Abstract. The Last Interglacial period (LIG) is a period with increased summer insolation at high northern latitudes, which results in strong changes in the terrestrial and marine cryosphere. Understanding the mechanisms for this response via climate modelling and comparing the models' representation of climate reconstructions is one of the objectives set up by the Paleoclimate Modelling Intercomparison Project for its contribution to the sixth phase of the Coupled Model Intercomparison Project. Here we analyse the results from 16 climate models in terms of Arctic sea ice. The multi-model mean reduction in minimum sea ice area from the pre industrial period (PI) to the LIG reaches 50 % (multi-model mean LIG area is 3.20×106 km2, compared to 6.46×106 km2 for the PI). On the other hand, there is little change for the maximum sea ice area (which is 15–16×106 km2 for both the PI and the LIG. To evaluate the model results we synthesise LIG sea ice data from marine cores collected in the Arctic Ocean, Nordic Seas and northern North Atlantic. The reconstructions for the northern North Atlantic show year-round ice-free conditions, and most models yield results in agreement with these reconstructions. Model–data disagreement appear for the sites in the Nordic Seas close to Greenland and at the edge of the Arctic Ocean. The northernmost site with good chronology, for which a sea ice concentration larger than 75 % is reconstructed even in summer, discriminates those models which simulate too little sea ice. However, the remaining models appear to simulate too much sea ice over the two sites south of the northernmost one, for which the reconstructed sea ice cover is seasonal. Hence models either underestimate or overestimate sea ice cover for the LIG, and their bias does not appear to be related to their bias for the pre-industrial period. Drivers for the inter-model differences are different phasing of the up and down short-wave anomalies over the Arctic Ocean, which are associated with differences in model albedo; possible cloud property differences, in terms of optical depth; and LIG ocean circulation changes which occur for some, but not all, LIG simulations. Finally, we note that inter-comparisons between the LIG simulations and simulations for future climate with moderate (1 % yr−1) CO2 increase show a relationship between LIG sea ice and sea ice simulated under CO2 increase around the years of doubling CO2. The LIG may therefore yield insight into likely 21st century Arctic sea ice changes using these LIG simulations.

The Last Interglacial (LIG), a warmer period 130-116 ka before present, is a potential analog for future climate change. Stronger LIG summertime insolation at high northern latitudes drove Arctic land summer temperatures 4-5 • C higher than the preindustrial era. Climate model simulations have previously failed to capture these elevated temperatures, possibly because they were unable to correctly capture LIG sea-ice changes. Here, we show the latest version of the fully-coupled UK Hadley Center climate model (HadGEM3) simulates a more accurate Arctic LIG climate, including elevated temperatures. Improved model physics, including a sophisticated sea-ice melt-pond scheme, result in a complete simulated loss of Arctic 1 sea ice in summer during the LIG, which has yet to be simulated in past generations of models. This ice-free Arctic yields a compelling solution to the longstanding puzzle of what drove LIG Arctic warmth and supports a fast retreat of future Arctic summer sea ice. Both land air temperatures and sea surface temperatures in high northern latitudes were considerably warmer during the LIG (≈ 130 000-116 000 years before present) 1-5 and global sea level was likely 6-9 m higher than present 6, 7. Previous climate model simulations of the LIG, forced by appropriate greenhouse gas (GHG) and orbital changes, have failed to capture the observed high temperatures 8-11. This suggests that these models may not have accurately captured Arctic key climate processes in warmer climates. Whilst knowledge of past Arctic temperatures is robust thanks to the available observations 2, 10 , interpretation of Arctic sea ice changes during the LIG has previously been afflicted by uncertainty 8, 10, 12, 13. Water-isotope measurements from ice cores have been interpreted to suggest that, alongside the Arctic warming, there was a reduction in mean annual sea ice area 8. Microfauna in LIG marine sediments recovered from boreholes on the Beaufort Sea Shelf have been interpreted as implying a lack of perennial Arctic sea ice cover 14 , as have planktonic foraminifera recovered from some Arctic marine cores 15, 16. Similarly, ostracodes on the Lomonosov and Mendeleyev Ridges and Morris Jesup Rise have been interpreted as indicative of minimum sea ice coverage during peak LIG warmth 17. On the other hand, measurements of the recently-developed sea ice proxy IP25 (a carbon-25 highly-branched isoprenoid lipid), when combined with terrestrial and open-water phytoplankton biomarkers, have been interpreted as evidence of perennial LIG ice cover in the central

