Mitigation choices impact carbon budget size compatible with low temperature goals

Assessing the policy and technology dependence of cumulative carbon emissions ideally makes use of emission scenarios with an internally consistent evolution of all radiatively active species over time. We use the IAM framework MESSAGE (Model for Energy Supply Strategy Alternatives and their General Environmental impact) (Riahi et al 2007) developed at the International Institute for Applied Systems Analysis (IIASA). At the core of MESSAGE lies a systems-engineering optimization model of the global energy economy, which includes a large portfolio of energy technologies for the assessment of technological transitions over the 21st century (see for example: Riahi et al 2012, McCollum et al 2013). The modelling framework contains a detailed representation of the most important GHG-emitting sectors, a representation of land-use emissions, a link to the macro-economic model MACRO (Messner and Schrattenholzer 2000), and a link to IIASA's GAINS (GHG and Air Pollution Interactions and Synergies) air-quality model (Amann et al 2009), which allows the consistent tracking of (non-GHG) air pollutants over time.

In brief, putting a limit on the cumulative amount of GHG emissions over the 21st century results in a transformation of the energy, industry and land-use sector, modelled by MESSAGE. The carbon price implied by this transformation is used by MACRO to iteratively determine possible price-driven reductions of energy demand. Finally, the transformation of the energy, industry and land-use system also impact possible sources of air pollutants. The presence and activity of these sources (like coal-fired power plants or diesel engines) is translated into actual emissions by the MESSAGE-GAINS linkage, which assumes that current legislation is successfully implemented throughout the 21st century (Rogelj et al 2014b). A detailed description of the model setup used in this study is provided in Rogelj et al (2013b).

We use a set of more than 500 scenarios that were originally developed in a study focusing on the relative importance of uncertainties (technological, societal, political and geophysical) for the cost of limiting warming below particular temperature levels with a given probability (Rogelj et al 2013b). The scenarios differ from each other in their underlying assumptions: (i) the stringency of mitigation action (reflected in the equivalent carbon price applied to GHG emissions over time), (ii) variations in the availability of key mitigation technologies, and (iii) future energy demand.

GHG emissions follow cost-minimizing trajectories that stay within varying global cumulative emission limits, with emissions of non-CO₂ gases weighted using 100-year Global Warming Potentials (GWPs: IPCC, 2007). Delayed mitigation cases are not considered here, in other words all scenarios started optimizing concerted, global efforts to mitigate climate change from 2010 onward. Delaying action can significantly constrain the achievability of low temperature stabilization targets and would imply higher overall costs (Luderer et al 2013, Rogelj et al 2013b, Clarke et al 2014). However, carbon budgets to 2100 would remain similar even in delayed-action scenarios (see, Collins et al 2013).

Our scenario set contains five technology variations which span a wide range of mitigation options and which are categorized in two groups: three technology-restricting cases and two cases that assume breakthroughs in mitigation technologies (Rogelj et al 2013a, Rogelj et al 2013b). Mitigation technologies are defined here as technologies or practices that significantly reduce GHG emissions by replacing technologies or practices that would generate more GHG emissions otherwise. The technology restricting cases are (1) no new nuclear (assuming a phase-out over the 21st century), (2) limited land-based mitigation measures (limiting the global bio-energy potential and not considering forest management as a mitigation option), and (3) no Carbon Capture and Storage (no CCS—assuming that CCS will never become available at large scale during the 21st century). The technological breakthrough cases are (1) advanced transport (allowing more than 50% of the transport sector's final energy consumption being electrified by the end of the century), and (2) advanced non-CO₂ mitigation (assuming that mitigation technologies for non-CO₂ GHGs, like CH₄, improve significantly throughout the 21st century beyond current levels). The latter two cases are considered breakthroughs because, on the one hand, the transport sector is particularly difficult to decarbonize given its high reliance on liquid fossil fuels and because, on the other hand, currently only limited mitigation potentials for some parts of the agricultural sector have been identified (Smith et al 2014).

Other drivers that can influence future GHG emission pathways are population (O'Neill et al 2012) or economic projections (of world gross domestic product (GDP)). These are not explicitly varied in our scenarios but are taken directly from the Global Energy Assessment (Riahi et al 2012). Such socio-economic and demographic developments can lead to different levels of future energy demand, depending, however, on how efficiently energy is transformed into services going forward. In order to cover a broad scenario space in terms of energy-demand, we assume a range of projections that roughly spans the entire range of outcomes that would otherwise be obtained when widely varying population and GDP projections in consistent ways (Nakicenovic and Swart 2000, Rogelj et al 2013b).

