 41 GtCO₂, are 18% and 24% higher than 2010 CO₂ emissions levels from fossil fuels and industrial processes (at 33 GtCO₂). This compares to the total GHG emissions levels estimated by The United Nations Environment Programme (UNEP)’s 2014 Emissions Gap report [23] in the least stringent version of the Cancun pledges, at 12% and 20% higher than 2010 GHG emissions. However, as shown in Appendix A, the 2020 and 2030 fossil and industry CO₂ estimates for the weak interpretation of the Cancun pledges in this study compare fairly closely to those in the Assessment of Climate Change Mitigation Pathways and Evaluation of the Robustness of Mitigation Cost Estimates (AMPERE) study [6] in which two of the three models in this inter-comparison (WITCH (World Induced Technical Change Hybrid) and MESSAGE (Model for Energy Supply Strategy Alternatives and their General Environmental Impact)) participated. It should be noted that—although the model inter-comparison undertaken in this study pre-dated the signing of the Paris Agreement [24] in December 2015, the Intended Nationally Determined Contributions (INDCs) of countries made in the run-up Paris 21st Conference of the Parties (COP 21) in 2015 sum to a total GHG emissions level of approximately 55 GtCO₂e in 2030, marginally higher than the Cancun pledges estimate of about 53 GtCO₂e in 2020 [23]. As such, the 2030 “action from 2030” scenario, with 41 GtCO₂ in 2030 compared to 39 GtCO₂ in 2020, represents a useful approximation to the case where action in line with the INDCs is undertaken to 2030, before global coordinated mitigation action to the LTTGs is enacted.

Where the potential for end-use electrification has been limited, this has been done to allow only moderate increases in the share of electricity in the end-use (i.e., transport, buildings and industry) sectors over and above current shares. This reflects barriers to the increasing penetration of electricity end-use technologies such as heat pumps, electric vehicles, as well as electric process heating in the industrial manufacturing sectors. Details of how these electrification caps have been derived are given in Appendix B.

Three different IAMs have been inter-compared in order to explore variations in key input assumptions around future technology costs, fossil fuel supply and costs, as well as energy efficiency improvement potential:

• The Imperial College London Grantham Institute’s TIMES IAM (TIAM-Grantham) [25,26];
• The International Institute for Applied Systems Analysis (IIASA)’s MESSAGE model (MESSAGE-GLOBIOM (Global Biosphere Management Model)) [14,27,28,29];
• The Centro Euro-Mediterraneo sui Cambiamenti Climatici (CMCC)’s WITCH model [30].

Appendix C provides a brief description of each model, and

Table 2. Integrated assessment models (IAMs) in this study and their key features. Notes: Key input assumptions around technology costs are shown in Figure 8; CCS: carbon capture and storage; BECCS: bioenergy with carbon capture and storage (a key “negative emissions” technology); PV: photovoltaics; and CSP: concentrated solar power.

Model	New Nuclear	CCS	BECCS	Solar (PV and CSP)	Wind (on and offshore)	Time Step (years)	Base Year	Solution Approach
TIAM-Grantham [25,26]	Yes	Yes	Yes	Yes	Yes	10	2012	Inter-temporal optimisation
MESSAGE-GLOBIOM [14,27,28,29]	Yes	Yes	Yes	Yes	Yes	10	2010	Inter-temporal optimisation and recursive dynamic
WITCH [30]	Yes	Yes	Yes	Yes	Yes	5	2010	Inter-temporal optimisation

its key features. In order to limit the degree of differentiation, population and economic growth assumptions have been equalised across models, taken from the shared SSP2 scenario [3]. The SSPs have been developed to provide a standardised set of assumptions for the integrated assessment model and impacts, adaptation and vulnerability (IAV) communities. The storylines underlying each SSP range from relatively conservative assumptions on population growth, economic growth and other factors driving the degree of challenge for mitigation and adaptation, to drivers which make either or both of these objectives highly challenging. For this study, population and economic growth driers from SSP2 have been selected (specifically the Organisation for Economic Cooperation and Development (OECD) variant which provides a median level of GDP growth throughout the century), as it is considered the most closely associated with recent socio-economic growth patterns [31]. This helps to assess the feasibility of meeting the stringent targets even in the face of future energy demand growth based on current trends in socio-economic growth.

The IAM scenarios have been limited to an assessment of the impacts of reducing CO₂ emissions from energy systems (resulting from the combustion of fossil fuels) and industrial process (principally from the chemistry of the cement production process). Since future temperature change will depend not just on CO₂ emissions from these sources, but also from a) CO₂ emissions from land use and b) non-CO₂ emissions from a variety of sources such as agriculture, waste and industrial manufacturing, these sources must also be assessed in any future climate scenario. This has been done by deriving estimated emissions from other GHG sources in scenarios consistent with different LTTGs using data from the Representative Concentration Pathways (RCPs) as well as IIASA’s Greenhouse Gas Air Pollution Interactions and Synergies (GAINS) model. Figure 1 summarises the modelling steps to arrive at this temperature change level, with a full description in Appendix D.

Table 3. Indicators for degree of challenge in achieving mitigation scenarios.

