the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
A perspective on the next generation of Earth system model scenarios: towards representative emission pathways (REPs)
Malte Meinshausen
Carl-Friedrich Schleussner
Kathleen Beyer
Greg Bodeker
Olivier Boucher
Josep G. Canadell
John S. Daniel
Aïda Diongue-Niang
Fatimah Driouech
Erich Fischer
Piers Forster
Michael Grose
Gerrit Hansen
Zeke Hausfather
Tatiana Ilyina
Jarmo S. Kikstra
Joyce Kimutai
Andrew King
June-Yi Lee
Chris Lennard
Tabea Lissner
Alexander Nauels
Glen P. Peters
Anna Pirani
Gian-Kasper Plattner
Hans Pörtner
Joeri Rogelj
Maisa Rojas
Joyashree Roy
Bjørn H. Samset
Benjamin M. Sanderson
Roland Séférian
Sonia Seneviratne
Christopher J. Smith
Sophie Szopa
Adelle Thomas
Diana Urge-Vorsatz
Guus J. M. Velders
Tokuta Yokohata
Tilo Ziehn
Zebedee Nicholls
Abstract. In every IPCC Assessment cycle, a multitude of scenarios are assessed, with different scope and emphasis throughout the various Working Group and Special Reports and their respective chapters. Within the reports, the ambition is to integrate knowledge on possible climate futures across the Working Groups and scientific research domains based on a small set of ‘framing pathways’, such as the so-called RCP pathways from the Fifth IPCC Assessment report (AR5) and the SSP-RCP scenarios in the Sixth Assessment Report (AR6). This perspective, initiated by discussions at the IPCC Bangkok workshop in April 2023 on the “Use of Scenarios in AR6 and Subsequent Assessments”, is intended to serve as one of the community contributions to highlight needs for the next generation of framing pathways that is being advanced under the CMIP umbrella for use in the IPCC AR7. Here we suggest a number of policy research objectives that such a set of framing pathways should ideally fulfil, including mitigation needs for meeting the Paris Agreement objectives, the risks associated with carbon removal strategies, the consequences of delay in enacting that mitigation, guidance for adaptation needs, loss and damage, and for achieving mitigation in the wider context of Societal Development goals. Based on this context we suggest that the next generation of climate scenarios for Earth System Models should evolve towards ‘Representative Emission Pathways’ (REPs) and suggest key categories for such pathways. These ‘framing pathways’ should address the most critical mitigation policy and adaptation needs over the next 5–10 years. In our view the most important categories are those relevant in the context of the Paris Agreement long-term goal, specifically an immediate action (low overshoot) 1.5 °C pathway, and a delayed action (high overshoot) 1.5 °C pathway. Two other key categories are a pathway category approximately in line with current (as expressed by 2023) near- and long-term policy objectives, and a higher emissions category that is approximately in line with “current policies” (as expressed by 2023). We also argue for the scientific and policy relevance in exploring two ‘worlds that could have been’. One of these categories has high emission trajectories well above what is implied by current policies, and the other has very low emission trajectories that assume that global mitigation action in line with limiting warming to 1.5 °C without overshoot had begun in 2015. Finally, we note that timely provision of new scientific information on pathways is critical to inform the development and implementation of climate policy. For the second Global Stocktake under the Paris Agreement in 2028, and to inform subsequent development of Nationally Determined Contributions (NDCs) up to 2040, scientific inputs are required well before 2028. These needs should be carefully considered in the development timeline of community modelling activities including those under CMIP7.
- Preprint
(3661 KB) - Metadata XML
- BibTeX
- EndNote
Malte Meinshausen et al.
Status: open (until 07 Dec 2023)
-
CC1: 'A value-of-information lens is needed', Robert Kopp, 12 Sep 2023
reply
I thank the authors for this manuscript but am concerned the proposal does not fully grapple with the costs of identifying a scenario as a high-priority scenario. Running high-priority scenarios through dozens of Earth system models represents a major commitment of scientists' time – time that cannot be spent otherwise advancing the scientific frontier – and should not be treated as cost-free. There should be good reason to think that the ESM scenarios are sufficiently distinct that the differences among them will be meaningful and will truly provide insights that can only be gleaned from an ESM.
In this regard, I am particularly concerned about the overlap between the DAPD, IAPD, and IA2015 scenarios. I am skeptical that ESMs will tell us enough here that cannot be gleaned by taking a warming-level lens to climate model emulator output, and would strongly urge that one or more of the better-resourced climate modeling groups demonstrate the value of such scenarios before they become a global recommendation of CMIP7.
The warming-level framing (that a 2°C world is climatologically a 2°C world regardless of whether it happens in 2060 or 2100) has significant limitations. From a human perspective, this is certainly not the case – the impacts of 2°C to a world with 2060 socioeconomics is quite different from a world with 2200 socioeconomics. But that’s irrelevant to the Earth system model.
