Last time, we had SSP naming in the scenarios but we should recognize that climate models are first of all dependent on greenhouse gas forcing and not so on the underlying socio-economic conditions. The latter do impact land-use (and thus biophysical forcing) and aerosols forcing. . But for most forcing levels, these are globally of secondary importance compared to the forcing from well mixed greenhouse gases. The SSP/RCP matrix in fact encourages users to use climate outcomes for different SSPs. An exception may be on the low side of the range of radiative forcing, where land use and air pollution emissions may be mportant too (research is still needed to determine this in the context of CMIP-type ensemble experiments). If we can more clearly decouple the climate futures from socioeconomic development, successive impact analysis may consider how their non-univocal couplings determine the end result (impact/risk). In the strawman proposal, in principle we want to create scenarios that are relevant to finding different climate futures, so the range of radiative forcing is our main concern, then lin parallel the IAM community can think how these levels of forcings can be the results of different socio-economic futures. However, there will always be underlying socio-economic assumption for the specific scenario that is chosen to be run by ESMs, and, at the moment, they are being updated. There is a process ongoing to update population, GDP, energy and land use of GDPs of the SSPs.
ScenarioMIP wants to run plausible scenarios and therefore wants to run the highest plausible scenario , as the current proposal includes (still to be determined how high that will be, and we are expecting it to be determined by a convergence of trends that would characterize it as low likelihood). Tipping points are a question for the climate modelling community - current simulations do not represent tipping points. By providing a high scenario (also with a potential extension beyond 2100) we will enable the modelling community to explore tipping points but we cannot guarantee that the models will simulate them.
In our current proposal we indicate we would like at least two scenarios relevant to 1.5C and also one for well below 2C. Especially the latter might be fairly close to end of century net-zero in total greenhouse gases (the other 2 are likely to be earlier at net-zero). We are also considering another scenario not on the proposal, which would branch off from the middle scenario relatively late.
Of course, the analysis that these runs will support can be focused anywhere along this timeline. The ScenarioMIP simulations will produce climate model output covering the entire 21st century and beyond in some cases, and the analysis can then be focused on the 2040s or 2050s as desired.
We do not think of these as two competing approaches. Along the simulations run under these scenarios there will be times when the various temperature levels will be reached, and model output from those times can be used to calculate impacts at those particular Global Warming Levels (GWL). By running scenarios as proposed we produce plausible trajectories to get to those GWL.
We are mostly considering CO2 when discussing the emissions-driven mode. For some other gases this may be more complex as models need to represent the atmospheric chemistry that produces these gases from precursors and this needs to be discussed further. This said, all forcings will also be made available as atmospheric concentrations, so that the possibility of running ScenarioMIP experiments in concentrations-driven mode will be available as an option.
We all have our own opinions on what is politically realistic. We want to concentrate on what is technically plausible and had a discussion on whether 1.5C is still possible. When these scenarios come out in around AR7 around 2028-2030 as part of CMIP7 we will have lost a period of climate action so for CO2 we will be at a higher concentration than some CMIP6 scenarios. In Reading, weconcluded that the possibility to strengthen methane mitigation could still keep 1.5C scenarios plausible, certainly in combination with the fact that these scenarios are very policy relevant to Paris Agreement. Question is now to the IAM community, can they provide scenarios with relative late start to climate action but then with very stringent CO2 and CH4 policy to remain below 1.5C as the most ambitious scenario. We would then contrast that with the scenario that is slower via negative emissions and come back to 1.5C. The low scenario task force will be examining these questions.
This has also been part of discussions and as far as we can this is a logical additional assumption to add to the scenarios. The world probably does not get much better if it implements strict climate policy but misses many other SDGs, e.g. through extensive use of land-based solutions without concerns of food security. So, considering SDGs may have consequences for land-based assumptions we make, and we need to determine how to build this in.
By running in emissions mode we already test the feasibility as run in both the IAM and then in the ESMs. On land-based mitigation, many IAMs in response to the Paris Agreement have started to produce scenarios which are quite aggressive on land use mitigation. In general, models have started to be a bit more cautious and look at other ways to reach the low emission goals. Part of idea of ScenarioMIP low scenario is what is the lowest plausible scenario, and this will be discussed in the Low Scenario task force and then discussed further once the IAM runs are initiated.