Indicator	Relevance	Example of Challenge
Does the model “solve”	Models contain a wide range of technologies and significant energy efficiency improvement capability. Lack of solution implies more ambitious technology deployment and efficiency improvements must be achieved in reality [1].	All models provide an analytical solution for all scenarios explored, although for 2 °C scenario with global action delayed to 2030, TIAM-Grantham reaches its $10,000/tCO2 limit by 2100, indicating this is at its own model-defined feasibility limit (See Section 3.2).
CO2 price and rate of increase	Very high CO2 prices would imply energy services are very expensive. Very rapid decadal rises in CO2 price imply rapid adjustments to energy prices, indicating a limited availability of low-carbon technologies to provide rapid mitigation possibilities at reasonable costs. Both of these could be socially unacceptable and/or result in economic instability [33].	For the 2 °C scenario with global action delayed to 2030, two models (TIAM-Grantham and WITCH) see decadal CO2 price increases of greater than $1,000/tCO2 (See Section 3.2).
Mitigation cost	High mitigation cost implies more expensive energy, which indicates a lack of available, reasonable cost mitigation technologies, and which is likely to lead to resistance from households and businesses.	WITCH mitigation cost for 2 °C scenario with global action delayed to 2030 costs almost 10% of 21st century GDP. This may be unacceptably high (see Section 3.3).
Rate of decarbonis-ation	No sustained periods of historical decarbonization globally since the beginning of the 20th century. At a country level rates of up to 3% per year during periods of policy to achieve a rapid shift away from oil [6].	WITCH and TIAM-Grantham both show average annual CO2 reduction rates in excess of 10% per year over the decade 2030-2040, in 2 °C scenario with global action delayed to 2030 (See Section 3.4).
Rate of energy intensity improvements	Very rapid energy efficiency improvements across the economy would require a widespread shift to a range of technologies prone to behavioural barriers [34] and would also require avoidance of significant rebound effects [34].	WITCH sees almost flat final energy demand globally over the 21st century in the 2 °C scenario with action delayed to 2020. This compares to a more-than-doubling of final energy demand in the reference scenario (see Section 3.4).
Technology deployment rates	Significant decadal increases in particular technologies must be questioned on the grounds of real-world ability to develop and scale up supply chains and access skills and labour, and financial and material resources [10,35].	In the 2 °C scenario with delayed action to 2020, the most striking deployment rates over the period 2020–2030 are for nuclear (830 GW in WITCH, more than twice current deployed capacity), gas with CCS (800 GW in TIAM-Grantham), biomass with CCS (520 GW in WITCH), and onshore wind (480 GW in MESSAGE-GLOBIOM, approximately current installed capacity) (See Section 3.4).
Idling of high-carbon assets	Early retirement (as evidenced by sustained zero capacity factors of coal plants within their lifetime) means potentially significant economic losses for coal-fired electricity generators. This will lead to resistance from utilities to idle these plants [9].	In the 2 °C scenario with delayed action to 2030, TIAM-Grantham has 780 GW of zero capacity factor coal plants in 2040, of which 315 GW has 20 or more years of remaining life. In the 2 °C scenario with delayed action to 2020, TIAM-Grantham has 1400 GW of idle coal plant by 2030, of which almost 1200 GW has 7 years of remaining life (See Section 3.5).
Quantity of CO2 captured and stored	Implies successful large-scale deployment of CCS, overcoming technical, economic, legal and other barriers for CO2 transport and storage [36].	MESSAGE-GLOBIOM and TIAM-Grantham see over 30 GtCO2/year captured by 2080 in the 2 °C scenario with delayed action to 2020 (see Section 3.6).
Timing of net global negative CO2 emissions	Very large-scale deployment of negative emissions technologies (e.g., BECCS) poses technical, regulatory, infrastructure, economic challenges [37,38,39].	All three models see global CO2 emissions at negative levels by 2080 in the 2 °C scenario with delayed action to 2030 (see Section 3.6), with CCS deployed from the 2020s onwards.

outlines the different dimensions of feasibility explored. None of these dimensions is definitive in determining the degree of feasibility of any given scenario. In particular, the mitigation cost and carbon prices only provide macroeconomic metrics of energy system decarbonisation cost. In reality, the costs of mitigation, through rising energy and fuel prices, are likely to be felt differently across different socio-economic groups and in different regions (for example see [32]). The models used here therefore provide only a high-level interpretation of the economic costs of mitigation. Nevertheless, taken together, they provide an important set of indicators of how challenging each mitigation scenario is likely to be.

3. Results

3.1. Overview of Results

Global CO₂ emissions in the scenarios with mitigation action starting in 2020, as well as the unmitigated reference scenarios, are shown in Figure 2.

This figure highlights the very different pathways that the different temperature change goals require, particularly from 2020 onwards, with the 2 °C pathways all seeing immediate rapid reductions in CO₂ emissions. The 3 °C and above scenarios see continuing increases in emissions through the 2020s, whilst the picture for 2.5 °C is somewhat more mixed, with a range of decarbonisation rates, from insignificant (as for TIAM-Grantham) to very significant (as for WITCH).

3.2. Can the Models Achieve the Different Temperature Goals?

If global coordinated mitigation action is delayed until 2030, two models (WITCH, MESSAGE-GLOBIOM) can still technically meet the 21st century CO₂ budget. The TIAM-Grantham model can only solve by relying in the last decade of the century on a theoretical “backstop” technology which mitigates CO₂ at a cost of $10,000/tCO₂. Its results have been included here for illustrative purposes only, since the level of backstop technology is an arbitrary choice and does not indicate scenario impossibility in an absolute sense. In principle it would be possible to specify a lower-cost backstop technology if it were considered feasible to deploy measures such as air capture or other CO₂ removal technologies at lower costs.

