the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 Feb 2025
The Detection and Attribution Model Intercomparison Project (DAMIP v2.0) contribution to CMIP7
Abstract. The first version of the Detection and Attribution Model Intercomparison Project (DAMIP v1.0) coordinated key simulations exploring the role of individual forcings in past, current and future climate as part of the Coupled Model Intercomparison Project, Phase 6 (CMIP6). The simulations have been used extensively in the literature for detection and attribution of long-term changes, constraining projections of climate change, extreme event attribution, and understanding drivers of past and future simulated climate changes. Attribution studies using DAMIP v1.0 simulations underpinned prominent assessments of human-induced warming in the Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment Report. Here we describe the set of DAMIP v2.0 simulations, proposed for the next phase of CMIP, CMIP7. Detection and attribution studies rely on preindustrial control simulations and historical simulations which will be part of the Diagnostic, Evaluation and Characterization of Klima (DECK) set of simulations for CMIP7. In addition, DAMIP v2.0 identifies three highest priority single forcing experiments for CMIP7 to be run as Fast Track simulations in support of the Seventh Assessment Report of the IPCC – namely simulations with natural forcings only, anthropogenic well-mixed greenhouse gases only, and anthropogenic aerosols only. Beyond this, the DAMIP v2.0 experimental design includes full column ozone-only simulations and land-use-only simulations, such that the set of individual forcings experiments, when considered together, represents the full set of historical forcings. While concentration driven simulations are prioritized for attribution of past changes, emissions-driven versions of the DAMIP experiments are also proposed to support understanding of the influence of carbon-cycle feedbacks on the simulated responses to individual forcings.
- Preprint
(885 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-4086', Anonymous Referee #1, 02 Mar 2025
The manuscript describes the next phase of DAMIP (v2.0) as part of the CMIP7 contribution. The DAMIP suite of simulations make detection and attribution studies possible. The updates to the DAMIP since the predecessor (v1.0) from the CMIP6 cycle are substantial enough to warrant the new manuscript describing the new suite of simulations.
This manuscript is well written and is substantially ready for publication. However, there are improvements to the manuscript that can substantially improve the clarity and accessibility of the manuscript.These are minor in nature and the manuscript can be accepted for publication after these minor edits.
The following are suggested changes:
1. Line 83: The word "times" to appear after preindustrial.
2. Line 132: The concept of additivity should be introduces/explained more as it is not common knowledge.
3. Lines 134-137: This should be rephrased for clarity.
4. Lines 186-189: This should be rephrased for clarity.
5. Lines 241-245: Shorter sentences should be used for clarity.
6. Lines 404-406: Rephrase. Replace "not been completely true" with something along the lines of "there are differences".
7. Line 865: The figure 2 does not include any representation of the simulations with Medium scenario forcings up to 2035. The figure should be revisited as it is not obvious what the different boxes are (upper/lower as well as boxes with no connects to other boxes) The yellow boxes could be read in two ways - CMIP6 or Tier2. So better clarity is needed.Citation: https://doi.org/10.5194/egusphere-2024-4086-RC1 -
AC1: 'Reply on RC1', Nathan Gillett, 17 Mar 2025
Thanks to the reviewer for the positive review of the manuscript. We have responded to all the requests for changes and clarification of the text. The reviewer's comments are shown in bold below, and our responses are shown in plain text.
The following are suggested changes:
1. Line 83: The word "times" to appear after preindustrial.
Suggested change made.
2. Line 132: The concept of additivity should be introduces/explained more as it is not common knowledge.An example to explain the meaning of testing additivity has been added “For example, aerosol-only simulations and all-but-aerosol simulations can be used together with simulations including all forcings to test whether the response to aerosols and the response to other forcings add to give the response to all forcings combined (e.g. Simpson et al., 2023).”