<p>The Last Interglacial (LIG) is a period of great importance as an analog for future climate change. Global sea level was 6-9 m higher than present. Stronger LIG summertime insolation at high northern latitudes drove Arctic land summer temperatures around 4-5 K higher than during the preindustrial era. Climate-model simulations have previously failed to capture these elevated temperatures. This may be because these models failed to correctly capture LIG sea ice changes.</p><p>Here, we show that the latest version of the UK Hadley Center coupled ocean-atmosphere climate model (HadGEM3) simulates a much improved Arctic LIG climate, including the observed high temperatures. Improved model physics in HadGEM3, including a sophisticated sea ice melt-pond scheme, results in the first-ever simulation of the complete loss of Arctic sea ice in summer during the LIG.</p><p>Our ice-free Arctic yields a compelling solution to the long-standing puzzle of what drove LIG Arctic warmth. The LIG simulation result is a new independent constraint on the strength of Arctic sea ice decline in climate-model projections, and provides support for a fast retreat of Arctic summer sea ice in the future.</p>

Last Interglacial (LIG), stable water isotope values (18 O) measured in Greenland deep ice cores are at least 2.5‰ higher compared to the present day. Previous isotopic climate simulations of the LIG do not capture the observed Greenland 18 O increases. Here, we use the isotope-enabled HadCM3 (UK Met Office coupled atmosphere-ocean general circulation model) to investigate whether a retreat of Northern Hemisphere sea ice was responsible for this model-data disagreement. Our results highlight the potential significance of sea ice changes on the LIG Greenland isotopic maximum. Sea ice loss in combination with increased sea surface temperatures, over the Arctic, affect 18 O: water vapour enriched in heavy isotopes and a shorter distillation path may both increase 18 O values over Greenland. We show, for the first time, that simulations of the response to Arctic sea ice reduction are capable of producing the likely magnitude of LIG 18 O increases at NEEM, NGRIP, GIPS2 and Camp Century ice core sites. However, we may underestimate 18 O changes at the Renland, DYE3 and GRIP ice core locations. Accounting for possible ice sheet changes is likely to be required to produce a better fit to the ice core measurements.

Abstract. Palaeoclimate model simulations are an important tool to improve our understanding of the mechanisms of climate change. These simulations also provide tests of the ability of models to simulate climates very different to today. Here we present the results from two brand-new simulations using the latest version of the UK's physical climate model, HadGEM3-GC3.1; they are the mid-Holocene (∼6 ka) and Last Interglacial (∼127 ka) simulations, both conducted under the auspices of CMIP6/PMIP4. This is the first time this version of the UK model has been used to conduct palaeoclimate simulations. These periods are of particular interest to PMIP4 because they represent the two most recent warm periods in Earth history, where atmospheric concentration of greenhouse gases and continental configuration are similar to the pre-industrial period but where there were significant changes to the Earth's orbital configuration, resulting in a very different seasonal cycle of radiative forcing. Results for these simulations are assessed firstly against the same model's pre-industrial control simulation (a simulation comparison, to describe and understand the differences between the pre-industrial – PI – and the two palaeo simulations) and secondly against previous versions of the same model relative to newly available proxy data (a model–data comparison, to compare all available simulations from the same model with proxy data to assess any improvements due to model advances). The introduction of this newly available proxy data adds further novelty to this study. Globally, for metrics such as 1.5 m temperature and surface rainfall, whilst both the recent palaeoclimate simulations are mostly capturing the expected sign and, in some places, magnitude of change relative to the pre-industrial, this is geographically and seasonally dependent. Compared to newly available proxy data (including sea surface temperature – SST – and rainfall) and also incorporating data from previous versions of the model shows that the relative accuracy of the simulations appears to vary according to metric, proxy reconstruction used for comparison and geographical location. In some instances, such as mean rainfall in the mid-Holocene, there is a clear and linear improvement, relative to proxy data, from the oldest to the newest generation of the model. When zooming into northern Africa, a region known to be problematic for models in terms of rainfall enhancement, the behaviour of the West African monsoon in both recent palaeoclimate simulations is consistent with current understanding, suggesting a wetter monsoon during the mid-Holocene and (more so) the Last Interglacial, relative to the pre-industrial era. However, regarding the well-documented “Saharan greening” during the mid-Holocene, results here suggest that the most recent version of the UK's physical model is still unable to reproduce the increases suggested by proxy data, consistent with all other previous models to date.