Here we use the reduced-complexity carbon-cycle and climate model MAGICC (Meinshausen et al 2011a, Meinshausen et al 2011b) in a probabilistic setup (Meinshausen et al 2009) to translate emissions to atmospheric concentrations and then to temperature change. The model setup is updated such that the marginal climate sensitivity distribution is consistent with the findings of the Fourth Assessment Report (AR4) of the IPCC (Rogelj et al 2012; IPCC, 2007). This model setup also remains consistent with the more recent AR5 assessment (Rogelj et al 2014a). For each of the 500+ socio-economic scenarios created with the MESSAGE model, MAGICC is used to compute the temperature increase relative to pre-industrial (1850–1875) for a set of 600 geophysical ensemble members (Meinshausen et al 2011b). The 'implied probabilities' (Smith and Stern, 2011) used in this study are defined by the proportion of geophysical ensemble members exceeding a particular temperature limit at a given point in time.

Cumulative CO₂ emissions from 1870 to 2011 are estimated at about 1810 billion metric tonnes of carbon-dioxide (GtCO₂) in our simulations. This is just slightly lower than the median (1890 GtCO₂; ∼4% lower) but well within the likely range (1630–2150 GtCO₂) of historical cumulative carbon emissions reported by the IPCC AR5 (IPCC 2014b). Uncertainties in cumulative CO₂ emissions over this time period are about ±12% (90% range derived from Le Quéré et al 2013).

We find that to limit warming to below 2 °C with 66% chance (or alternatively, 75 or 50%) cumulative CO₂ emissions from 2011 to 2100 need to be limited to 670–1050 GtCO₂ (rounded to the nearest 10 GtCO₂, range of upper and lower bound estimates over our entire scenario set; see table 1). For two other probability levels (75 and 50%), the cumulative amounts are 610–830 and 990–1450, respectively. The range of cumulative emissions is markedly smaller for the 75% probability level because such climate protection levels can only be reached under a subset of our full set of scenario variations. The spread of cumulative CO₂ emissions in each of the ranges reported above (for example, the 670–1050 GtCO₂ range for a 66% chance) stems from non-CO₂ forcing variations, which are driven by differences in technology and energy-demand assumptions in the range of scenarios. Geophysical uncertainties are reflected by the fact that any given carbon budget can limit warming only at a given probability level (with levels of 50, 66 and 75% reported in this study).

This means that if the world implements a 1000 GtCO₂ carbon budget over the 2011–2100 period, the probability of global temperature increase in fact remaining below 2 °C could be about 50% or more than 75%, depending on the technology choices and the pathway that the world follows in meeting this budget (figure 1). In the next section we further explore what drives these differences in carbon budgets and the probability with which they constrain warming below given limits.

Zoom In Zoom Out Reset image size

All our technology limiting cases result in a smaller potential for CO₂ emission reductions by 2100. At the same time, our technological breakthrough cases increase the CO₂ mitigation potential (advanced transport) or leave it unchanged (advanced non-CO₂ mitigation).

However, changes in the CO₂ and non-CO₂ mitigation potential do not translate into a proportional actual change in corresponding CO₂ and non-CO₂ emissions. We find a significant and complex interaction between non-CO₂ abatement and CO₂ emissions budgets, even though many non-CO₂ GHGs are emitted by sources which are relatively independent of sources of CO₂. This is because in MESSAGE, as in most IAM frameworks that assume a multi-gas approach to climate change mitigation, both reductions in CO₂ and in other Kyoto-GHGs are driven by the same carbon price, which is translated between gases via 100-year GWPs.

When mitigation of CO₂ is made relatively harder (by limiting CO₂ mitigation technologies), carbon prices will be higher to achieve a similar level of CO₂ emission reductions. These higher prices will result in relatively larger concurrent emissions reductions of CH₄ and other non-CO₂ gases; the lower radiative forcing from those gases hence allows slightly greater cumulative emissions of CO₂. Conversely, larger CH₄ reductions can also be the result of a higher non-CO₂ mitigation potential (for example, this is the case in our advanced non-CO₂ mitigation case for agriculture). Peak warming is largely determined by the cumulative emissions of CO₂, but is also influenced by the annual emissions of shorter-lived GHGs like CH₄ leading up to the peak (Smith et al 2012). Consequently, stronger abatement of non-CO₂ gases will also result in a slightly smaller amount of CO₂ reductions needed to meet the same peak temperature goal and hence allow greater cumulative CO₂ emissions, resulting in slightly lower carbon prices.