In addition to the model solution considerations, two models (WITCH and TIAM-Grantham) show very large CO₂ price shocks, as shown in Figure 3. In the WITCH model, the CO₂ price increases from zero to $1400/tCO₂ between 2030 and 2040, whilst in the TIAM-Grantham model, the CO₂ price increases by more than $1000/tCO₂ per decade from 2060 onwards. Such decadal rises in CO₂ prices (with $1000/tCO₂ equivalent to an increase of $270/bbl in the price of crude oil) have been suggested to be a useful indication of scenario infeasibility, as they would represent substantial shocks to the global energy-economic system [33]. In the MESSAGE-GLOBIOM model, the CO₂ price increases more gradually, but this is largely as a result of much lower CO₂ emissions growth in the period 2010–2030.

3.3. What is the Cost of Mitigation?

The measures of mitigation cost (as shown in Figure 4) reported by each of the three models is different. TIAM-Grantham reports the annual change in global welfare compared to the reference, as defined by the sum of changes in consumer and producer surplus, which is essentially the change in energy system cost once changes in energy service supply and demand (that result from changes in energy prices) have been accounted for. MESSAGE-GLOBIOM links the changes in energy prices from its energy-technology module to an aggregated macro-economic growth model, in order to investigate the changes in production and consumption of all goods and services (i.e., not just energy, as in TIAM-Grantham) that result from the mitigation scenario. WITCH reports a “policy cost”, which results from a more detailed macro-economic model, taking into account fully the general equilibrium effects of climate policies.

There is no simple relationship between how the mitigation cost is calculated and the magnitude of the cost, i.e., the degree to which a mitigation cost including a more complete set of macro-economic feedbacks leads to a larger or smaller cost compared to a cost based purely on the energy system technology costs [40]. However, mitigation costs calculated by only analysing energy system costs tend to be lower. In addition, technology availability and cost is a key determinant of mitigation costs across models. As can be seen from Figure 4, the relative mitigation costs between scenarios (indicated by the shape of the cost curves) are broadly similar across the three models, with an increasingly sharp rise in cost between the 3 °C and 2.5 °C, and the 2.5 °C and 2 °C scenarios, and with delayed global mitigation action and technology limitations leading to increased mitigation costs for the 2 °C scenarios in particular. The magnitude of mitigation costs is similar in TIAM-Grantham and MESSAGE-GLOBIOM, but in general much higher in WITCH.

The TIAM-Grantham and MESSAGE-GLOBIOM models’ mitigation costs for the 2°C scenario with immediate action and delayed action to 2020 (in a range of about 1.3%–1.7% of present value GDP to 2100) are similar to those found in previous AVOID studies which used variants of these models to assess regional mitigation costs for China and India [41,42,43]. The higher costs for the WITCH model reflect its macro-economic structure, which includes a production function with energy supply technologies “nested” together and with limited substitutability, which may be too rigid to reflect longer-term possibilities for low-carbon technologies to replace high-carbon technologies in the energy supply sectors. In addition, there are limited mitigation options in the transport sector within the model. Combined, these tend to result in much higher mitigation costs.

Across all three models, the global cost range for achieving the 2 °C scenarios spans 1.1%–10% of present value GDP to 2100 (equivalent to $34–288 trillion). This order of magnitude difference has been reported in previous modelling exercises, notably Clarke et al. [44] whose Energy Modelling Forum 22 (EMF 22) study showed present value mitigation costs for a 450 ppm scenario ranging from $12–120 trillion over the century.

3.4. How Fast Does the Energy System Decarbonise?

Table 4. Average annual rate of change of global CO₂ in decade following start of global mitigation.

Scenario	TIAM-Grantham	MESSAGE-GLOBIOM	WITCH
2C immediate	−2.2%	−0.9%	−6.0%
2C delay to 2020	−5.2%	−1.9%	−8.7%
2C delay to 2030	−10.8% 1	−6.6%	−14.2%
2.5C immediate	+1.0%	+0.4%	−1.5%
2.5C delay to 2020	−0.1%	+0.4%	−3.5%
2.5C delay to 2030	−2.0%	−0.8%	−5.7%
3C immediate	+2.0%	+1.0%	+1.0%
3C delay to 2020	+1.4%	+1.4%	+0.6%
3C delay to 2030	+1.1%	+0.9%	-0.2%
4C immediate	+1.1%	+1.1%	+2.3%
4C delay to 2020	+1.7%	+1.7%	+2.6%
4C delay to 2030	+1.4%	+1.4%	+2.7%

Notes: ¹ TIAM-Grantham relies on a hypothetical “backstop” technology removing CO₂ at a cost of 2005US$ 10,000/tCO₂ in 2100, in order to provide a solution for this scenario.

shows the average annual rate of global CO₂ emissions reductions in the decade following the start of global mitigation action, for each temperature goal. Energy system decarbonisation rates are very rapid in the most delayed 2 °C scenario, in which global coordinated mitigation action towards the 2 °C goal doesn’t begin until 2030. The most drastic decarbonisation decade is that following the start of such mitigation action (2030–2040) which sees global CO₂ emissions fall by an average 7%–14% per annum. Where action is delayed until 2020, the 2020–2030 decade sees average annual CO₂ emissions reductions of 2%–8% per annum.

For the higher temperature goals, rates of decarbonisation are much less rapid. For the 2.5 °C scenarios, two models (TIAM-Grantham and MESSAGE-GLOBIOM) show emissions continuing to rise in the immediate action scenarios and in the case of MESSAGE-GLOBIOM in the delay to 2020 scenario as well. The highest decarbonisation rate is for the WITCH model (−5.7% per year) when action is delayed until 2030. For the 3 °C and 4 °C goals, in almost all modelled scenarios, CO₂ emissions actually continue to grow in the decade following the start of global mitigation action.

