by Jacques Hagoort
Why the IPCC carbon budgets in SR1.5 are over conservative, and the CO2 reduction pathways are too stringent.
Abstract Carbon Budgets specify the total amount of CO2 that can be emitted before global warming exceeds a certain threshold. Since the introduction of the Paris Climate Agreement in 2015, prominently featuring the 1,5 and 2˚C global warming limits, Carbon Budgets have become the cornerstone of global warming mitigation policy by CO2 reduction. In the recent IPCC special report on global warming of 1,5˚C from 2018 (SR15), the Carbon Budgets have been substantially upgraded compared with the ones reported in the preceding IPCC Fifth Assessment Report from 2013 (AR5). We have analyzed the new method for estimating Carbon Budgets in SR15 and found it seriously wanting, leading to non-physical future global warming profiles. The net result is Carbon Budget estimates that are over-conservative leading to timeframes for the reduction of CO2 emission to net-zero that are too stringent. A simple alternative calculation method without the shortcomings of the SR15 method yields substantially larger Carbon Budgets and thus more lenient timeframes for net-zero emission. Assuming a linear emission reduction pathway, we estimate that net-zero emission for a global warming limit of 1,5˚C is reached in 2070 instead of 2043 as per the SR15 budget. In the case of a warming limit of 2˚C, net-zero happens in 2125 rather than in 2079.
Since the Fifth Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC) came out in 2013, Carbon Budget has emerged as a widely accepted and powerful concept in both Climate Science and Climate Policy (IPCC, 2013a). It stems from the observation that in climate model projections, global warming since pre-industrial times shows an approximate straight line when plotted against cumulative CO2 emission, irrespective of the emission scenario. Hence, to limit global warming to a certain threshold there is a limit to the allowable cumulative CO2 emission. This defines the Carbon Budget. Given the straight line, Carbon Budgets can be readily calculated for an agreed global warming limit, such as 1,5 or 2˚C as stipulated in the Paris Climate Agreement.
The metric to characterize global warming straight lines is the Transient Climate Response to Cumulative Emission (TCRE). It is defined as the warming in ˚C per emission of 1000GtC (1 GtC = 109 tonne of Carbon) and is essentially a measure for the slope of the straight line. The TCREs in AR5 are based on global warming projections of Earth Systems Models (ESM), a subset of some 20 climate models that are capable of simulating the global carbon cycle. They show a rather wide spread: from about 1,25 to 3,25˚C/(1000GtC) for the 5-95% confidence interval with a central value of 2,44˚C/(1000GtC). See Figure SPM.10 in the Summary for Policymakers of AR5 WG1 (IPCC, 2013b). These TCREs represent total warming (CO2 warming and the rest) and for this reason are sometimes referred to as effective TCRE as distinct from the TCRE for warming due to CO2 only. Naturally, the CO2-only TRCEs are smaller than the effective TCREs. According to the expert judgement of the AR5 author team the CO2-only TCREs are distributed normally around a central value of 1,65˚C/(1000GtC) with a standard deviation (SD) of 0,85˚C/(1000GtC).
The best-estimate Carbon Budgets per 01-01-2011 presented in AR5 were based on the central effective TCRE of 2,44˚C/(1000GtC). See Table 2.2 of the AR5 Synthesis Report (IPCC, 2014). Following the publication of AR5 in 2013, a steady stream of climate studies appeared that challenged the official AR5 Carbon Budgets because they were based exclusively on climate model projections and ignored observational climate data (Hausfather, 2018a). As the ESMs used for the Carbon Budget calculations in AR5 tend to overestimate global warming and underestimate cumulative emission, the AR5 Carbon Budgets were deemed too conservative (Hausfather, 2018b). The recent IPCC Special Report on Global Warming of 1,5˚C (SR15) assessed the various Carbon Budget studies and came up, not unexpectedly, with a rather drastic upward revision of the AR5 Carbon Budgets (IPCC, 2018a). For example, the Carbon Budget for a warming limit of 1,5˚C increased by a factor five. Exactly how these revised budgets were arrived at, however, was not entirely clear from the report (Lewis, 2018). Fortunately, a recent publication by the responsible author team of SR15 has shed more light on the underlying calculation procedure (Rogelj et al., 2019). It even provides a theoretical framework for assessing Carbon Budgets so that future changes can be more easily traced and assessed.
To better understand the new calculation method used in SR15, we have tried to reproduce the Carbon Budgets reported in SR15. While successful, in the process we have come upon a number of serious shortcomings, which casts doubt on the reliability of the reported Carbon Budgets. This article sets out our reservations about the new SR15 method and shows what to do about it. The outline is as follows. First, we present a brief summary of the SR15 method. Next, we show how the SR15 method has been applied and has led to the reported Carbon Budgets. We then highlight the shortcomings and present an alternative method to overcome these. Finally, we discuss the implications of the alternative Carbon Budgets for the mitigation of global warming by CO2 reduction.