In the research of the mitigation and impacts community, assumptions of socio-economic development and shared policy assumptions (SPAs) are critical. We know the costs of reaching certain targets depends strongly on socio-economic assumptions and the climate damage depends on ability of society to adapt to climate change. On the climate modelling side this is less important theme, the most dominant forcing is GHGs with some impact of land use and aerosols that appear noisy and non-robust across different ESMs. In CMIP6 we decided to decouple so the ESMs did not have to run the full matrix of options of SSP and RCPs - we want to maintain this assumption. The use of specific SSPs and SPAs will return to be critical when ESM runs are used in the IAM and impacts and vulnerability research community. In CMIP6, we should not have named the scenarios with the SSPx.y as we perpetuated the idea that the SSP used to produce the specific pathway of emissions and land-use forcings by the IAMs was integral to the climate model outcomes and that was not the case - but it is not feasible to ask a climate model to run the same radiative forcing scenario developed by different SSPs and we had to choose one. Assumption is that it is not critical to climate model outcome and only becomes critical when analysing impacts and adaptation outcomes. The point of SSP-RCP matrix is to integrate SSPs and RCPs “a posteriori” and that remains true, we should try this time around to better avoid giving the impression that you can integrate only in a very restricted way and try to promote the assumption that there is flexibility in the coupling.
In the proposal and timeline presented there are windows of community engagement. This may involve discussions over the socio-economic assumptions but this will not be the major focus.
There will be many climate models/ESMs running these scenarios. On the IAM side, more than likely that many teams will also be involved, with one selected and proposed to be run by the climate models.
Practically, that will be the case in emissions-driven mode - different ESMs will produce different CO2 concentrations and different warming outcomes. In emissions-driven mode there will be no need to provide a concentration range. However, we will also be providing concentrations for those models to be able to run in concentration mode and here we would aim for a best estimate rather than range of concentrations to control one element of uncertainty. The idea of providing a concentration range is well motivated but goes against the cost of running these experiments through ESMs.
Emissions driven is likely more true to the real uncertainty than how it was done in CMIP6 and wider uncertainty may be seen as an advantage. In the impacts community it may present challenges and they will need to think how to handle additional uncertainty and will also cause issues with naming as the actual radiative forcing/warming may come out much higher or lower than the scenario was meant to be, when run in emissions driven mode. We feel as if that outcome is a better reflection of where our science (and the uncertainties) are. In CMIP6 there was already a range of temperature outcomes and although will be larger in emissions-driven context we have faced this before.
No, there are not.
ScenarioMIP have asked the IAM groups to explore extending their modelling to 2125. Still unclear if feasible but that is the ask.
There are three main goals for ScenarioMIP: 1) contribution to climate science, 2) contribution to other science communities and 3) policy support. The latter two goals imply additional criteria regarding plausibility, a logical combination of forcing elements and the logical evolution over time. For the climate impact community, it is critical to have a useful set of climate model outcomes. Earlier, the different goals of ScenarioMIP has been presented as a possible trade-off (as also suggested in the question) - but in reality, the ScenarioMIP scenarios are also the most popular for climate research.
In the near term there is a process under way to update the current SSPs in terms of their population, education, GDP, and other assumptions that are inputs to the IAMs. In the longer term it would be possible to consider larger revisions to the SSPs. This process was discussed at the last Scenarios Forum held in 2022, including the degrowth topic. There is no formal process now, but research groups are encouraged to experiment with revised SSPs, which may then be adopted more broadly. Further, the climate outcomes can in most cases be combined with a range of socio-economic assumptions.
In CMIP6 scenarios were extended from 2100. We will also here use stylized extensions. The question is just at what year to start the stylised extensions. Given the fact that we are further in the future now, it seems logical to see whether the period of IAM model runs can be prolonged.
The ScenarioMIP workshop report gives an optimistic timeline for which ESM output could possibly feed into the Global Stocktake in 2028. However, they also acknowledge this timeline is very tight and alternatives such as using CMIP6 output and emulators may be considered. In terms of the decadal prediction runs, our understanding is that they plan to run updated predictions with their current models/systems.
As always, modelling centres are free to choose to run what they deem to be the most appropriate.
ScenarioMIP has stated they will provide both emissions and concentrations driven scenarios but will promote emissions driven.
Yes, for CMIP7 there are plans to extend simulations to 2125, with some potentially going out to 2300.
The full archive of CMIP6 data can be accessed on ESGF nodes.
The full archive of CMIP6 data can be accessed on ESGF nodes in NetCDF format. Google Cloud and AWS only have a subset of the CMIP6 data, so we encourage you to always check ESGF to find all the CMIP6 datasets in the same place. Additionally, ESGF is linked to the Errata service, which allows users to see logged issues with datasets and updated versions.