the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Feb 2023
Exploring Holocene temperature trends and a potential summer bias in simulations and reconstructions
Abstract. Proxy-based reconstructions and climate model simulations of surface temperature trends during the Holocene disagree: While reconstructions show a cooling during the mid- and late Holocene, climate models show a continuous warming – a contradiction known as the Holocene temperature conundrum. Despite extensive research, the reason for the disagreement remains unclear. Both, missing processes in the models as well as biases in the proxies and the resulting reconstructions are possible sources of the conundrum. Here we compare our TransEBM v1.2 climate simulation as well as additional climate models of different complexity and Holocene temperature trends from the Temperature12k dataset (Kaufman et al., 2020b), with regards to model-data and model-model agreement. We show that models of all complexities disagree with mid-Holocene temperature trends in reconstructions and that this disagreement is almost independent of proxy and archive type. While, models show the highest agreement with summer temperature trends in reconstructions, our study shows that a trivial summer bias in proxies is not sufficient to explain the conundrum. Further effort to disentangle seasonal biases in proxies and the testing of potential misrepresentations in climate models, like anthropogenic land-use, in form of sensitivity experiments are needed to resolve the Holocene conundrum.
- Preprint
(6543 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2023-86', Anonymous Referee #1, 13 Mar 2023
The manuscript compares the temperature trends simulated by climate models of various complexities with paleo data over the Holocene. Many explanations have been suggested to explain the disagreements between models and data, the authors addressing specifically the impact of potential seasonal biases. They describe in details the seasonal and spatial distribution of the trends in the selected models and in data. This description is very clear. The paper is well written and easy to follow. I thus have no minor comment or suggestion to improve the presentation of the manuscript. However, there are two major points to consider in a revised version of the text.
1/ The added value of the study is not clearly explained and the authors should insist more on this in the conclusion, which is very short in the current version of the manuscript. The first paragraph of the conclusion summarizes the description of the trends presented in the previous sections. The second (and final) paragraph starts by a quite mild sentence: ‘Regarding the Holocene conundrum, it follows that a simple seasonal proxy bias is unlikely as a full explanation’ and then present some general suggestions for improvements or new studies. The fact that seasonal proxy biases might play a role but could not explain the full model-data disagreement is already around for some time (see the recent review of Kaufmann and Broadman, 2023) and the authors should explain more clearly the new contribution they bring to the debate.
2/ The authors analyze relatively old simulations that have been discussed in several studies. The selected data base has also already been used in model-data comparisons. A new simulation is included (TransEBM) but it has in general a lower agreement than the other ones with observations (see for instance Figure 5). This new simulation might be helpful to understand some of the characteristics of the other models but this is not developed in the current version of the manuscript. Furthermore, the set of selected experiments is not designed to test hypotheses, such as the potential role of vegetation or of the volcanic forcing for instance, as done in some other studies. Several transient Holocene have been performed recently. Some only cover parts of the Holocene or might not be publicly available but a larger set of experiments would provide additional information for the discussion (see for instance Gravgaard Askjær et al. 2022, in particular their Figure 3).
References:
Gravgaard Askjær T., Q.Zhang, F. Schenk, F.Charpentier Ljungqvist, Z. Lu, C. M. Brierley, P. O. Hopcroft, J. Jungclaus, X. Shi, G. Lohmann, W. Sun, J. Liu, P. Braconnot, B.L. Otto-Bliesner , Z. Wu, Q. Yin, Y. Kang, H. Yang, 2022. Multi-centennial Holocene climate variability in proxy records and transient model simulations. Quat. Sci. Rev. https://doi.org/10.1016/j.quascirev.2022.107801
Kaufman D.S. and E. Broadman, 2023. Revisiting the Holocene global temperature conundrum. Nature 614, 425-435 . https://doi.org/10.1038/s41586-022-05536-w
Citation: https://doi.org/10.5194/egusphere-2023-86-RC1 - AC1: 'Reply on RC1', Christian Wirths, 17 Jun 2023
-
RC2: 'Comment on egusphere-2023-86', Anonymous Referee #2, 22 Mar 2023
Summary
Wirths et al. summarise a comparison of simulations and reconstructions of temperature trends over three phases of the Holocene. They confirm important differences between the warming seen in models and the cooling reconstructed in temp12k. They conclude that a simple seasonal bias in the temp12k proxy records is unlikely to be able to explain the differences with the model runs.