It is also true from a physical perspective that the climate continues changing after temperature stabilizes, as the oceans continue to warm. So a world that stabilizes at 2 C in 2070 looks physically different from that same world in 2150. Likewise, a world that reaches 1.5°C in 2150 after high overshoot will not be identical to one that stabilizes at 1.5°C with low or no overshoot. This is important to investigate – but I don’t think the threshold has been met for saying it’s a high priority for every modeling team participating in CMIP7 to run multiple emissions scenarios that indirectly probe this. In the absence of clear evidence, the payoff is not convincing to me.
I am also concerned by the deprioritization of the very high emissions (TEWA) scenario. While indeed this scenario should not be portrayed as a reference scenario, stress-testing Earth system models is important for understanding the behavior of the Earth system and characterizing deep uncertainty. There is likely to be greater scientific insight gleaned from one very high emissions scenarios than three very low emissions scenarios, and from a scientific perspective, it is quite useful to stress-test every model with such a scenario.
I recognize that my comments here point to some tension between a WG1 perspective and a WG2/3 perspective. This should be acknowledged. It is indeed important to produce the physical projections needed for impact and mitigation analysis. But I am unconvinced that standardized scenarios run with full-complexity Earth system models, which have substantial opportunity-costs, are always the right approach, and I would like to see a critical value-of-information lens brought to the scenario discussion before such an approach is adopted.
Citation: https://doi.org/10.5194/gmd-2023-176-CC1 -
CC2: 'Reply on CC1', Zebedee R. Nicholls, 12 Sep 2023
reply
Thanks for a very nice comment. Some excellent points. I particularly like the idea of "that one or more of the better-resourced climate modelling groups demonstrate the value of such scenarios before they become a global recommendation of CMIP7". I hope this idea can be fed to the CMIP7/ScenarioMIP steering committees, who then have the fun job of working out if this is feasible on the timelines required or not :) Without that information, it is indeed an interesting yet tricky debate to have, with pros, cons and unknowns.
Citation: https://doi.org/10.5194/gmd-2023-176-CC2 -
CC3: 'Reply on CC1', Malte Meinshausen, 13 Sep 2023
reply
Thanks. A worthwhile consideration we thought about in the author team....
Two points:
a) to some degree it is a question of balance and focus. Do we want to focus on extreme scenarios (e.g. take SSP5-8.5) to advance scientific insights w.r.t the climate system behaviour under high warming or do we want to focus on scenarios that have a closer link to being of relevance for societies (and all scenarios cater to both aspects, just to different degrees of course). The trillion-dollar decisions in this and the next decade are made against the backdrop of 1.5C versus 1.7C versus 2C and if we do not populate that space with scientific WGI & WGII studies on relative impacts, then IPCC AR7 can again inform only in relatively general statements ("every bit of warming matters") rather than drilling down into the specifics. See below the relevant text excerpt from our study.
"Against a backdrop of ‘the emission world avoided’ context provided by high-end emission scenarios, it is paramount for decision makers to also understand the implications of stronger mitigation efforts in terms of climate benefits and avoided impacts. Whether we follow a scenario that delays mitigation efforts by 10, 20 or 30 years and reaches net-zero CO2 emissions 315 by 2050 or 2060 or 2070 makes trillion-dollar differences in terms of directing government incentives and private capital (Riahi et al., 2022; van der Wijst et al., 2023), but also in terms of adaptation costs, limits to adaptation, irreversible loss and economic and non-economic costs of anticipated losses and damages (Pörtner et al., 2022). While natural variability in any single year influences global-mean temperatures by ±0.25°C (Box 4.1 in IPCC AR6 WGI, i.e., Lee et al., 2021), climate extremes (Seneviratne et al., 2021) and impacts that reflect long-term, cumulative climate changes (e.g. glacier melt or sea 320 level rise) can be substantially different between a scenario peaking at 1.6°C or 1.8°C in the middle of the century (Mengel et al., 2018; Pfleiderer et al., 2018). "
b) While maybe not from the ESM output per se, but further down the line in the cause-effect WGI-WGII model chain, there could be ample new science on the impacts of, e.g., slight versus higher overshoot of 1.5C. At the moment, such (often more WGII focussed) science is simply not possible given that the first piece of the chain (the ESM output) is not available. Thus, am not sure that we would solve the consideration of where to place resources with simply "one ESM model test run", given that the chain of impact models would need to be employed.Citation: https://doi.org/10.5194/gmd-2023-176-CC3 -
CC4: 'Reply on CC3', Robert Kopp, 13 Sep 2023
reply
I agree on the value of understanding the difference of impacts in different 1.5-2 C scenarios. My question is whether more (expensive) ESM scenarios will provide you that information. Much of the difference in impacts will have to do with the state of the human system when different warming levels occur — not something you necessarily need new ESM simulations to address. Yes, emulation has limitations, but before spending millions of dollars of scientist time on new simulation, it seems reasonable to do some experiments to see whether those limitations are large relative to internal variability and structural uncertainty.
In the quote from the paper, you highlight the difference between 1.6 and 1.8 C in terms of sea level rise. This I think points to a limit of the “run everything through ESM” approaches. ESMs of course neither generally include all sea level processes nor sample sea level uncertainty adequately (compare the differences in medians in Mengel et al 2018 to the overall range). Further, putting aside ice sheet instability temperature thresholds that ESM scenarios are not optimized for probing, global mean sea level change is closely related to integrated temperature — so yes, an extended period over 1.5 C adds up, but not in a way that an ESM is needed to quantify.
