Originally published by our Content Partner: Intergovernmental Panel on Climate Change (other articles)

Palaeoclimate

This chapter should be cited as:
Jansen, E., J. Overpeck, K.R. Briffa, J.-C. Duplessy, F. Joos, V. Masson-Delmotte, D. Olago, B. Otto-Bliesner, W.R. Peltier, S. Rahmstorf, R. Ramesh, D. Raynaud, D. Rind, O. Solomina, R. Villalba and D. Zhang, 2007: Palaeoclimate. In: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.

Executive Summary

What is the relationship between past greenhouse gas concentrations and climate?

• The sustained rate of increase over the past century in the combined radiative forcing from the three well-mixed greenhouse gases carbon dioxide (CO₂), methane (CH₄), and nitrous oxide (N₂O) is very likely unprecedented in at least the past 16 kyr. Pre-industrial variations of atmospheric greenhouse gas concentrations observed during the last 10 kyr were small compared to industrial era greenhouse gas increases, and were likely mostly due to natural processes.
• It is very likely that the current atmospheric concentrations of CO₂ (379 ppm) and CH₄ (1,774 ppb) exceed by far the natural range of the last 650 kyr. Ice core data indicate that CO₂ varied within a range of 180 to 300 ppm and CH₄ within 320 to 790 ppb over this period. Over the same period, antarctic temperature and CO₂ concentrations covary, indicating a close relationship between climate and the carbon cycle.
• It is very likely that glacial-interglacial CO₂ variations have strongly amplified climate variations, but it is unlikely that CO₂ variations have triggered the end of glacial periods. Antarctic temperature started to rise several centuries before atmospheric CO₂ during past glacial terminations.
• It is likely that earlier periods with higher than present atmospheric CO₂ concentrations were warmer than present. This is the case both for climate states over millions of years (e.g., in the Pliocene, about 5 to 3 Ma) and for warm events lasting a few hundred thousand years (i.e., the Palaeocene-Eocene Thermal Maximum, 55 Ma). In each of these two cases, warming was likely strongly amplified at high northern latitudes relative to lower latitudes.

What is the significance of glacial-interglacial climate variability?

• Climate models indicate that the Last Glacial Maximum (about 21 ka) was 3°C to 5°C cooler than the present due to changes in greenhouse gas forcing and ice sheet conditions. Including the effects of atmospheric dust content and vegetation changes gives an additional 1°C to 2°C global cooling, although scientific understanding of these effects is very low. It is very likely that the global warming of 4°C to 7°C since the Last Glacial Maximum occurred at an average rate about 10 times slower than the warming of the 20th century.
• For the Last Glacial Maximum, proxy records for the ocean indicate cooling of tropical sea surface temperatures (average likely between 2°C and 3°C) and much greater cooling and expanded sea ice over the high-latitude oceans. Climate models are able to simulate the magnitude of these latitudinal ocean changes in response to the estimated Earth orbital, greenhouse gas and land surface changes for this period, and thus indicate that they adequately represent many of the major processes that determine this past climate state.
• Last Glacial Maximum land data indicate significant cooling in the tropics (up to 5°C) and greater magnitudes at high latitudes. Climate models vary in their capability to simulate these responses. It is virtually certain that global temperatures during coming centuries will not be significantly influenced by a natural orbitally induced cooling. It is very unlikely that the Earth would naturally enter another ice age for at least 30 kyr.
• During the last glacial period, abrupt regional warmings (likely up to 16°C within decades over Greenland) and coolings occurred repeatedly over the North Atlantic region. They likely had global linkages, such as with major shifts in tropical rainfall patterns. It is unlikely that these events were associated with large changes in global mean surface temperature, but instead likely involved a redistribution of heat within the climate system associated with changes in the Atlantic Ocean circulation.
• Global sea level was likely between 4 and 6 m higher during the last interglacial period, about 125 ka, than in the 20th century. In agreement with palaeoclimatic evidence, climate models simulate arctic summer warming of up to 5°C during the last interglacial. The inferred warming was largest over Eurasia and northern Greenland, whereas the summit of Greenland was simulated to be 2°C to 5°C higher than present. This is consistent with ice sheet modeling suggestions that large-scale retreat of the south Greenland Ice Sheet and other arctic ice fields likely contributed a maximum of 2 to 4 m of sea level rise during the last interglacial, with most of any remainder likely coming from the Antarctic Ice Sheet.

What does the study of the current interglacial climate show?

• Centennial-resolution palaeoclimatic records provide evidence for regional and transient pre-industrial warm periods over the last 10 kyr, but it is unlikely that any of these commonly cited periods were globally synchronous. Similarly, although individual decadal-resolution interglacial palaeoclimatic records support the existence of regional quasi-periodic climate variability, it is unlikely that any of these regional signals were coherent at the global scale, or are capable of explaining the majority of global warming of the last 100 years.
• Glaciers in several mountain regions of the Northern Hemisphere retreated in response to orbitally forced regional warmth between 11 and 5 ka, and were smaller (or even absent) at times prior to 5 ka than at the end of the 20th century. The present day near-global retreat of mountain glaciers cannot be attributed to the same natural causes, because the decrease of summer insolation during the past few millennia in the Northern Hemisphere should be favourable to the growth of the glaciers.
• For the mid-Holocene (about 6 ka), GCMs are able to simulate many of the robust qualitative large-scale features of observed climate change, including midlatitude warming with little change in global mean temperature (<0.4°C), as well as altered monsoons, consistent with the understanding of orbital forcing. For the few well-documented areas, models tend to underestimate hydrological change. Coupled climate models perform generally better than atmosphere-only models, and reveal the amplifying roles of ocean and land surface feedbacks in climate change.
• Climate and vegetation models simulate past northward shifts of the boreal treeline under warming conditions. Palaeoclimatic results also indicated that these treeline shifts likely result in significant positive climate feedback. Such models are also capable of simulating changes in the vegetation structure and terrestrial carbon storage in association with large changes in climate boundary conditions and forcings (i.e., ice sheets, orbital variations).
• Palaeoclimatic observations indicate that abrupt decadalto centennial-scale changes in the regional frequency of tropical cyclones, floods, decadal droughts and the intensity of the African-Asian summer monsoon very likely occurred during the past 10 kyr. However, the mechanisms behind these abrupt shifts are not well understood, nor have they been thoroughly investigated using current climate models.

How does the 20th-century climate change compare with the climate of the past 2,000 years?