This indicates that cumulative CO₂ emissions consistent with a specific temperature limit are contingent on the future abatement potential (and actual abatement) of non-CO₂ gases. In our set of scenarios, the non-CO₂ mitigation potential is varied by the assumed availability and future improvements of technologies to reduce emissions from agriculture and some energy technologies.

Our scenario results indicate that the effect of limiting key mitigation technologies on cumulative CO₂ emissions compatible with limiting warming to below 2 °C is about ±10%, although this estimate varies in sign and magnitude over time (figure 1).

If the availability of CO₂ mitigation technologies is restricted in the long term, then 2 °C-consistent cumulative CO₂ emission budgets over the 2011–2100 period are generally up to 10% higher than when assuming a full mitigation portfolio, because the required higher carbon prices will stimulate more mitigation of non-CO₂ species. Another consequence of the higher carbon prices is that more CO₂ mitigation is carried out in the near-term and hence cost-effective cumulative carbon emissions until 2050 are up to 10% lower.

Assuming that the full portfolio of mitigation technologies is available, the cumulative 2011–2100 CO₂ budget in our scenarios in line with 2°C is 940 GtCO₂ (66% chance) and 1280 GtCO₂ (50% chance), respectively (table 1, figure 1). Limiting the construction of new nuclear plants has virtually no effect on the 2011–2100 CO₂ budgets, which are changed by less than 1%. However, limiting land-based mitigation options increases carbon prices to achieve this budget by about 30% and increases the compatible CO₂ budget by about 5% (see also section 3.3). In the absence of CCS, our model is unable to limit warming to below 2 °C with at least a 66% chance under our default (intermediate) levels of future energy demand. Only if demand is lower are these higher-probability options still possible without CCS (see, Rogelj et al 2013b, and further below). For budgets that give a 50% chance of staying below 2 °C, not relying on the future availability of CCS technologies increases carbon prices to achieve these budgets by about 90% and in turn increases 2011–2100 CO₂ budgets by about 12%.

In the nearer term (2011–2050), 2 °C-consistent CO₂ budgets are slightly lower instead of higher when assuming technology limitations (table 1) for the reasons given above. Limiting nuclear availability reduces the 2011–2050 CO₂ budget by about 3%, limiting land-based mitigation options does so by 5–6%, and assuming no large-scale CCS does so by about 9%. This is because higher carbon prices associated with limited technology scenarios mean that more costly mitigation options are taken up earlier (before 2050) rather than later.

Qualitatively similar effects are found under our technological breakthrough cases. However, depending on the sector affected, breakthroughs can result in an increase or a decrease of the carbon budget for a given temperature limit. Assuming breakthroughs in the transportation sector yields 2011–2050 and 2011–2100 CO₂ budgets that are about 7% higher and 10% lower, respectively, for having a 66% chance of staying below 2 °C. This is because lower carbon prices stimulate less near-term CO₂ abatement but allow more CO₂ mitigation in the long term in this sensitivity case. Assuming breakthroughs in the mitigation of non-CO₂ GHGs implies CO₂ budgets which are consistently higher by 12%, both in the 2011–2050 and the 2011–2100 period, because the contribution of non-CO₂ gases to peak warming is reduced, and hence slightly more CO₂-induced warming is permissible.

In theory, similar shifts in emission budgets are expected for a 1.5 °C limit (table 1). However, the number of scenarios reaching such a stringent level of climate protection in our set is too limited to extract robust quantitative results on these shifts.

The level of energy-demand growth over the 21st century, is another factor that is varied in our scenario ensemble. We consider three levels of future energy demand in our scenarios: low, intermediate, and high, consistent with the 'Efficiency', 'Mix', and 'Supply' scenario families, respectively, of the framework of the Global Energy Assessment (Riahi et al 2012). Also here, relationships that are qualitatively similar to the ones discussed above are found.