As recently as 2010, decarbonisation rates in excess of 3% per annum were deemed to be “extreme”, based on a review of models at that time [13]. More recent analysis includes scenarios with delayed action beginning in 2030, in which average decarbonisation rates over the period 2030–2050 are also very high (5.9%–8.5%) [6]. This results from the models’ ability to rapidly substitute low-carbon for carbon-intensive technologies—a rapidity which can only be slowed by imposing explicit constraints on the models. Hence, the increasingly rapid rates of decarbonisation observed in the most recent assessments are a facet of the requirement to decarbonise at that rate in order to meet a given CO₂, GHG or other emissions or climate target, given that emissions have continued to rise over time. Such rates have been compared to historic decarbonisation rates across countries, noting that countries such as France and Sweden achieved rates of 2%–3% per annum following the early 1970s oil crisis, but that at both a national and global scale, sustained rates as high as recently modelled are “unprecedented” [6]. A detailed analysis of the energy system changes across the century helps shed light on where the greatest challenges lie if such historic decarbonisation rates are to be exceeded.

3.5. How Does the Energy System Change over the Century?

For the 2 °C scenario with mitigation action delayed until 2020, all models depend on a wide range of technologies and measures to meet the 2 °C goal, although to different extents for different technologies. Figure 5 shows that the fossil fuel share of primary energy reduces to 48%–62% by 2050 and to 22%–32% by 2100, compared to a level of more than 80% since 1970 [45]. Although total primary energy supply will increase by 2100, total fossil fuel supply will shrink.

As shown in Figure 6, the models show a broad range of primary energy supply reduction in the mitigation scenarios, with a 2100 value of 1150–1450 EJ/year in the reference reducing to 550–1250 EJ/year in the 2 °C scenario with delayed action to 2020. In the most extreme case, the WITCH model sees primary energy intensity of global GDP reduce from 7.8 MJ/$2005 in 2010 to 1.0 MJ/$2005 GDP by 2100—an average annual reduction of 2.3% per year. By contrast, TIAM-Grantham shows a reduction rate of 1.3% per year, and MESSAGE-GLOBIOM 1.7% per year. However, the annual average rates of reduction in the first decade following the start of global coordinated mitigation action are particularly high, ranging from 2.4% (TIAM-Grantham) to 6.8% (WITCH). These projected rates compare to historical primary energy reduction rates of 1.2% per year since 1970 [46]. Whilst these efficiency improvements are technically possible and reflected in other studies with a focus on maximising energy efficiency potential [46], it is unclear whether such a sector-wide, global improvement in energy efficiency is socially and politically realistic.

In the model with the highest energy intensity of GDP by 2100 (TIAM-Grantham), the 2 °C goal is achievedthrough a very significant shift of the energy system from fossil fuel-based to a mix of low-carbon sources dominated by wind, solar and biomass, as shown in Figure 6.

In each model, the electricity sector sees a fundamental shift from a system dominated by fossil fuel (mostly coal), nuclear and hydro in 2010 to a broad mix of renewables, nuclear and coal and gas with CCS by 2100, as shown in Figure 7. The increase in electricity generation in the TIAM-Grantham model is particularly striking, with a ten-fold increase in electricity generation between 2012 and 2100, reflecting that, in the latter half of the century, electricity increases as a share of final energy from 24% in 2050 (compared to about 18% today [47]) to 66% in 2100, dominated by buildings (88%) and industry (75%).

There is some variation between models in terms of the electricity generation technologies favoured. The period to 2050 sees a rapid penetration of CCS, which is already responsible for almost half of power generation globally by 2030 in the TIAM-Grantham model, and about 30% of generation in WITCH and MESSAGE-GLOBIOM. Nuclear takes a significant share of generation in WITCH and MESSAGE-GLOBIOM by 2100, whilst it is far less rapidly deployed in TIAM-Grantham, particularly compared to solar PV and CSP, as well as onshore wind. Although for all models nuclear is one of the more expensive technologies in capital cost terms (see Figure 8), its relatively large-scale deployment in WITCH and MESSAGE-GLOBIOM reflects the technology’s potential for supplying low-carbon, base-load power. In contrast, solar PV and wind are constrained in the models by the intermittency and variability of the resource.

Table 5. Maximum absolute ramp-up rates of low-carbon technologies in 2 °C scenarios. Notes: Only power generation technologies deployed at a rate greater than 30 GW per year on average (i.e., 300 GW per decade) have been shown; no exogenous constraints have been imposed on technology deployment rates in these scenarios.

Scenario	Technology	Growth Rate
2 °C with delay to 2020	Gas with CCS	800 GW in 2020–2030 (TIAM-Grantham)
	Biomass with CCS	520 GW in 2020–2030 (WITCH)
	Nuclear	830 GW in 2020–2030 (WITCH)
	Onshore wind	480 GW in 2020–2030 (MESSAGE-GLOBIOM)
2 °C with delay to 2030	Gas with CCS	1600 GW in 2030–2040 (TIAM-Grantham)
	Biomass with CCS	1000 GW in 2030–2040 (TIAM-Grantham)
	Nuclear	640 GW in 2030–2040 (WITCH)
	Onshore wind	750 GW in 2030–2040 (MESSAGE-GLOBIOM)
	Solar PV	1300 GW in 2030–2040 (TIAM-Grantham)
	Solar CSP	950 GW in 2030–2040 (TIAM-Grantham)
2 °C with delay to 2020 and CCS delayed until 2050	Gas without CCS	780 GW in 2020–2030 (TIAM-Grantham)
	Biomass without CCS	480 GW in 2020–2030 (TIAM-Grantham)
	Nuclear	1050 GW in 2020–2030 (WITCH)
	Offshore wind	320 GW in 2020–2030 (WITCH)
	Solar PV	380 GW in 2020–2030 (MESSAGE-GLOBIOM)
	Solar CSP	550 GW in 2020–2030 (TIAM-Grantham)
2 °C with delay to 2020 and weak electrification	Gas with CCS	900 GW in 2020–2030 (TIAM-Grantham)
	Biomass with CCS	540 GW in 2020–2030 (WITCH)
	Nuclear	780 GW in 2020–2030 (WITCH)
	Onshore wind	440 GW in 2020–2030 (MESSAGE-GLOBIOM)

