Elsevier

Quaternary Science Reviews

Volume 99, 1 September 2014, Pages 224-243
Quaternary Science Reviews

Temperature trends during the Present and Last Interglacial periods – a multi-model-data comparison

https://doi.org/10.1016/j.quascirev.2014.06.031Get rights and content

Highlights

  • First model-data comparison of interglacial multi-millennial temperature trends.

  • Relation between Present and Last Interglacial insolation and NH temperature trends.

  • Multi-millennial temperature trends consistent between models and data for NH, mismatch for SH.

Abstract

Though primarily driven by insolation changes associated with well-known variations in Earth's astronomical parameters, the response of the climate system during interglacials includes a diversity of feedbacks involving the atmosphere, ocean, sea ice, vegetation and land ice. A thorough multi-model-data comparison is essential to assess the ability of climate models to resolve interglacial temperature trends and to help in understanding the recorded climatic signal and the underlying climate dynamics. We present the first multi-model-data comparison of transient millennial-scale temperature changes through two intervals of the Present Interglacial (PIG; 8–1.2 ka) and the Last Interglacial (LIG; 123–116.2 ka) periods. We include temperature trends simulated by 9 different climate models, alkenone-based temperature reconstructions from 117 globally distributed locations (about 45% of them within the LIG) and 12 ice-core-based temperature trends from Greenland and Antarctica (50% of them within the LIG). The definitions of these specific interglacial intervals enable a consistent inter-comparison of the two intervals because both are characterised by minor changes in atmospheric greenhouse gas concentrations and more importantly by insolation trends that show clear similarities.

Our analysis shows that in general the reconstructed PIG and LIG Northern Hemisphere mid-to-high latitude cooling compares well with multi-model, mean-temperature trends for the warmest months and that these cooling trends reflect a linear response to the warmest-month insolation decrease over the interglacial intervals. The most notable exception is the strong LIG cooling trend reconstructed from Greenland ice cores that is not simulated by any of the models. A striking model-data mismatch is found for both the PIG and the LIG over large parts of the mid-to-high latitudes of the Southern Hemisphere where the data depicts negative temperature trends that are not in agreement with near zero trends in the simulations. In this area, the positive local summer insolation trend is counteracted in climate models by an enhancement of the Southern Ocean summer sea-ice cover and/or an increase in Southern Ocean upwelling. If the general picture emerging from reconstructions is realistic, then the model-data mismatch in mid and high Southern Hemisphere latitudes implies that none of the models is able to resolve the correct balance of these feedbacks, or, alternatively, that interglacial Southern Hemisphere temperature trends are driven by mechanisms which are not included in the transient simulations, such as changes in the Antarctic ice sheet or meltwater-induced changes in the overturning circulation.

Introduction

It has long been recognised that changes in the astronomical configuration of the Earth's orbit are a primary driver of long-term climate variations (Milankovitch, 1941). Past changes in the latitudinal and seasonal distribution of incoming top-of-the-atmosphere solar radiation (referred to as insolation in this manuscript) can be accurately calculated (Berger, 1978; Berger and Loutre, 1991, Laskar et al., 2004). The response of the climate system to this astronomical forcing includes a diversity of feedbacks involving the atmosphere, ocean, sea ice, vegetation and land ice (Braconnot et al., 2012, PALAEOSENS Project Members, 2012). Palaeoclimate information can be used to assess the realism of the representation of these feedbacks within numerical climate models. Systematic model-data comparisons performed within the Paleoclimate Modelling Intercomparison Project (PMIP) have focused on time-slice experiments such as those performed by Braconnot et al. (2007) for the middle of the Present Interglacial period (PIG-period) and by Lunt et al. (2013) for the thermal maximum of the Last Interglacial period (LIG-period; note that throughout the text PIG-period and LIG-period refer loosely to the two interglacial periods, while PIG and LIG will be used to indicate the specific target intervals defined in Section 2.1). During the last decade, progress has been made in the documentation of past climate variability based on time series from geological archives (e.g. deep-sea sediments, ice cores, lake sediments) resulting in a large number of proxy-based temperature reconstructions for the PIG-period (e.g. Wanner et al., 2008, Lohmann et al., 2013, Marcott et al., 2013) and, to a lesser extent, for the LIG-period (Andersen et al., 2004, CAPE Last Interglacial Project Members et al., 2006, Turney and Jones, 2010, McKay et al., 2011), with improved age models (e.g. Waelbroeck et al., 2008, Masson-Delmotte et al., 2011b). In parallel, increasing computational capacity, as well as the development of more computationally efficient climate models, has made it feasible to perform multi-millennial climate simulations (Bakker et al., 2013, Langebroek and Nisancioglu, 2013, Lohmann et al., 2013). These lines of progress now allow an investigation of PIG-period and LIG-period multi-millennial temperature trends, e.g. multi-millennial linear temperature changes, based on multiple reconstructions and simulations. The assessment of the ability of climate models to reproduce climate trends during periods warmer than today is strongly motivated by the context of projected global warming. Indeed, while there is no direct analogy between the physics of greenhouse gas versus astronomical forcing, the LIG-period is characterised by Northern Hemisphere (NH) continental summer temperatures that are similar to those expected for the coming centuries (Masson-Delmotte et al., 2010b).