In scenarios assuming a high future energy demand, the single most important influencing factor in our scenarios is that CH₄ emissions from the waste sector are about 50% higher in the second half of the century. Therefore, CO₂ emission budgets consistent with a given temperature limit become smaller. If CO₂ mitigation is assumed to be easier because of a lower future energy demand, emissions can be reduced at relatively lower cost (i.e., lower carbon price, consumption loss, etc). However, the lower associated carbon price in the energy sector also results—for similar reductions in CO₂—in less concurrent CH₄ mitigation in low energy-demand scenarios, at least in cases where mitigation of CO₂ and CH₄ is linked. On aggregate, this results in a greater relative need to reduce carbon emissions in order to achieve the same climate outcome with a given budget.

Due to the carbon-price linkage described above, 2 °C-consistent budgets of CO₂ emissions from 2011 until 2100 are smaller by about 17% under high energy-demand assumptions relative to our intermediate future energy-demand case and for a 50% chance. They remain virtually the same, however, (<1% smaller) under low energy-demand assumptions (figure 1). For a 66% chance, a low energy-demand future would imply a CO₂ budget that is about 6% smaller.

Having explored how the efficiency of 2 °C-compatible CO₂ budgets can change with technology variations, which themselves are influenced by mitigation choices, we now focus on the question of how mitigation choices influence the costs of keeping emissions within a given budget (figure S3).

The carbon price associated with keeping temperature rise to below 2 °C with 66% chance assuming a full mitigation technology portfolio and intermediate future energy demand is about 37 USD in 2020. (Note that we here report year-2020 carbon prices discounted back with a discount rate of 5% to 2011, the first year the model assumes the carbon price to be effective, and report these in year-2005 USD.) The corresponding discounted total mitigation costs are about 23.5 trillion USD (table 2). Discounted total mitigation costs are computed over the entire century and relative to a scenario in absence of any mitigation.

Costs vary strongly with technology availability and future energy demand (table 2). Carbon prices to limit emissions to within the same 2 °C-consistent CO₂ budget over the entire 21st century (940–1280 GtCO₂, table 1, figure S3) decrease by about 30–40% or increase more than an order of magnitude, when varying the availability of mitigation technologies. Technology limitation cases consistently increase mitigation costs while technology breakthroughs decrease costs. When varying future energy-demand projections, consistent carbon prices can decrease by about 70% (low energy demand) or up by 450% (high energy demand).

Total mitigation costs see qualitatively similar shifts (table 2). Underlying drivers that can be actively influenced by policies (for example, energy efficiency improvements) therefore appear robust means to limit mitigation costs in both the short and long term.

In a final step of our assessment of the impact of policy choices on the efficiency of cumulative CO₂ emissions for limiting warming below a given temperature level in a multi-gas context, we integrate all findings and explore robust features in the relationship between costs, technology dependencies and emission budgets. We find important difference between the medium term (until 2050) and the long term (until 2100).

We find that for limiting warming to below 2 °C with a given probability, lower carbon prices or mitigation costs lead to higher compatible carbon emissions budget until 2050 (figure 2). This relationship holds over all our mitigation technology variations and for each energy-demand sensitivity case. This is because high availability of technologies or low energy-demand scenarios are generally associated with lower carbon prices, which means that relatively less costly mitigation is carried out in the near term resulting in higher cumulative CO₂ emissions until 2050. This also implies that emissions targets can be achieved at lower overall cost.

Zoom In Zoom Out Reset image size

However, looking further until the end of the century this relationship changes. Within each energy-demand category, scenarios that assume limited availability of technology have, as noted above, larger consistent cumulative CO₂ emission budgets until the end of the century, as much higher carbon prices drive more abatement of non-CO₂ gases. However, these higher prices nonetheless result in higher overall mitigation costs despite the slightly greater carbon budgets. Vice versa, scenarios with breakthroughs in the electrification of the transport sector give smaller CO₂ budgets because they are able to remove more CO₂ from the atmosphere by 2100, but they are able to achieve these smaller budgets at lower costs. A different relationship arises where the technology breakthrough is for advanced mitigation of non-CO₂ gases. This case results in both higher cumulative CO₂ emission budgets over the 21st century and lower overall costs, as more low-cost abatement is provided from non-CO₂ sectors. This further confirms the important influence of long-term non-CO₂ abatement potentials for allowable cumulative CO₂ emissions and overall abatement costs.