shows the deployment rates of key low-carbon technologies in the decade following the start of global mitigation action in the 2 °C scenarios with action starting in 2020 and 2030. The table is limited to show only those technologies requiring a build rate of greater than 30 GW per year on average (i.e., 300GW or more per decade). Rates of 30 GW per year have been achieved in key technologies including solar PV, nuclear and (on and offshore) wind, which is why deployment rates below this level are not deemed particularly challenging.

Table 5 indicates that a major challenge will include achieving hundreds of GW of installed CCS and nuclear capacity, with large-scale deployment starting as early as 2020 in the 2 °C scenario with action starting in 2020. Whilst these technology choices are not prescriptive, but rather indicate what would be deployed in a least-cost scenario without specific deployment constraints, they nevertheless highlight the potential importance of CCS and nuclear in achieving rapid decarbonisation of an energy system deeply reliant on fossil fuel combustion. Table 5 also shows the power generation technologies deployed in a 2 °C scenario with delayed action to 2020, where CCS is not available until 2050 as well as where electrification rates are capped. The former scenario indicates the increased importance of nuclear and the importance of gas and biomass generation (without CCS) as well as solar (PV and CSP). The latter scenario, in which electricity demand is lower than the other scenarios, still sees significant requirements for CCS (with gas and biomass), wind and nuclear power. Hence, as relatively unproven technologies, there is an immense benefit to successfully demonstrating both CCS and biomass (with and without CCS) power generation.

Such rapid deployment rates of specific technologies are common to studies of this kind, with recent model inter-comparisons focused specifically on this issue showing median deployment rates of wind of between 600–1500 GW per decade, solar 1700 GW per decade and nuclear just below 500 GW per decade during the period 2030–2050 in 2 °C-consistent (in this case 450 ppm) scenarios with delayed action to 2030 [10,35]. On the demand side, the energy mix across end-use sectors changes significantly over time, as shown in Figure 9. Although economic growth is harmonised across models, they can obtain different compositions of growth by sector (i.e., by industrial, commercial and agricultural services). This, as well as differing energy efficiency improvement rates, explains why MESSAGE-GLOBIOM and TIAM-Grantham have different energy demand growth rates in the industrial and transport sectors. WITCH does not have a sectoral split for final energy demand although does separate out the light duty vehicles sector, as represented in Figure 9d.

The figure shows that in all three models, total final energy demand shifts to electricity over the century, most markedly in the TIAM-Grantham model, in which electricity increases from 17% of total final energy in 2012 to 66% in 2100. This includes the virtual complete electrification of the buildings sector (about 90% of final energy by 2100, a proportion also reflected in the MESSAGE-GLOBIOM model) and industry sector (about 75% of final energy by 2100). In the transport sector, all models show a significant shift from oil over the course of the century, with TIAM-Grantham favouring hydrogen (fuel cell) vehicles and MESSAGE-GLOBIOM showing a more balanced split between gas, electricity, hydrogen and biofuels, by 2100.

3.6. What Does Rapid Mitigation Imply for Coal-Fired Power Stations?

Even where global mitigation action begins in 2020, there are likely to be significant stranded coal plants as a result of rapid decarbonisation to meet the long term temperature goal of 2 °C, with average capacity factors falling to between 0 and 0.5 by 2030 (compared to 0.65 currently), as shown in Figure 10.

In two models (WITCH and TIAM-Grantham) the capacity factors fall to approximately zero, implying the early scrapping of 1400 GW of coal capacity by 2030. This is equivalent to scrapping 80% of existing economically viable coal capacity. Idling of coal plant has been explored in a previous study using a variant of the MESSAGE model with a broadly 2 °C-consistent goal, finding that an average of 350 GW of coal plant would be stranded on average over the period 2030–2050 if global mitigation action were delayed to 2030 [9]—a similar magnitude to the 450 GW of idled coal plant in MESSAGE-GLOBIOM in this study’s 2 °C scenario with delayed action until 2030.

3.7. How Important is CO₂ Capture in Achieving the Most Stringent Mitigation Scenarios?

To achieve the 2 °C goal, all models show a significant role for CO₂ capture technologies, as illustrated in Figure 11. This peaks by 2080 in two models (TIAM-Grantham and MESSAGE-GLOBIOM) where 30–35 GtCO₂/year (approximately the current CO₂ emissions level) is being captured. In theory there is a sufficiently large global geological storage potential to accommodate this cumulative level of sequestration, which in the TIAM-Grantham model (which has the highest cumulative level of sequestration) reaches 1900 GtCO₂ by 2100, compared to estimates of storage of at least 2000 GtCO₂ globally, with potentially much more [50,51]. This does, however, highlight the importance of CCS, which must be sufficiently developed to be deployed at scale as soon as possible. With delayed CCS, mitigation costs increase very significantly, with half a percentage point of GDP lost over the century (as shown in Figure 4). This compares to an almost doubling of mitigation cost if there is no CCS at all [6].