• It is very likely that the average rates of increase in CO₂, as well as in the combined radiative forcing from CO₂, CH₄ and N₂O concentration increases, have been at least five times faster over the period from 1960 to 1999 than over any other 40-year period during the past two millennia prior to the industrial era.
• Ice core data from Greenland and Northern Hemisphere mid-latitudes show a very likely rapid post-industrial era increase in sulphate concentrations above the preindustrial background.
• Some of the studies conducted since the Third Assessment Report (TAR) indicate greater multi-centennial Northern Hemisphere temperature variability over the last 1 kyr than was shown in the TAR, demonstrating a sensitivity to the particular proxies used, and the specific statistical methods of processing and/or scaling them to represent past temperatures. The additional variability shown in some new studies implies mainly cooler temperatures (predominantly in the 12th to 14th, 17th and 19th centuries), and only one new reconstruction suggests slightly warmer conditions (in the 11th century, but well within the uncertainty range indicated in the TAR).
• The TAR pointed to the ‘exceptional warmth of the late 20th century, relative to the past 1,000 years’. Subsequent evidence has strengthened this conclusion. It is very likely that average Northern Hemisphere temperatures during the second half of the 20th century were higher than for any other 50-year period in the last 500 years. It is also likely that this 50-year period was the warmest Northern Hemisphere period in the last 1.3 kyr, and that this warmth was more widespread than during any other 50- year period in the last 1.3 kyr. These conclusions are most robust for summer in extratropical land areas, and for more recent periods because of poor early data coverage.
• The small variations in pre-industrial CO₂ and CH₄ concentrations over the past millennium are consistent with millennial-length proxy Northern Hemisphere temperature reconstructions; climate variations larger than indicated by the reconstructions would likely yield larger concentration changes. The small pre-industrial greenhouse gas variations also provide indirect evidence for a limited range of decadal- to centennial-scale variations in global temperature.
• Palaeoclimate model simulations are broadly consistent with the reconstructed NH temperatures over the past 1 kyr. The rise in surface temperatures since 1950 very likely cannot be reproduced without including anthropogenic greenhouse gases in the model forcings, and it is very unlikely that this warming was merely a recovery from a pre-20th century cold period.
• Knowledge of climate variability over the last 1 kyr in the Southern Hemisphere and tropics is very limited by the low density of palaeoclimatic records.
• Climate reconstructions over the past millennium indicate with high confidence more varied spatial climate teleconnections related to the El Niño-Southern Oscillation than are represented in the instrumental record of the 20th century.
• The palaeoclimate records of northern and eastern Africa, as well as the Americas, indicate with high confidence that droughts lasting decades or longer were a recurrent feature of climate in these regions over the last 2 kyr.

What does the palaeoclimatic record reveal about feedback, biogeochemical and biogeophysical processes?

• The widely accepted orbital theory suggests that glacial-interglacial cycles occurred in response to orbital forcing. The large response of the climate system implies a strong positive amplification of this forcing. This amplification has very likely been influenced mainly by changes in greenhouse gas concentrations and ice sheet growth and decay, but also by ocean circulation and sea ice changes, biophysical feedbacks and aerosol (dust) loading.
• It is virtually certain that millennial-scale changes in atmospheric CO₂ associated with individual antarctic warm events were less than 25 ppm during the last glacial period. This suggests that the associated changes in North Atlantic Deep Water formation and in the large-scale deposition of wind-borne iron in the Southern Ocean had limited impact on CO₂.
• It is very likely that marine carbon cycle processes were primarily responsible for the glacial-interglacial CO₂ variations. The quantification of individual marine processes remains a difficult problem.
• Palaeoenvironmental data indicate that regional vegetation composition and structure are very likely sensitive to climate change, and in some cases can respond to climate change within decades.

6.1 Introduction

This chapter assesses palaeoclimatic data and knowledge of how the climate system changes over interannual to millennial time scales, and how well these variations can be simulated with climate models. Additional palaeoclimatic perspectives are included in other chapters.

Palaeoclimate science has made significant advances since the 1970s, when a primary focus was on the origin of the ice ages, the possibility of an imminent future ice age, and the first explorations of the so-called Little Ice Age and Medieval Warm Period. Even in the first IPCC assessment (IPCC, 1990), many climatic variations prior to the instrumental record were not that well known or understood. Fifteen years later, understanding is much improved, more quantitative and better integrated with respect to observations and modelling.

After a brief overview of palaeoclimatic methods, including their strengths and weaknesses, this chapter examines the palaeoclimatic record in chronological order, from oldest to youngest. This approach was selected because the climate system varies and changes over all time scales, and it is instructive to understand the contributions that lower-frequency patterns of climate change might make in influencing higher-frequency patterns of variability and change. In addition, an examination of how the climate system has responded to large changes in climate forcing in the past is useful in assessing how the same climate system might respond to the large anticipated forcing changes in the future.

Cutting across this chronologically based presentation are assessments of climate forcing and response, and of the ability of state-of-the-art climate models to simulate the responses. Perspectives from palaeoclimatic observations, theory and modelling are integrated wherever possible to reduce uncertainty in the assessment. Several sections also assess the latest developments in the rapidly advancing area of abrupt climate change, that is, forced or unforced climatic change that involves crossing a threshold to a new climate regime (e.g., new mean state or character of variability), often where the transition time to the new regime is short relative to the duration of the regime (Rahmstorf, 2001; Alley et al., 2003; Overpeck and Trenberth, 2004).

6.2 Palaeoclimatic Methods

6.2.1 Methods – Observations of Forcing and Response

The field of palaeoclimatology has seen significant methodological advances since the Third Assessment Report (TAR), and the purpose of this section is to emphasize these advances while giving an overview of the methods underlying the data used in this chapter. Many critical methodological details are presented in subsequent sections where needed. Thus, this methods section is designed to be more general, and to give readers more insight to and confidence in the findings of the chapter. Readers are referred to several useful books and special issues of journals for additional methodological detail (Bradley, 1999; Cronin, 1999; Fischer and Wefer, 1999; Ruddiman and Thomson, 2001; Alverson et al., 2003; Mackay et al., 2003; Kucera et al., 2005; NRC, 2006).

6.2.1.1 How are Past Climate Forcings Known?

Time series of astronomically driven insolation change are well known and can be calculated from celestial mechanics (see Section 6.4, Box 6.1). The methods behind reconstructions of past solar and volcanic forcing continue to improve, although important uncertainties still exist (see Section 6.6).

6.2.1.2 How are Past Changes in Global Atmospheric Composition Known?

Perhaps one of the most important aspects of modern palaeoclimatology is that it is possible to derive time series of atmospheric trace gases and aerosols for the period from about 650 kyr to the present from air trapped in polar ice and from the ice itself (see Sections 6.4 to 6.6 for more methodological citations). As is common in palaeoclimatic studies of the Late Quaternary, the quality of forcing and response series are verified against recent (i.e., post-1950) measurements made by direct instrumental sampling. Section 6.3 cites several papers that reveal how atmospheric CO₂ concentrations can be inferred back millions of years, with much lower precision than the ice core estimates. As is common across all aspects of the field, palaeoclimatologists seldom rely on one method or proxy, but rather on several. This provides a richer and more encompassing view of climatic change than would be available from a single proxy. In this way, results can be cross-checked and uncertainties understood. In the case of pre-Quaternary carbon dioxide (CO₂), multiple geochemical and biological methods provide reasonable constraints on past CO₂ variations, but, as pointed out in Section 6.3, the quality of the estimates is somewhat limited.

6.2.1.3 How Precisely Can Palaeoclimatic Records of Forcing and Response be Dated?

Much has been researched and written on the dating methods associated with palaeoclimatic records, and readers are referred to the background books cited above for more detail. In general, dating accuracy gets weaker farther back in time and dating methods often have specific ranges where they can be applied. Tree ring records are generally the most accurate, and are accurate to the year, or season of a year (even back thousands of years). There are a host of other proxies that also have annual layers or bands (e.g., corals, varved sediments, some cave deposits, some ice cores) but the age models associated with these are not always exact to a specific year. Palaeoclimatologists strive to generate age information from multiple sources to reduce age uncertainty, and palaeoclimatic interpretations must take into account uncertainties in time control.

There continue to be significant advances in radiometric dating. Each radiometric system has ranges over which the system is useful, and palaeoclimatic studies almost always publish analytical uncertainties. Because there can be additional uncertainties, methods have been developed for checking assumptions and cross verifying with independent methods. For example, secular variations in the radiocarbon clock over the last 12 kyr are well known, and fairly well understood over the last 35 kyr. These variations, and the quality of the radiocarbon clock, have both been well demonstrated via comparisons with age models derived from precise tree ring and varved sediment records, as well as with age determinations derived from independent radiometric systems such as uranium series. However, for each proxy record, the quality of the radiocarbon chronology also depends on the density of dates, the material available for dating and knowledge about the radiocarbon age of the carbon that was incorporated into the dated material.