The SR15 method
Figure 1 schematically illustrates the SR15 method for estimating Carbon Budgets as described by Rogelj et al. It shows an (x, y) diagram with the global warming since pre-industrial times on the y-axis and the cumulative CO2 emission since pre-industrial times on the x-axis. The crux of the SR15 method is the observed total global warming (i.e. CO2 warming and the rest) at a certain cumulative CO2 emission, indicated by the blue bullet point. It serves as a calibration point for the linear warming relationships. The light-blue line begins in the origin (zero global warming at zero cumulative CO2 emission) and ends in the calibration point. It describes past total warming but plays no role in the SR15 method proper. The uninterrupted dark-blue straight line begins in the calibration point and describes the best-estimate warming caused by CO2 only. The slope of this CO2-only line is of course less steep than the slope of the light-blue total warming line. Uncertainty in future warming is accounted for by maximum and minimum CO2-only warming straight lines that also begin in the calibration point. The Carbon Budget for a certain total warming limit follows from the intersection of the CO2only straight lines with the horizontal red line at the level of the allowable total warming limit minus the non-CO2 warming. The dark-grey horizontal bar shows the Carbon Budget for the intersection with the best-estimate warming straight line and thus represents the best-estimate Carbon Budget relative to the calibration point. Likewise, the intersections of the horizontal red line with the maximum and minimum warming straight lines result in minimum and maximum values of the Carbon Budget.
Figure 1 – Schematic of SR15 method
The calibration point used in SR15 is the observed average total warming in the period 2006-2015 relative to the average warming in the period 1850 – 1900 at the average cumulative emission during 2005 – 2016 relative to the average cumulative emission during 1850 – 1900. The average warming in 1850 – 1900 is considered representative of the beginning of the industrial era. The warming depends on how global warming is defined. In climate models, global warming is customarily expressed as the average of near-surface air temperatures (SAT) everywhere. The warming reported in observational global temperature series, however, is a blend of near-surface air temperatures over land and sea surface water temperatures (SAT/SST). The observational (SAT/SST) global warming in the period 2006 – 2015 is 0,87˚C. The corresponding SAT global temperature is slightly higher: 0,97˚C. See table 1.1 of Chapter 1 of SR15 (IPCC, 2018b). The average cumulative emission during 2005 – 2016 relative to 1850 – 1900 is 1958 GtCO2 (1 GtC = 3,664 GtCO2). The cumulative emission at the calibration point follows from the CO2 emissions database of the Global Carbon Project (GCP), a consortium of international climate research groups that keeps track of historical global carbon emissions (Le Quéré et al., 2018). The CO2 emissions comprise emissions by the burning of fossil fuels, by the industry, and by changes in land use. In line with the expert judgement of the AR5 author team, the SR15 method assumes that the TCREs for warming by CO2-only are normally distributed with a central value of 1,65˚C/(1000GtC) and a standard deviation (SD) of 0,85˚C/(1000GtC).
For the minimum and maximum TCRE, SR15 takes the 33th and 66th percentile of the CO2-only TCRE distribution following the IPCC practice that started with AR5. The 33th and 66th percentiles are interpreted by the IPCC as corresponding to a chance of meeting the warming limits of 1 out of 3 and 2 out of 3, respectively. The expected non-CO2 warming in the future at the global warming limits is estimated from the results of projections of climate models of reduced complexity calibrated against the full-fledged ESM climate models. For the 1,5˚C warming limit this non-CO2 warming is about 0,1˚C and for the 2˚C limit it is twice as much.
Table 1 summarizes the results of the Carbon Budget calculations reported in SR15 (Table 2.2) for the reference date of 01-01-2018 (IPCC, 2018c). They are equal to the calculated Carbon Budgets relative to the calibration point minus the cumulative emission from 01-01-2011 to 01-01-2018 of 290 GtCO2. The Carbon Budgets for the 50th percentile represent best estimates and are the main outcome of the SR15 budget calculations. The 33th and 66th percentile Carbon Budgets illustrate the sensitivity of the Carbon Budgets to the selected TRCE for CO2-only warming.
Table 1 – SR15 Carbon Budgets (CB) at 01-01-2018 for different TCRE percentiles
The error margins of the calculated budgets are rather large. In the case of the 1,5˚C warming limit the relative standard deviation (RSD), the standard deviation relative to the best estimate, for SAT and SAT/SST is 67 and 63%, respectively. For the 2˚C case the RSD for SAT and SAT/SSS is 54 and 53%, respectively.