3. Lines 134-137: This should be rephrased for clarity.
This sentence was complicated by the inclusion of an unnecessary condition concerning additivity. This has now been removed and the text has been modified to improve clarity. The sentence now reads “However, if the objective of an analysis is to characterize the response to one particular forcing, then individual forcing simulations will lead to reduced sampling uncertainties, because they do not require a difference between two sets of simulations to be taken. each of which has its own sampling uncertainties.”
4. Lines 186-189: This should be rephrased for clarity.This sentence was separated into two shorter sentences and modified to increase readability and improve clarity. This text now reads “hist-O3 simulations should not be carried out using models with interactive gas-phase chemistry. This is because ozone is simulated interactively in response to changes in ozone depleting substances, methane and other species in such models, and the concentrations of ozone depleting substances, methane and other species do not change in these hist-O3 simulations.”
5. Lines 241-245: Shorter sentences should be used for clarity.We have separated this text into shorter sentences and made other changes to improve clarity and readability. This text now reads “While much of the time evolution of biomass burning emissions has occurred as a result of human activity, the historical simulation includes observed year-to-year variations in biomass burning partly driven by natural variability. However, it is not easy to separate human-induced changes in biomass burning from naturally-induced changes. Therefore we request that modelling centres specify constant biomass burning emissions as in the piControl in this hist-nat simulation.”
6. Lines 404-406: Rephrase. Replace "not been completely true" with something along the lines of "there are differences".‘this has not been completely true’ replaced with ‘there have been some differences’.
7. Line 865: The figure 2 does not include any representation of the simulations with Medium scenario forcings up to 2035. The figure should be revisited as it is not obvious what the different boxes are (upper/lower as well as boxes with no connects to other boxes). The yellow boxes could be read in two ways - CMIP6 or Tier2. So better clarity is needed.The figure has been revised to state explicitly that the historical simulations are extended to 2035 using the Medium scenario. Text has also been added to caption to clarify this, as well as the meaning of the different boxes:
‘Blue boxes are Tier 1 experiments, yellow boxes are Tier 2 experiments, green boxes are Tier 3 experiments, and white boxes contain descriptive text. All historical simulations except historical-CMIP6 run from 1850 to 2035 using the Medium scenario forcings from 2022, while historical-CMIP6 runs from 1850 to 2035 using the SSP2-4.5 scenario from 2015. The historical or esm-historical simulation shown in the top row uses CMIP7 historical forcings which can be decomposed into the sets of forcings used in each of the simulations in the second row. Three of these simulations are extended using the Medium scenario from 2036 to 2100 as shown in the third row. The fourth row depicts two additional experiments that are complementary to those in the second row.’
Citation: https://doi.org/10.5194/egusphere-2024-4086-AC1
-
AC1: 'Reply on RC1', Nathan Gillett, 17 Mar 2025
-
CC1: 'Comment on egusphere-2024-4086', Gareth S. Jones, 25 Mar 2025
The detection and attribution model intercomparison project is an important part of CMIP6 and data from their experiments have been used in hundreds of studies.
I have some comments, that I hope the authors of this important proposal will consider.
+ Encouragement of model involvement (Lines 71-73, 498-499)
I really like Figure 1 that shows how many models produced DAMIP experiments for CMIP6 and how many studies used them.
I estimate that for Chapter 10 in IPCC 2013 they used 44 models with historical simulations, and 16 with historicalNat experiments. But for Chapter 3 in IPCC 2021 the numbers were 59 models with historical, and 14 models with hist-nat experiments. This is a bit disappointing, I hope we can do better for CMIP7.
The experiments are extremely useful for understanding the responses within individual models, something institutions should be interested in.
Can there be more outreach to encourage more institutions involvement, which will benefit their own model development, as well as be useful for DAMIP.
+ Design of hist-nat (Lines 238-247, 252-254)
I am surprised to see the design for hist-nat to not include stratospheric ozone changes due to influences of solar and volcanic factors, and put those forcing factors in the hist-O3 with the anthropogenic O3 factors.
