Compendium

Paleoclimate and Deep Time

35 min read

Paleoclimate and Deep Time

Paleoclimate science reads natural archives — air bubbles in glacial ice, planktonic and benthic shells in marine sediment, drip-stone calcite, annually banded tree rings, lake varves, peat, coral aragonite, leaf wax biomarkers — to reconstruct the temperature, ice volume, atmospheric composition, ocean chemistry, hydrology, and circulation of past climate states. Deep-time intervals extending back hundreds of millions of years sample the climate system under CO2 forcings and continental geometries radically different from today, while the Quaternary record (last 2.58 Myr) resolves the orbital pacing of glacial-interglacial cycles and the abrupt millennial-scale instabilities that operate even under steady boundary conditions. This note compiles the proxy infrastructure, the canonical Quaternary and pre-Quaternary records, the leading databases, and the open mechanistic debates that connect paleoclimate evidence to present-day projection.

See also

1. The proxy infrastructure

A climate proxy is a natural recorder whose composition, growth rate, isotopic ratio, or morphology covaries with a target climate variable through a calibrated transfer function. The transfer function may be empirical (regression against modern observations along a spatial gradient), mechanistic (a forward model of the recording process), or both. Proxy interpretation requires that the recorded signal be identifiable above noise, the calibration be portable to the time interval of interest, and the chronology be well constrained.

1.1 Ice cores

Polar and high-altitude ice preserves a layered record of accumulating snow whose isotopic composition (δ18O, δD) reflects condensation temperature, whose entrained dust and chemistry record continental aridity and atmospheric transport, and whose entrapped air bubbles preserve direct samples of the past atmosphere.

  • Greenland deep cores. GRIP (Greenland Ice Core Project, summit, 1990–1992, 3 029 m), GISP2 (Greenland Ice Sheet Project 2, summit, 1989–1993, 3 053 m), NorthGRIP (North Greenland Ice Core Project, 75.1°N 42.3°W, 1996–2003, 3 085 m), NEEM (North Greenland Eemian Ice Drilling, 77.5°N 51.1°W, 2007–2012, 2 540 m), EGRIP (East Greenland Ice-core Project, 75.6°N 36.0°W, 2015–2023, ~2 670 m through Northeast Greenland Ice Stream). NorthGRIP reaches ~123 ka BP, capturing the Eemian interglacial (the Marine Isotope Stage 5e analog for warmer-than-present climate, ~127–116 ka).
  • Antarctic deep cores. Vostok (78.5°S 106.8°E, Russian, 1970s–1998, 3 623 m, 420 ka), EPICA Dome C (75.1°S 123.4°E, European Project for Ice Coring in Antarctica, 1996–2004, 3 270 m, 800 ka, the longest continuous ice-core record), EPICA Dronning Maud Land (EDML, 2001–2006, 2 774 m, 150 ka), WAIS Divide (West Antarctic Ice Sheet, 79.5°S 112°W, 2006–2011, 3 405 m, 68 ka with annual resolution to 31 ka), Talos Dome, Dome Fuji (Japanese, 3 028 m, 720 ka). The ongoing Beyond EPICA-Oldest Ice project at Little Dome C (Italian-EU consortium, drilling 2021–2026) targets continuous ice to ~1.5 Ma to span the Mid-Pleistocene Transition.
  • Tropical and mountain ice. Quelccaya (Peru, 5 670 m), Sajama (Bolivia, 6 542 m), Huascarán (Peru, 6 048 m), Guliya and Dunde (Tibetan Plateau), Kilimanjaro — Lonnie Thompson program at Byrd Polar and Climate Research Center. Mountain ice extends only to ~25 ka in most sites, but provides tropical and subtropical signal complementary to polar archives.

Dating. Annual layer counting via visible stratigraphy, electrical conductivity, dust, chemistry, and δ18O — applicable in high-accumulation sections (Holocene Greenland, late-Pleistocene WAIS Divide). Deeper or lower-accumulation ice dated by ice-flow modelling tied to absolute marker horizons (Toba 74 ka tephra, dated U-Th coral terraces, the 10Be peak at the Laschamp geomagnetic excursion ~41 ka). The Greenland Ice Core Chronology 2005 (GICC05) and its extensions (GICC05modelext) provide annual layer counts through ~60 ka. The Antarctic Ice Core Chronology 2012 (AICC2012, Bazin et al. 2013, Climate of the Past 9, and Veres et al. 2013) synchronizes Antarctic deep cores via 10Be and methane to Greenland for ages back to ~800 ka.

Air bubbles and the gas record. Snow consolidates to firn over ~50–120 m (firn-air column, depth scaling with site temperature and accumulation). Air remains in open porosity through this depth until bubble close-off at the lock-in zone (typical pore close-off density 815 kg m−3). Because gases continue to diffuse through firn while ice ages by layer burial, gas ages are younger than ice ages at the same depth by the Δage offset (10–7 000 yr depending on site). The CO2, CH4, N2O, and noble-gas records are extracted by crushing or melt-refreezing techniques; high-resolution continuous-flow systems (laser spectrometers feeding a melt manifold) measure δ18O and chemistry at sub-annual resolution.

1.2 Foraminifera

Marine sediments accumulate planktonic and benthic foraminifera whose calcite tests preserve the chemistry of the water in which they grew. The benthic δ18O composite (Lisiecki and Raymo 2005 Paleoceanography 20, the “LR04 stack”) averages 57 globally distributed cores spanning the past 5.3 Myr, supplying the standard chronology for the Plio-Pleistocene.

  • δ18O. Reflects a convolved signal of seawater δ18O (governed by global ice volume — ice locks up 16O preferentially, enriching the ocean during glacials) and calcification temperature (Urey 1947 paleothermometer, calibrated ~0.21 ‰ per °C cooling for calcite). The benthic δ18O signal at depths >2 km is dominated by ice volume because deep ocean temperature varies little; planktonic δ18O carries mixed temperature and salinity signal.
  • Mg/Ca paleothermometry. Foraminiferal calcite incorporates Mg2+ in temperature-dependent proportion (Nürnberg-Bijma-Hemleben 1996; Lea-Mashiotta-Spero 1999 Paleoceanography 14). Calibrations have an exponential form Mg/Ca = B exp(A · T), with species-specific A ≈ 0.09–0.1 °C−1 and B ≈ 0.3–0.7 mmol mol−1. Combining δ18O with independent Mg/Ca temperature isolates the δ18O_seawater (ice-volume + salinity) component. Complications: dissolution corrections, secondary calcite cleaning protocols (Barker-Greaves-Elderfield 2003), salinity sensitivity.
  • B/Ca and δ11B. Boron isotopes in foraminiferal calcite track seawater pH because the borate ion B(OH)4− is preferentially incorporated and its abundance relative to boric acid B(OH)3 is a function of pH (Hemming-Hanson 1992; Foster-Rae 2016 Annual Review of Earth and Planetary Sciences 44). With auxiliary constraints on a second parameter of the carbonate system (alkalinity, or DIC, or boron-based [CO3 2−] via B/Ca), one recovers paleo-CO2.
  • Cd/Ca, Ba/Ca, Zn/Ca. Trace-metal proxies for nutrient and circulation tracers (Boyle 1988).
  • εNd in fossil fish teeth and Fe-Mn oxide coatings. Tracks the deep-water mass mixture because Atlantic and Pacific endmembers differ by ~10 ε-units in 143Nd/144Nd.

1.3 Speleothems

Stalagmites and flowstones precipitate from cave drip water as calcite or aragonite, with δ18O reflecting precipitation δ18O (and thus moisture source and rainfall amount), δ13C reflecting soil productivity and vegetation type (C3 vs C4), and trace elements (Mg/Ca, Sr/Ca, Ba/Ca, U/Ca) reflecting hydrology and host-rock interaction. U-Th dating by mass spectrometry yields chronologies precise to ±0.5% over the past ~600 ka — the most precise absolute chronology available in paleoclimate.