6.2.1.4 How Can Palaeoclimatic Proxy Methods Be Used to Reconstruct Past Climate Dynamics?

Most of the methods behind the palaeoclimatic reconstructions assessed in this chapter are described in some detail in the aforementioned books, as well as in the citations of each chapter section. In some sections, important methodological background and controversies are discussed where such discussions help assess palaeoclimatic uncertainties.

Palaeoclimatic reconstruction methods have matured greatly in the past decades, and range from direct measurements of past change (e.g., ground temperature variations, gas content in ice core air bubbles, ocean sediment pore-water change and glacier extent changes) to proxy measurements involving the change in chemical, physical and biological parameters that reflect – often in a quantitative and well-understood manner – past change in the environment where the proxy carrier grew or existed. In addition to these methods, palaeoclimatologists also use documentary data (e.g., in the form of specific observations, logs and crop harvest data) for reconstructions of past climates. While a number of uncertainties remain, it is now well accepted and verified that many organisms (e.g., trees, corals, plankton, insects and other organisms) alter their growth and/or population dynamics in response to changing climate, and that these climate-induced changes are well recorded in the past growth of living and dead (fossil) specimens or assemblages of organisms. Tree rings, ocean and lake plankton and pollen are some of the best-known and best-developed proxy sources of past climate going back centuries and millennia. Networks of tree ring width and density chronologies are used to infer past temperature and moisture changes based on comprehensive calibration with temporally overlapping instrumental data. Past distributions of pollen and plankton from sediment cores can be used to derive quantitative estimates of past climate (e.g., temperatures, salinity and precipitation) via statistical methods calibrated against their modern distribution and associated climate parameters. The chemistry of several biological and physical entities reflects well-understood thermodynamic processes that can be transformed into estimates of climate parameters such as temperature. Key examples include: oxygen (O) isotope ratios in coral and foraminiferal carbonate to infer past temperature and salinity; magnesium/calcium (Mg/Ca) and strontium/ calcium (Sr/Ca) ratios in carbonate for temperature estimates; alkenone saturation indices from marine organic molecules to infer past sea surface temperature (SST); and O and hydrogen isotopes and combined nitrogen and argon isotope studies in ice cores to infer temperature and atmospheric transport. Lastly, many physical systems (e.g., sediments and aeolian deposits) change in predictable ways that can be used to infer past climate change. There is ongoing work on further development and refinement of methods, and there are remaining research issues concerning the degree to which the methods have spatial and seasonal biases. Therefore, in many recent palaeoclimatic studies, a combination of methods is applied since multi-proxy series provide more rigorous estimates than a single proxy approach, and the multi-proxy approach may identify possible seasonal biases in the estimates. No palaeoclimatic method is foolproof, and knowledge of the underlying methods and processes is required when using palaeoclimatic data.

The field of palaeoclimatology depends heavily on replication and cross-verification between palaeoclimate records from independent sources in order to build confidence in inferences about past climate variability and change. In this chapter, the most weight is placed on those inferences that have been made with particularly robust or replicated methodologies.

6.2.2 Methods – Palaeoclimate Modelling

Climate models are used to simulate episodes of past climate (e.g., the Last Glacial Maximum, the last interglacial period or abrupt climate events) to help understand the mechanisms of past climate changes. Models are key to testing physical hypotheses, such as the Milankovitch theory (Section 6.4, Box 6.1), quantitatively. Models allow the linkage of cause and effect in past climate change to be investigated. Models also help to fill the gap between the local and global scale in palaeoclimate, as palaeoclimatic information is often sparse, patchy and seasonal. For example, long ice core records show a strong correlation between local temperature in Antarctica and the globally mixed gases CO₂ and methane, but the causal connections between these variables are best explored with the help of models. Developing a quantitative understanding of mechanisms is the most effective way to learn from past climate for the future, since there are probably no direct analogues of the future in the past.

At the same time, palaeoclimate reconstructions offer the possibility of testing climate models, particularly if the climate forcing can be appropriately specified, and the response is sufficiently well constrained. For earlier climates (i.e., before the current ‘Holocene’ interglacial), forcing and responses cover a much larger range, but data are more sparse and uncertain, whereas for recent millennia more records are available, but forcing and response are much smaller. Testing models with palaeoclimatic data is important, as not all aspects of climate models can be tested against instrumental climate data. For example, good performance for present climate is not a conclusive test for a realistic sensitivity to CO₂ – to test this, simulation of a climate with a very different CO₂ level can be used. In addition, many parametrizations describing sub-grid scale processes (e.g., cloud parameters, turbulent mixing) have been developed using present-day observations; hence climate states not used in model development provide an independent benchmark for testing models. Palaeoclimate data are key to evaluating the ability of climate models to simulate realistic climate change.

In principle the same climate models that are used to simulate present-day climate, or scenarios for the future, are also used to simulate episodes of past climate, using differences in prescribed forcing and (for the deep past) in configuration of oceans and continents. The full spectrum of models (see Chapter 8) is used (Claussen et al., 2002), ranging from simple conceptual models, through Earth System Models of Intermediate Complexity (EMICs) and coupled General Circulation Models (GCMs). Since long simulations (thousands of years) can be required for some palaeoclimatic applications, and computer power is still a limiting factor, relatively ‘fast’ coupled models are often used. Additional components that are not standard in models used for simulating present climate are also increasingly added for palaeoclimate applications, for example, continental ice sheet models or components that track the stable isotopes in the climate system (LeGrande et al., 2006). Vegetation modules as well as terrestrial and marine ecosystem modules are increasingly included, both to capture biophysical and biogeochemical feedbacks to climate, and to allow for validation of models against proxy palaeoecological (e.g., pollen) data. The representation of biogeochemical tracers and processes is a particularly important new advance for palaeoclimatic model simulations, as a rich body of information on past climate has emerged from palaeoenvironmental records that are intrinsically linked to the cycling of carbon and other nutrients.

