EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Large-scale features and evaluation of the PMIP4-CMIP6 midHolocene simulations

Chris Brierley1, Anni Zhao1, Sandy Harrison2, Pascale Braconnot3, and the PMIP4 Community*
Chris Brierley et al.
  • 1University College London, Geography, London, United Kingdom of Great Britain and Northern Ireland (
  • 2Department of Geography and Environmental Science, University of Reading, Reading, RG6 6AB, UK
  • 3Laboratoire des Sciences du Climat et de l’Environnement-IPSL, Unité Mixte CEA-CNRS-UVSQ, Université Paris-Saclay, Orme des Merisiers, Gif-sur-Yvette, France
  • *A full list of authors appears at the end of the abstract

The mid-Holocene (6,000 years ago) is a standard experiment for the evaluation of the simulated response of global climate models using paleoclimate reconstructions. The latest mid-Holocene simulations are a contribution by the Palaeoclimate Model Intercomparison Project (PMIP4) to the current phase of the Coupled Model Intercomparison Project (CMIP6). Here we provide an initial analysis and evaluation of the results of the experiment for the mid-Holocene. We show that state-of-the-art models produce climate changes that are broadly consistent with theory and observations, including increased summer warming of the northern hemisphere and associated shifts in tropical rainfall.  Many features of the PMIP4-CMIP6 simulations were present in the previous generation (PMIP3-CMIP5) of simulations. The PMIP4-CMIP6 ensemble for the mid-Holocene has a global mean temperature change of -0.3 K, which is -0.2 K cooler that the PMIP3-CMIP5 simulations predominantly as a result of the prescription of realistic greenhouse gas concentrations in PMIP4-CMIP6. Neither this difference nor the improvement in model complexity and resolution seems to improve the realism of the simulations. Biases in the magnitude and the sign of regional responses identified in PMIP3-CMIP5, such as the amplification of the northern African monsoon, precipitation changes over Europe and simulated aridity in mid-Eurasia, are still present in the PMIP4-CMIP6 simulations. Despite these issues, PMIP4-CMIP6 and the mid-Holocene provide an opportunity both for quantitative evaluation and derivation of emergent constraints on climate sensitivity and feedback strength.

PMIP4 Community:

Charles J. R. Williams, David J. R. Thornalley, Xiaoxu Shi, Jean-Yves Peterschmitt, Rumi Ohgaito, Darrell S. Kaufman, Masa Kageyama, Julia C. Hargreaves, Michael P. Erb, Julien Emile-Geay, Roberta D’Agostino, Deepak Chandan, Matthieu Carré, Partrick Bartlein, Weipeng Zheng, Zhongshi Zhang, Qiong Zhang, Hu Yang, Evgeny M. Volodin, Robert A. Tomas, Cody Routson, W. Richard Peltier, Bette Otto-Bliesner, Polina A. Morozova, Nicholas P. McKay, Gerrit Lohmann, Allegra N. Legrande, Chuncheng Guo, Jian Cao, Esther Brady, James D. Annan, and Ayako Abe-Ouchi

How to cite: Brierley, C., Zhao, A., Harrison, S., and Braconnot, P. and the PMIP4 Community: Large-scale features and evaluation of the PMIP4-CMIP6 midHolocene simulations, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-8911,, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 28 Apr 2020
  • CC1: AMOC in mid-Holocene, Didier Swingedouw, 01 May 2020

    Hi Chris,

    This is a very nice presentation. I like the idea of using video to make things more vivid.

    I had just a comment concerning the AMOC response: I was quite surprised to see such a weak  and unconsistent response among PMIP4 models. This is relatively different  to what happened in PMIP3 where Gainusa-Bogdan et al. (2020) found an enhancement in most models, although it was also quite weak.

    Indeed, reconstructions from THornalley et al. (2013) and Ayache et al. (2018) indicated a potentially stronger AMOC at 6ka. The absence of such a feature in PMIP4 might limit the northward shift of the ITCZ and greening of the Sahara. This might be something to consider for sensitiivity experiments within PMIP.

    Best wishes,

    Didier Swingedouw


    Ayache M. Swingedouw D., Mary Y., Eynaud F., Colin C. (2018) AMOC variability over the Holocene: A new reconstruction based on multiple proxy-based SST records. Global and Planetary Changes 170, pp. 172-189.

    Gainusa Bogdan A. , Swingedouw D., Yiou P., Cattiaux J. and Codron F. (2020) AMOC as the key driver of the spread in Mid-Holocene winter temperature patterns over Europe in PMIP3 models. Global and Planetary Changes .

    Thornalley, D.J.R., Blaschek, M., Davies, F.J., Praetorius, S., Oppo, D.W., McManus, J.F., Hall, I.R., Kleiven, H., Renssen, H., McCave, I.N., 2013. Long-term variations in Iceland–Scotland overflow strength during the Holocene Geoscientific Instrumentation Methods and Data Systems long-term variations in Iceland–Scotland overflow strength during the Holocene. Clim. Past 95194, 2073–2084.

  • AC1: Comment on EGU2020-8911, Chris Brierley, 06 May 2020

    Hello Didier,

    Thanks for the complements about the video.

    I wasn't aware of the Gainusa-Bogdan et al. (2020) paper. Intriguingly the plot in our preprint where we suggest little impact on the AMOC contains a load of PMIP3 models (Fig. 9 of From looking at the figures in Gainusa-Bogdan et al. (2020), we don't have the 3 models with the largest changes CNRM-CM5, FGOALS-s2 and GISS-E2-R - becuase they haven't lodged msftmyz on the ESGF. Nonetheless even those models appear to have increases in max AMOC of only 1-2 Sv, perhaps wouldn't alter the conclusions much. If you happen to have the streamfunction fields handy, I'll try to include those models on the figure during the revisions.


    • CC2: Reply to AC1, Didier Swingedouw, 06 May 2020

      Hi Chris,

      Yes, I do have the streamfunction data. I think Alina directly asked them to the modelling center.

      I can provide them to you without any problem. I will come back to you with a private message. You're right that this might not affect that much the relationship of your paper. This is also because of the large spread of the mean state AMOC among the models, because 2 Sv is starting to be quite big (about 0.1 PW more heat transport at 26°N) and represents more than 10% of present-day AMOC mean state.

      But apparently not that clear in new models.

      Thank for this nice analysis.

      Best regards,