  • Hulu Cave (China, Nanjing). Wang-Cheng-Edwards 2001 Science 294 produced the seminal record of the East Asian summer monsoon back to 75 ka with sub-decadal resolution; Cheng-Edwards-Sinha 2016 Nature 534 extended the Asian monsoon stack to 640 ka, providing the Speleothem Asian Monsoon record now used to tune ice-core chronologies via methane synchronization.
  • Sanbao Cave (China). Wang-Cheng-Edwards 2008 Nature 451 extended back to 224 ka and demonstrated direct phase locking of the monsoon to summer insolation at 65°N.
  • Soreq Cave (Israel). Bar-Matthews-Ayalon 2003 record of Mediterranean climate back ~250 ka.
  • South American records. Botuverá (Brazil), Pacupahuain (Peru, Kanner-Burns-Cheng-Edwards 2012 Science 335) record South American summer monsoon.

1.4 Tree rings

Annually banded tree rings record growing-season temperature (ring width and maximum latewood density in cool sites) and moisture (ring width in semi-arid sites). The International Tree-Ring Data Bank (NOAA Paleoclimatology) hosts >4 000 chronologies. Cross-dating against living trees pushes chronologies back via subfossil wood preserved in glacial moraines, lakebeds, and bogs. The longest continuous chronologies extend ~12.6 ka (Hohenheim master pine and oak chronology, Friedrich-Remmele-Kromer 2004 Radiocarbon 46) and provide the 14C calibration for the IntCal radiocarbon timescale (Reimer et al. 2020, Radiocarbon 62, IntCal20).

  • Bristlecone pine (Pinus longaeva). White Mountains California, Methuselah Grove — single trees >4 800 years old.
  • Foxtail pine, Sierra juniper, Douglas-fir. Western US drought reconstructions feeding the North American Drought Atlas (Cook-Meko-Stahle-Cleaveland 1999 Journal of Climate 12; Cook-Anchukaitis-Buckley-D’Arrigo-Jacoby-Wright 2010 Science 328) and Living Blended Drought Atlas (Cook-Smerdon-Williams-Cook-Coats-Anchukaitis-Hofmann-Peterson-Marvel 2018).
  • Tropical species (teak, conifers in Indonesia, Tectona grandis). Less reliably annual but rings useable with auxiliary calibration.
  • Quantitative wood anatomy. Cell-by-cell measurements (Fonti-vonArx 2010) extend tree-ring information beyond ring width.

1.5 Pollen and macrofossils

Lake and bog sediments preserve windborne pollen whose taxonomic composition tracks the regional vegetation, which in turn tracks the regional climate (temperature, growing-degree-days, precipitation seasonality). Modern analog technique and Weighted Average Partial Least Squares (WA-PLS, ter Braak-Juggins 1993 Hydrobiologia 269) regress modern surface samples of pollen against modern climate to invert fossil assemblages into climate estimates.

  • Neotoma Paleoecology Database. Open archive, >10 000 sites globally; Williams-Grimm-Blois-Charles-Davis-Goring-Graham-Smith-Anderson-Arroyo-Cabrera-Booth-Brewer-Brunelle-Carrara-Davis-deBoer-Dietl-Fisher-Howat-Hostetler-Hwang-Klein-Manchester-Mauri-McMichael-Newby-Newton-Noss-Quamen-Russell-Sax-Selden-Smith-Wickham-Williams-Yang 2018 Quaternary Research 89.
  • Plant macrofossils. Identifiable seeds, leaves, needles preserved in lake sediment and packrat middens (the Packrat Midden Database, Strickland 1997+) yield species-level identifications complementing pollen.

1.6 Corals

Tropical scleractinian corals (Porites, Diploria, Montastraea) precipitate skeletal aragonite at temperature- and salinity-dependent rates with annual density banding. δ18O records sea surface temperature minus seawater δ18O (rainfall and freshwater advection); Sr/Ca is a comparatively clean SST proxy (Beck-Edwards-Ito-Taylor-Recy-Rougerie-Joannot-Henin 1992 Science 257); Ba/Ca tracks river runoff and upwelling. Uranium-thorium dating provides absolute chronology over the past 500 ka; coral terraces from interglacial high-stands (Barbados, Huon Peninsula, Vanuatu) constrain sea-level history.

  • Palmyra, Christmas Island, Maiana, Tarawa, Bunaken, Madang. Central and western tropical Pacific corals provide ENSO reconstruction (Cobb-Charles-Cheng-Edwards 2003 Nature 424; Tudhope-Chilcott-McCulloch-Cook-Chappell-Ellam-Lea-Lough-Shimmield 2001 Science 291) extending discontinuously to MIS 5e.
  • Coral reef terraces. Barbados (Fairbanks 1989 Nature 342) and Huon (Chappell-Polach 1991 Nature 349) provided the original sea-level timescale through deglaciation.

1.7 Lake varves

Glacial-fed and saline lakes deposit visibly banded annual couplets (varves) of light-coloured silt (spring meltwater) and dark organic matter or biogenic carbonate (summer-fall). Lake Suigetsu (Japan, Bronk Ramsey-Staff-Bryant-Brock-Kitagawa-vanderPlicht-Schlolaut-Marshall-Brauer-Lamb-Payne-Tarasov-Haraguchi-Gotanda-Yonenobu-Yokoyama-Tada-Nakagawa 2012 Science 338) provided the marine-independent 14C calibration to 52.8 ka via varves and tephra layers. Cariaco Basin (anoxic Venezuelan coastal basin) deposits varves through the last deglaciation. Lake El’gygytgyn (Russia, Brigham-Grette-Melles-Minyuk-Andreev-Tarasov-DeConto-Koenig-Nowaczyk-Wennrich-Rosén-Haltia-Hopfenberg-Cherepanova-Snyder-Sauerbrey-Niessen-Francke 2013 Science 340) preserved a continuous lacustrine record across the Mid-Pleistocene Transition because it sits in a meteor crater that escaped glacial scour.

1.8 Loess

Wind-blown silt accumulating on land downwind of glacial source areas preserves grain-size, magnetic susceptibility, and biomarker records of past aridity and monsoon strength. The Chinese Loess Plateau (CLP) sequences (Heller-Liu 1986 Geophysical Journal International 87) record the last ~22 Myr; the Luochuan section spans 2.5 Myr of glacial-interglacial dust deposition.

1.9 Biomarkers and organic geochemistry

Sedimentary lipid biomarkers from haptophyte algae, archaea, and land plants survive in marine and lake sediments and carry temperature, salinity, and vegetation signal.