6.3 The Pre-Quaternary Climates

6.3.1 What is the Relationship Between Carbon Dioxide and Temperature in this Time Period?

Figure 6.1. (Top) Atmospheric CO₂ and continental glaciation 400 Ma to present. Vertical blue bars mark the timing and palaeolatitudinal extent of ice sheets (after Crowley, 1998). Plotted CO₂ records represent five-point running averages from each of the four major proxies (see Royer, 2006 for details of compilation). Also plotted are the plausible ranges of CO₂ from the geochemical carbon cycle model GEOCARB III (Berner and Kothavala, 2001). All data have been adjusted to the Gradstein et al. (2004) time scale. (Middle) Global compilation of deep-sea benthic foraminifera 18O isotope records from 40 Deep Sea Drilling Program and Ocean Drilling Program sites (Zachos et al., 2001) updated with high-resolution records for the Eocene through Miocene interval (Billups et al., 2002; Bohaty and Zachos, 2003; Lear et al., 2004). Most data were derived from analyses of two common and long-lived benthic taxa, Cibicidoides and Nuttallides. To correct for genus-specific isotope vital effects, the 18O values were adjusted by +0.64 and +0.4 (Shackleton et al., 1984), respectively. The ages are relative to the geomagnetic polarity time scale of Berggren et al. (1995). The raw data were smoothed using a five-point running mean, and curve-fitted with a locally weighted mean. The 18O temperature values assume an ice-free ocean (–1.0‰ Standard Mean Ocean Water), and thus only apply to the time preceding large-scale antarctic glaciation (~35 Ma). After the early Oligocene much of the variability (~70%) in the 18O record reflects changes in antarctic and Northern Hemisphere ice volume, which is represented by light blue horizontal bars (e.g., Hambrey et al., 1991; Wise et al., 1991; Ehrmann and Mackensen, 1992). Where the bars are dashed, they represent periods of ephemeral ice or ice sheets smaller than present, while the solid bars represent ice sheets of modern or greater size. The evolution and stability of the West Antarctic Ice Sheet (e.g., Lemasurier and Rocchi, 2005) remains an important area of uncertainty that could affect estimates of future sea level rise. (Bottom) Detailed record of CO₂ for the last 65 Myr. Individual records of CO₂ and associated errors are colour-coded by proxy method; when possible, records are based on replicate samples (see Royer, 2006 for details and data references). Dating errors are typically less than ±1 Myr. The range of error for each CO₂ proxy varies considerably, with estimates based on soil nodules yielding the greatest uncertainty. Also plotted are the plausible ranges of CO₂ from three geochemical carbon cycle models.

Pre-Quaternary climates prior to 2.6 Ma (e.g., Figure 6.1) were mostly warmer than today and associated with higher CO₂ levels. In that sense, they have certain similarities with the anticipated future climate change (although the global biology and geography were increasingly different further back in time). In general, they verify that warmer climates are to be expected with increased greenhouse gas concentrations. Looking back in time beyond the reach of ice cores, that is, prior to about 1 Ma, data on greenhouse gas concentrations in the atmosphere become much more uncertain. However, there are ongoing efforts to obtain quantitative reconstructions of the warm climates over the past 65 Myr and the following subsections discuss two particularly relevant climate events of this period.

How accurately is the relationship between CO₂ and temperature known? There are four primary proxies used for pre-Quaternary CO₂ levels (Jasper and Hayes, 1990; Royer et al., 2001; Royer, 2003). Two proxies apply the fact that biological entities in soils and seawater have carbon isotope ratios that are distinct from the atmosphere (Cerling, 1991; Freeman and Hayes, 1992; Yapp and Poths, 1992; Pagani et al., 2005). The third proxy uses the ratio of boron isotopes (Pearson and Palmer, 2000), while the fourth uses the empirical relationship between stomatal pores on tree leaves and atmospheric CO₂ content (McElwain and Chaloner, 1995; Royer, 2003). As shown in Figure 6.1 (bottom panel), while there is a wide range of reconstructed CO₂ values, magnitudes are generally higher than the interglacial, pre-industrial values seen in ice core data. Changes in CO₂ on these long time scales are thought to be driven by changes in tectonic processes (e.g., volcanic activity source and silicate weathering drawdown; e.g., Ruddiman, 1997). Temperature reconstructions, such as that shown in Figure 6.1 (middle panel), are derived from O isotopes (corrected for variations in the global ice volume), as well as Mg/Ca in forams and alkenones. Indicators for the presence of continental ice on Earth show that the planet was mostly ice-free during geologic history, another indication of the general warmth. Major expansion of antarctic glaciations starting around 35 to 40 Ma was likely a response, in part, to declining atmospheric CO₂ levels from their peak in the Cretaceous (~100 Ma) (DeConto and Pollard, 2003). The relationship between CO₂ and temperature can be traced further back in time as indicated in Figure 6.1 (top panel), which shows that the warmth of the Mesozoic Era (230–65 Ma) was likely associated with high levels of CO₂ and that the major glaciations around 300 Ma likely coincided with low CO₂ concentrations relative to surrounding periods.

6.3.2 What Does the Record of the Mid-Pliocene Show?

The Mid-Pliocene (about 3.3 to 3.0 Ma) is the most recent time in Earth’s history when mean global temperatures were substantially warmer for a sustained period (estimated by GCMs to be about 2°C to 3°C above pre-industrial temperatures; Chandler et al., 1994; Sloan et al., 1996; Haywood et al., 2000; Jiang et al., 2005), providing an accessible example of a world that is similar in many respects to what models estimate could be the Earth of the late 21st century. The Pliocene is also recent enough that the continents and ocean basins had nearly reached their present geographic configuration. Taken together, the average of the warmest times during the middle Pliocene presents a view of the equilibrium state of a globally warmer world, in which atmospheric CO₂ concentrations (estimated to be between 360 to 400 ppm) were likely higher than preindustrial values (Raymo and Rau, 1992; Raymo et al., 1996), and in which geologic evidence and isotopes agree that sea level was at least 15 to 25 m above modern levels (Dowsett and Cronin, 1990; Shackleton et al., 1995), with correspondingly reduced ice sheets and lower continental aridity (Guo et al., 2004).

Both terrestrial and marine palaeoclimate proxies (Thompson, 1991; Dowsett et al., 1996; Thompson and Fleming, 1996) show that high latitudes were significantly warmer, but that tropical SSTs and surface air temperatures were little different from the present. The result was a substantial decrease in the lower-tropospheric latitudinal temperature gradient. For example, atmospheric GCM simulations driven by reconstructed SSTs from the Pliocene Research Interpretations and Synoptic Mapping Group (Dowsett et al., 1996; Dowsett et al., 2005) produced winter surface air temperature warming of 10°C to 20°C at high northern latitudes with 5°C to 10°C increases over the northern North Atlantic (~60°N), whereas there was essentially no tropical surface air temperature change (or even slight cooling) (Chandler et al., 1994; Sloan et al., 1996; Haywood et al., 2000, Jiang et al., 2005). In contrast, a coupled atmosphere-ocean experiment with an atmospheric CO₂ concentration of 400 ppm produced warming relative to pre-industrial times of 3°C to 5°C in the northern North Atlantic, and 1°C to 3°C in the tropics (Haywood et al., 2005), generally similar to the response to higher CO₂ discussed in Chapter 10.

Figure 6.2.The Palaeocene-Eocene Thermal Maximum as recorded in benthic (bottom dwelling) foraminifer (Nuttallides truempyi) isotopic records from sites in the Antarctic, south Atlantic and Pacific (see Zachos et al., 2003 for details). The rapid decrease in carbon isotope ratios in the top panel is indicative of a large increase in atmospheric greenhouse gases CO₂ and CH₄ that was coincident with an approximately 5°C global warming (centre panel). Using the carbon isotope records, numerical models show that CH₄ released by the rapid decomposition of marine hydrates might have been a major component (~2,000 GtC) of the carbon flux (Dickens and Owen, 1996). Testing of this and other models requires an independent constraint on the carbon fluxes. In theory, much of the additional greenhouse carbon would have been absorbed by the ocean, thereby lowering seawater pH and causing widespread dissolution of seafloor carbonates. Such a response is evident in the lower panel, which shows a transient reduction in the carbonate (CaCO₃) content of sediments in two cores from the south Atlantic (Zachos et al., 2004, 2005). The observed patterns indicate that the ocean’s carbonate saturation horizon rapidly shoaled more than 2 km, and then gradually recovered as buffering processes slowly restored the chemical balance of the ocean. Initially, most of the carbonate dissolution is of sediment deposited prior to the event, a process that offsets the apparent timing of the dissolution horizon relative to the base of the benthic foraminifer carbon isotope excursion. Model simulations show that the recovery of the carbonate saturation horizon should precede the recovery in the carbon isotopes by as much as 100 kyr (Dickens and Owen, 1996), another feature that is evident in the sediment records.

