the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
More modest peak temperatures during the Last Interglacial for both Greenland and Antarctica suggested by multi-model isotope simulations
Abstract. The Last Interglacial (LIG) period, approximately 130,000 to 115,000 years ago, represents one of the warmest intervals in the past 800,000 years. Here we simulate water isotopes in precipitation in Antarctica and the Arctic during the LIG, using three isotope-enabled atmosphere-ocean coupled climate models: HadCM3, MPI-ESM-wiso, and GISS-E2.1. These models were run following the Paleoclimate Modelling Intercomparison Project, phase 4 (PMIP4) protocol for the LIG at 127 ka (kilo-years ago), supplemented by a 3000-year Heinrich Stadial 11 (H11) experiment run with HadCM3. The long H11 simulation has meltwater from the Northern Hemisphere applied to the North Atlantic which causes large-scale changes in ocean circulation including cooling in the North Atlantic and Arctic and warming in the Southern and Global Ocean. We find that the standard 127 ka simulations do not capture the observed Antarctic warming and sea ice reduction in the Southern Ocean and Antarctic regions, but they capture around half of the warming in the Arctic. The H11 simulations align better with observations: they capture more than 80 % of the warming, sea ice loss, and δ18O changes for both Greenland and Antarctica. Decomposition of seasonal δ18O drivers highlights the dominant role of sea-ice retreat and associated changes in precipitation seasonality in influencing isotopic values in all simulations, alongside a small common response to orbital forcing. We use the H11 and multi-model 127 k simulations together to infer LIG surface air temperature (SAT) changes based on ice core measurements. The peak inferred LIG Greenland SAT increase is +2.89 ± 1.32 K at the NEEM ice core site. This is less than half the previously inferred warming. Peak inferred LIG Antarctic SAT increases are +4.39 ± 1.45 K at EDC, dropping to +1.67 ± 3.67 K at TALDICE. These calculated warming values are from climate effects alone, and do not take account of any ice flow or site elevation related impacts. Coastal sites in Greenland and Antarctica appear to have experienced less warming compared with higher central regions.
- Preprint
(23380 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 26 Mar 2025)
-
RC1: 'Comment on egusphere-2025-288', Anonymous Referee #1, 02 Feb 2025
reply
This paper studies the climate/d18O response at the LIG with the focus on the Arctic and Antarctic regions. The authors used 4 isotope-enabled climate models under 127 climate forcing, and one model with a long hosing simulating the H11 event. The major conclusion seems to be the model response is too small relative to the ice core observations in both temperatura and d18O, except perhaps the H3000-overshoot. It seems to me this paper has two major points. First, there is a systematic model-data inconsistency, with the model of less signal than in observation. Second, the H11 events indeed tends to reduce the model-data inconsistency, and therefore may be an important factor in the real world LIG resposne. It is a useful paper that summrizes the current state-of-the-art modeling of climate/d18O on LIG. Nevertheless, I think the paper can be further improved before publications.
Major concern:
Comparison with LGM: The LIG model-data comparsion will be better compared in the context of LGM model-data comparison of the same models. (I assume these models have done the LGM experiments before LIG experiments). Are all models has the similar inconsistency with observations? This LGM comparison has two advantages. First, observational data should have more uncertainty at the LIG than at LGM, while the model uncertainty is the same at LIG and LGM. Second, LGM clearly has no influence of H events (because it is well separated from H1 and H2). So, if all the models also have less signals in temperature/d18O at LGM than in observations, it is more likely that the model-data inconsistency is caused by the model deficiency. Otherwise, H11 may be a more important factor in reconciling the model-data discrepancy.
Minor concerns:
- The strategy to use cross-model T-d18O slope is no guarantee the model slope is more correct than real world. Even if the slope is the same as in real world, the small signal in both temperatura and d18O are consistent with the data-model inconsistency anyway. It is an interesting try, but does not give much information.
- The major result is model-data inconsistency, not just in the Arctic and Antarctic, but also for the global mean temperatura. The paper title, however, implies a bias in observation.
- The paper should address the inversion layer problem more explicitly. This is potentially serious in the Antactica and has been shown recently to disrupt the T-d18O slope dramatically, at least, at LGM (Liu et al. (2023).
Liu, Z. et al., 2023: Reconstructing past Antarctic temperature using present seasonal d18O-inversion layer temperature: Unified Slope Equations and application. J. Clim., 36, 2933-2957,
Citation: https://doi.org/10.5194/egusphere-2025-288-RC1
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
150 | 31 | 13 | 194 | 1 | 1 |
- HTML: 150
- PDF: 31
- XML: 13
- Total: 194
- BibTeX: 1
- EndNote: 1
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1