  • Alkenones and the U^K’_37 index. Long-chain ketones synthesized by haptophyte algae (Emiliania huxleyi, Gephyrocapsa oceanica) have a temperature-dependent degree of unsaturation; U^K’_37 = C37:2/(C37:2 + C37:3) calibrates to SST with sensitivity ~0.033 per °C across the 0–29°C range (Müller-Kirst-Ruhland-vonStorch-Rosell-Melé 1998 Geochimica et Cosmochimica Acta 62). Used to constrain Neogene CO2-temperature relationships and Plio-Pleistocene tropical SST.
  • TEX_86. Isoprenoid glycerol dialkyl glycerol tetraethers (GDGTs) from marine Thaumarchaeota; ratio of crenarchaeol and isoprenoid variants tracks SST in the >20°C range where alkenones saturate (Schouten-Hopmans-Schefuß-SinningheDamsté 2002 Earth and Planetary Science Letters 204). The TEX_86^H calibration of Kim-vanderMeer-Schouten-Helmke-Willmott-Sangiorgi-Koç-Hopmans-SinningheDamsté 2010 Geochimica et Cosmochimica Acta 74 covers tropical applications.
  • MBT’/CBT in soils and lakes. Branched GDGTs produced by soil bacteria record terrestrial mean annual air temperature (Weijers-Schouten-vandenDonker-Hopmans-SinningheDamsté 2007; Naafs-Inglis-Zheng-Amesbury-Biester-Bindler-Blewett-Burrows-delCastillo-Chambers-Cohen-Evershed-Feakins-Gallego-Sala-Gallego-Torres-Hatcher-Honorio-Coronado-Hughes-Huguet-Ineson-Kayama-Kim-Korenkova-Lähteenoja-Magyari-Marchant-McClymont-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost-Pancost 2017 Geochimica et Cosmochimica Acta 208).
  • Plant leaf wax δ2H and δ13C. Long-chain n-alkanes (C27, C29, C31) from terrestrial plants preserved in marine and lake sediments record source-water δ2H (precipitation isotopes) and vegetation type via δ13C (C3 vs C4 plant fraction). Tierney-Pausata-deMenocal 2017 Science Advances 3 used African leaf-wax records to constrain Mediterranean Holocene rainfall.
  • Black carbon and PAHs. Quantify fire history; combined with charcoal counts in lake sediments (Global Charcoal Database, Marlon-Kelly-Daniau-Vanniere-Power-Bartlein-Harrison-Brewer-Daniau-Inoue-Lavoie-Maezumi-Magi-Mansouri-Niklasson-Robin-Trauernicht 2016 Biogeosciences 13).

1.10 Geochemical proxies for ancient CO2

Reconstructing atmospheric CO2 before the 800-ka ice-core limit relies on indirect proxies, each with a distinct calibration and uncertainty.

  • Boron isotopes (δ11B) in marine carbonates. Foster-Royer-Lunt 2017 Nature Communications 8 compiled a Cenozoic CO2 record. Recent benthic boron records (Anagnostou-John-Edgar-Foster-Ridgwell-Inglis-Pancost-Lunt-Pearson 2016 Nature 533) place early Eocene CO2 at 1 400 ± 470 ppm.
  • Alkenone δ13C. Pagani-Freeman-Arthur 1999 Science 285; sensitivity of haptophyte 13C fractionation to dissolved CO2 enables paleo-CO2 estimation through Cenozoic, with sensitivity declining at high CO2.
  • Stomatal indices. Density of stomatal pores on fossil leaves (or stomatal index = stomata / (stomata + epidermal cells)) correlates inversely with atmospheric CO2 (Beerling-Royer 2011 Nature Geoscience 4). Useful in the Neogene–Quaternary range; deep-time use challenged by species turnover.
  • Paleosol δ13C. Carbonate nodules in paleosols carry a δ13C signal that depends on respired-soil CO2 and atmospheric CO2 (Cerling 1991, 1999 American Journal of Science 291). Used through the Phanerozoic but with large uncertainties.
  • Phytane and isorenieratane δ13C. Photosynthetic carbon isotope fractionation depends on aqueous CO2; combined with paleo-temperature reconstruction recovers paleo-pCO2 (Witkowski-Bijl-Andringa-Kim-vanderMeer-Schouten-SinningheDamsté 2018 Science Advances 4).
  • Liverwort δ13C. Modern Conocephalum calibration extends to coal-ball liverworts in deep time (Fletcher-Brentnall-Anderson-Berner-Beerling 2008 Nature Geoscience 1).

The CenCO2PIP synthesis (Cenozoic CO2 Proxy Integration Project, Hönisch-Royer-Breecker-Polissar-Bowen-Henehan-Cui-Steinthorsdottir-McElwain-Kohn-Pearson-Tindall-Pancost-Inglis-Anagnostou-Foster-Bachem-Affek-Beerling-Cui-Ehlert-Eldrett-Greenwood-Hupp-Knorr-Kowalczyk-Liu-Naafs-O’Brien-Pagani-Petrizzo-Rae-Raitzsch-Ridgwell-Sluijs-Wade-Witkowski-Wright-Yang-Zachos 2023 Science 382) merges these proxies through 66 Myr with a unified Bayesian framework.

2. Quaternary cycles

The Quaternary period (2.58 Ma – present) is defined by repeated growth and decay of Northern Hemisphere ice sheets at orbital frequencies. The benthic δ18O stack reveals 103 Marine Isotope Stages (MIS) of alternating glacial (odd-numbered when reaching low δ18O ice-volume maxima — note MIS 1 is the Holocene interglacial; conventions are mixed) and interglacial conditions.

2.1 Orbital forcing — the Milankovitch theory

Insolation at top of atmosphere varies with three orbital parameters of Earth’s geometry relative to the Sun:

  • Eccentricity (e). Earth’s orbital ellipticity oscillates between e ≈ 0.000055 (near circular) and e ≈ 0.0679 (most elliptical) with dominant periods of 95, 125, and 405 kyr — the latter being the longest-period robust periodicity (the so-called grand 405-kyr cycle, used as the chronometer for Mesozoic and Cenozoic astrochronology; Laskar-Robutel-Joutel-Gastineau-Correia-Levrard 2004 Astronomy and Astrophysics 428).
  • Obliquity (ε). Tilt of Earth’s rotation axis relative to orbital plane oscillates between 22.1° and 24.5° on a 41-kyr cycle.
  • Climatic precession (e sin ω). Procession of equinoxes combined with eccentricity modulation. Dominant periods 19 and 23 kyr. Determines the seasonal phasing of Earth-Sun distance (currently NH winter at perihelion → mild NH winters, cool NH summers — favorable for NH ice growth).

The classical Milankovitch (1941) hypothesis posits that summer insolation at high northern latitudes (the canonical 65°N June insolation curve) controls ice-sheet growth and decay by setting the rate of summer ablation. Hays-Imbrie-Shackleton 1976 Science 194 (“Variations in the Earth’s Orbit: Pacemaker of the Ice Ages”) demonstrated the spectral matches in marine sediment records, establishing astronomical pacing.

2.2 The Mid-Pleistocene Transition

Before ~1.25 Ma, glacial-interglacial cycles operated at a 41-kyr (obliquity) pace. After ~0.7 Ma, cycles shifted to a quasi-100-kyr pace with larger-amplitude, asymmetric sawtooth structure (gradual ice buildup, abrupt deglaciation). The transition spanned ~1.25–0.7 Ma and is called the Mid-Pleistocene Transition (MPT) or “100-kyr problem” because the 100-kyr eccentricity forcing is weak — too weak to drive the observed glacial-interglacial amplitude directly. Proposed mechanisms include:

  • Regolith hypothesis. Clark-Pollard 1998 Paleoceanography 13 proposed that progressive erosion of unconsolidated regolith from beneath North American ice sheets exposed bedrock with higher friction, allowing ice sheets to grow thicker and thus to persist through obliquity insolation maxima.
  • CO2 thresholds. Hönisch-Hemming-Archer-Siddall-McManus 2009 Science 324 and Chalk-Hain-Foster-Rohling-Sexton-Badger-Cherry-Hasenfratz-Haug-Jaccard-Martínez-García-Pälike-Pancost-Wilson 2017 PNAS 114 found mean atmospheric CO2 declined across the MPT, enabling longer glacial intervals.
  • Combination tone. Maslin-Brierley 2015 Quaternary Science Reviews 122 reviewed proposals that the 100-kyr cycle is a combination tone of obliquity and precession harmonics arising from a nonlinear climate response, not a direct eccentricity signal.