Main comments:
A main assumption within this work is that the seasonal and annual reconstructions extracted from temp12k are able to separately resolve the seasons. I am no expert in this dataset, but this seems unlikely to be the case. In fact if we can take the separate seasonal records at face value then we can immediately conclude that the seasonal bias is not the answer to the ‘Holocene temperature conundrum’. My best guess would be that the annual and summer reconstructions are probably both somewhat a mixture of several seasons, consistent with line 144: “reconstructions tend to show similar patterns in annual and summer temperature trends over all periods”.
In the abstract it is stated that: “our study shows that a trivial summer bias in proxies is not sufficient to explain the conundrum”. I believe this is based on the comparison of summer simulations with annual records but it’s not clear because, apart from the caveat around the records mentioned above, the modelled summer signal would be unlikely to resemble a summer-biased annual record. It seems more likely that it might look like a weighted combination of two seasons. Given this, I think the main finding around seasonality could benefit from futher elaboration.
Minor comments:
It’s not always clear how the present study’s use of the EBM adds to the existing debate. Instead the results seem to highlight where the EBM is significantly different from the other models and these already span a fairly large gradient of complexity. Perhaps one conclusion that could be strengthened is that a model like TransEBM is not greatly informative for this type of problem where seasonal differences are large?
The MPI-ESM simulations by Bader et al (2020) show a cooling mode as discussed in your introduction. It would be good to clarify here whether the simulation with MPI-ESM analysed here shows a similar result as it does not look to be the case from figure 1 or B3. If this is not the case, does this support your hypothesis about ocean spin up temperature being important or is there some other reason that can be identified?
It seems like a major difference between TransEBM and the other 3D models arises in the polar regions. Could you elaborate on this?
Line 67: “Sea-ice extent is linearly interpolated between the Last Glacial Maximum (LGM) and present-day states given in Zhuang et al. (2017).”
Line : “The sea-ice was interpolated using the same method. For sea-ice, the distributions given by Zhuang et al. (2017) were used as fix points for present-day and LGM conditions.”
Related to the point above, the role that sea-ice plays in the EBM is not clear from the sentences above. I suggest you add a paragraph briefly summarising TransEBM itself (in addition to forcings) to the Appendix.
Line 91: Could you spell out in more detail how you extract a seasonal reconstruction from temp12k?
Is JJA the best choice of season for the southern hemisphere? Related to this, can you replace summer with northern hemisphere summer (JJA) in the rest of the text.
Line 129: expect for TransEBM which shows a high latitude cooling and warming over the North Atlantic and the North American
East Coast.
Can you discuss why TransEBM has hte opposite sign over the ice-sheet?
Line 228: Another relevant reference here is Dallmeyer et al (2022): https://doi.org/10.1038/s41467-022-33646-6.
This is a bit of a minor point, but to me the figures would make more sense if ordered the panels in the modelling figures (3 & 4) from low complexity to high? Maybe TransEBM, LOVECLIM, CCSM3, MPI-ESM (going by number of levels for example)?
Technical corrections:
Line 15: perhaps spell out cmp - compare?
Line 42: “a feature of” perhaps better as “an artifact of”?
Erb et al 2022 is now published.
Ziegler, E. and Rehfeld, K this reference is incomplete.
Citation: https://doi.org/10.5194/egusphere-2023-86-RC2 - AC2: 'Reply on RC2', Christian Wirths, 17 Jun 2023
-
RC3: 'Comment on egusphere-2023-86', Anonymous Referee #3, 27 Apr 2023
This paper uses existing datasets from climate models and proxy databases to investigate Holocene temperature evolution. The paper attempts to compare and contrast these pre-existing data sources but fails to put forth new conclusions or findings. Thus, a major restructuring with new evidence and support would be required for publication of this work. My main reasons for this distinction are as follows:
- The paper fails to put forth novel findings: While it is true that the authors compare and contrast pre-existing modeling and proxy datasets, the findings presented in this paper are not novel in comparison to previous literature. Furthermore, many of the claims made by the paper are not substantiated by robust evidence or are not discussed in adequate detail. Therefore, the paper reads closer to a review paper rather than a journal article.
- Use of TransEBM is not justified: The authors highlight simulations with TransEBM, a simple energy balance model that has previously been published, as their primary contribution in the modeling space. However, it is never discussed why a low complexity model like TransEBM is suitable or advantageous for the type of analysis conducted here. In fact, as a reader I am left convinced that use of this model is entirely unnecessary as it underperforms the three other models, two of which are even out of date at this point in time.
- More thorough literature review must be done: In the concluding remarks, the paper makes strong claims, for example that climate models are likely missing boundary conditions such as land cover. Previous studies have investigated these topics at length (for example, see Thompson et al., 2022, Science Advances); however, there is no mention of these articles here. The authors should do a more thorough literature review of the state of science surrounding the Holocene temperature conundrum before attempting to publish on this topic.