I think the paper perhaps too readily dismisses the utility of smart emulation methods (eg Tebaldi et al 2022’s STITCHES methodology, https://esd.copernicus.org/articles/13/1557/2022/). It might be a better use of researchers’ time to have a smaller number of standardized ESM scenarios combined with standardized emulated scenarios that could also be used by impact models. The comparison between the two approaches should at least be made on quantitative grounds rather than intuition.
Citation: https://doi.org/10.5194/gmd-2023-176-CC4
-
CC4: 'Reply on CC3', Robert Kopp, 13 Sep 2023
reply
-
CC2: 'Reply on CC1', Zebedee R. Nicholls, 12 Sep 2023
reply
-
CC5: 'Comment on gmd-2023-176', Alexandre Magnan, 19 Sep 2023
reply
Dear colleagues,
First of all, many thanks to the authors for these thoughts on how to improve the scenario approach in the AR7. That is a critical question that the scientific community should seriously reflect on over the coming years, and that relates to how to capture the various components of climate action together (mitigation + adaptation) and project them into the future… and in a way that speaks in the end to decision-making at large. The scientific and methodological challenges are enormous, and so that is extremely valuable to have papers that give a try, such as this one.
I concur with the authors that we need a pathways approach that shows the range of potential futures, beyond only the ones that one could judge today as “realistic”. As scientists, our role is probably to speak to policy today, yes, but also lay foundations for thinking beyond present day-driven path dependencies and allow for imagining potential radical shifts in one way or another.
The proposal to classify scenarios into NFA, DASMT, DAPD, IAPD + TEWA and IA2015 is interesting, but being not myself a climate modeler, I will let experts to comment on the technical aspects and the overall relevance of such a framing.
On the multi-“Representative x Pathways” (RxP) approach, it makes sense in principle as we need, along with emission/warming projections, insights on what other aspects of climate action, especially adaptation-related, could look like in the future. And the authors are right to advocate for not stopping at 2100, though challenging from a social science perspective (but still, the role of the scientific community is to try envisaging the very long-term).
My two general comments rather refer to the bottom part of Figure 1. Maybe the authors will see them as being a bit out of the scope of this Discussion paper, sorry for that, but that’s an attempt on my side to expand on the ideas suggested by the authors. These are just food for thought; I don’t have neither turnkey solutions nor lessons to give; so please take the below as complementary reflections and not as pure (negative) criticism :-)
- Figure 1 seems to suggest that there is a deterministic path from RTPs to RSPs and then RACs. That doesn’t not align with what the adaptation science says and that views adaptation has being driven by more than warming-related and economic-related drivers. I know drawing figures means being reductionist in a way, so forgive this comment if your intention was not to advocate for a “deterministic view”.
- One question at one point will be: how to combine the various pathway scenarios into a kind of synthetic view of climate action options for the future? That is, what could be the common language/metric across the various RxP, or more specifically, the language/metric to be used to move from Representative Socioeconomic Pathways to Representative Adaptation Pathways, and then from these latter to Representative Impact/Risk Pathways ? In other words : how to connect information on Hazards (influenced by REPs, but maybe not only) to information on Exposure and Vulnerability? Answering this question could be decisive from a cross-Working Group perspective and in terms of laying foundations for the AR7 Synthesis Report. The question itself seems trivial from a theoretical perspective (the Risk propeller diagram shows connections !), but in practice the answer is far from being an easy one… Using quantified indicators for designing adaptation pathways that are susceptible to land on clear information on Exposure and Vulnerability, would be a way to come up with numbers and create bridges with the outcomes of REPs (with a meeting point around Risk/Hazards, as suggested in Figure 1); but the history of developing quantitative indicators for adaptation proved to be unsatisfying. Especially because that often means reducing adaptation aspects to parameters that can be quantified and informed with existing databases, which usually ends up with selecting GDP-related dimensions (e.g. income, proportion of population/activities at risk) and leaving aside more qualitative ones (e.g., risk perception, power relations). There is therefore, at least to me, a danger with moving towards a quantitative indicator-based approach, for example through IAMs, what the direct arrow from RSPs to RAPs suggests under the lines. Other methods rather use expert judgments to assess adaptation progress/gaps in a more integrative way (i.e. not only GDP- and warming-driven) and offer more relevant ways to think the future of adaptation, but then how to connect them to RSPs on the one hand, and to RIPs on the other one? Again, I don’t have any clear answer to this, but such a question should also drive the way scenarios are developed in view of feeding the AR7.
