the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Jan 2026
DeepMIP-Eocene-p2: Experimental design for Phase 2 of the early Eocene component of the the CMIP7/PMIP7 Deep-time Model Intercomparison Project (DeepMIP-Eocene)
Abstract. Warm, high-CO2 climates of Earth's past provide an opportunity to evaluate climate models under extreme forcing, and to explore mechanisms that lead to such warmth. One such time period is the early Eocene, when global mean surface temperatures were 10–17 °C higher than preindustrial, and CO2 concentrations were ~1500 ppmv. In this paper we present the experimental design for Phase 2 of the Eocene component of the Deep-time Model Intercomparison project (DeepMIP-Eocene-p2). The aim is to provide a framework for modelling groups to carry out a common set of simulations, thereby facilitating exploration of inter-model dependencies. The focus is on the early Eocene Climatic Optimum (EECO, ~53.3–49.1 million years ago). Relative to Phase 1 of DeepMIP-Eocene, we provide a new paleogeography (topography, bathymetry) derived from several recent independent reconstructions that focused on different regions, a new vegetation derived by merging paleobotanical data with vegetation model simulations, and a new CO2 specification derived from recent reevaluations of proxy data. The core set of simulations consists of a preindustrial control, an abrupt increase to 4x preindustrial CO2 concentrations from this preindustrial control, a standard control EECO simulation at 5x preindustrial CO2 concentrations, and an EECO simulation with preindustrial CO2 concentrations. In addition to these core simulations, we suggest a suite of optional sensitivity studies, which allow the impact of various factors to be explored, such as topography/bathymetry, greenhouse gases, land-surface parameters, astronomical and solar forcings, and internal model parameters. The updated boundary conditions and guidance on initialisation and spinup in Phase 2 will allow more robust model-data comparisons, more accurate insights into mechanisms influencing early Eocene climate, and increased relevance for informing future climate change projections.
- Preprint
(3621 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 25 Mar 2026)
-
RC1: 'Review of egusphere-2025-6135 by Chris Brierley', Chris Brierley, 17 Feb 2026
reply
The authors present a very nice experimental protocol that is well-justified. I look forward to its publication, but have I couple of suggestions.Paleogeography - I am generally happy with both the field and your explanation of it. However, I note that there appears both an isolated very high peak as well as a 'lake' at the join between Greenland and (what I think is) Scandavia. This feels unrealistic to me. Can you either smooth them out if that's the case, or explicitly discuss the geological evidence for them?Figure 4. Can you please harmonise the shape and color between the panels? It is the ice cover that switches with savannah that is most confusing.ESGF. On L386 you state that participants should upload their simulation to the DeepMIP archive. Is there a reason you do not also allow the possibility of uploading data created using CMIP7 models onto the ESGF? [Now is the time to establish the relevant controlled vocabulary, and you are within the CMIP7 special issue]. This would have the advantage of combining the Eocene with the other PMIP experiments.Experiment name: Are you sure that you have selected the most helpful experiment names?
- The precedent from the wider PMIP and CMIP efforts does not include phase numbering. Bear in mind that all phase 2 data is stored within a specific directory (L391).
- No period is included in the name. This is especially pertinent, as I understood the Miocene is also included in DeepMIP. Since you focus only on the EECO, why not choose either 'eocene' or 'EECO' instead of deepmip
- I find the inclusion of 'stand' to be unhelpful. I can understand defining subsidiary experiments, with a 'sens' flag. Surely this implicitly assumes that other than the named feature being changed, everything else is as set as standard - inconsistent with the requirement of a 'stand' flag.
- If you do choose to allow ESGF inclusion, your main experiment name will be the longest around (and will not make sense to people not involved in deepMIP).
Please be aware that equilibrium-4xCO2 could be exactly the simulation (but at a different point) as abrupt-4xCO2. How is CMIP7 DECK treating this possibility?Sect 3. You state this section includes 'plans for analysis', but it doesn't really. Either expand on this a little, or perhaps just removed them from the section title.Sect 2.3.4: Can you please shout out to the relevant sensitivity experiments listed in Table 1L148: can you please move 'red' to before 'line'? This would fit the same structure as description of the other lines in the figure.L154: can you clarify the word 'records'? I believe that you mean that two compilations ingest multiple of the same individual readings. But it could mean that the error bars on the two timeseries overlap.e.g. L178, L255: there are a couple of instances of the wrong \cite command, leaving the bracket in the after, rather than before the authors.L239: Consider removing `field of'L241: I appreciate you stating that Herold et al provide river routing. Can you comment of this fields relevance, given the different topography?L263: add CO2 after 6xPIL294: "in, that" -> "in and that"Sect 2.4: I am happy with the explanation of all the various methods. Can you please reiterate that whatever approach is selected should be documented in the simulation publication?L402: Can you please spell out that 'std' stands for 'standard deviation'L402 and L404: 'timeseries' seems to be doubly defined, and I don't understand the distinction.L407: You ask for temperatures at 3500m depth. Can you clarify if you just want the layer containing 3500m, or to vertically interpolating the full profile to get the value at 3500m for intercomparison?ReplyCitation: https://doi.org/10.5194/egusphere-2025-6135-RC1 -
RC2: 'Comment on egusphere-2025-6135', Catherine Bradshaw, 07 Mar 2026
reply
General Comments
This manuscript presents the experimental design for Phase 2 of the DeepMIP–Eocene initiative, outlining updated boundary conditions, simulation protocols, and scientific motivations for investigating early Eocene climate. Overall, the paper is clearly written, well‑structured, and highly valuable to the paleoclimate modelling community. Justification for focusing solely on the EECO is strong, the boundary condition updates are explained in a logical and transparent manner, and the connection to Phase 1 findings is thoughtfully presented. The manuscript also provides a clear link between deep‑time model intercomparison and broader climate questions such as equilibrium climate sensitivity and the evaluation of model behaviour under high‑CO₂ states.