Shortcomings of the SR15 method
Figure 2 displays the total warming since pre-industrial times expressed in both SAT and SAT/SST versus cumulative CO2 emission as inferred from the SR15 Carbon Budget estimation method and its results. For both SAT and SAT/SST the warming consists of two straight lines: one up to the calibration point for past warming and one thereafter for future warming.
In the observational period up to the calibration point the SAT line has an equivalent slope of 1,81˚C/(1000GtC). The equivalent slope of the SAT/SST line is 1,63˚C/(1000GtC), lower than the one of the SAT line and commensurate with the lower calibration temperature.
The straight lines beyond the calibration point follow directly from the reported SR15 Carbon Budgets for 1,5 and 2˚C which are indicated in Fig. 2 by the plus symbols. The slopes of the two warming lines are identical and equivalent to a TRCE of 1,99˚C/(1000GtC). The intercepts differ by 0,1˚C, equal to the difference between SAT and SAT/SST warming at the calibration point.
Figure 2 – Total global warming relationships as per SR15 method
Figure 2 reveals three evident shortcomings of the SR15 method. First, there is a discontinuity of the total warming at the calibration point, which is physically impossible. Second, there is not only a discontinuity in the warming but also in the warming slope. In the case of SAT the equivalent slope jumps from 1,81 to 1,99˚C/(1000GtC) and for SAT/SST from 1,63 to 1,99˚C/(1000GtC). It means that global warming in the future is stronger, i.e. more warming for the same amount of CO2 emission, than in the past before the calibration point. This may not be impossible but at this point there is no evidence for such a change. If we go by the climate model projections reported in AR5, the simulated warming curves are concave down rather than concave up, which means that we may expect a decrease in warming strength as opposed to an increase. See Figure SPM.10 of the SPM of AR5 WG1 (IPCC, 2013b). Third, the difference between future SAT and SAT/SST warming is constant and equal to the difference at the calibration point. This is highly peculiar, if not impossible. The difference is zero at zero warming and zero cumulative emission and has increased to 0,1˚C at the calibration point. Why would this increase stop at precisely the calibration point?
The discontinuity in total warming is due to the CO2-only warming straight line starting at the calibration point for total warming (see Fig. 1), while in actual fact this line should have started below this point. According to AR5, the temperature difference between central total warming and central CO2-only warming at the calibration point is 0,42 ˚C (= (2,441,65)/(1000×3,664)×1958). See again Figure SPM.10 of the SPM of AR5 WG1 (IPCC, 2013b). In reality this difference might be overstated but there must be a marked difference between total and CO2-only warming. The discontinuity in total warming slope at the calibration point is caused by too steep a slope of the CO2-only straight line of 1,65˚C/(1000GtC). In essence the calibration point narrows the suite of possible total warming straight lines spanning an effective TCRE range 1,25 to 3,25˚C/(1000GtC) to a single straight line with an effective TRCE of 1,81˚C/(1000GtC) for SAT and of 1,63˚C/(1000GtC) for SAT/SST. The effective TCRE of the calibrated SAT straight line of 1,81˚C/(1000GtC) is considerably lower than the original central estimate of the AR5 range of 2,44˚C/(1000GtC). The implication is that the slope of the best estimate of the CO2-only straight line, which is based on the same AR5 climate model projections, must also be less steep and thus lower than the assumed 1,65˚C/(1000GtC). A proportional reduction (=1,81/2,44) would result in a CO2-only TCRE of 1,22˚C/(1000GtC). That the slope of the CO2-only line is too steep is also obvious from the SAT/SST slope for total warming of 1,63˚C/(1000GtC), which is already lower than the CO2-only slope of 1,65˚C/(1000GtC), a physical no-no.
In addition to the inconsistencies in the profiles for total warming, the incorporation of uncertainty in the estimated Carbon Budgets is another issue. In the SR15 method, the uncertainty is accounted for exclusively through the uncertainty in the climate-model derived TCREs for CO2-only warming (the Min and Max lines in Fig. 1). The uncertainty in the CO2-only TCRE is substantial and leads to an equally substantial uncertainty in the Carbon Budgets. Subsequently, to make allowance for this uncertainty, the final estimates of the Carbon Budgets are significantly downgraded from the best-estimate Carbon Budgets.