- The scale issue : while it makes perfect sense to develop REPs at the global level (+ possibly a regional perspective), developing RIPs and RAPs requires to further consider sub-national context-specificities (i.e. intra-national variability). So that hoping for globally relevant “climate change scenarios” (i.e. at the crossroads of multiple RxPs in Figure 1) could reveal problematic — as Shared Socioeconomic Pathways revealed to be (a point that has been raised during the IPCC AR6 Scenario workshop in April 2023). On the other hand, we cannot expect from the IPCC to consider a wide range of sub-national context-specific RxPs… which would anyway also raise the “aggregation” issue that I mention above. Here again, no solution on my side, but that’s a debate that needs to happen between the IPCC and the policy community: who are the decision-makers that the IPCC primarily wants to talk to ? And what is the level of information and scenario outcomes these policy-makers truly need? My understanding from the recent years is: global-level information is not enough, but local-level information is too much detailed; so where to put the cursor for the IPCC? I guess this question is at the heart of the design of multi-perspective scenarios/pathways in the perspective of the AR7.
Citation: https://doi.org/10.5194/gmd-2023-176-CC5 -
CC6: 'Reply on CC5', Malte Meinshausen, 20 Sep 2023
reply
Thank you Alexandre for these very pertinent questions and the food for thought. I can think of some sentences that would point to (rather than solve) these key issues for the manuscript (as e.g. inserting "various methods, including qualitative expert judgment" or similar into the green part of the figure 1). However, ...
.... maybe your comments provide the incentive for some bright minds to chime in, especially from the WGII and WGIII domains. If not for the manuscript, it would be great to start thinking about the options to address these two all-important issues. For one, how to link the various RxPs up (without it ending up be a large hodgepodge that impedes synthesis at the end) and secondly, how to integrate the richness and diversity of the national perspective (both on the adaptation and mitigation side).
Looking forward to the further discussion. Hopefully some here, maybe a bit elsewhere.Citation: https://doi.org/10.5194/gmd-2023-176-CC6
-
RC1: 'Comment on gmd-2023-176', Andy Reisinger, 11 Oct 2023
reply
GENERAL COMMENTS
The perspective by Meinshausen et al proposes a set of representative emission pathways (REPs) to be used to drive the next set of Earth System Model experiments, with a view to provide a core set of climate framing pathways for the IPCC 7th Assessment Cycle. The perspective is well written and makes a cogent argument supported by a set of policy- and science-related criteria, having drawn on a wide and diverse range of authors.
I have only one high-level concern, which is whether the manuscript sufficiently differentiates the broader narrative value of REPs from the narrower question of what REPs should be used to run ESMs. I.e. it is one question what REPs would best serve as climate framing pathways to support discussions under the UNFCCC and Paris Agreement/GST2, to provide a range of climate futures for the next IPCC assessment cycle, and to guide research across a broad range of domains; it is another question whether all those REPs necessarily need to be run specifically through ESMs (as a resource-intensive community effort) to produce the scientific climate knowledge that those broader processes need. As noted in a community comment by Robert Kopp, some of the differences in impacts and risks in the half-degree space between 1.5 and 2 degrees (or between ‘somewhere below’ and ‘slightly above’ 2 degrees under different confidence levels in the achievement of current long-term targets) are not necessarily best resolved by ESMs but rather by more specialised models that either probe a specific biogeophysical domain (such as ice sheet responses to warming) or the interaction between human, ecological and biogeophysical drivers of impacts and risks.
This does raise the question whether the resources needed to run all REPs through ESMs as a community effort are indeed spent well enough, and to what extent a (slightly) smaller or different set of REPs might be equally justifiable as far as ESM runs are concerned, while treating it as a separate, related but distinct, question what REPs should be used to frame alternative climate futures more generally to support climate policy processes in the UNFCCC and IPCC and to drive broader research efforts across multiple domains.
However, I accept that this cannot be answered based solely on the unique scientific knowledge that the ESM runs would provide. ESM runs have signalling and narrative power. E.g. even within the IPCC, ESM-run scenarios will very likely pre-structure the range of alternate climate futures within which the assessment across WGII and WGIII (or within any Special Report) occurs. Any REP that is not covered by ESMs risks playing a lesser role not only in the WGI domain but also in the assessment of differences in risks, and of adaptation and mitigation responses considered by WGII and WGIII in the next IPCC assessment – or at least any scenario not run through ESMs would have to work much harder for its justification in the overall narrative, including in the IPCC approval process. This means a strong, but separate(!), argument can be made to run an emission pathway through an ESM even if it was not strictly necessary from a scientific perspective to have that scenario modelled through an ESM. The authors set up criteria from both policy and science domains to argue for their selection of REPs, but there is no discussion of whether the policy and science criteria may in some cases lead to contradictory conclusions about a preferred set of REPs; and if so, based on what considerations the authors reach their final recommendation.
So my overall sense is that the manuscript could work harder to disentangle and disclose the different motivations and arguments relating to those different objectives, and the degree to which different REP choices and different modelling tools using those REPs might best respond to the authors’ set of criteria, and to reflect more critically and transparently on whether there might be a difference between the proposed set of REPs to serve as broad climate framing pathways in general, and REPs that are run specifically through an ESM community effort. Many of my specific comments point to areas where I think this clarity could be increased. But in the end, I fully accept that the balancing of those considerations does rely on judgement, meaning that reasonable people can look at the same set of facts and nonetheless come to different conclusions.
Despite my multiple (exhaustive but hopefully not exhausting) specific comments below, I'm happy to regard this as a minor revision, since I consider the manuscript to be already of high value - but there's an opportunity for a clearer discussion of those differences to lift its value and utility further.