I think this is a major step backwards. It will no longer be possible to compare anthropogenic only and natural only simulated climate, without adding even more caveats than we do already.
I would be reassured if the authors could demonstrate that the natural O3 changes in the stratosphere have ignorable influence on climate in the simulations, but I fear that is not the case for all diagnostics of interest, especially in the stratosphere (Shindell et al, 2013).
As it currently stands the design does not "allow the attribution of observed climate change to natural, greenhouse gas and other anthropogenic forcings" (Line 253).
+ Usefulness of interactive CO2 simulations in D&A (Lines 145-159, 326-345)
Could the authors give some examples how interactive CO2 experiments, with emissions of CO2 rather than prescribed CO2, be useful in D&A studies. I see how such experiments can be very interesting and useful in other fields.
I have had conversations with several proponents of the experiments, but I remain unconvinced of their use in D&A studies.
A main assumption in D&A is linearity of responses when combining forcings. Has that been demonstrated in interactive CO2 experiments?The example shown in Figure 3 is claimed to show the concentrations of CO2 in CO2 emissions simulations are "relatively close" to that in prescribed concentration simulations. I think that is debatable in the shown model. Also other models give even wider discrepancies between the concentrations in the two set ups (Figure 7 in Sanderson et al, 2024), with equivalently wider surface temperature responses.
This would mean adding a radiative forcing uncertainty to one of the few forcing factors with low uncertainties.
To avoid confusion, these experiments should be separated from DAMIP.
+ Extension from 2021 to 2035 (Lines 222, 234-236)
Is the proposed extension of 14 years that helpful? This is much longer than we did for CMIP6. Do we have a sense of how much the SSP2-4.5 forcings differed with reality up to present day? That would provide some evidence to support the proposal.
The CMIP7 forcings task team, are debating about how to update the historical forcings year on year. If they succeed this might mitigate somewhat the requirement to extend to 2035. Has there been any discussion with that team?
+ Variant ID use suggestion (Lines 201, 424-429)
It might be complicated to use the "f" value in the variant ID to indicate specific forcing set ups across all models, in the way suggested. Some institutions have used that to indicate their unique forcing set ups (e.g., HadGEM3, UKESM, GISS-E2, CAMS-CSM1-0) in CMIP6, and the CMIP7 forcings task team have discussed recommending models use the "f" value to be associated with the input forcings version number. Has this proposal been discussed with other MIPs and the CMIP7 forcing team?
+ Additivity/linearity of 5 experiments (Lines 231-232)
The recommended hist-GHG, hist-nat, hist-aer, hist-lu and hist-O3 experiments are chosen as they are hoped that summed up they are equal to the historical. It would be helpful to say that with everything else being equal that the variance of the climate response of such a sum would be 5 times greater than the variance of the climate response of historical, due to internal variability. For a multi-model mean this might seem less important, but for examining responses for individual models it could make it more difficult to interpret results.
+ Tuning to the historical record
This is a bit of a controversial subject, but it would be helpful to mention the issue that some models have included the historical observed temperature changes - in one way or another - in their model development cycle (Hourdin et al, 2017). Some colleagues have suggested that the DAMIP experiments are then more important, as single forcing experiments are less likely to be impacted by any circular reasoning.
+ Specific comments:L73-75: I am curious. How are experiments added to DAMIP?
L170, L330, L335, and elsewhere: It would be helpful to give section numbers for IPCC references when talking about something very specific. It can be difficult to find what is being referred to.
L185-189: I presume the authors mean the experiment name "esm-hist-GHG" and not "hist-GHG" here.
L234: "... from 1850 to 2035 using ...", should be "... from 1850 to 2021 using ..."
L280-281: There are difficulties even if not specifying stratospheric ozone in isolation. A prescribed troposphere and stratosphere ozone can still cause issues as the tropopause in the model may not correspond to the input O3, causing unwanted effects (Hardiman et al. (2019)).