- Analysis of seasonal bias is not robust: The analysis of seasonal bias in this paper is performed in a simple manner – annual or summer modeled results are compared with annual or summer proxies. Given the statements made about seasonal bias in proxies, a more thorough analysis that considers each proxy and the inherent assumptions present must be done in order to make the claims made by the authors.
Citation: https://doi.org/10.5194/egusphere-2023-86-RC3 - AC3: 'Reply on RC3', Christian Wirths, 17 Jun 2023
Status: closed
-
RC1: 'Comment on egusphere-2023-86', Anonymous Referee #1, 13 Mar 2023
The manuscript compares the temperature trends simulated by climate models of various complexities with paleo data over the Holocene. Many explanations have been suggested to explain the disagreements between models and data, the authors addressing specifically the impact of potential seasonal biases. They describe in details the seasonal and spatial distribution of the trends in the selected models and in data. This description is very clear. The paper is well written and easy to follow. I thus have no minor comment or suggestion to improve the presentation of the manuscript. However, there are two major points to consider in a revised version of the text.
1/ The added value of the study is not clearly explained and the authors should insist more on this in the conclusion, which is very short in the current version of the manuscript. The first paragraph of the conclusion summarizes the description of the trends presented in the previous sections. The second (and final) paragraph starts by a quite mild sentence: ‘Regarding the Holocene conundrum, it follows that a simple seasonal proxy bias is unlikely as a full explanation’ and then present some general suggestions for improvements or new studies. The fact that seasonal proxy biases might play a role but could not explain the full model-data disagreement is already around for some time (see the recent review of Kaufmann and Broadman, 2023) and the authors should explain more clearly the new contribution they bring to the debate.
2/ The authors analyze relatively old simulations that have been discussed in several studies. The selected data base has also already been used in model-data comparisons. A new simulation is included (TransEBM) but it has in general a lower agreement than the other ones with observations (see for instance Figure 5). This new simulation might be helpful to understand some of the characteristics of the other models but this is not developed in the current version of the manuscript. Furthermore, the set of selected experiments is not designed to test hypotheses, such as the potential role of vegetation or of the volcanic forcing for instance, as done in some other studies. Several transient Holocene have been performed recently. Some only cover parts of the Holocene or might not be publicly available but a larger set of experiments would provide additional information for the discussion (see for instance Gravgaard Askjær et al. 2022, in particular their Figure 3).
References:
Gravgaard Askjær T., Q.Zhang, F. Schenk, F.Charpentier Ljungqvist, Z. Lu, C. M. Brierley, P. O. Hopcroft, J. Jungclaus, X. Shi, G. Lohmann, W. Sun, J. Liu, P. Braconnot, B.L. Otto-Bliesner , Z. Wu, Q. Yin, Y. Kang, H. Yang, 2022. Multi-centennial Holocene climate variability in proxy records and transient model simulations. Quat. Sci. Rev. https://doi.org/10.1016/j.quascirev.2022.107801
Kaufman D.S. and E. Broadman, 2023. Revisiting the Holocene global temperature conundrum. Nature 614, 425-435 . https://doi.org/10.1038/s41586-022-05536-w
Citation: https://doi.org/10.5194/egusphere-2023-86-RC1 - AC1: 'Reply on RC1', Christian Wirths, 17 Jun 2023
-
RC2: 'Comment on egusphere-2023-86', Anonymous Referee #2, 22 Mar 2023
Summary
Wirths et al. summarise a comparison of simulations and reconstructions of temperature trends over three phases of the Holocene. They confirm important differences between the warming seen in models and the cooling reconstructed in temp12k. They conclude that a simple seasonal bias in the temp12k proxy records is unlikely to be able to explain the differences with the model runs.
Main comments:
A main assumption within this work is that the seasonal and annual reconstructions extracted from temp12k are able to separately resolve the seasons. I am no expert in this dataset, but this seems unlikely to be the case. In fact if we can take the separate seasonal records at face value then we can immediately conclude that the seasonal bias is not the answer to the ‘Holocene temperature conundrum’. My best guess would be that the annual and summer reconstructions are probably both somewhat a mixture of several seasons, consistent with line 144: “reconstructions tend to show similar patterns in annual and summer temperature trends over all periods”.
In the abstract it is stated that: “our study shows that a trivial summer bias in proxies is not sufficient to explain the conundrum”. I believe this is based on the comparison of summer simulations with annual records but it’s not clear because, apart from the caveat around the records mentioned above, the modelled summer signal would be unlikely to resemble a summer-biased annual record. It seems more likely that it might look like a weighted combination of two seasons. Given this, I think the main finding around seasonality could benefit from futher elaboration.