SPECIFIC COMMENTS
L77-78: this sentence condenses a broader use (use of climate framing pathways in the IPCC AR7) and the specific evaluation of those pathways through CMIP/ESMs; consistent with my general comment, I wonder if authors could try to recognise more that there may be a distinction between those two uses, and the implications of that.
L94: “well before 2028” in my view underplays the urgency, set out in lines 746ff. Suggest consider slightly rephrasing this.
L101-102: framing pathways can shape how the next IPCC assessment approaches alternative climate futures in general; or they can simply mean “how ESMs can be run with a consistent set of drivers”. I suggest authors disentangle these very different meanings of ‘framing’ in different places of the manuscript. L113-114 acknowledges this point, but I don’t see this distinction elaborated in the manuscript and how different objectives (the broad and the narrow use of ‘framing’) might lead to different conclusions. Sections 5.2 and 5.6 to some extent speak to this issue, but these sections come across as defending the authors’ REP choice rather than working through the different implications and disclosing judgements that lead authors to reach their conclusions.
L131-132: this might be worth elaborating a bit: what is meant by the IPCC mandate here? What is the role of ESM runs in setting up an envelope of climate futures more broadly (beyond the immediate value-add from the specific knowledge gained)? Do ESM runs hold a special role in defining this envelope not (just) because of the knowledge they provide but perhaps also because they will be seen as unassailable cornerstone in the intersection between science and policy?
L136ff: I find the distinction between pathways and scenarios (especially as stated in lines 141-145) contradictory and confusing; line 143 says that pathways do not provide any explicit assumptions about socio-economics or policy, but then lines 143/144 say that pathways may describe quantified socio-economic futures. My interpretation is that pathways tend to be more uni-dimensional, whereas scenarios describe a complex set of interconnected and internally consistent drivers, assumptions and outputs. So pathways can cover any type of variable (input or output) that can be extracted from scenarios – emission pathways, temperature pathways, perhaps aggregated into climate pathways, but also energy-specific pathways, or socio-economic pathways. So it’s not whether socio-economics is in or out, it’s the uni- (or simply narrow) dimensionality of pathways compared to the multi-dimensional and internally consistent set of assumptions, drivers and outputs that characterise scenarios. I’m not saying the authors have to adopt this way of distinguishing pathways and scenarios, it’s just a suggestion that may be consistent with what the authors had in mind but the current text does not get this across. Box 1 is slightly less confusing, but it could also be clearer about the distinction between pathway vs scenario.
L158ff: this section could do more to acknowledge the overlap between climate and ecosystem/land-use models, which can fall into the gap between ‘climate’ and ‘socio-economic’ domain. These are arguably becoming more important, especially in the context of overshoot and land demand for CDR. Also somewhere in this section might be the place to more explicitly and in more detail acknowledge and map out the space that more detailed and specialised biogeophysical models can and must play to complement ESM runs to understand the difference that half a degree could make for some key dimensions of climate-related risks – and the degree to which such models must have ESM runs as their input, or whether they could also do their job using other inputs that are less resource-intensive to produce. This is critical to the question whether the set of REPs proposed as ‘framing’ pathways in a broader sense necessarily and in its entirety must be run through ESMs.
Figure 1: As noted in a community comment by Alex Magnan, there is no linear path from REPs to RIPs and RAPs, or between RTPs, RSPs and RAPs. It may not be worth spending too much time re-doing the figure, but if the authors can think of a way to represent the cluster of pathways on the left hand side as a dynamically interconnected set of issues rather than a linear progression from one to the other, that might be very useful (since the figure may well be used in lots of presentations to show the overall REP and scenario logic). It may also provide an opportunity to highlight the much smaller geographical scale, and dependence on nuanced socio-economic assumptions (including e.g. the potential for different regional-/national-level SPAs nested within a single global-scale SSP that will then drive diverse regional-/national-scale risks), that will be relevant for RIRTs, RAPs, and RTPs.
L298ff: I don’t disagree with the points made in this section, but it’s not clear whether all those issues rely on a community ESM modelling effort. The knowledge needs related to many of those questions rely on modelling approaches very different to ESMs (and, I would argue, in some cases not at all dependent on ESMs). On substance, I would add a criterion relating to the increasing need to consider overshoot pathways from a policy perspective (i.e. how to achieve a decline in temperature after a peak has been reached, and the extent to which this ought to influence near-term policies when temperatures are still on an upward trend).
L325-333: A couple of questions: it’s not clear why high-end global warming outcomes are of particular importance for a loss and damage conversation, given that loss and damage (as far as I understand) relates to actual losses and damages, not projected risks? Secondly, precisely because the range of near-term warming is dominated by climate uncertainties rather than differences between emission scenarios, this seems to argue against placing too much emphasis on a wide range of REPs if the goal is to inform loss and damage conversations and near-term adaptation needs? I don’t disagree with the overall thrust of the discussion in this section overall, but the arguments in those lines don’t seem to stack up well.