While the paper is comprehensive, there are areas where clarity could be improved or where additional elaboration might benefit readers, as described below.
Specific CommentsIntroduction
L20: The Mallory reference is perhaps belittling the importance of learning about processes and mechanisms that led to things being the way they were/are even if they are not so directly relevant for future climate change. I suggest removing.
L76-78: Do we have any idea why there are such differences in CO2 estimate for the best agreement with the data?
L97-98: Can you perhaps infer that this may be the cause of some disagreement between models?
L98: Over what period were the simulations run but still did not reach equilibrium?
L63-92: I wonder if this section could be reorganised into consistencies between models and data, consistencies between models but disagreement with data, inconsistencies between models (some of which may agree with the data), and inconsistencies between models and data so that is clearer what the main certainties and uncertainties are?Figure 1: I presume the divergence in the colour scheme around the EOT is to highlight the shift from the greenhouse world to the icehouse world - might be worth adding those labels to the figure to explain why that was chosen as the divergence point.
Experimental Design
L122: Include Eocene/EECO in the simulation naming to avoid confusion with the other DeepMIP simulations. I believe MioMIP is using Deepmip-Mio, perhaps here therefore they should be Deepmip-Eo or Deepmip-EECO?
L254-244: Ambiguous as to whether you are suggesting that this runoff routing field could be used for Phase 2 or not
L270: What criteria is used to determine if regions have "sufficient" data coverage? A comment on the vegetation data availability for this period would also be useful
L330-331: Is the best-fit GCM CESM1.2 for Thomspon in BIOME4 the same as the best-fit GCM CESM1.2 for Brugger in LPJ-GUESS? If so, worth commenting on that.
L336-338: Are the discrepancies between the BIOME4 version of the vegetation and the data because these are where the model has been corrected towards the data in the LPJ-GUESS version of the vegetation? i.e. because the BIOME4 version has not undergone the hybrid proxy reconstruction correction step? Or are these different regions?
L362-363: You suggest that groups can use initial conditions from the Phase 1 model database, but in L98 you state that many models did not reach full equilibrium but the end of their simulations. Are there some models in the Phase 1 data that you would therefore recommend are NOT used for the initial conditions for Phase 2? Similarly, are there any in the database that suffered from crashing early or running away that should NOT be used? Or are these not in the database?
Figure 4: It is very difficult to see the details of the legends for (a) and (b), and the distinction between some of the greens is so small it is nearly impossible to discern which is which on the map. It also appears that the same colour has been used for Ice in (a) as Savannah in (b) and (c). Can you try to use a more distinctive colour scheme?Figure 5: Similar comments on the colour scheme to Figure 4.
Summary and Plans for analysis
L416: This section is very short and the title is misleading. Perhaps Section 2.5 might fit better here. Maybe Summary and Next Steps or similar might be a better descriptor?
L424: Are there plans to version control the proxy database so it is clear which dataset is being used for model-data comparisons and to ensure consistency between the analyses by different groups?Figure A1. The colour scheme used may prove difficult for some colour blindness conditions.
Technical CorrectionsAbstract
L1-2: No commas required
L2: Define when the early Eocene is
L8: Word missing, a new vegetation... scheme/distribution/boundary condition
L10-12: Pluralisation of concentrations seems a little odd in the context of the model simulations where preindustrial CO2 is a constant and 4 x preindustrial CO2 is therefore also a constant.Introduction
L38: Word missing, we/DeepMIP extend...
L38: Comma after Quaternary not required
L46: Word missing, also indicate
L47: No comma required before 'and a vegetation consistent'
L47: Duplication of wording, perhaps change 'a generally warmer and wetter climate than modern' to something like 'those warmer and wetter conditions'?
L53: Is Herold the only reference for all these differences? Or do you mean to say Herold and references therein?
L55: No comma required before 'and has been used'
L61: No comma required before 'and have been discussed'
L70: Pluralisation of concentrations again
L73: Unnecessary 'and' before 'drier conditions'Experimental Design
L120: Maybe introduce PI=preindustrial when talking about CO2 concentrations earlier in the manuscript
L157 & 161: Inconsistent wording, runaway versus 'runs away'. Maybe also explain what this term means.
L173: Typo, Appemdix
L200: Wrong wording?, should be 'or between' rather than 'as well as between'
L205-206: Incorrect wording, 'As the paleogeographic reconstructions used here all have modified' should be 'As all the paleogeographic reconstructions used here have modified'
L222: Commas needed around the word 'particularly'
L227: No comma needed after the word 'resolved'
L286-289: The long sentence is a little confusing. Perhaps reword to something like 'The recommendations for all other boundary conditions remains unchanged from Phase 1 (refer to Lunt et al. 2017). Using the section numbering from Lunt et al. (2017), detailed recommendations are given for soils and lakes in Section 4.2.2,' etcCitation: https://doi.org/10.5194/egusphere-2025-6135-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 421 | 217 | 22 | 660 | 15 | 23 |
- HTML: 421
- PDF: 217
- XML: 22
- Total: 660
- BibTeX: 15
- EndNote: 23
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1