Minor comments:
It’s not always clear how the present study’s use of the EBM adds to the existing debate. Instead the results seem to highlight where the EBM is significantly different from the other models and these already span a fairly large gradient of complexity. Perhaps one conclusion that could be strengthened is that a model like TransEBM is not greatly informative for this type of problem where seasonal differences are large?
The MPI-ESM simulations by Bader et al (2020) show a cooling mode as discussed in your introduction. It would be good to clarify here whether the simulation with MPI-ESM analysed here shows a similar result as it does not look to be the case from figure 1 or B3. If this is not the case, does this support your hypothesis about ocean spin up temperature being important or is there some other reason that can be identified?
It seems like a major difference between TransEBM and the other 3D models arises in the polar regions. Could you elaborate on this?
Line 67: “Sea-ice extent is linearly interpolated between the Last Glacial Maximum (LGM) and present-day states given in Zhuang et al. (2017).”
Line : “The sea-ice was interpolated using the same method. For sea-ice, the distributions given by Zhuang et al. (2017) were used as fix points for present-day and LGM conditions.”
Related to the point above, the role that sea-ice plays in the EBM is not clear from the sentences above. I suggest you add a paragraph briefly summarising TransEBM itself (in addition to forcings) to the Appendix.
Line 91: Could you spell out in more detail how you extract a seasonal reconstruction from temp12k?
Is JJA the best choice of season for the southern hemisphere? Related to this, can you replace summer with northern hemisphere summer (JJA) in the rest of the text.
Line 129: expect for TransEBM which shows a high latitude cooling and warming over the North Atlantic and the North American
East Coast.
Can you discuss why TransEBM has hte opposite sign over the ice-sheet?
Line 228: Another relevant reference here is Dallmeyer et al (2022): https://doi.org/10.1038/s41467-022-33646-6.
This is a bit of a minor point, but to me the figures would make more sense if ordered the panels in the modelling figures (3 & 4) from low complexity to high? Maybe TransEBM, LOVECLIM, CCSM3, MPI-ESM (going by number of levels for example)?
Technical corrections:
Line 15: perhaps spell out cmp - compare?
Line 42: “a feature of” perhaps better as “an artifact of”?
Erb et al 2022 is now published.
Ziegler, E. and Rehfeld, K this reference is incomplete.
Citation: https://doi.org/10.5194/egusphere-2023-86-RC2 - AC2: 'Reply on RC2', Christian Wirths, 17 Jun 2023
-
RC3: 'Comment on egusphere-2023-86', Anonymous Referee #3, 27 Apr 2023
This paper uses existing datasets from climate models and proxy databases to investigate Holocene temperature evolution. The paper attempts to compare and contrast these pre-existing data sources but fails to put forth new conclusions or findings. Thus, a major restructuring with new evidence and support would be required for publication of this work. My main reasons for this distinction are as follows:
- The paper fails to put forth novel findings: While it is true that the authors compare and contrast pre-existing modeling and proxy datasets, the findings presented in this paper are not novel in comparison to previous literature. Furthermore, many of the claims made by the paper are not substantiated by robust evidence or are not discussed in adequate detail. Therefore, the paper reads closer to a review paper rather than a journal article.
- Use of TransEBM is not justified: The authors highlight simulations with TransEBM, a simple energy balance model that has previously been published, as their primary contribution in the modeling space. However, it is never discussed why a low complexity model like TransEBM is suitable or advantageous for the type of analysis conducted here. In fact, as a reader I am left convinced that use of this model is entirely unnecessary as it underperforms the three other models, two of which are even out of date at this point in time.
- More thorough literature review must be done: In the concluding remarks, the paper makes strong claims, for example that climate models are likely missing boundary conditions such as land cover. Previous studies have investigated these topics at length (for example, see Thompson et al., 2022, Science Advances); however, there is no mention of these articles here. The authors should do a more thorough literature review of the state of science surrounding the Holocene temperature conundrum before attempting to publish on this topic.
- Analysis of seasonal bias is not robust: The analysis of seasonal bias in this paper is performed in a simple manner – annual or summer modeled results are compared with annual or summer proxies. Given the statements made about seasonal bias in proxies, a more thorough analysis that considers each proxy and the inherent assumptions present must be done in order to make the claims made by the authors.
Citation: https://doi.org/10.5194/egusphere-2023-86-RC3 - AC3: 'Reply on RC3', Christian Wirths, 17 Jun 2023
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 624 | 325 | 45 | 994 | 25 | 26 |
- HTML: 624
- PDF: 325
- XML: 45
- Total: 994
- BibTeX: 25
- EndNote: 26
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1