Figure 12 highlights the degree to which global CO₂ emissions become negative as a result of delays to global coordinated mitigation action. In the scenario with delayed action to 2020, one of the models (TIAM-Grantham) has net negative emissions by 2070, whilst MESSAGE-GLOBIOM has net negative emission by 2080. In the scenario with delayed action to 2030, all three models show net negative emissions by 2080, with TIAM-Grantham have more significant net negative emissions by 2070, and MESSAGE-GLOBIOM net negative emissions by 2070. As shown in Figure 12, across the three models, net negative emissions happen between 5 (TIAM-Grantham) and 25 (WITCH) years earlier with the 10-year delay in mitigation action.

To a large extent this reflects the RCP2.6 scenario originally presented in the literature, with net negative emissions by around 2070, even where mitigation action begins immediately [52]. This conclusion is also reflected in other assessments such as the UNEP Emissions Gap report, whose scenarios have net zero emissions achieved between 2060 and 2080 [23].

A significant driver of net negative emissions is bio-energy with CCS (BECCS) technology, in which net sequestration of atmospheric CO₂ occurs, through the use of biomass to generate electricity or produce biofuels, with capture of CO₂ in these processes. Figure 13 shows the growing importance of BECCS over the century in each model, in the 2 °C scenario with global mitigation action delayed until 2020. The economic and biophysical challenges of deploying large quantities of BECCS and other negative emissions technologies indicate that those scenarios which are highly reliant on BECCS are likely to face greater challenges [37,38].

3.8. A Matrix of Feasibility Indicators to Assess the Challenges of Different Mitigation Scenarios

The results presented and discussed in the previous sub-sections of Section 3 highlight a number of challenges to achieving the mitigation scenarios, in particular those with the most stringent temperature goal (i.e., 2 °C) and with the most delayed action or constrained technologies.

Table 6. Relative degree of challenge presented by mitigation scenarios which achieve a 2 °C median warming in 2100. Notes: Green = least challenging, red = most challenging; colours do not indicate absolute level of challenge, only relative level to each-other. “Overall” column is purely a coloured assessment of relative challenge.

Scenario	Models Solve	CO2 Prices	CO2 Rate of Change	Mitigation Cost	Idling of Coal Plant	Technology Deployment Rates (Max Across Models)	Energy Intensity Improvement	CO2 Captured	Negative Emissions	Overall
Delay to 2020	All models solve	1 model shows >$1000/tCO2 increase in CO2 price per decade (in period 2080–2100)	2020–2030 period sees 2%–9% average annual CO2 reductions	2 models have cost as 1.3%–1.7% of 21st century GDP. 1 model 8.0% of 21st century GDP	2 models have 1400 GW of idle coal plant by 2030	Over 300 GW each of nuclear, gas CCS, biomass CCS, onshore wind in 2020–2030	2.4%–6.8% annual fall in primary energy/unit GDP in 2020–2030	2 models have >30 GtCO2 captured in 2080	2 models see net negative emissions by 2080	-
Delay to 2020, late CCS	All models solve	1 model shows >$1000/tCO2 increase in CO2 price per decade in period 2070–2100. and CO2 price almost $10,000/tCO2 by 2100	2020–2030 period sees rate 2%–7% average annual CO2 reductions	2 models have cost as 1.9%–2.3% of 21st century GDP, 1 model 8.6% of 21st century GDP	2 models have 1400 GW of idle coal plant by 2030	Over 300 GW each of gas, biomass, nuclear, solar (PV, CSP) and offshore wind in 2020–2030	3.0%–8.3% annual fall in primary energy/unit GDP in 2020–2030	1 model has >30 GtCO2 captured by 2060	All models see net negative emissions by 2090	-
Delay to 2020, weak electrific-ation	All models solve	1 model shows >$1000/tCO2 increase in CO2 price per decade in period 2070–2100, and CO2 price almost $9000/tCO2 by 2100	2020–2030 period sees rate 2%–9% average annual CO2 reductions	2 models have cost as 1.6%–2.2% of 21st century GDP, 1 model 8.5% of 21st century GDP	2 models have 780–1400 GW of idle coal plant by 2030	Over 300 GW each of gas CCS, biomass CCS, onshore wind and nuclear in 2020–2030	2.6%–7.1% annual fall in primary energy/unit GDP in 2020–2030	2 models have >30 GtCO2 captured in 2080	2 models see net negative emissions by 2080	-
Delay to 2030	Only two out of three models solve	All models show >$1000/tCO2 increase in CO2 price per decade in period 2090–2100. 2 models show CO2 price >$7000/tCO2 by 2100	2030–2040 period sees rate 7%–14% average annual CO2 reductions	2 models have cost as 2.2% of 21st century, 1 model 9.6% of 21st century GDP	2 models have 800 GW of idle coal plants by 2040	Over 300 GW of gas CCS, biomass CCS, solar (PV, CSP), onshore wind and nuclear in 2030–2040	1.9%–8.9% annual fall in primary energy/unit GDP in 2030–2040	1 model has >30 GtCO2 captured by 2060	All models see net negative emissions by 2080	-

sets out a (subjective) judgement on the degree of challenge associated with achieving each of the 2 °C scenarios explored in this model inter-comparison exercise. The 2 °C scenario with immediate action (in which action started from the models’ base years of 2010 or 2012) is excluded from this analysis, since it has been included purely as a hypothetical scenario, which is in fact no longer attainable. The table suggests that the 2 °C scenario with action delayed to 2030 is the most challenging when considering the full range of criteria. It is a clear indication for the need to commence global mitigation action towards a 2 °C-consistent CO₂ budget as early as possible in the decade 2020–2030. This is all the more pertinent given that, as stated in Section 2, the delayed action until 2030 scenario is (at a global level of effort) broadly commensurate with the INDC pledges already made in the Paris Agreement process. The clear indication is that a ramping up of ambition in the 2020–2030 period is critical to increasing the feasibility of achieving a 2 °C target.