The PIG-period (∼12 ka until the present-day; throughout this manuscript we will use [ka] to indicate kiloyears before 1950) and the LIG-period (∼130–115 ka) are both warm interglacial periods with distinct δ18O values in both ice-core and deep-sea-sediment records. Extensive research has resulted in a large number of proxy-based temperature reconstructions for the PIG-period (e.g. Wanner et al., 2008, Lohmann et al., 2013, Marcott et al., 2013) and (to a lesser extent) for the LIG-period (Andersen et al., 2004, CAPE Last Interglacial Project Members et al., 2006, Turney and Jones, 2010, McKay et al., 2011). For most mid-to-high latitude NH regions, the temperature evolution of the PIG-period is characterised by a gradual 1 °C decrease of the summer temperature from the early part of the interglacial to the pre-industrial situation. This multi-millennial temperature trend has been attributed to insolation changes (Wanner et al., 2008, Renssen et al., 2009). However, there are significant spatial differences in the magnitude and timing of the early PIG-period temperature maximum, and therefore in the PIG-period temperature trends, which are related for instance to the cooling effect of remnants of NH glacial ice sheets (Renssen et al., 2009). Lohmann et al. (2013) have recently shown that climate models underestimate reconstructed sea surface temperature (SST) changes over part of the PIG-period and that overall the model-data disagreement becomes smaller if seasonal biases in the proxy-based temperature trends are taken into account.

Climate reconstructions from the LIG-period show that maximum temperatures were roughly 2–5 °C higher than at present in, respectively, the mid- and high-latitudes of the NH (CAPE Last Interglacial Project Members et al., 2006, Turney and Jones, 2010, McKay et al., 2011, Sánchez-Goñi et al., 2012) during the early part of the LIG-period (>∼126 ka; Sánchez-Goñi et al., 2012, Shackleton et al., 2003). In Antarctica, LIG-period peak warmth (temperature anomalies of at least 4 °C; Sime et al., 2008, Masson-Delmotte et al., 2011b), was shown to occur earlier than at NH high latitudes (Masson-Delmotte et al., 2010a, Govin et al., 2012). However, the number of climate reconstructions available for the LIG-period is much smaller than for the PIG-period. Moreover, mainly because the LIG-period lies outside the time span covered by 14C dating techniques or ice-core layer counting, climate reconstructions are hampered by important issues related to assigning absolute ages and, as a consequence, to aligning different time series.