The estimated lack of tropical warming is a result of basing tropical SST reconstructions on marine microfaunal evidence. As in the case of the Last Glacial Maximum (see Section 6.4), it is uncertain whether tropical sensitivity is really as small as such reconstructions suggest. Haywood et al. (2005) found that alkenone estimates of tropical and subtropical temperatures do indicate warming in these regions, in better agreement with GCM simulations from increased CO₂ forcing (see Chapter 10). As in the study noted above, climate models cannot produce a response to increased CO₂ with large high-latitude warming, and yet minimal tropical temperature change, without strong increases in ocean heat transport (Rind and Chandler, 1991).

The substantial high-latitude response is shown by both marine and terrestrial palaeodata, and it may indicate that high latitudes are more sensitive to increased CO₂ than model simulations suggest for the 21st century. Alternatively, it may be the result of increased ocean heat transports due to either an enhanced thermohaline circulation (Raymo et al., 1989; Rind and Chandler, 1991) or increased flow of surface ocean currents due to greater wind stresses (Ravelo et al., 1997; Haywood et al., 2000), or associated with the reduced extent of land and sea ice (Jansen et al., 2000; Knies et al., 2002; Haywood et al., 2005). Currently available proxy data are equivocal concerning a possible increase in the intensity of the meridional overturning cell for either transient or equilibrium climate states during the Pliocene, although an increase would contrast with the North Atlantic transient deep-water production decreases that are found in most coupled model simulations for the 21st century (see Chapter 10). The transient response is likely to be different from an equilibrium response as climate warms. Data are just beginning to emerge that describe the deep ocean state during the Pliocene (Cronin et al., 2005). Understanding the climate distribution and forcing for the Pliocene period may help improve predictions of the likely response to increased CO₂ in the future, including the ultimate role of the ocean circulation in a globally warmer world.

6.3.3 What Does the Record of the Palaeocene- Eocene Thermal Maximum Show?

Approximately 55 Ma, an abrupt warming (in this case of the order of 1 to 10 kyr) by several degrees celsius is indicated by changes in 18O isotope and Mg/Ca records (Kennett and Stott, 1991; Zachos et al., 2003; Tripati and Elderfield, 2004). The warming and associated environmental impact was felt at all latitudes, and in both the surface and deep ocean. The warmth lasted approximately 100 kyr. Evidence for shifts in global precipitation patterns is present in a variety of fossil records including vegetation (Wing et al., 2005). The climate anomaly, along with an accompanying carbon isotope excursion, occurred at the boundary between the Palaeocene and Eocene epochs, and is therefore often referred to as the Palaeocene-Eocene Thermal Maximum (PETM). The thermal maximum clearly stands out in high-resolution records of that time (Figure 6.2). At the same time, ¹³C isotopes in marine and continental records show that a large mass of carbon with low ¹³C concentration must have been released into the atmosphere and ocean. The mass of carbon was sufficiently large to lower the pH of the ocean and drive widespread dissolution of sea floor carbonates (Zachos et al., 2005). Possible sources for this carbon could have been methane (CH₄) from decomposition of clathrates on the sea floor, CO₂ from volcanic activity, or oxidation of sediments rich in organic matter (Dickens et al., 1997; Kurtz et al., 2003; Svensen et al., 2004). The PETM, which altered ecosystems worldwide (Koch et al., 1992; Bowen et al., 2002; Bralower, 2002; Crouch et al., 2003; Thomas, 2003; Bowen et al., 2004; Harrington et al., 2004), is being intensively studied as it has some similarity with the ongoing rapid release of carbon into the atmosphere by humans. The estimated magnitude of carbon release for this time period is of the order of 1 to 2 × 10¹⁸ g of carbon (Dickens et al., 1997), a similar magnitude to that associated with greenhouse gas releases during the coming century. Moreover, the period of recovery through natural carbon sequestration processes, about 100 kyr, is similar to that forecast for the future. As in the case of the Pliocene, the high-latitude warming during this event was substantial (~20°C; Moran et al., 2006) and considerably higher than produced by GCM simulations for the event (Sluijs et al., 2006) or in general for increased greenhouse gas experiments (Chapter 10). Although there is still too much uncertainty in the data to derive a quantitative estimate of climate sensitivity from the PETM, the event is a striking example of massive carbon release and related extreme climatic warming.

6.4 Glacial-Interglacial Variability and Dynamics

6.4.1 Climate Forcings and Responses Over Glacial-Interglacial Cycles

Figure 6.3. Variations of deuterium (δD; black), a proxy for local temperature, and the atmospheric concentrations of the greenhouse gases CO₂ (red), CH₄ (blue), and nitrous oxide (N₂O; green) derived from air trapped within ice cores from Antarctica and from recent atmospheric measurements (Petit et al., 1999; Indermühle et al., 2000; EPICA community members, 2004; Spahni et al., 2005; Siegenthaler et al., 2005a,b). The shading indicates the last interglacial warm periods. Interglacial periods also existed prior to 450 ka, but these were apparently colder than the typical interglacials of the latest Quaternary. The length of the current interglacial is not unusual in the context of the last 650 kyr. The stack of 57 globally distributed benthic δ¹⁸O marine records (dark grey), a proxy for global ice volume fluctuations (Lisiecki and Raymo, 2005), is displayed for comparison with the ice core data. Downward trends in the benthic δ¹⁸O curve reflect increasing ice volumes on land. Note that the shaded vertical bars are based on the ice core age model (EPICA community members, 2004), and that the marine record is plotted on its original time scale based on tuning to the orbital parameters (Lisiecki and Raymo, 2005). The stars and labels indicate atmospheric concentrations at year 2000.

Palaeoclimatic records document a sequence of glacial-interglacial cycles covering the last 740 kyr in ice cores (EPICA community members, 2004), and several million years in deep oceanic sediments (Lisiecki and Raymo, 2005) and loess (Ding et al., 2002). The last 430 kyr, which are the best documented, are characterised by 100-kyr glacial-interglacial cycles of very large amplitude, as well as large climate changes corresponding to other orbital periods (Hays et al., 1976; Box 6.1), and at millennial time scales (McManus et al., 2002; NorthGRIP, 2004). A minor proportion (20% on average) of each glacial-interglacial cycle was spent in the warm interglacial mode, which normally lasted for 10 to 30 kyr (Figure 6.3). There is evidence for longer interglacial periods between 430 and 740 ka, but these were apparently colder than the typical interglacials of the latest Quaternary (EPICA community members, 2004). TheHolocene, the latest of these interglacials, extends to the present.

The ice core record indicates that greenhouse gases co-varied with antarctic temperature over glacial-interglacial cycles, suggesting a close link between natural atmospheric greenhouse gas variations and temperature (Box 6.2). Variations in CO₂ over the last 420 kyr broadly followed antarctic temperature, typically by several centuries to a millennium (Mudelsee, 2001). The sequence of climatic forcings and responses during deglaciations (transitions from full glacial conditions to warm interglacials) are well documented. High-resolution ice core records of temperature proxies and CO₂ during deglaciation indicates that antarctic temperature starts to rise several hundred years before CO₂ (Monnin et al., 2001; Caillon et al., 2003). During the last deglaciation, and likely also the three previous ones, the onset of warming at both high southern and northern latitudes preceded by several thousand years the first signals of significant sea level increase resulting from the melting of the northern ice sheets linked with the rapid warming at high northern latitudes (Petit et al., 1999; Shackleton, 2000; Pépin et al., 2001). Current data are not accurate enough to identify whether warming started earlier in the Southern Hemisphere (SH) or Northern Hemisphere (NH), but a major deglacial feature is the difference between North and South in terms of the magnitude and timing of strong reversals in the warming trend, which are not in phase between the hemispheres and are more pronounced in the NH (Blunier and Brook, 2001).