2.3 Glacial-interglacial cycles and the Last Glacial Maximum

The Last Glacial Maximum (LGM) is conventionally dated to 26.5–19 ka BP (Clark-Dyke-Shakun-Carlson-Clark-Wohlfarth-Mitrovica-Hostetler-McCabe 2009 Science 325). At its peak:

  • Global mean SST. Cooler than pre-industrial by 3.6 ± 0.4 °C (Tierney-Zhu-King-Malevich-Hakim-Poulsen 2020 Nature 584, using PALEOSENS and proxy assimilation). Earlier reconstructions (CLIMAP 1981, MARGO 2009 Nature Geoscience 2) had given ~2 °C cooling, since revised upward.
  • Atmospheric CO2. ~190 ppm (vs ~280 ppm pre-industrial).
  • Atmospheric CH4. ~370 ppb (vs ~720 ppb pre-industrial).
  • Sea level. ~125–135 m below present (Clark-Mix 2002 Quaternary Science Reviews 21; updated estimates of 130 ± 5 m in Lambeck-Rouby-Purcell-Sun-Sambridge 2014 PNAS 111).
  • Ice sheets. Laurentide (eastern and central North America, ~3 km thick at center over Hudson Bay), Cordilleran (western North America), Fennoscandian (Europe), Barents-Kara (Eurasian Arctic), Patagonian, expanded Antarctic margin. Total LGM ice volume ~50 × 10^6 km^3 vs ~26 × 10^6 km^3 at present.

The Last Interglacial (LIG, MIS 5e, 129–116 ka BP) provides a natural analog for moderate warmth. Global mean surface temperature was ~1–2°C above pre-industrial (Capron-Govin-Stone-Masson-Delmotte-Mulitza-Otto-Bliesner-Rasmussen-Sime-Waelbroeck-Wolff 2014 Quaternary Science Reviews 103). Sea level reached 6–9 m above present (Dutton-Carlson-Long-Milne-Clark-DeConto-Horton-Rahmstorf-Raymo 2015 Science 349) — a sobering benchmark for projections of future ice loss under 1.5–2°C warming.

2.4 Holocene climate variability

The Holocene (11.7 ka – present) is conventionally subdivided by Walker-Berkelhammer-Björck-Cwynar-Fisher-Long-Lowe-Newnham-Rasmussen-Weiss 2018 into Greenlandian (11.7–8.2 ka), Northgrippian (8.2–4.2 ka), and Meghalayan (4.2 ka – present) stages bounded by the 8.2 ka event and 4.2 ka event. The Holocene Climate Optimum (~9–5 ka) saw NH summer insolation peaks ~8% above present, driving northern monsoon expansion (a “Green Sahara” with lake records of widespread fluvial activity, deMenocal-Ortiz-Guilderson-Adkins-Sarnthein-Baker-Yarusinsky 2000 Quaternary Science Reviews 19) and tundra retreat. Holocene global mean temperature evolution remains debated — Marcott-Shakun-Clark-Mix 2013 Science 339 inferred mid-Holocene warmth followed by gradual cooling; Bova-Rosenthal-Liu-Bagaria-Yan 2021 Nature 589 argued that proxy seasonal bias produces an apparent Holocene temperature peak that is absent when seasonality is corrected, leaving a monotonic warming through the Holocene.

3. Abrupt and millennial-scale events

Within steady glacial backgrounds, the climate system exhibits abrupt shifts on decadal to centennial timescales — clear evidence of nonlinear thresholds and multiple regimes.

3.1 Dansgaard-Oeschger oscillations

NGRIP δ18O reveals 25 stadial-interstadial cycles between 115 and 14 ka. Each Dansgaard-Oeschger (DO) event begins with abrupt (decades-to-centuries) warming over Greenland of 8–15°C, sustained warm interstadial for centuries to millennia, followed by gradual cooling back to stadial state. Named after Willi Dansgaard and Hans Oeschger; documented in Dansgaard-Johnsen-Clausen-Dahl-Jensen-Gundestrup-Hammer-Hvidberg-Steffensen-Sveinbjornsdottir-Jouzel-Bond 1993 Nature 364. Mechanisms involve:

  • AMOC switching. Stommel 1961 Tellus 13 first showed that the thermohaline circulation has multiple stable states. Modern coupled-model simulations (Ganopolski-Rahmstorf 2001 Nature 409) reproduce DO-like oscillations as transitions between strong and weak Atlantic Meridional Overturning Circulation under glacial boundary conditions.
  • Sea-ice feedback. Li-Battisti-Bitz-Battisti 2010 Quaternary Science Reviews 29 emphasized the role of Nordic Seas sea-ice cover in modulating Greenland air temperature.

3.2 Heinrich events

Six Heinrich events (H1–H6) between 60 and 16 ka are identified by ice-rafted debris (IRD) layers in North Atlantic sediments downwind of the Hudson Strait (Heinrich 1988 Quaternary Research 29; Hemming 2004 Reviews of Geophysics 42). Each represents a massive Laurentide ice-sheet purge that delivered freshwater to the North Atlantic, shutting down AMOC for centuries, producing the coldest stadials within each DO cycle. Heinrich Stadial 1 (HS1, ~17.5–14.7 ka) is the deglacial example; it preceded the abrupt Bølling-Allerød warming.

3.3 The deglacial sequence

The transition from LGM to Holocene proceeded in steps:

  • HS1 (Heinrich Stadial 1, 17.5–14.7 ka). Cold North Atlantic, weak AMOC, dry northern monsoons, warm Antarctic and tropical SH (the bipolar seesaw, Crowley 1992; EPICA Community Members 2006 Nature 444 documented the Greenland-Antarctic phasing).
  • Bølling-Allerød (14.7–12.9 ka). Abrupt NH warming of ~10°C in Greenland over decades; AMOC resumption. Coincident sea-level rise pulse Meltwater Pulse 1A (MWP-1A) of 14–18 m in <500 yr (Deschamps-Durand-Bard-Hamelin-Camoin-Thomas-Henderson-Okuno-Yokoyama 2012 Nature 483).
  • Younger Dryas (12.9–11.7 ka). Abrupt return to near-glacial conditions in NH, traditionally attributed to freshwater discharge from Lake Agassiz routing through St. Lawrence (Broecker-Kennett-Flower-Teller-Trumbore-Bonani-Wolfli 1989 Nature 341), though the routing path is contested (Murton-Bateman-Dallimore-Teller-Yang 2010 Nature 464 favored Mackenzie outlet). YD termination at 11.65 ka BP marks the Pleistocene-Holocene boundary at the NGRIP δ18O step (Walker-Johnsen-Rasmussen-Popp-Steffensen-Gibbard-Hoek-Lowe-Andrews-Björck-Cwynar-Hughen-Kershaw-Kromer-Litt-Lowe-Nakagawa-Newnham-Schwander 2009 GSSP definition).
  • 8.2 ka event. A ~150-year cold and dry anomaly recorded in Greenland ice and proxies across the North Atlantic and northern monsoons, triggered by catastrophic drainage of glacial Lake Agassiz-Ojibway through Hudson Strait (Barber-Dyke-Hillaire-Marcel-Jennings-Andrews-Kerwin-Bilodeau-McNeely-Southon-Morehead-Gagnon 1999 Nature 400; Alley-Ágústsdóttir 2005 Quaternary Science Reviews 24). Magnitude: ~3°C Greenland cooling, ~0.5–1°C in adjacent North Atlantic.
  • 4.2 ka event. Drought anomaly recorded in the Mediterranean, Middle East, and South Asia; coincident with the Akkadian Empire collapse and other societal disruptions (Weiss 1993 Science 261; Weiss 2017 Megadrought and Collapse review). Severity and global synchrony disputed.