In addition to proxy-based reconstructions, transient climate-model experiments are helpful for analysing the PIG-period and LIG-period climates. If forced by the best estimate of millennial-scale astronomical and greenhouse gas (GHG) forcing scenarios at hand, simulations allow us to i) provide a global picture of the surface temperature evolution and ii) to investigate the role of external forcings and climate feedbacks in shaping the seasonal evolution of temperature. Earlier model-data comparisons focused on equilibrium simulations (Lunt et al., 2013). Beyond this approach, the transient nature of the simulations presented here enables us to i) describe the dynamical temperature response to the evolution of the astronomical forcing in the PIG-period and the LIG-period and ii) to take into account feedbacks from parts of the climate system which have a long (>1 ka) response time such as the deep ocean. A model inter-comparison of transient simulations covering the later part of the PIG-period, 6 ka to present, has been described in an earlier review study by Wanner et al. (2008). Their main findings are a decrease in summer temperatures at mid and high NH latitudes while winter temperatures decrease north of 60°N and increase at the mid latitudes of the NH. Results from the equatorial region and the Southern Hemisphere (SH) tend to differ between the different simulations. In a more recent model inter-comparison of transient PIG-period simulations, Varma et al. (2012) found a consistent cooling trend over Antarctica and the adjacent ocean between 7 and 0.25 ka. Model inter-comparison studies for LIG-period simulations have been undertaken within the framework of the PMIP3 project. The time slice experiments reviewed by Lunt et al. (2013) for the early part of the LIG-period, between 130 and 125 ka, show that annual mean changes compared to present-day only demonstrate a robust warming in the Arctic region and a cooling in the African and Indian monsoon regions. The seasonal signal tends to be more robust and shows a summer warming over the mid-to-high latitudes of the NH and an overall cooling of DJF with the exception of the Arctic where a warming is simulated in winter. Lunt et al. (2013) have shown that LIG-period equilibrium simulations are in poor agreement with reconstructed LIG-period annual mean and seasonal temperature changes. A number of transient LIG-period simulations, all of them included in this manuscript, have been inter-compared by Bakker et al. (2013). When the transiently simulated temperatures for the early part of the LIG-period are compared with the time slice experiments many similarities are found. However, Bakker et al. (2013) stress that the temperature evolution in specific regions is strongly altered by highly model-dependent feedbacks within the climate system related to the Meridional Overturning Circulation (MOC), sea ice, remnants of glacial continental ice sheets and monsoon dynamics, a conclusion which is in line with that found for the Mid-Holocene by Otto-Bliesner et al. (2006). The simulated transient LIG temperatures have, however, not yet been compared to reconstructed temperatures or to simulated transient PIG-period temperatures.

In this study we therefore combine PIG-period and LIG-period transient simulations from 9 different climate models including fast coupled ocean-atmosphere general circulation models (GCMs) and intermediate complexity models (EMICs). Including two different interglacial periods in the inter-comparison enables us to describe patterns in interglacial temperature trends that are likely valid for interglacial periods in general, because climate change features specific to either the PIG or the LIG can be isolated. Here one can think, for instance, of climatic changes caused by changes in the global ocean circulation. Moreover, the comparison of PIG and LIG allows us to address the question whether interglacial climate reacts, to a first order, linear to the orbital forcing. The fact that the PIG-period and LIG-period climates are simulated with the same models offers the opportunity to analyse the impact of the forcings within each model. Including results from various climate models that differ from one another in a number of ways (e.g. model sensitivity, setup and resolution, the degree to which different processes are parameterised and the extent of coupling between the various subcomponents of the climate system), yields robust multi-model results that are largely model independent. These multi-model transient temperature simulations are combined with a new compilation of reconstructed temperature time-series for the PIG-period and the LIG-period. Because of the highly complex nature of such a large multi-model-data comparison, we limit the analysis to the first order temperature changes, e.g. multi-millennial linear temperature trends. Furthermore, our focus will be on specific sub-intervals of the PIG- and LIG-periods as defined in Section 2.1. We will address the following questions:

  • 1)

    Are there temperature trends that are common to reconstructions and simulations?

  • 2)

    Are the magnitudes as well as the spatial patterns of the temperature trends different during the PIG and LIG?

  • 3)

    What are the caveats linked to proxy record interpretation and/or missing feedbacks in the simulations?

  • 4)

    Are the PIG and LIG temperature trends linearly related to the astronomical forcing? If not, can we identify the non-linear processes operating on multi-millennial time scales?

Section snippets

Defining Present and Last Interglacial target intervals

To consistently compare both reconstructed and simulated PIG-period and LIG-period temperature trends we need to define target intervals. Our main criterion in defining these intervals is that the rate of change of the underlying astronomical forcing should be comparable between the two interglacial intervals. To meet this criterion, an alignment of the PIG-period and the LIG-period is needed. For this purpose, we identified the timing of the maximum change in the climatic precession (defined

Simulated temperature trends

To investigate the temporal variations of simulated interglacial temperatures, we calculate linear temperature trends over the defined PIG and LIG intervals. To make the comparison of model results with findings from proxy-based reconstructions feasible, we aim to construct multi-model mean (MMM) temperature trends. As for the reconstructed temperature trends, the simulated temperature trends are linear least squares fits to the 50-year average temperature data. To ensure this MMM captures all

Summarising and concluding remarks

From the comparison of a large number of reconstructed and simulated temperature time-series for the PIG and the LIG intervals (8–1.2 ka and 123–116.2 ka respectively) we find overall agreement on a negative NH warmest-month temperature trend in both intervals. The simulated warmest-month temperature trends appear linearly related to the warmest-month insolation trends and to changes in climatic precession, with the same positive relation at different NH extra-tropical latitudes and for both