On a general note, it is debatable whether best-estimate Carbon Budgets should be downgraded at all. The warming limits of 1,5 and 2˚C are no hard physical boundaries but negotiated political compromises laid down in the Paris Climate Agreement. In the absence of any specification in the Agreement, the Carbon Budgets belonging to these warming limits should be, by legal convention, best estimates, i.e. taken at the 50th percentile of the range of possible Carbon Budgets. Taking a stricter estimate, e.g. at the (arbitrary) 66th percentile as in SR15, amounts to an unstated lowering of the negotiated warming limits, disregarding the letter and spirit of the Paris Climate Agreement.
Regardless of the legitimacy of the downgrading, the actual evaluation of the uncertainty in the Carbon Budget estimates in SR15 is flawed on two counts. First, by attributing the uncertainty exclusively to the uncertainty in CO2-only TCRE, the uncertainty in the calibration point and in the estimates of the non-CO2 warming at the warming limits is ignored. As we shall demonstrate in the next section, errors in the calibration point may have a significant effect on the error in the final Carbon Budget and should be part of the overall error analysis. This also holds for the errors in the estimates of non-CO2 warming. Second, the uncertainty in CO2-only TCRE is vastly overstated. Using a calibration point drastically lessens the uncertainty in the effective TCRE. That is exactly what a calibration point is all about. Because the CO2-only TCRE is directly related to the effective TCRE, the uncertainty in the CO2-only TCRE is considerably reduced as well.
All in all, the SR15 method is seriously flawed and used as-is leads to over-conservative Carbon Budgets. The discontinuity in warming effectively lifts the total warming straight line and thus lowers the Carbon Budgets. The inflated warming slopes increase the warming strength of CO2 and thus decrease the Carbon Budgets. Finally, the choice for the 66th instead of the 50th percentile reduces the Carbon Budget even more.
We propose an alternative method for estimating Carbon Budgets that preserves the strong point of the SR15 method (calibration) and does away with its shortcomings (discontinuities): simply extrapolate the calibrated total warming straight line of the past (light-blue line in Fig. 2) to the future. This goes back to the initial proposition in AR5 to estimate the Carbon Budgets from a single total warming straight line. The only difference is that the slope of the line does not depend on climate model projections as in AR5 but on the observational record as boiled down in the calibration point. What is retained from the climate model projections is the notion that mean global temperatures are a function of cumulative CO2 emission and that this functional relationship is approximately linear.
Figure 3 – Total global warming relationships as per alternative method
Figure 3 depicts the total warming straight lines for both SAT and SAT/SST as defined by the calibration points. The SAT line is steeper with a TRCE of 1,81˚C/(1000GtC) compared with the SAT/SST line with a TCRE of 1,63˚C/(1000GtC). Carbon Budgets follow straightforwardly from the intersection of the straight warming lines with the horizontal line corresponding to the warming limit.
Table 2 – Alternative and SR15 best-estimate Carbon Budgets (CB) at 01-01-2018 and Carbon Budget Emission Ratio (CBER)
Table 2 lists the alternative Carbon Budgets relative to 01-01-2018 for the warming limits of 1,5 and 2˚C along with the comparable best-estimate SR15 Carbon Budgets. Also shown is the Carbon Budget/Emission Ratio (CBER) for the calculated Carbon Budgets. It is the ratio of Carbon Budget at a certain reference date and the annual CO2 emission rate just before the reference date and indicates the number of years a given budget will last if emission continues at the then current annual emission rate. Here the reference date is 01-01-2018 and the current emission rate in 2017 is 41 GtCO2/year. As expected, the alternative budgets are substantially larger than the SR15 budgets. What is also worth noting is that the difference between SAT and SAT/SST in the new budgets is much more pronounced than in the SR15 budgets.
The IPCC has traditionally used the blended SAT/SST as a measure for global warming. See Chapter 1 of SR15 (IPCC, 2018b). In keeping with this tradition, the relevant Carbon Budgets are 1130 and 2250 GtCO2 for a warming limit of 1,5 and 2˚C, respectively. The Carbon Budget for 1,5˚C is almost 50% larger than the comparable SR15 best estimate and about twice as much as the 66th percentile estimate of SR15 (see Table 1). At the 2017 emission rate, the 1,5 and 2˚C Carbon Budgets will be consumed in 27 and 55 years, respectively.
The Carbon Budgets in Table 2 are best estimates. The uncertainty (error) in these estimates is governed by the errors in the calibration temperature and in the calibration cumulative emission. According to Chapter 1 of SR15 (IPPC, 2018) the relative standard deviation (RSD) of the SAT/SST calibration temperature is 13,8% (=0,12/0,87×100). We assume the same RSD for the SAT calibration temperature. We have conservatively set the RSD of the cumulative emission to 1%. Using the Error Propagation Law, we then calculate in the case of SAT for the RSDs in the Carbon Budgets at the calibration point 39 and 27% for the 1,5 and 2˚C warming limits, respectively. For SAT/SST the RSDs become 33 and 24% for the 1,5 and 2˚C warming limits, respectively. The SAT errors are larger than the SAT/SST errors because the SAT calibration temperature is closer to the warming limit. The uncertainty in the Carbon Budgets due to uncertainty in the calibration point is appreciable but not as much as in the reported SR15 budgets due to the uncertainty in the CO2-only TCRE.