3.4 The Common Era reconstructions

The PAGES 2k Consortium has produced multiple iterations of last-2000-year temperature reconstructions:

  • PAGES 2k 2013 (Nature Geoscience 6). Continental-scale regional composites.
  • PAGES 2k 2017 (Scientific Data 4, “A global multiproxy database for temperature reconstructions of the Common Era”). Standardized proxy compilation.
  • PAGES 2k 2019 (Nature Geoscience 12). Showed the 20th-century warming is the only period in the past 2 000 yr in which all reconstructions show warming exceeding multi-decadal noise globally.

Notable Common Era anomalies include the Medieval Climate Anomaly (~950–1250 CE, regional rather than synchronous global warmth), the Little Ice Age (~1300–1850 CE, NH cool with regional sub-events), and volcanic forcing pulses (1257 Samalas eruption, the largest of the past 7 000 yr; 1452/3 Kuwae; 1600 Huaynaputina; 1815 Tambora — “year without a summer” 1816).

4. Deep-time analogs

Beyond the Quaternary, intervals of higher CO2 and warmer climate offer analogs — albeit with different boundary conditions (continental configuration, biota, ocean gateways) — for projected near-future climate.

4.1 Pliocene warm period (mPWP, 3.264–3.025 Ma)

The Mid-Piacenzian Warm Period (mPWP) is the most recent interval with CO2 sustained in the modern target range. Reconstructions:

  • CO2. 374 (320–470) ppm based on δ11B (de la Vega-Chalk-Wilson-Smith-Foster 2020 Scientific Reports 10), broadly consistent with alkenone and stomatal estimates.
  • Global mean surface temperature. ~3.2 (2.5–4.4) °C above pre-industrial (PlioMIP2, Haywood-Tindall-Dowsett-Dolan-Foley-Hunter-Hill-Chan-Abe-Ouchi-Stepanek-Lohmann-Chandan-Peltier-Tan-Contoux-Ramstein-Li-Zhang-Guo-Nisancioglu-Zhang-Li-Otto-Bliesner-Brady-Kamae-Chandler-Sohl-Hopcroft-Sherriff-Tindall-Liu-Hill-Lunt-Mahowald 2020 Climate of the Past 16).
  • Sea level. 17 ± 6 m above present (Dumitru-Austermann-Polyak-Fornós-Asmerom-Ginés-Ginés-Onac 2019 Nature 574 using flowstones in Mallorca caves, and Grant-Naish-Dunbar-Stocchi-Kominz-Kamp-Tapia-McKay-Levy-Patterson 2019 Nature 574).
  • Sea ice. Strongly reduced Arctic seasonal cover; no Greenland ice sheet at peak.

mPWP is the canonical “equilibrium ~3°C-warmer” analog. The PlioMIP2 model intercomparison (Haywood et al. 2020) tested whether models calibrated against modern observations can reproduce mid-Pliocene proxy patterns. Persistent biases — particularly insufficient high-latitude amplification and tropical Pacific east-west SST gradient — fed into the PMIP-CMIP6 dialogue about Earth System Sensitivity (see §6).

4.2 Mid-Miocene Climatic Optimum (MMCO, ~17–14.5 Ma)

CO2 estimates ~400–600 ppm (Sosdian-Greenop-Hain-Foster-Pearson-Lear 2018 Earth and Planetary Science Letters 498 via δ11B; Tanner-Lunt-Farnsworth-Lunt-Markwick-Pearson 2020 Climate of the Past 16 via leaf gas exchange). Global mean temperature ~4–8°C above modern; the East Antarctic Ice Sheet was probably more dynamic but still present; tropical SSTs ~25–32°C. The Mi-3 cooling event (~13.8 Ma) ended the MMCO and ushered in the late-Cenozoic icehouse trajectory.

4.3 Eocene Climatic Optimum (~52–50 Ma)

Polar continents were ice-free, tropical vegetation extended to ~60° latitude (palms in southern Australia and Wyoming, alligators on Ellesmere Island at 78°N paleolatitude — Eberle-Greenwood 2012 GSA Bulletin 124). CO2 ~1 000–2 000 ppm. The shallow latitudinal SST gradient (pole-to-equator <17°C vs ~28°C today) remains a challenge for climate models — the “equable climate problem” of Huber-Sloan 2001 Geophysical Research Letters 28 and Lunt-Huber-Anagnostou-Baatsen-Caballero-DeConto-Dijkstra-Donnadieu-Evans-Feng-Foster-Gasson-vonderHeydt-Hollis-Inglis-Jones-Kiehl-Kirtland-Turner-Korty-Kozdon-Krapp-Kutzbach-Lear-Liu-Lohmann-Lohmann-Losada-Lunt-Lunt-Marwick-Mills-Morris-Otto-Bliesner-Polly-Salzmann-Schubert-Tigchelaar-Tindall-Upchurch-Valdes-Williams-Wilson-Winguth-Zachos 2017 Climate of the Past 13).

4.4 Paleocene-Eocene Thermal Maximum (PETM, 56.0 Ma)

A geologically rapid (~5 kyr injection) release of >3 000 Pg of isotopically light carbon (δ13C excursion of ~3‰) drove ~5°C global warming sustained for ~100 kyr (McInerney-Wing 2011 Annual Review of Earth and Planetary Sciences 39; Zachos-Dickens-Zeebe 2008 Nature 451). Possible sources: methane hydrate dissociation, North Atlantic Igneous Province volcanism (Storey-Duncan-Swisher 2007 Science 316), thawing of permafrost soils, oceanic anoxia release. Ocean acidification caused dissolution of carbonate sediments (the Paleocene-Eocene carbonate compensation depth shoaled by ~2 km; Zachos-Röhl-Schellenberg-Sluijs-Hodell-Kelly-Thomas-Nicolo-Raffi-Lourens-McCarren-Kroon 2005 Science 308). The PETM is the closest geologic analog for rapid anthropogenic CO2 release, though anthropogenic emissions are an order of magnitude faster (Zeebe-Ridgwell-Zachos 2016 Nature Geoscience 9 estimated PETM injection rate at ~1.1 Pg C yr−1 vs current ~10 Pg C yr−1).

4.5 Snowball Earth

The Cryogenian period (720–635 Ma) hosted at least two near-global glaciations — the Sturtian (717–660 Ma) and Marinoan (~650–635 Ma). Evidence: equatorial tillites (Kirschvink 1992; Hoffman-Kaufman-Halverson-Schrag 1998 Science 281 “A Neoproterozoic Snowball Earth”); cap carbonates above glacial diamictites with extreme δ13C excursions; banded iron formations indicating anoxic deep ocean. Mechanism: ice-albedo runaway triggered when ice line moves within ~30° of equator. Termination: CO2 buildup from volcanism over millions of years (~0.12 bar required to overwhelm albedo) drove abrupt deglaciation. The hypothesis is contested by “slushball Earth” variants (Hyde-Crowley-Baum-Peltier 2000 Nature 405) that preserve a thin equatorial open-ocean band.

4.6 Phanerozoic CO2-climate compilations

GEOCARB and GEOCARBSULF (Berner 1991, 2004, 2006 American Journal of Science), COPSE (Bergman-Lenton-Watson 2004), and the empirical compilation of Foster-Royer-Lunt 2017 reconstruct CO2 over 500+ Myr by combining mass balance for silicate weathering–volcanic outgassing, organic carbon burial, and isotopic constraints. Royer 2014 Treatise on Geochemistry 6 reviewed sensitivity of long-term CO2 to weathering parameters. The Veizer δ18O carbonate compilation (Veizer-Ala-Azmy-Bruckschen-Buhl-Bruhn-Carden-Diener-Ebneth-Godderis-Jasper-Korte-Pawellek-Podlaha-Strauss 1999 Chemical Geology 161 and Veizer-Prokoph 2015 Earth-Science Reviews 146) provides a Phanerozoic seawater temperature trajectory used in calibrations.