Acknowledgements

We thank Dr. Henning Bauch and the anonymous referees for their constructive criticism of the manuscript. This is Past4Future contribution no. 53. The research leading to these results has received funding from the European Union's Seventh Framework programme (FP7/2007–2013) under grant agreement no. 243908, “Past4Future. Climate change – Learning from the past climate”. B.M. thanks the Past Interglacials and Ocean2k working groups of the Past Global Changes (PAGES) project for discussions and

References (106)

  • F.G. Prahl et al.

    Further evaluation of long-chain alkenones as indicators of paleoceanographic conditions

    Geochim. Cosmochim. Acta

    (1988)
  • S.P. Ritz et al.

    Noble gases as proxies of mean ocean temperature: sensitivity studies using a climate model of reduced complexity

    Quat. Sci. Rev.

    (2011)
  • D.M. Roche et al.

    Evaluating Southern Ocean sea ice for the Last Glacial Maximum and pre-industrial climates: PMIP-2 models and data evidence

    Quat. Sci. Rev.

    (2012)
  • A. Rosell-Melé et al.

    Seasonality of temperature estimates as inferred from sediment trap data

    Quat. Sci. Rev.

    (2013)
  • A. Rosell-Melé et al.

    Atlantic core-top calibration of the U37K index as a sea-surface palaeotemperature indicator

    Geochim. Cosmochim. Acta

    (1995)
  • W.F. Ruddiman et al.

    A test of the overdue-glaciation hypothesis

    Quat. Sci. Rev.

    (2005)
  • N.J. Shackleton et al.

    Marine Isotope Substage 5e and the Eemian Interglacial

    Glob. Planet. Change

    (2003)
  • M.A. Sicre et al.

    Mid-latitude Southern Indian Ocean response to Northern Hemisphere Heinrich events

    Earth Planet. Sci. Lett.

    (2005)
  • C. Sonzogni et al.

    Core-top calibration of the alkenone index vs sea surface temperature in the Indian Ocean

    Deep Sea Res. Part II Top. Stud. Oceanogr.

    (1997)
  • B. Stenni et al.

    The deuterium excess records of EPICA Dome C and Dronning Maud Land ice cores (East Antarctica)

    Quat. Sci. Rev.

    (2010)
  • K. Andersen et al.

    High-resolution record of Northern Hemisphere climate extending into the last interglacial period

    Nature

    (2004)
  • P. Bakker et al.

    Last interglacial temperature evolution – a model inter-comparison

    Clim. Past

    (2013)
  • L. Bazin et al.

    An optimized multi-proxy, multi-site Antarctic ice and gas orbital chronology (AICC2012): 120-800 ka

    Clim. Past Discuss.

    (2012)
  • M. Berger et al.

    The sensitivity of the Arctic sea ice to orbitally induced insolation changes: a study of the mid-Holocene Paleoclimate Modelling Intercomparison Project 2 and 3 simulations

    Clim. Past

    (2013)
  • P. Braconnot et al.

    Results of PMIP2 coupled simulations of the Mid-Holocene and Last Glacial Maximum – part 1: experiments and large-scale features

    Clim. Past

    (2007)
  • P. Braconnot et al.

    Evaluation of climate models using palaeoclimatic data

    Nat. Clim. Change

    (2012)
  • S.L. Bradley et al.

    Combining ice core records and ice sheet models to explore the evolution of the East Antarctic Ice sheet during the Last Interglacial period

    Glob. Planet Change

    (2012)
  • S.C. Brassell et al.

    Molecular stratigraphy: a new tool for climatic assessment

    Nature

    (1986)
  • W.S. Broecker

    Paleocean circulation during the Last Deglaciation: a bipolar seesaw?

    Paleoceanography

    (1998)
  • CAPE Last Interglacial Project Members: Anderson, P., Bennike, O., Bigelow, N., Brigham-Grette, J., Duvall, M.,...
  • P.U. Clark et al.

    Sea-level fingerprinting as a direct test for the source of global meltwater pulse IA

    Science

    (2002)
  • M. Claussen et al.

    Simulation of an abrupt change in Saharan vegetation in the Mid-Holocene

    Geophys. Res. Lett.

    (1999)
  • W.D. Collins et al.