Box 6.1: Orbital Forcing

It is well known from astronomical calculations (Berger, 1978) that periodic changes in parameters of the orbit of the Earth around the Sun modify the seasonal and latitudinal distribution of incoming solar radiation at the top of the atmosphere (hereafter called ‘insolation’). Past and future changes in insolation can be calculated over several millions of years with a high degree of confidence (Berger and Loutre, 1991; Laskar et al., 2004). This box focuses on the time period from the past 800 kyr to the next 200 kyr. Over this time interval, the obliquity (tilt) of the Earth axis varies between 22.05° and 24.50° with a strong quasi-periodicity around 41 kyr. Changes in obliquity have an impact on seasonal contrasts. This parameter also modulates annual mean insolation changes with opposite effects in low vs. high latitudes (and therefore no effect on global average insolation). Local annual mean insolation changes remain below 6 W m–2. Box 6.1, Figure 1. (Left) December to February (top), annual mean (middle) and June to August (bottom) latitudinal distribution of present-day (year 1950) incoming mean solar radiation (W m–2). (Right) Deviations with respect to the present of December to February (top), annual mean (middle) and June to August (bottom) latitudinal distribution of incoming mean solar radiation (W m–2) from the past 500 kyr to the future 100 kyr (Berger and Loutre, 1991; Loutre et al., 2004). The eccentricity of the Earth’s orbit around the Sun has longer quasi-periodicities at 400 and around 100 kyr, and varies between values of about 0.002 and 0.050 during the time period from 800 ka to 200 kyr in the future. Changes in eccentricity alone modulate the Sun-Earth distance and have limited impacts on global and annual mean insolation. However, changes in eccentricity affect the intra-annual changes in the Sun-Earth distance and thereby modulate significantly the seasonal and latitudinal effects induced by obliquity and climatic precession. Associated with the general precession of the equinoxes and the longitude of perihelion, periodic shifts in the position of solstices and equinoxes on the orbit relative to the perihelion occur, and these modulate the seasonal cycle of insolation with periodicities of about 19 and about 23 kyr. As a result, changes in the position of the seasons on the orbit strongly modulate the latitudinal and seasonal distribution of insolation. When averaged over a season, insolation changes can reach 60 W m–2 (Box 6.1, Figure 1). During periods of low eccentricity, such as about 400 ka and during the next 100 kyr, seasonal insolation changes induced by precession are less strong than during periods of larger eccentricity (Box 6.1, Figure 1). High-frequency variations of orbital variations appear to be associated with very small insolation changes (Bertrand et al., 2002a). The Milankovitch theory proposes that ice ages are triggered by minima in summer insolation near 65°N, enabling winter snowfall to persist all year and therefore accumulate to build NH glacial ice sheets. For example, the onset of the last ice age, about 116 ± 1 ka (Stirling et al., 1998), corresponds to a 65°N mid-June insolation about 40 W m–2 lower than today (Box 6.1, Figure 1). Studies of the link between orbital parameters and past climate changes include spectral analysis of palaeoclimatic records and the identification of orbital periodicities; precise dating of specific climatic transitions; and modelling of the climate response to orbital forcing, which highlights the role of climatic and biogeochemical feedbacks. Sections 6.4 and 6.5 describe some aspects of the state-of-the-art understanding of the relationships between orbital forcing, climate feedbacks and past climate changes.

Greenhouse gas (especially CO₂) feedbacks contributed greatly to the global radiative perturbation corresponding to the transitions from glacial to interglacial modes (see Section 6.4.1.2). The relationship between antarctic temperature and CO₂ did not change significantly during the past 650 kyr, indicating a rather stable coupling between climate and the carbon cycle during the late Pleistocene (Siegenthaler et al., 2005a). The rate of change in atmospheric CO₂ varied considerably over time. For example, the CO₂ increase from about 180 ppm at the Last Glacial Maximum to about 265 ppm in the early Holocene occurred with distinct rates over different periods (Monnin et al., 2001; Figure 6.4).

6.4.1.1 How Do Glacial-Interglacial Variations in the Greenhouse Gases Carbon Dioxide, Methane and Nitrous Oxide Compare with the Industrial Era Greenhouse Gas Increase?

Figure 6.4.The concentrations and radiative forcing by (a) CO₂, (b) CH₄ and (c) nitrous oxide (N₂O), and (d) the rate of change in their combined radiative forcing over the last 20 kyr reconstructed from antarctic and Greenland ice and firn data (symbols) and direct atmospheric measurements (red and magenta lines). The grey bars show the reconstructed ranges of natural variability for the past 650 kyr (Siegenthaler et al., 2005a; Spahni et al., 2005). Radiative forcing was computed with the simplified expressions of Chapter 2 (Myhre et al., 1998). The rate of change in radiative forcing (black line) was computed from spline fits (Enting, 1987) of the concentration data (black lines in panels a to c). The width of the age distribution of the bubbles in ice varies from about 20 years for sites with a high accumulation of snow such as Law Dome, Antarctica, to about 200 years for low-accumulation sites such as Dome C, Antarctica. The Law Dome ice and firn data, covering the past two millennia, and recent instrumental data have been splined with a cut-off period of 40 years, with the resulting rate of change in radiative forcing shown by the inset in (d). The arrow shows the peak in the rate of change in radiative forcing after the anthropogenic signals of CO₂, CH₄ and N₂O have been smoothed with a model describing the enclosure process of air in ice (Spahni et al., 2003) applied for conditions at the low accumulation Dome C site for the last glacial transition. The CO₂ data are from Etheridge et al. (1996); Monnin et al. (2001); Monnin et al. (2004); Siegenthaler et al. (2005b; South Pole); Siegenthaler et al. (2005a; Kohnen Station); and MacFarling Meure et al. (2006). The CH₄ data are from Stauffer et al. (1985); Steele et al. (1992); Blunier et al. (1993); Dlugokencky et al. (1994); Blunier et al. (1995); Chappellaz et al. (1997); Monnin et al. (2001); Flückiger et al. (2002); and Ferretti et al. (2005). The N₂O data are from Machida et al. (1995); Battle et al. (1996); Flückiger et al. (1999, 2002); and MacFarling Meure et al. (2006). Atmospheric data are from the National Oceanic and Atmospheric Administration’s global air sampling network, representing global average concentrations (dry air mole fraction; Steele et al., 1992; Dlugokencky et al., 1994; Tans and Conway, 2005), and from Mauna Loa, Hawaii (Keeling and Whorf, 2005). The globally averaged data are available from http://www.cmdl.noaa.gov/.

The present atmospheric concentrations of CO₂, CH₄ and nitrous oxide (N₂O) are higher than ever measured in the ice core record of the past 650 kyr (Figures 6.3 and 6.4). The measured concentrations of the three greenhouse gases fluctuated only slightly (within 4% for CO₂ and N₂O and within 7% for CH₄) over the past millennium prior to the industrial era, and also varied within a restricted range over the late Quaternary. Within the last 200 years, the late Quaternary natural range has been exceeded by at least 25% for CO₂, 120% for CH₄ and 9% for N₂O. All three records show effects of the large and increasing growth in anthropogenic emissions during the industrial era.