Climate policy implications
Carbon Budgets form the core of global warming mitigation policies by CO2 reduction. They define how fast CO2 emissions are to be reduced to net-zero to prevent global warming from exceeding the agreed global warming limits of 1,5 and 2˚C. An often used reduction scenario assumes a simple linear reduction to zero from the current emission rate, called a ‘stylized CO2 reduction pathway’ by the IPCC. Of course such a scenario is not realistic, it is merely a thought experiment that provides uncomplicated but useful insight into the timeframe for CO2 reduction and thus into the urgency of CO2 mitigation. In some countries (e.g. The Netherlands) climate policy is based on such a simple scenario.
To illustrate the implications of the results of our analysis for climate policy we have constructed two stylized CO2 reduction pathways: (1) a base case with the 66th percentile SAT/SST Carbon Budgets as reported in SR15 and (2) the alternative case with the best-estimate SAT/SST Carbon Budgets calculated by the above alternative method. We have chosen the blended SAT/SST Carbon Budgets to conform to the IPCC practice of expressing global warming in SAT/SST. We assume that the reduction pathways start in 2020, the year that the Paris Climate Agreement enters force. The IPCC budgets at 01-01-2020 for the 1,5 and 2˚C warming limits are 488 and 1608 GtCO2, respectively, equal to the 66th percentile of the corresponding SAT/SST Carbon Budgets at 01-01-2018 (see Table 1), minus the estimated total emission in 2018 and 2019 of 82 GtCO2. Likewise, the alternative Carbon Budgets per 01-01-2020 for the 1,5 and 2˚C warming limits are 1048 and 2468 GTCO2, respectively, equal to the SAT/SST Carbon Budgets 21 01-01-2018 (see Table 2) and corrected for the CO2 emission in 2018 and 2019.
Figure 4 depicts the pathways for the base-case SR15 budgets. The orange bullets denote the historical CO2 emission rates from 2000 up to and including 2017 and the assumed emission rates in 2018 and 2019. The green and blue straight lines represent the linear emission paths for the 1,5 and 2˚C warming limits, respectively. Net-zero emission for the 1,5˚C warming limit happens in 2043 and for the 2˚C limit in 2079. At the end of the term of the Paris Agreement in 2030, the reduction in CO2 emission is then 42% of the emission in 2010 for the 1,5˚C warming limit and 12% for the 2˚C limit. The 1,5˚C reduction pathway broadly agrees with the main conclusion in Chapter 2 of SR15: “decline of about 45% from 2010 levels and reaching net-zero around 2050” (IPCC, 2018c). It corroborates the central message of SR15 that limiting global warming to 1,5˚C is within reach, in theory, but would require an extraordinary effort in emissions reduction.
Figure 4 – CO2 reduction pathways for the SR15 base-case Carbon Budgets
Figure 5 shows the CO2 reduction pathways for the larger Carbon Budgets estimated with the alternative method described above. As expected, the pathways to net-zero emission are substantially longer. The net-zero point for the 1,5˚C warming limit happens in 2070 and in 2124 for the 2˚C limit. Logically, the associated reductions in 2030 are not as stringent as in the SR15 base case: 16% and 4%. Hence the larger and more appropriate Carbon Budgets derived in this study offer considerably more latitude in meeting the Paris warming limits than the over-conservative SR15 budgets. As a consequence, the need for reducing CO2 does not seem as urgent as conveyed in SR15.
Figure 5 – CO2 reduction pathways for the alternative Carbon Budgets of this study
- The method for estimating Carbon Budgets in SR15 has serious shortcomings, leading to non-physical future global warming profiles.
- Used as-is the method gives rise to over-conservative Carbon Budgets and thus to too stringent timeframes to reach net-zero emission.
- A simple alternative method without the shortcomings yields substantially larger Carbon Budgets and so more lenient mitigation timeframes.
- For a linear emission reduction pathway, net-zero emission for a global warming limit of 1,5˚C is reached in 2070 instead of in 2043 as per the SR15 budget. For a warming limit of 2˚C, net-zero emission happens in 2125 rather than in 2079.
Moderation note: As with all guest posts, please keep your comments civil and relevant.