    The community climate system model version 3 (CCSM3)

    J. Clim.

    (2006)
  • M.H. Conte et al.

    Global temperature calibration of the alkenone unsaturation index (UK37) in surface waters and comparison with surface sediments

    Geochem. Geophys. Geosyst.

    (2006)
  • J.C. Duplessy et al.

    The deep ocean during the Last Interglacial period

    Science

    (2007)
  • N.R. Edwards et al.

    Uncertainties due to transportparameter sensitivity in an efficient 3-D ocean-climate model

    Clim. Dyn.

    (2005)
  • H. Goosse et al.

    Description of the Earth system model of intermediate complexity LOVECLIM version 1.2

    Geosci. Model Dev.

    (2010)
  • C. Gordon et al.

    The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments

    Clim. Dyn.

    (2000)
  • A. Govin et al.

    Evidence for northward expansion of Antarctic Bottom Water mass in the Southern Ocean during the last glacial inception

    Paleoceanography

    (2009)
  • A. Govin et al.

    Persistent influence of ice sheet melting on high northern latitude climate during the early Last Interglacial

    Clim. Past

    (2012)
  • M. Gröger et al.

    Changes in the hydrological cycle, ocean circulation, and carbon/nutrient cycling during the last interglacial and glacial transition

    Paleoceanography

    (2007)
  • P.B. Holden et al.

    Interhemispheric coupling, the West Antarctic Ice Sheet and warm Antarctic interglacials

    Clim. Past

    (2010)
  • P. Huybers

    Antarctica's orbital beat

    Science

    (2009)
  • P. Huybers et al.

    Antarctic temperature at orbital timescales controlled by local summer duration

    Nat. Geosci.

    (2008)
  • K. Izumi et al.

    Consistent large-scale temperature responses in warm and cold climates

    Geophys. Res. Lett.

    (2013)
  • P.D. Jones et al.

    Surface air temperature and its changes over the past 150 years

    Rev. Geophys.

    (1999)
  • P.D. Jones et al.

    Systematic optimisation and climate simulations of FAMOUS, a fast version of HadCM3

    Clim. Dyn.

    (2005)
  • S. Joussaume et al.

    Sensitivity of paleoclimate simulation results to season definitions

    J. Geophys. Res.

    (1997)
  • J. Jouzel et al.

    Paleoclimates: what do we learn from deep ice cores?

    WIREs Clim. Change

    (2010)
  • J. Jouzel et al.

    Magnitude of isotope/temperature scaling for interpretation of central Antarctic ice cores

    J. Geophys. Res.

    (2003)
  • Cited by (48)

    • General characteristics of climate change over China and associated dynamic mechanisms during the Last Interglacial based on PMIP4 simulations

      2022, Global and Planetary Change
      Citation Excerpt :

      Furthermore, cloud feedback in the simulations is of high uncertainty and therefore also a source of model bias (Liu et al., 2014). Although model–data discrepancies exist, the PMIP4 models still reveal warming in summer and cooling in winter over China during the LIG in response to changes in Earth's orbit, as indicated by previous studies (Bakker et al., 2014). For the annual mean precipitation, the wetter conditions over northern China, southeastern China, and the Tibetan Plateau in the PMIP4 models are consistent with data from multiproxies.

    • Modeling climate-vegetation interactions during the last interglacial: The impact of biogeophysical feedbacks in North Africa

      2020, Quaternary Science Reviews
      Citation Excerpt :

      Knowledge of the LIG climate thus provides valuable insights on the responses of the climate system under relatively “warm” forcing conditions. The LIG climate is therefore of interest to the Paleoclimate Modeling Intercomparison Project (PMIP, Otto-Bliesner et al., 2017) and has been studied extensively in both reconstructions (Andersen et al., 2004; CAPE Last Interglacial Project Members, 2006; Turney and Jones, 2010; McKay et al., 2011) and simulations (Schurgers et al., 2006, 2007; Gröger et al., 2007; Fischer and Jungclaus, 2010; Govin et al., 2012; Lunt et al., 2013; Nikolova et al., 2013; Otto-Bliesner et al., 2013; Bakker et al., 2013, 2014; Langebroek and Nisancioglu, 2014; Pedersen et al., 2017). However, PMIP3 simulations underestimated the proxy-based reconstructed LIG temperature response.

    View all citing articles on Scopus
    1

    Now at Swedish Meteorological and Hydrological Institute (SMHI), Norrköping, Sweden.

    View full text