Table 7. Relative degree of challenge of mitigation scenarios with global coordinated action starting in 2020, achieving median warming of 2–4 °C in 2100. Notes: Green = least challenging, red = most challenging; colours do not indicate absolute level of challenge, only relative level to each-other. “Overall” column is purely a coloured assessment of relative challenge.

Scenario	Models Solve	CO2 Prices	CO2 Rate of Change (2020–2030)	Mitigation Cost	Idling of Coal Plant	Technology Deployment Rates (Max Across Models)	Energy Intensity Improvement	CO2 Captured	Negative Emissions	Overall
2 °C	All models solve	1 model shows >$1000/tCO2 increase in CO2 price per decade (in period 2080–2100)	2%–9% average annual CO2 reductions	2 models have cost as 1.3%–1.7% of 21st century GDP. 1 model 8.0% of 21st century GDP	2 models have 1400 GW of idle coal plant by 2030	Over 300 GW each of nuclear, gas CCS, biomass CCS, onshore wind in 2020–2030 period	2.4%–6.8% annual fall in primary energy/unit GDP in 2020–2030 period	2 models have >30 GtCO2/year captured in 2080	2 models see net negative emissions by 2080	-
2.5 °C	All models solve	CO2 price range $504–1573/t CO2 by 2100 with maximum decadal increase $607/t CO2 (TIAM-Grantham) in 2090–2100	Between a 0.4% increase and 3.5% reduction in average annual CO2	2 models have cost as 0.5% of 21st century GDP, 1 model 4.0% of 21st century GDP	1 model has reduced capacity factor equivalent to 370 GW of idle coal plant by 2030. Other models have no idling by 2030	Over 300 GW each of gas with CCS, gas (w/out CCS), biomass and onshore wind in 2020–2030 period	2.0%–4.5% annual fall in primary energy/unit GDP in 2020–2030 period	2 models have >30 GtCO2/year captured by 2090	1 model has net negative emissions by 2100	-
3 °C	All models solve	CO2 price range $126–382/tCO2 by 2100	0.6%–1.4% average annual CO2 increase	2 models have cost as 0.1%–0.2% of 21st century GDP, 1 model 1.7% of 21st century GDP	All models see no drop in coal capacity factor, so no idling, by 2030	Over 300 GW each of gas, biomass and onshore wind in 2020–2030 period	1.5%–2.6% annual fall in primary energy/unit GDP in 2020–2030 period	2 models have >30 GtCO2/year captured in 2100	No net negative emissions (lowest 2100 emissions level is 7 GtCO2)	-
4 °C	All models solve	CO2 price range $16–104/tCO2 by 2100	1.7%–2.6% average annual CO2 increase	2 models have cost as 0.02%–0.03% of 21st century, 1 model 0.5% of 21st century GDP	All models see no drop in coal capacity factor, so no idling, by 2030	Over 300 GW of gas and biomass in 2020–2030 period	1.2%–1.9% annual fall in primary energy/unit GDP in 2020–2030 period	By 2100, range of capture across models is 8–16 GtCO2/year	No net negative emissions (lowest 2100 emissions level is 45 GtCO2)	-

shows the same multi-dimension comparison for the different temperature goals explored in this study, in each case for a scenario in which global coordinated mitigation action begins in 2020. This highlights that the degree of relative challenge of the 2 °C scenario across almost all dimensions of feasibility as measured in this study contrasts starkly with the higher LTTG scenarios. Even the 2.5 °C temperature goal has several challenging aspects, including non-trivial carbon prices and mitigation costs, potentially rapid near-term technology deployment rates across a range of low-carbon technologies, as well as potential idling of coal plants in the near-term and negative emissions in the long-term. By contrast, a global CO₂ pathway which limits median warming in 2100 to 3 °C or above looks eminently achievable (which is encouraging given that scenarios with low or no mitigation action could lead to median 2100 temperature changes in excess of 4 °C [1]).

What about the Paris Agreement’s longer-term aims to achieve a “well below 2 °C” limit to global warming [52]? A sensitivity analysis using just the TIAM-Grantham model serves to highlight the additional difficulty of achieving long-term temperature change of less than 2 °C.

Table 8. Relative degree of challenge of mitigation scenarios with global coordinated action starting in 2020, achieving median warming of ≤2 °C in 2100. Notes: Green = least challenging, red = most challenging; colours do not indicate absolute level of challenge, only relative level to each-other. “Overall” column is purely a coloured assessment of relative challenge.