L333-339: these points are useful but are left hanging. Given that the authors ultimately recommend including something like SSP3-7.0 in the REPs, I’m missing a conclusion that says something like “on balance, we feel that because of small ensemble sizes, we will learn more if we use the median results from SSP3-7.0 and treat those as proxy (potentially with scaling) for high-end climate outcomes under current policies”. Without clarity about why SSP3-7.0 should be included, SSP3-7.0 might just become the new RCP8.5 for future arguments about the validity and utility of high-end emission scenarios. Of course there is also the separate rationale for SSP3-7.0 being the “world avoided”, but is that the main reason to include it, or is it (also) because it can and should be used as a proxy for high-end warming under current policies? Or is it also because we will learn something about earth system responses to climate forcing that is important to learn, regardless of its relevance from a narrative space?
L346ff: these policy criteria are all useful, but it is not clear to what extent they rely on ESMs to provide the answers that policymakers need. A critical reflection on differences (where they exist) between policy/narrative value, and scientific value, of different REPs would in my view significantly lift the value of this manuscript for the scientific community processes that ultimately have to reach a decision on REPs for ESMs.
L346ff: From a policy perspective, one point missing is the question of the extent to which there is flexibility in treating non-CO2 mitigation differently to CO2 mitigation, within a given climate goal and over different time scales. Existing emission scenarios in the “below 2°C” space tend to assume a similar stringency for non-CO2 mitigation as for CO2, and hence both SSP1-1.9 and SSP1-2.6 have rapid and deep reductions of CH4 (and other air pollutants), well beyond simply co-mitigation from reduced fossil fuel use. But that’s not what we’re seeing in the policy world, where non-CO2 policies tend to be constructed very differently and with much less stringency. So from a policy perspective it would be useful to probe the feasibility and climate consequences of a pathway where governments do meet net-zero goals for CO2, but don’t apply the same stringency to non-CO2 mitigation as is implied in current scenarios.
L417:422: I would have thought that a more fundamental question is to what extent is climate change itself reversible? That’s what ESMs answer. The reversibility of impacts and risks needs much more complex scenarios and models that inevitably depend on socio-economic assumptions (and/or ecosystem models that may not rely on ESMs as input to answer questions about reversibility). From a policy perspective, I would also add questions about whether an intention to achieve net negative emissions in future will/should affect the way that we approach mitigation in the near term?
L433: (and L45): given the general push of the authors to look beyond 2100, I found the distinction between “up to 2100” and “beyond 2100” a bit jarring here. “Beyond 2100” to me does not signal multi-century scales relating to sea level rise, which is a key (science and policy) question.
L470ff: One point missing from this list, in my view, is better representation and intercomparison of the role of non-CO2 emissions in driving climate and uncertainties (see also L544), especially for CH4. How well do we really understand the feedbacks that influence CH4 chemistry in a warming climate? One can of course work around this by driving ESMs with prescribed concentrations, just so that more models can run the full set of forcings, but this potentially hides a non-trivial element of uncertainty. More specific results might also help calibrate emulators better for climate responses to CH4 emissions. So progressing this space would strike me as a highly relevant science question.
L471-478: I agree with the point being made here, but a single scenario will not help us clarify this. What we would need are two scenarios with similar levels of CO2 mitigation, and then different efforts to reduce non-CO2 emissions. This suggests that answering this sort of question requires a different emission pathway architecture, but might then also require more limited ESM intercomparisons, or the use of other models. So it’s not clear how this point relates to the goal of this manuscript, which is to propose a set of REPs to be used to drive ESMs as a community effort.
L488ff: I would have thought that better understanding of feedbacks is relevant not just under overshoot but also for current policy scenarios? This would also specifically include risks to carbon stocks in terrestrial ecosystems, especially for carbon stocks that may have been enhanced deliberately as mitigation measure and claimed as removal to counterbalance continued emissions.
L498ff: I don’t disagree with the question, but it’s not clear that ESMs are best placed to resolve those differences in the half-degree space?
L506ff: I don’t disagree with the question, but it’s not clear that ESMs are best placed to answer those questions, given the crucial role of socio-economic conditions and policy design to understand the consequences of land-based CDR?
L545ff: There is no linear logical way to get from criteria to specific scenarios, and I’m very happy to accept that the authors made (and had to make) some judgements to get from their criteria to this list. So I’m offering just some thoughts on other scenarios that could (in my view) equally have been in this list. I don’t expect the authors to rebut those ideas one by one, it’s more an invitation that if they have a clear reason why they did not choose one of those other scenarios, it might be worth including that rationale in the manuscript so that others benefit from their thinking.
- In terms of comparability with CMIP6, I would see some value in having two backward compatible REPs. The obvious choice for this would be TEWA and IA2015 (SSP3-7.0 and SSP1-1.9), since both those scenarios could justifiably run with (by now) counterfactual emissions prior to 2023. This would provide a robust way to separate changes in model behaviour in the next generation of ESMs from changes in assumed future and past actual emissions.