5. Data archives and modelling consortia

  • NOAA Paleoclimatology Program (NCEI). Open archive at ncei.noaa.gov/products/paleoclimatology hosting ice-core, tree-ring, coral, speleothem, marine and lake sediment, and historical climate records. Distribution under World Data Service for Paleoclimatology.
  • PAGES (Past Global Changes). International coordination office, Bern, Switzerland. Working groups on 2k Network (Common Era), PALSEA (paleo sea level), CVAS (climate variability across scales), PaleoVar, Iso2k, Ocean2k, Temp12k.
  • Iso2k (Konecky-McKay-Churakova(Sidorova)-Comas-Bru-DeLong-Falster-Fischer-Jones-Jonkers-Kaufman-Leduc-Managave-Martrat-Opel-Orsi-Partin-Sayani-Thomas-Thompson-Tyler-Abram-Atwood-Cartapanis-Conroy-Curran-Dee-Deininger-Divine-Kern-Porter-Stevenson-vonGunten-Wahl 2020 Earth System Science Data 12). Global isotope-based hydroclimate database, >700 records.
  • PMIP (Paleoclimate Modelling Intercomparison Project), now in its fourth phase (PMIP4) within CMIP6. Standardized boundary conditions for mid-Holocene (6 ka), Last Interglacial (127 ka), Last Glacial Maximum (21 ka), Last Millennium (850–1849 CE), and Mid-Pliocene Warm Period (3.2 Ma) experiments. Kageyama-Braconnot-Harrison-Haywood-Jungclaus-Otto-Bliesner-Peterschmitt-Abe-Ouchi-Albani-Bartlein-Brierley-Crucifix-Dolan-Fernandez-Donoso-Fischer-Hopcroft-Ivanovic-Lambert-Lunt-Mahowald-Peltier-Phipps-Roche-Schmidt-Tarasov-Valdes-Zhang-Zhou 2018 Geoscientific Model Development 11.
  • PlioMIP2. Pliocene experiment within PMIP4 (Haywood et al. 2020).
  • DeepMIP. Eocene experiment within PMIP4 (Lunt et al. 2017, 2021 Climate of the Past 17).
  • CenCO2PIP (Cenozoic CO2 Proxy Integration Project). 2023 Science synthesis above.
  • LIPDverse (Linked Paleo Data). Standard format for paleoclimate timeseries (McKay-Emile-Geay 2016 Climate of the Past 12) supporting Iso2k, Temp12k, PaleoStorm.
  • Neotoma Paleoecology Database. Pollen, vertebrate fossils, plant macrofossils, diatoms, charcoal.
  • NOAA International Tree-Ring Data Bank (ITRDB). >4 000 chronologies.
  • PANGAEA. General geoscience data archive (Bremen) hosting many marine and lake paleoclimate datasets.
  • Stratigraphy.org (International Commission on Stratigraphy). Authoritative geological time scale (GTS 2020).

6. Paleo-constrained climate sensitivity

Paleoclimate provides the only empirical estimates of climate sensitivity at forcings comparable to or larger than anthropogenic. Two distinct sensitivity metrics matter:

  • Equilibrium Climate Sensitivity (ECS). Global mean surface temperature change in response to a sustained doubling of CO2, holding ice sheets, vegetation, and aerosols at present-day values. CMIP6 model range: 1.8–5.6°C (Zelinka-Myers-McCoy-Po-Chedley-Caldwell-Ceppi-Klein-Taylor 2020 Geophysical Research Letters 47). Sherwood-Webb-Annan-Armour-Forster-Hargreaves-Hegerl-Klein-Marvel-Rohling-Watanabe-Andrews-Braconnot-Bretherton-Foster-Hausfather-vonderHeydt-Knutti-Mauritsen-Norris-Proistosescu-Rugenstein-Schmidt-Tokarska-Zelinka 2020 Reviews of Geophysics 58 (the “World Climate Research Programme assessment”) combined process, historical, and paleo evidence for ECS = 2.6–3.9°C (66% range).
  • Earth System Sensitivity (ESS). Same forcing but with ice sheets, vegetation, and slow feedbacks free to equilibrate. PALAEOSENS Project (Rohling-PALAEOSENS Project Members 2012 Nature 491) estimated ESS roughly 50% larger than ECS based on Cenozoic data; mid-Pliocene comparison (PlioMIP2 + PRISM4) gives ESS ~5–8°C per CO2 doubling.

The discrepancy between proxy SSTs in past warm intervals (e.g., LGM cool, PETM hot) and the warming simulated by CMIP-class GCMs at corresponding CO2 has been called the “paleo-vs-historical” sensitivity gap. Sherwood et al. 2020 explicitly used the LGM cold extreme to anchor the low end of the ECS distribution and the mid-Pliocene and Eocene to anchor the high end.

7. Open problems

  • MPT mechanism. No consensus on the dominant trigger of the obliquity-to-100kyr transition; combinations of CO2 trends, regolith erosion, and nonlinear ice-sheet dynamics remain in play.
  • Tropical Pacific zonal SST gradient in the Pliocene. Proxies indicate a weak east-west gradient (a “permanent El Niño-like” state, Wara-Ravelo-Delaney 2005 Science 309); models produce a gradient closer to modern. The bias has implications for ENSO under future warming (Tierney-Haywood-Feng-Bhattacharya-Otto-Bliesner 2019 Geophysical Research Letters 46).
  • Equable Eocene climates. Models systematically underestimate continental interior winter warmth at high CO2; mechanisms (convective cloud feedback, polar stratospheric clouds, ocean heat transport) remain debated (Hollis-Dunkley-Jones-Anagnostou-Bijl-Cramwinckel-Cui-Dickens-Edgar-Eley-Evans-Foster-Frieling-Inglis-Kennedy-Mansor-Markwick-Naafs-Pancost-Pancost-Pearson-Royer-Sluijs-Steinig-Stickley-Sutton-Vahlenkamp-Wade-Wilson-Wing-Witkowski-Lunt 2019 Geoscientific Model Development 12 — the DeepMIP database).
  • PETM trigger and rate. The mass and rate of carbon injection that produced the δ13C excursion remain uncertain by a factor of 3, depending on assumed source δ13C ( −60‰ for biogenic methane, ~−25‰ for organic carbon, ~−5‰ for mantle CO2).
  • Common Era reconstructions. Differences across PAGES 2k iterations and the Mann-Bradley-Hughes 1998 “hockey stick” lineage of debates highlight sensitivity to proxy selection, statistical method (PCA, composite-plus-scale, Bayesian hierarchical models), and uncertainty propagation.
  • 8.2 ka and 4.2 ka event mechanisms. Magnitude and global signature of the 4.2 ka event in particular remain contested; its definition as a chronostratigraphic boundary (Walker et al. 2018) is critiqued by Voosen 2018 Science 361.
  • Ice-sheet stability in Pliocene. Models disagree on whether the East Antarctic Ice Sheet partly collapsed during the mPWP; resolving this is critical for understanding multi-meter sea-level commitment under sustained warmth (DeConto-Pollard 2016 Nature 531).

8. Data assimilation and paleo-reanalysis

A growing subfield merges proxy networks with climate-model dynamics through data-assimilation techniques originally developed for numerical weather prediction.

8.1 Ensemble Kalman filter approaches

Steiger-Hakim-Steig-Battisti-Roe 2014 Journal of Climate 27 (Last Millennium Reanalysis, LMR) used offline ensemble Kalman filter assimilating PAGES 2k proxies into model priors from PMIP3 last-millennium simulations to produce annually resolved, gridded fields of temperature, precipitation, sea-level pressure for 0–2000 CE. Hakim-Emile-Geay-Noone-Anchukaitis-Tardif-Steiger-Perkins 2016 Earth and Space Science Open Archive extended LMR with iterative methods.