Scenario	Model Solves	CO2 Prices	CO2 Rate of Change (2020–2030)	Mitigation Cost	Idling Of Coal Plant	Technology Deployment Rates (2020–2030)	Energy Intensity Improvement	CO2 Captured	Negative Emissions	Overall
1100 GtCO2 (1.85 °C)	Yes	>$1000/tCO2 increase in CO2 price per decade (in period 2070-2100)	7.2% average annual CO2 reductions	2.5% of 21st century GDP	1400 GW of idle coal plant by 2030	>1300 GW gas CCS	2.5% annual fall in primary energy/unit GDP in 2020–2030 period	>30 GtCO2/year captured in 2060	Net negative emissions by 2070	-
						>500 GW biomass CCS
						300 GW onshore wind
1340 GtCO2 (2 °C)	Yes	>$1000/tCO2 increase in CO2 price per decade (in period 2080-2100)	5.2% average annual CO2 reductions	1.7% of 21st century GDP	1400 GW of idle coal plant by 2030	>800 GW Gas CCS	2.4% annual fall in primary energy/unit GDP in 2020–2030 period	>30 GtCO2/year captured in 2070	Net negative emissions by 2080	-
						>400 GW biomass CCS
						300 GW onshore wind

shows a direct comparison between the 2 °C scenario with global mitigation action beginning in 2020, and a further scenario in which a lower temperature change goal is achieved from the same 2020 starting point, in line with a cumulative fossil fuel combustion and industrial process CO₂ emissions level of 1100 GtCO₂ over the 21st century, compared to 1340 GtCO₂ for the 2 °C scenario. This results in a median temperature change in 2100 of 1.85 °C, according to the analytical framework set out in Appendix D. It is therefore arguably not well below 2 °C and certainly some way off 1.5 °C, but it represents the lowest feasible scenario that can be attained in this set-up of the TIAM-Grantham model (i.e., with the socio-economic drivers and technology availability in the model that is used in the rest of this study).

As shown in Table 8, the implications of this more stringent mitigation scenario are even more challenging than those for the 2 °C scenario: earlier onset of rapid CO₂ price increases; significantly higher mitigation costs; significantly higher initial rates of decarbonization, and marginally higher initial rates of energy intensity reduction; much higher initial deployment rates of low-carbon energy technologies; and earlier onset of significant carbon capture, with global net negative emissions a decade earlier than for the 2 °C scenario. As such, achieving even marginally more mitigation compared to the 2 °C scenario requires significant changes to the energy system in the TIAM-Grantham model. More recent analysis of the well below 2 °C reinforces that this challenge is likely to be felt across many dimensions including costs, stringency of near-term mitigation and reliance on negative emissions technologies [53,54,55]. In addition, these challenges are only quantified here and in these other studies across economic and technical dimensions, rather than across political and social dimensions. A key area of further research will be to better characterize these political and social dimensions so that we can judge more conclusively how likely it is that we can transform the energy system as quickly and fundamentally as necessary to achieve the Paris Agreement’s aims.

4. Discussion

Consideration of the different metrics associated with the modelled mitigation scenarios, as combined into the matrix presented in Section 3.8, highlights the following critical points affecting the feasibility of meeting the 2 °C goal:

• Ensuring that mitigation action at a global level in line with the target begins as soon as possible, given the significant costs of delays, particularly to 2030, which implies the need for a ramping up of ambition over and above the currently submitted INDCs;
• Achieving sustained energy efficiency improvements over the course of the century and very rapid near-term improvements, which though technically feasible, would be unlikely to occur without very effective policies;
• Ensuring commercial-scale deployment of CCS is feasible as soon as technically and economically possible, such that hundreds of GW of CCS power stations can be deployed in the coming decades;
• Developing supply chains for other low-carbon technologies such as wind, biomass, solar and nuclear to ensure that hundreds of GW globally can be deployed each decade in the near future;
• Demonstrating the different aspects of BECCS technology and/or other negative emissions technologies so that global CO₂ emissions can become first neutral and then net-negative in the latter half of the century;
• Increasing the penetration of electricity-using heating, transport and industrial process technologies throughout the end-use sectors;
• Managing the political economy issues that would be associated with the early idling of coal-fired power stations without CCS fitted.

A comparison of scenarios aimed at achieving a range of LTTGs (between a 2 °C and 4 °C median warming in 2100) also highlights that achieving the 2 °C goal (even if global coordinated mitigation action were to begin in 2020) is highly challenging compared to less stringent temperature goals. The analysis suggests that even a 2.5 °C temperature change may be relatively challenging, in terms of mitigation costs, required rates of deployment of key low-carbon technologies, and in some cases possible idling of coal plants in the near-term, plus negative emissions in the long-term.

Finally, a sensitivity analysis of the TIAM-Grantham model to going below 2 °C (specifically to 1.85 °C) suggests that almost all dimensions of feasibility explored here look significantly more challenging than even the 2 °C goal. This is of direct relevance to the current United Nations Framework Convention on Climate Change (UNFCCC) process which is seeking to raise ambition compared to Parties’ current Nationally Determined Contributions, in line with achieving a long-term temperature change of well below 2 °C and towards 1.5 °C [24].

In conclusion, the challenges associated with achieving a below 2 °C limit to temperature change are made clearer by highlighting the many relevant outputs of modelled scenarios against each-other, and suggest that the Paris Agreement targets will be extremely challenging. Application of this approach when modelling future low-carbon pathways aimed at achieving the most ambitious temperature limit of the Paris Agreement, at 1.5 °C, is recommended. Scenarios which outline what might be required to meet such stringent mitigation goals are of limited value without a clear and systematic assessment of the feasibility and degree of challenge involved in meeting those goals. This study makes a first attempt to systematize this feasibility assessment, whilst accepting that further research which more explicitly includes political and social dimensions should also be pursued to arrive at a more complete picture of how realistic our long-term climate change goals actually are.