Variations in atmospheric CO₂ dominate the radiative forcing by all three gases (Figure 6.4). The industrial era increase in CO₂, and in the radiative forcing (Section 2.3) by all three gases, is similar in magnitude to the increase over the transitions from glacial to interglacial periods, but started from an interglacial level and occurred one to two orders of magnitude faster (Stocker and Monnin, 2003). There is no indication in the ice core record that an increase comparable in magnitude and rate to the industrial era has occurred in the past 650 kyr. The data resolution is sufficient to exclude with very high confidence a peak similar to the anthropogenic rise for the past 50 kyr for CO₂, for the past 80 kyr for CH₄ and for the past 16 kyr for N₂O. The ice core records show that during the industrial era, the average rate of increase in the radiative forcing from CO₂, CH₄ and N₂O is greater than at any time during the past 16 kyr (Figure 6.4). The smoothing of the atmospheric signal (Schwander et al., 1993; Spahni et al., 2003) is small at Law Dome, a high-accumulation site in Antarctica, and decadal-scale rates of change can be computed from the Law Dome record spanning the past two millennia (Etheridge et al., 1996; Ferretti et al., 2005; MacFarling Meure et al., 2006). The average rate of increase in atmospheric CO₂ was at least five times larger over the period from 1960 to 1999 than over any other 40-year period during the two millennia before the industrial era. The average rate of increase in atmospheric CH₄ was at least six times larger, and that for N₂O at least two times larger over the past four decades, than at any time during the two millennia before the industrial era. Correspondingly, the recent average rate of increase in the combined radiative forcing by all three greenhouse gases was at least six times larger than at any time during the period AD 1 to AD 1800 (Figure 6.4d).

6.4.1.2 What Do the Last Glacial Maximum and the Last Deglaciation Show?

Past glacial cold periods, sometimes referred to as ‘ice ages’, provide a means for evaluating the understanding and modelling of the response of the climate system to large radiative perturbations. The most recent glacial period started about 116 ka, in response to orbital forcing (Box 6.1), with the growth of ice sheets and fall of sea level culminating in the Last Glacial Maximum (LGM), around 21 ka. The LGM and the subsequent deglaciation have been widely studied because the radiative forcings, boundary conditions and climate response are relatively well known.

The response of the climate system at the LGM included feedbacks in the atmosphere and on land amplifying the orbital forcing. Concentrations of well-mixed greenhouse gases at the LGM were reduced relative to pre-industrial values (Figures 6.3 and 6.4), amounting to a global radiative perturbation of –2.8 W m^–2 – approximately equal to, but opposite from, the radiative forcing of these gases for the year 2000 (see Section 2.3). Land ice covered large parts of North America and Europe at the LGM, lowering sea level and exposing new land. The radiative perturbation of the ice sheets and lowered sea level, specified as a boundary condition for some LGM simulations, has been estimated to be about –3.2 W m^–2, but with uncertainties associated with the coverage and height of LGM continental ice (Mangerud et al., 2002; Peltier, 2004; Toracinta et al., 2004; Masson-Delmotte et al., 2006) and the parametrization of ice albedo in climate models (Taylor et al., 2000). The distribution of vegetation was altered, with tundra expanded over the northern continents and tropical rain forest reduced (Prentice et al., 2000), and atmospheric aerosols (primarily dust) were increased (Kohfeld and Harrison, 2001), partly as a consequence of reduced vegetation cover (Mahowald et al., 1999). Vegetation and atmospheric aerosols are treated as specified conditions in some LGM simulations, each contributing about –1 W m^–2 of radiative perturbation, but with very low scientific understanding of their radiative influence at the LGM (Claquin et al., 2003; Crucifix and Hewitt, 2005). Changes in biogeochemical cycles thus played an important role and contributed, through changes in greenhouse gas concentration, dust loading and vegetation cover, more than half of the known radiative perturbation during the LGM. Overall, the radiative perturbation for the changed greenhouse gas and aerosol concentrations and land surface was approximately –8 W m^–2 for the LGM, although with significant uncertainty in the estimates for the contributions of aerosol and land surface changes (Figure 6.5).

Understanding of the magnitude of tropical cooling over land at the LGM has improved since the TAR with more records, as well as better dating and interpretation of the climate signal associated with snow line elevation and vegetation change. Reconstructions of terrestrial climate show strong spatial differentiation, regionally and with elevation. Pollen records with their extensive spatial coverage indicate that tropical lowlands were on average 2°C to 3°C cooler than present, with strong cooling (5°C–6°C) in Central America and northern South America and weak cooling (<2°C) in the western Pacific Rim (Farrera et al., 1999). Tropical highland cooling estimates derived from snow-line and pollen-based inferences show similar spatial variations in cooling although involving substantial uncertainties from dating and mapping, multiple climatic causes of treeline and snow line changes during glacial periods (Porter, 2001; Kageyama et al., 2004), and temporal asynchroneity between different regions of the tropics (Smith et al., 2005). These new studies give a much richer regional picture of tropical land cooling, and stress the need to use more than a few widely scattered proxy records as a measure of low latitude climate sensitivity (Harrison, 2005).

Frequently Asked Question 6.1 What Caused the Ice Ages and Other Important Climate Changes Before the Industrial Era?