8.2 PHYDA and other reanalyses

PHYDA (Paleo Hydrodynamics Data Assimilation, Steiger-Smerdon-Cook-Cook 2018 Scientific Data 5) extends LMR with hydroclimate proxies; Tardif-Hakim-Perkins-Horlick-Erb-Emile-Geay-Anderson-Steig-Noone 2019 Climate of the Past 15 developed an online ensemble-square-root-filter variant. Comparable efforts include Coupled Model Reanalysis (Goosse-Crespin-Dubinkina-Loutre-Mann-Renssen-Sallaz-Damaze-Shindell 2012 Climate Dynamics 38).

8.3 Last Glacial Maximum proxy assimilation

Tierney-Zhu-King-Malevich-Hakim-Poulsen 2020 Nature 584 assimilated marine SST proxies into iCESM (isotope-enabled CESM) at LGM boundary conditions to constrain global mean cooling at 6.1 ± 0.4°C (rather than the earlier ~3°C estimates) and to infer high climate sensitivity. Osman-Tierney-Zhu-Tardif-Hakim-King-Poulsen 2021 Nature 599 extended through the full deglaciation, producing the LGMR (Last Glacial Maximum Reanalysis) covering 24–0 ka.

8.4 Limitations

Assimilation depends on model priors that may carry systematic bias (the “equable Eocene” problem of §7); proxy forward models with temperature-dependence assumptions; and assumed observation error structures. Sensitivity tests across model priors and proxy networks are essential. King-Tierney-Anchukaitis-Tardif-Hakim 2024 PNAS Nexus explored prior sensitivity systematically.

9. Stable-isotope hydrology and paleo-precipitation

Stable-water-isotope ratios (δ18O, δ2H/δD) preserve information about the hydrologic cycle on timescales from synoptic to deep time.

9.1 The Rayleigh distillation model

Cooling of a vapour parcel preferentially removes the heavier isotopologue (H218O, HDO) into condensate, leaving residual vapour progressively depleted. For closed-system Rayleigh distillation, residual vapour isotope ratio follows R/R0 = f^(α-1), where f is the remaining vapour fraction and α is the equilibrium fractionation factor. Real atmospheric trajectories deviate due to mixing, kinetic fractionation during evaporation and ice formation, and post-condensation exchange (Galewsky-Steen-Larsen-Field-Worden-Risi-Schneider 2016 Reviews of Geophysics 54).

9.2 The Global Network of Isotopes in Precipitation

The IAEA/WMO GNIP database (1961+) records monthly δ18O and δ2H in precipitation at ~1 100 stations globally. The Meteoric Water Line δ2H = 8 × δ18O + 10 (Craig 1961 Science 133) and deuterium excess (d = δ2H − 8 × δ18O) carry source-region and recycling information.

9.3 Isotope-enabled climate models

iCESM (Brady-Stevenson-Bailey-Liu-Noone-Nusbaumer-Otto-Bliesner-Tabor-Tomas-Wong-Zhang-Zhu 2019 Journal of Advances in Modeling Earth Systems 11), LMDZ-iso (Risi-Bony-Vimeux-Jouzel 2010 Journal of Geophysical Research 115), HadCM3-iso, ECHAM5/MPI-iso. Allow direct simulation of paleo-precipitation δ18O for comparison with speleothem and ice-core records.

9.4 Triple-oxygen isotopes and clumped isotopes

Δ17O = δ17O − 0.528 × δ18O (mass-independent fractionation) carries stratospheric input information; clumped isotopes Δ47 (CO2 with one 13C-18O bond) provide carbonate formation temperature independent of fluid δ18O (Ghosh-Adkins-Affek-Balta-Guo-Schauble-Schrag-Eiler 2006 Geochimica et Cosmochimica Acta 70; Bernasconi-Müller-Bergmann-Breitenbach-Fernandez-Hodell-Jaggi-Meckler-Millan-Ziegler 2018 Geochemistry, Geophysics, Geosystems 19 international standardisation).

10. Volcanic forcing records

Stratospheric sulfate aerosol from explosive volcanic eruptions cools the climate for years; the cooling is recorded in tree rings, ice cores, and corals.

10.1 Ice-core sulfate records

Bipolar synchronisation of GISP2 + Vostok ice-core sulfate signals identifies eruption events (Sigl-Winstrup-McConnell-Welten-Plunkett-Ludlow-Büntgen-Caffee-Chellman-Dahl-Jensen-Fischer-Kipfstuhl-Kostick-Maselli-Mekhaldi-Mulvaney-Muscheler-Pasteris-Pilcher-Salzer-Schüpbach-Steffensen-Vinther-Woodruff 2015 Nature 523) — produced a new chronology of explosive eruptions for the past 2 500 yr. Identified the 536 CE / 540 CE / 547 CE eruption sequence that produced the Late Antique Little Ice Age (Büntgen-Myglan-Ljungqvist-McCormick-DiCosmo-Sigl-Jungclaus-Wagner-Krusic-Esper-Kaplan-deVaan-Luterbacher-Wacker-Tegel-Kirdyanov 2016 Nature Geoscience 9).

10.2 Quaternary mega-eruptions

  • Toba (74 ka). Largest known Quaternary eruption (VEI 8, magnitude ~7); produced ~2 800 km^3 of tephra and ~10^17 g sulfate; possible “bottleneck” in human population debated (Williams-Ambrose-vanderKaars-Ruehlemann-Chattopadhyaya-Pal-Chauhan 2009 Palaeogeography, Palaeoclimatology, Palaeoecology 284).
  • Campi Flegrei (39 ka). Italy; coincident with Heinrich Stadial 4 cold pulse.
  • Oruanui / Taupo (25.4 ka). New Zealand; VEI 8.
  • Mt. Mazama / Crater Lake (7.7 ka). Pacific Northwest; tephra used as chronostratigraphic marker.
  • Santorini / Thera (~1620 BCE). Bronze Age Aegean.

10.3 Holocene volcanism and societal impacts

The 1815 Tambora “year without a summer” 1816 produced famine across North America and Europe. The 1257 Samalas eruption (Lavigne-Degeai-Komorowski-Guillet-Robert-Lahitte-Oppenheimer-Stoffel-Vidal-Surono-Pratomo-Wassmer-Hajdas-Hadmoko-deBelizal 2013 PNAS 110) preceded the early-medieval cooling. Tephra chronologies (Plunkett-Pilcher 2018) date numerous eruptions through the Holocene.

11. Continental ice-sheet history

11.1 Antarctic ice-sheet stability

Antarctic glaciation onset at the Eocene-Oligocene boundary (33.7 Ma; Coxall-Wilson-Pälike-Lear-Backman 2005 Nature 433) — sustained East Antarctic ice sheet through subsequent global warm intervals (MMCO, mid-Pliocene). West Antarctic Ice Sheet (WAIS) is younger and less stable; possibly fully or partly collapsed during MIS 5e (Last Interglacial, with sea level 6–9 m above present, requiring contribution from WAIS).

Marine Ice Sheet Instability (Schoof 2007 Journal of Fluid Mechanics 573; Joughin-Smith-Medley 2014 Science 344) — grounding lines on reverse-sloping beds are unstable; once retreat begins it may proceed unimpeded. Thwaites and Pine Island glaciers (Amundsen Sea sector WAIS) are presently in suspected instability mode (Rignot-Mouginot-Morlighem-Seroussi-Scheuchl 2014 Geophysical Research Letters 41).