L333-335 What does this refer to? All I can find in the IPCC 2021 SPM that this could refer to is figure SPM.2. But that compares the attribution assessment from chapter 3, using concentrations of CO2 (and other forcings) in different models and approaches, with the outputs from simple climate models (chapter 7) driven by different radiative forcing changes, where the CO2 concentrations are constrained to match historic CO2 concentration changes (7.SM.2.2).
L347-350: While I disagree with the hist-nat design, it should be made clear if the esm-hist-nat has the similar no O3 changes from solar and volcanic implemented.
L419-421 It would be helpful to show how much estimated actual radiative forcings has diverged from ssp2-4.5 over the last decade or so, to give a sense of the expected future uncertainty.
Figure 2 (page 30) Tiers (1 to 3) are shown here, but "Tiers" for the CMIP7 are not mentioned in the main text.
ReferencesS.C. Hardiman et al., 2019, The Impact of Prescribed Ozone in Climate Projections Run With HadGEM3-GC3.1, JAMES
F. Hourdin et al., 2017, "The art and science of climate model tuning", BAMS
B.M. Sanderson et al., 2024, "The need for carbon-emissions-driven climate projections in CMIP7", GMD
D.T. Shindell et al., 2013, "Interactive ozone and methane chemistry in GISS-E2 historical and future climate simulations", ACP
Citation: https://doi.org/10.5194/egusphere-2024-4086-CC1 -
RC2: 'Comment on egusphere-2024-4086', Michael Wehner, 04 Apr 2025
One of the biggest criticisms of the CMIP project is that there are too many subprojects and a tendency to be “all things for all people”. Given human and hardware resource limitations, this puts tremendous pressure and stress on the climate modeling community. That being said, the DAMIP and SCENARIOMIP projects are among the most important components of the CMIP due to their relevance to policy and decision makers.
The paper is well written and explains the DAMIP project thoroughly. I do, however, have some strong opinions and I will use this venue to express them. My hope is that the authors will consider them.
First, the three Fast Track Tier 1 experiments, natural forcings only, anthropogenic well-mixed greenhouse gases only, and anthropogenic aerosols only are far more important than the Tier 2 and 3 experiments and modeling groups should strongly consider devoting their limited resources to the Tier 1 experiments.
Second, the Large Ensemble Single Forcing Model Intercomparison Project (LESFMIP) offers many new analysis opportunities. This is particularly true for attribution studies, where simple statistical comparisons of histograms from the different experiments can be insightful. Linear regression attribution algorithms have always been difficult to explain to those not directly involved in attribution while histogram comparisons offer simpler visual explanations and can make the case more strongly for the human influence, if any, on the climate system topics being considered. I recommend that the paper more strongly endorse large ensembles, especially for the Tier 1 experiments.
Third, no connection to HighResMIP2 is made in the paper. It has become quite clear that simulation of most types of extreme weather event requires far higher resolution than the standard resolution of CMIP6. While the HighResMIP1 organizers were focused on tropical cyclone, the improvement in simulation of extreme weather events is not limited to this one storm type. At horizontal resolutions of ~25km the simulated gradients of both temperature and moisture are far sharper and more realistic than at 100km or coarser. This is a critical feature in simulating intense storms and revealing super Clausius-Clapeyron scaling, if any, in changes in extreme precipitation. Better resolution of mountains also aids in simulating more realistic heatwaves and even blocking events can be improved in some regions.
My suggestion is that the authors consider my comments and make very minor revisions to accommodate them if they are so inclined.
Citation: https://doi.org/10.5194/egusphere-2024-4086-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 288 | 49 | 9 | 346 | 7 | 9 |
- HTML: 288
- PDF: 49
- XML: 9
- Total: 346
- BibTeX: 7
- EndNote: 9
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 117 | 33 |
| China | 2 | 43 | 12 |
| United Kingdom | 3 | 31 | 8 |
| Germany | 4 | 18 | 5 |
| Japan | 5 | 18 | 5 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 117