For the purpose of this assessment we collected a total of 223 emission pathways. Of these 126 were IAM emission pathways published by 15 modelling groups26, of which 113 explored low greenhouse gas concentration targets while taking into account some assumptions about technological and socio-economic inertia, whereas the remaining 13 represent scenarios without strong mitigation policy. These IAM pathways had varied rates of emission reductions across regions, sectors and gases in order to minimise costs. Of the 223 pathways, 97 were “stylized” pathways27 which did not make assumptions about technological and economic feasibility, but identified the emission pathways that corresponded to particular temperature targets based on carbon cycle and climate models.
We have evaluated the probability of each of the pathways meeting a 2° C and 1.5° C limit. In order to make results more comparable, we have adjusted the pathways so that they have the same emission levels in 2000 and 2005. Emissions for these years were taken from the multi-gas emissions inventory developed as part of the “Representative Concentration Pathways” (RCPs) scenario exercise (Granier et al. submitted, Meinshausen et al. submitted). When a particular pathway lacked the emissions of a particular substance (e.g. sulphate aerosols, organic carbon, black carbon or atmospheric ozone precursors), these data were taken from the RCP3-PD scenario (van Vuuren et al. submitted). It should be noted that the RCP-3PD scenario assumes strong environmental policies and this is consistent with the aim of this report to identify mitigation pathways that stay within a 2° C or 1.5° C limit. Ozone depleting substances controlled by the Montreal Protocol are assumed to follow a gradual phase-out during the twenty-first century.
The temperature calculations of the harmonised emission pathways were made more comparable by using a single model MAGICC 6.3 (Meinshausen et al. 2009, Meinshausen et al. 2008) to calculate the probabilistic temperature outcome up to 2100 for each emission pathway.
A joint probability distribution of the most important climate response uncertainties has been used, with climate sensitivity uncertainties closely reflecting the estimate provided by the IPCC (IPCC 2007c)28.This distribution gives the probability of a particular response of temperature to emissions. Because a probability distribution rather than a single number is used for the climate sensitivity factor, temperature outcomes are expressed in terms of probabilities, for example, “emission pathways with a medium chance of staying below a 2° C limit”. The emission pathways were put into different categories according to temperature limits (1.5° and 2° C), their probability of meeting the limit (50-66 per cent, greater than 66 per cent), the assumed technologies (e.g. negative emissions or not), and whether they are “stylized” or IAM pathways.
We also performed a sensitivity analysis by analysing 11 recalibrated versions of the climate model to explore alternative values of the climate sensitivity distribution that have been published (see Meinshausenet al. 2009). For emission pathways that give around a “medium” chance of meeting a 2° C limit during the twenty-first century, the sensitivity studies lead to a spread in the median projected temperature of only ±0.2° C.
It is important to note that although we have harmonised the pathways for comparability, some uncertainties remain, for example, about future levels of anthropogenic aerosols, soot and organic carbon.
The climate model used in this study has previously been validated and shown to credibly reproduce observed climate changes when driven by historic emissions or forcings. However, like other climate models it does not include all of the physical processes that could affect the real climate in future. For instance, there is no treatment of extra carbon release from melting permafrosts.
Our quantitative assessment of IAM pathways found a notable number and range of emissions that are consistent with the temperature limits of interest in this report, even after re-analysis. In the text we focus on the median and range of the “majority of results”, with the range corresponding to the 20th to 80th percentile of outcomes. Results at either end of this range are not necessarily invalid or incorrect, and are also discussed in the text.