- There is no emission pathway in this list that corresponds to a C3 category in IPCC WGIII. While I can see a narrative value of focusing on pathways that keep 1.5°C alive at least by 2100, it seems oddly dichotomous to characterise the lower end of the scenario space only by pathways that either miss “well below 2 degrees” entirely (because they end up at or above 2 degrees) or that do succeed in limiting warming to below 1.5°C in 2100. So I would regard a C3-type pathway as a highly policy relevant pathway. Not arguing that authors should add yet another 'close-by' REP to the list, but it might help the manuscript to recognise its absence and perhaps explain why the judgement was made not to have it.
- There is only one scenario with a substantial overshoot behaviour (DAPD). From an earth system perspective, I would expect significant value in modelling overshoot also at higher warming levels (exceeding and declining below 2 degrees), over long time scales, and with more substantial overshoot (simply to deal with noise). This could lose realism or political palatability of such a pathway, but it is an example where I would appreciate a clearer discussion of how different considerations in the science and policy space might lead to different sets of REPs, so that the authors can then transparently disclose their judgement in their final selection.
L617ff: This section is critical for the argument whether the set of REPs that is proposed for a broader narrative framing purpose necessarily needs to be run, in its entirety, through ESMs to address the questions that motivated the selection of REPs in the first place. I don’t think this section quite does justice to this need. I find L628-633 unclear and unconvincing: are the authors saying that we could rely on only small ensembles per model, but a wider set of models, to address signal-to-noise in close-by REPs? How does this approach handle the fact that CMIP contributions are still an ensemble of opportunity rather than necessarily providing robust statistics? Also in L631-636, regional differences and differences related to land-use may indeed be much larger than the global mean, but if we have only one or two REPs it will be very difficult to learn how much a regional difference is due to models or a reflection of the particular SLCF and land-use emission choices made within REPs. L645/646 argues that ESM runs of overshoot pathways are necessary to enable WGII to engage with impacts and risks under overshoot – I’m not sure this is the case, as the main questions that WGII needs to address to better understand risks under overshoot (but, importantly, also the avoided risks once temperatures start declining again) in my view are more fundamental about process understanding, reversibility and cumulative damages, not necessarily whether we have highly resolved ESM outputs to drive impact models. Lastly, as flagged earlier, the section leaves open whether we learn as much as we need to from those scenarios about earth system behaviour in response to net negative emissions, or whether there is a scientific case for a more extreme overshoot scenario. Such a scenario would be politically unpalatable and this may limit its value as a narrative ‘framing’ scenario, but it might provide critical knowledge from ESMs about processes and thresholds. Is that the case? If yes, how would the authors propose to resolve this tension?
L695ff: The focus on regional downscaling misses the broader question of whether a ESM run produced with a close-by REP could be scaled to infer regionally relevant information based on a more finite set of REPs. There is also the question whether other, more specialised models need separate ESM runs for close-by REPs, or whether they could be driven with a smaller set of ESM runs plus other information, and/or scaled ESM results from close-by REPs to approximate climate drivers for e.g. ice sheet models. In an ideal world, we would of course model all REPs through ESMs, but given resource constraints, are there short-cuts that would be valid to take – and are there short-cuts that would be a scientific mistake? (Or a political/narrative mistake?)
L783-795: I don’t disagree with the points being made, but I don’t think the manuscript has quite made the case that these needs are best and can only be addressed by running all those REPs through ESMs. What REPs best support framing of alternative (and foregone) futures under Paris and in the IPCC is not necessarily the same set as what REPs should be used to drive ESMs to learn what we need to and can only learn from ESMs. A more critical reflection on whether there are differences, and if so, whether those differences must be overcome to nonetheless end up with a single set, or whether there could be different sets or subsets of REPs that serve different purposes, would in my view lift the value of this (already highly valuable) manuscript.
TECHNICAL CORRECTIONS
Citations of Pirani et al (submitted), and Sanderson et al (in prep) should presumably not appear in the final version of this manuscript, but only published literature?
Citations to IPCC reports, and Glossary: please follow the IPCC guidance on how to cite those reports. IPCC reports as a whole should not be cited by first author et al, but as IPCC (with an editorial team). Also see guidance on how to cite the glossary.
Fix citation of WMO (currently it is World Meteorological, O.)
Please provide full citation details for the ScenarioMIP report (van Vuuren et al 2023)
L239: stray “e” before “REPs”
L339: suggest adding a paragraph mark to differentiate the discussion of high-end scenarios from the following discussion of a low-end scenario.
L380: the word “as” appears to be missing after “insofar”
L454: space missing in “consequencesincluding”
L727: “comp” -> “compounds” (?)
L805: can’t make sense of grammar here: “… in this critical and contribute to…”
Citation: https://doi.org/10.5194/gmd-2023-176-RC1 -
CC7: 'Comment on gmd-2023-176: On plausible economic scenarios and slow growth', Matthew Burgess, 14 Nov 2023
reply
I thank the authors for this work, which represents an important advance to the international climate change scenario framework. I have one small--but, I think, important--comment.
On line 175, the authors note that: "While the original SSP (Shared Socioeconomic Pathway) modelling exercise covered all the SSPs more or less evenly (Rogelj et al., 2018; Riahi et al., 2017), the SSP2 ‘middle of the road’ scenario has since been used by most modelling groups as a default socioeconomic pathway and represents more than 90% of the 1202 scenarios with a climate assessment in IPCC AR6 WG III scenarios database (Riahi et al., 2022)."