Climate on Earth has changed on all time scales, including long before human activity could have played a role. Great progress has been made in understanding the causes and mechanisms of these climate changes. Changes in Earth’s radiation balance were the principal driver of past climate changes, but the causes of such changes are varied. For each case – be it the Ice Ages, the warmth at the time of the dinosaurs or the fluctuations of the past millennium – the specific causes must be established individually. In many cases, this can now be done with good confidence, and many past climate changes can be reproduced with quantitative models. Global climate is determined by the radiation balance of the planet (see FAQ 1). There are three fundamental ways the Earth’s radiation balance can change, thereby causing a climate change: (1) changing the incoming solar radiation (e.g., by changes in the Earth’s orbit or in the Sun itself), (2) changing the fraction of solar radiation that is reflected (this fraction is called the albedo – it can be changed, for example, by changes in cloud cover, small particles called aerosols or land cover), and (3) altering the longwave energy radiated back to space (e.g., by changes in greenhouse gas concentrations). In addition, local climate also depends on how heat is distributed by winds and ocean currents. All of these factors have played a role in past climate changes. Figure 11: Schematic of the Earth’s orbital changes (Milankovitch cycles) that drive the ice age cycles. ‘T’ denotes changes in the tilt (or obliquity) of the Earth’s axis, ‘E’ denotes changes in the eccentricity of the orbit (due to variations in the minor axis of the ellipse), and ‘P’ denotes precession, that is, changes in the direction of the axis tilt at a given point of the orbit. (Source: Rahmstorf and Schellnhuber 2006) Starting with the ice ages that have come and gone in regular cycles for the past nearly three million years, there is strong evidence that these are linked to regular variations in the Earth’s orbit around the Sun, the so-called Milankovitch cycles (Figure 11). These cycles change the amount of solar radiation received at each latitude in each season (but hardly affect the global annual mean), and they can be calculated with astronomical precision. There is still some discussion about how exactly this starts and ends ice ages, but many studies suggest that the amount of summer sunshine on northern continents is crucial: if it drops below a critical value, snow from the past winter does not melt away in summer and an ice sheet starts to grow as more and more snow accumulates. Climate model simulations confirm that an Ice Age can indeed be started in this way, while simple conceptual models have been used to successfully ‘hindcast’ the onset of past glaciations based on the orbital changes. The next large reduction in northern summer insolation, similar to those that started past Ice Ages, is due to begin in 30,000 years. Although it is not their primary cause, atmospheric carbon dioxide (CO2) also plays an important role in the ice ages. Antarctic ice core data show that CO2 concentration is low in the cold glacial times (~190 ppm), and high in the warm interglacials (~280 ppm); atmospheric CO2 follows temperature changes in Antarctica with a lag of some hundreds of years. Because the climate changes at the beginning and end of ice ages take several thousand years, most of these changes are affected by a positive CO2 feedback; that is, a small initial cooling due to the Milankovitch cycles is subsequently amplified as the CO2 concentration falls. Model simulations of ice age climate (see discussion in Section 6.4.1) yield realistic results only if the role of CO2 is accounted for. During the last ice age, over 20 abrupt and dramatic climate shifts occurred that are particularly prominent in records around the northern Atlantic (see Section 6.4). These differ from the glacial-interglacial cycles in that they probably do not involve large changes in global mean temperature: changes are not synchronous in Greenland and Antarctica, and they are in the opposite direction in the South and North Atlantic. This means that a major change in global radiation balance would not have been needed to cause these shifts; a redistribution of heat within the climate system would have sufficed. There is indeed strong evidence that changes in ocean circulation and heat transport can explain many features of these abrupt events; sediment data and model simulations show that some of these changes could have been triggered by instabilities in the ice sheets surrounding the Atlantic at the time, and the associated freshwater release into the ocean. Much warmer times have also occurred in climate history – during most of the past 500 million years, Earth was probably completely free of ice sheets (geologists can tell from the marks ice leaves on rock), unlike today, when Greenland and Antarctica are ice-covered. Data on greenhouse gas abundances going back beyond a million years, that is, beyond the reach of antarctic ice cores, are still rather uncertain, but analysis of geological samples suggests that the warm ice-free periods coincide with high atmospheric CO2 levels. On million-year time scales, CO2 levels change due to tectonic activity, which affects the rates of CO2 exchange of ocean and atmosphere with the solid Earth. See Section 6.3 for more about these ancient climates. Another likely cause of past climatic changes is variations in the energy output of the Sun. Measurements over recent decades show that the solar output varies slightly (by close to 0.1%) in an 11-year cycle. Sunspot observations (going back to the 17th century), as well as data from isotopes generated by cosmic radiation, provide evidence for longer-term changes in solar activity. Data correlation and model simulations indicate that solar variability and volcanic activity are likely to be leading reasons for climate variations during the past millennium, before the start of the industrial era. These examples illustrate that different climate changes in the past had different causes. The fact that natural factors caused climate changes in the past does not mean that the current climate change is natural. By analogy, the fact that forest fires have long been caused naturally by lightning strikes does not mean that fires cannot also be caused by a careless camper. FAQ 4 addresses the question of how human influences compare with natural ones in their contributions to recent climate change.

The Climate: Long-range Investigation, Mapping, and Prediction (CLIMAP) reconstruction of ocean surface temperatures produced in the early 1980s indicated about 3°C cooling in the tropical Atlantic, and little or no cooling in the tropical Pacific. More pronounced tropical cooling for the LGM tropical oceans has since been proposed, including 4°C to 5°C based on coral skeleton records from off Barbados (Guilderson et al., 1994) and up to 6°C in the cold tongue off western South America based on foraminiferal assemblages (Mix et al., 1999). New data syntheses from multiple proxy types using carefully defined chronostratigraphies and new calibration data sets are now available from the Glacial Ocean Mapping (GLAMAP) and Multiproxy Approach for the Reconstruction of the Glacial Ocean surface (MARGO) projects, although with caveats including selective species dissolution, dating precision, nonanalogue situations, and environmental preferences of the organisms (Sarnthein et al., 2003b; Kucera et al., 2005; and references therein). These recent reconstructions confirm moderate cooling, generally 0°C to 3.5°C, of tropical SST at the LGM, although with significant regional variation, as well as greater cooling in eastern boundary currents and equatorial upwelling regions. Estimates of cooling show notable differences among the different proxies. Faunal-based proxies argue for an intensification of the eastern equatorial Pacific cold tongue in contrast to Mg/Ca-based SST estimates that suggest a relaxation of SST gradients within the cold tongue (Mix et al., 1999; Koutavas et al., 2002; Rosenthal and Broccoli, 2004). Using a Bayesian approach to combine different proxies, Ballantyne et al. (2005) estimated a LGM cooling of tropical SSTs of 2.7°C ± 0.5°C (1 standard deviation).

These ocean proxy synthesis projects also indicate a colder glacial winter North Atlantic with more extensive sea ice than present, whereas summer sea ice only covered the glacial Arctic Ocean and Fram Strait with the northern North Atlantic and Nordic Seas largely ice free and more meridional ocean surface circulation in the eastern parts of the Nordic Seas (Sarnthein et al., 2003a; Meland et al., 2005; de Vernal et al., 2006). Sea ice around Antarctica at the LGM also responded with a large expansion of winter sea ice and substantial seasonal variation (Gersonde et al., 2005). Over mid- and high-latitude northern continents, strong reductions in temperatures produced southward displacement and major reductions in forest area (Bigelow et al., 2003), expansion of permafrost limits over northwest Europe (Renssen and Vandenberghe, 2003), fragmentation of temperate forests (Prentice et al., 2000; Williams et al., 2000) and predominance of steppetundra in Western Europe (Peyron et al., 2005). Temperature reconstructions from polar ice cores indicate strong cooling at high latitudes of about 9°C in Antarctica (Stenni et al., 2001) and about 21°C in Greenland (Dahl-Jensen et al., 1998).

The strength and depth extent of the LGM Atlantic overturning circulation have been examined through the application of a variety of new marine proxy indicators (Rutberg et al., 2000; Duplessy et al., 2002; Marchitto et al., 2002; McManus et al., 2004). These tracers indicate that the boundary between North Atlantic Deep Water (NADW) and Antarctic Bottom Water was much shallower during the LGM, with a reinforced pycnocline between intermediate and particularly cold and salty deep water (Adkins et al., 2002). Most of the deglaciation occurred over the period about 17 to 10 ka, the same period of maximum deglacial atmospheric CO₂ increase (Figure 6.4). It is thus very likely that the global warming of 4°C to 7°C since the LGM occurred at an average rate about 10 times slower than the warming of the 20th century.

In summary, significant progress has been made in the understanding of regional changes at the LGM with the development of new proxies, many new records, improved understanding of the relationship of the various proxies to climate variables and syntheses of proxy records into reconstructions with stricter dating and common calibrations.

6.4.1.3 How Realistic Are Results from Climate Model Simulations of the Last Glacial Maximum?

Model intercomparisons from the first phase of the Paleoclimate Modelling Intercomparison Project (PMIP-1), using atmospheric models (either with prescribed SST or with simple slab ocean models), were featured in the TAR. There are now six simulations of the LGM from the second phase (PMIP-2) using Atmosphere-Ocean General Circulation Models (AOGCMs) and EMICs, although only a few regional comparisons were completed in time for this assessment. The radiative perturbation for the PMIP-2 LGM simulations available for this assessment, which do not yet include the effects of vegetation or aerosol changes, is –4 to –7 W m^–2. These simulations allow an assessment of the response of a subset of the models presented in Chapters 8 and 10 to very different conditions at the LGM.

Figure 6.5. The Last Glacial Maximum climate (approximately 21 ka) relative to the pre-industrial (1750) climate. (Top left) Global annual mean radiative influences (W m^–2) of LGM climate change agents, generall