While this quote accurately characterizes the literature, I encourage the authors to engage with, and comment on, the growing economic literature which suggests that a modal plausible economic scenario would be on the slow-growth end of the SSP range (SSP3 or SSP4), rather than in the middle of the range (SSP2). If this insight holds up, it would suggest that the range of economic scenarios needs to be revised downwards, compared to the SSPs, in future scenario frameworks.
Even before the COVID-19 pandemic, observed GDP per capita growth rates were more ~1%/year or more slower than all of the SSPs projected in most regions (see figures 2 and 3 in Burgess et al., 2020), with the largest divergences occurring in developing regions (Latin America and the Caribbean, Middle East and Africa) and not attributable to the 2008 recession. Looking ahead to 2100, the high-growth SSP scenarios' (SSP5 and SSP1) GDP per capita trajectories would require GDP per capita growth rates to increase by ~1.5-2x globally; they would have to do so almost overnight (to catch up to the scenarios' GDP per capita levels by 2100); and the growth rates would have to remain similarly high relative to regions' GDP per capita levels throughout the century (see figures 3a and 1c in Burgess et al., 2023). In Burgess et al. (2023), we show that two models on opposite ends of the complexity spectrum project a GDP per capita growth and inequality pathway similar to SSP4. We also show that one of our models would have slightly over-projected GDP per capita growth and convergence historically, despite outperforming short-term International Monetary Fund (IMF) forecasts. (Evaluating the historical performance of our other, more complex model is a larger undertaking, producing similar but not-yet-published results.) From this, we conclude that SSP4's GDP per capita trajectory might be on the fast-growth, fast-convergence end of a plausible range of 21st-century economic scenarios, rather than being on the slow-growth, slow-convergence end of the range, as it is currently in the SSP framework.
There are other recent studies, using methods quite different from ours, which reach similar conclusions: that the 21st-century economic outlook is more likely on the slow-growth end of the SSP range than in the middle (e.g., Buhaug & Vestby, 2019; Müller et al., 2022; Rennert et al., 2023). Of course, we acknowledge that economic pathways carry large uncertainty that is difficult to precisely quantify, and that there are a couple of studies which reach different conclusions from ours (Christensen et al., 2018; Startz, 2020). Given the importance of GDP per capita growth to a wide range of factors affecting mitigation and adaptation challenges, I encourage the authors to discuss the plausibility of socioeconomic scenarios in future iterations of their work. Again, I thank the authors for their very important contribution, which I think is a major advance to how we think about and use scenarios.
References:
Buhaug, H., & Vestby, J. (2019). On growth projections in the shared socioeconomic pathways. Global Environmental Politics, 19(4), 118-132.
Burgess, M. G., Ritchie, J., Shapland, J., & Pielke, R. (2020). IPCC baseline scenarios have over-projected CO2 emissions and economic growth. Environmental Research Letters, 16(1), 014016.
Burgess, M. G., Langendorf, R. E., Moyer, J. D., Dancer, A., Hughes, B. B., & Tilman, D. (2023). Multidecadal dynamics project slow 21st-century economic growth and income convergence. Communications Earth & Environment, 4(1), 220.
Christensen, P., Gillingham, K., & Nordhaus, W. (2018). Uncertainty in forecasts of long-run economic growth. Proceedings of the National Academy of Sciences, 115(21), 5409-5414.
Müller, U. K., Stock, J. H., & Watson, M. W. (2022). An econometric model of international growth dynamics for long-horizon forecasting. Review of Economics and Statistics, 104(5), 857-876.
Rennert, K., Errickson, F., Prest, B. C., Rennels, L., Newell, R. G., Pizer, W., ... & Anthoff, D. (2022). Comprehensive evidence implies a higher social cost of CO2. Nature, 610(7933), 687-692.
Startz, R. (2020). The next hundred years of growth and convergence. Journal of Applied Econometrics, 35(1), 99-113.
Citation: https://doi.org/10.5194/gmd-2023-176-CC7 -
CC8: 'Reply on CC7', Zebedee R. Nicholls, 15 Nov 2023
reply
Thanks for the comment, very interesting. In this paper we aren't going to be actually recommending any SSP in particular and the discussion of economic growth is outside the scope for the most part, but we can certainly note that the discussion of which SSP and how to set them up is ongoing.
There is actually a wider review of SSPs happening at the moment. If this is of interest to you and you're not already involved, I'd encourage you to reach out to those who are working on it. This website (https://data.ece.iiasa.ac.at/ssp/#/about) is probably as good a starting point as any, but if that gets you nowhere feel free to send me an email and I'll see who I can connect you with.
Citation: https://doi.org/10.5194/gmd-2023-176-CC8
-
CC8: 'Reply on CC7', Zebedee R. Nicholls, 15 Nov 2023
reply
Malte Meinshausen et al.
Malte Meinshausen et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
3,541 | 1,841 | 38 | 5,420 | 17 | 26 |
- HTML: 3,541
- PDF: 1,841
- XML: 38
- Total: 5,420
- BibTeX: 17
- EndNote: 26
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1