Marine Ice Cliff Instability (DeConto-Pollard 2016 Nature 531) — vertical ice cliffs above ~100 m become structurally unstable. Inclusion produces higher projected end-of-21st-century sea-level rise; later assessments (Edwards-Nowicki-Marzeion-Hock-Goelzer-Seroussi-Jourdain-Slater-Turner-Smith-McKenna-Simon-Abe-Ouchi-Gregory-Larour-Lipscomb-Payne-Shepherd-Agosta-Alexander-Albrecht-Anderson-Asay-Davis-Aschwanden-Barthel-Bliss-Calov-Chambers-Champollion-Choi-Cullather-Cuzzone-Dumas-Felikson-Fettweis-Fujita-Galton-Fenzi-Gladstone-Golledge-Greve-Hattermann-Hoffman-Humbert-Huss-Huybrechts-Immerzeel-Kleiner-Krapp-Larour-Le-Clec’h-Lee-Leguy-Little-Lowry-Malles-Martin-Maussion-Morlighem-O’Neill-Nias-Pattyn-Pelle-Price-Quiquet-Radić-Reese-Rounce-Rückamp-Sakai-Shafer-Schlegel-Shannon-Smith-Straneo-Sun-Tarasov-Trusel-VanBreedam-vandeWal-vanDeneylen-Vizcaino-Wal-Winkelmann-Zekollari-Zhao-Zhang-Zwinger 2021 Nature 593) downweighted MICI but did not exclude.

11.2 Greenland ice-sheet history

Greenland Ice Sheet evidence for Pliocene partial collapse (Schaefer-Finkel-Balco-Alley-Caffee-Briner-Young-Gow-Schwartz 2016 Nature 540 cosmogenic isotope evidence from sub-glacial bedrock). Future destabilisation threshold debated: Robinson-Calov-Ganopolski 2012 Nature Climate Change 2 estimated ~1.6°C above pre-industrial; later assessments place threshold in 1.5–3°C range. Inferred 1+ m sea-level commitment for sustained 2°C warming over multi-millennia.

11.3 Sea-level reconstruction

  • Coral terraces (Barbados, Huon, Tahiti).
  • Last Interglacial markers (eolianites, raised shorelines, in-situ corals; Hibbert-Rohling-Dutton-Williams-Chutcharavan-Zhao-Tamisiea 2016 Quaternary Science Reviews 145 compilation).
  • Glacial-isostatic-adjustment (GIA) correction (Mitrovica-Milne-Davis 2001 Geophysical Journal International 147; Peltier ICE-6G_C 2015 Journal of Geophysical Research 120).
  • Far-field sea-level sites least affected by GIA (Barbados, Tahiti, Seychelles).

11.4 Greenland Ice Core Project chronologies

Greenland Ice Core Chronology 2005 (GICC05; Rasmussen-Andersen-Svensson-Steffensen-Vinther-Clausen-Siggaard-Andersen-Johnsen-Larsen-Bigler-Röthlisberger-Fischer-Goto-Azuma-Hansson-Ruth 2006 Journal of Geophysical Research 111) annual layer counts through 60 ka with cumulative maximum counting error ~2 600 yr at 60 ka. Synchronised to AICC2012 Antarctic chronology via CH4 tie points.

12. Carbon cycle perturbations through deep time

12.1 Cenozoic CO2 evolution

The Cenozoic (66 Ma – present) trajectory of atmospheric CO2 declined from ~1 000+ ppm in the early Eocene to pre-industrial 280 ppm, modulated by silicate weathering responses to mountain building (Raymo-Ruddiman 1992 Nature 359 — Himalayan uplift hypothesis), organic carbon burial in marine sediments and high-latitude shelves, and volcanic outgassing variations associated with seafloor spreading and large igneous provinces.

Major Cenozoic events:

  • K-Pg boundary (66.0 Ma). Chicxulub asteroid impact + Deccan Traps eruption; mass extinction. Schoene-Eddy-Samperton-Keller-Adatte-Bowring-Khadri 2019 Science 363 dated Deccan main eruption phase to 600 kyr around K-Pg.
  • PETM (56.0 Ma). Discussed in §4.4.
  • Eocene Thermal Maxima (ETM2 at 53.5 Ma, ETM3 at 52.4 Ma, “X” event). Smaller analogs of PETM.
  • MECO (Middle Eocene Climatic Optimum, ~40 Ma). Brief CO2 + temperature spike.
  • Eocene-Oligocene boundary (33.9 Ma). Major cooling, onset of continental-scale Antarctic glaciation. CO2 threshold for ice-sheet inception inferred ~600–800 ppm (DeConto-Pollard 2003 Nature 421).
  • Mi-1 (23 Ma). Brief major glaciation early Miocene.
  • MMCO (17–14.5 Ma). Discussed in §4.2.
  • Pliocene cooling (3 Ma). Onset of NH glaciation.

12.2 Mesozoic CO2 and the carbonate compensation depth

Mesozoic CO2 levels (Hönisch et al. 2023 CenCO2PIP synthesis extended; Foster-Royer-Lunt 2017 Phanerozoic compilation) ranged from ~500 to several thousand ppm. The Cretaceous Ocean Anoxic Events (OAE1a at 120 Ma, OAE1b 113 Ma, OAE2 “Bonarelli” 93.9 Ma at the Cenomanian-Turonian boundary) show large carbon-isotope excursions associated with volcanism (Caribbean, Ontong Java large igneous provinces), warm climates, and global ocean anoxia (Schlanger-Jenkyns 1976 Geologie en Mijnbouw 55; Jenkyns 2010 Geochemistry, Geophysics, Geosystems 11). OAE2 produced ~5–8°C tropical SST and substantial sea-level rise.

12.3 Permian-Triassic and Triassic-Jurassic boundaries

The end-Permian extinction (~252 Ma) — largest known mass extinction (~95% marine species) — coincided with Siberian Traps volcanism producing rapid CO2 + SO2 release, ocean anoxia, warming of ~6–10°C (Joachimski-Lai-Shen-Jiang-Luo-Chen-Chen-Sun 2012 Geology 40). End-Triassic extinction (201 Ma) coincided with Central Atlantic Magmatic Province (CAMP) volcanism. End-Cretaceous (66 Ma) Deccan Traps add a third LIP-extinction correlation.

12.4 Carboniferous-Permian transition

Carboniferous CO2 fell to ~300–400 ppm under intense organic-carbon burial in coal swamps; combined with reduced silicate weathering at low CO2 produced the Late Paleozoic Ice Age (~330–260 Ma), the longest sustained icehouse of the Phanerozoic. Continental configuration (Pangaea) and elevated O2 (perhaps 30%, supporting giant Carboniferous insects) characterise the period.

13. Frontier methods

13.1 Compound-specific isotope analysis

Single biomarker compounds purified via gas chromatography and analysed for δ13C, δ2H, δ18O isolate single biological producers from bulk sediment averages. Standard for leaf-wax δ2H precipitation reconstruction (Sachse-Billault-Bowen-Chikaraishi-Dawson-Feakins-Freeman-Magill-McInerney-vanderMeer-Polissar-Robins-Sachs-Schmidt-Sessions-White-West-Kahmen 2012 Annual Review of Earth and Planetary Sciences 40). High-precision compound-specific Δ47 thermometry from biomarkers (Eiler 2007, 2013).

13.2 Cosmogenic nuclide dating

10Be, 26Al, 36Cl produced by spallation in surface rocks; ratio and concentration constrain exposure age and erosion rate (Lal 1991 Earth and Planetary Science Letters 104; Gosse-Phillips 2001 Quaternary Science Reviews 20). Applied to glacial-moraine chronologies, fault-scarp dating, paleo-precipitation via cave-deposit dating, and paleo-CO2 via meteoric 10Be flux variations.

13.3 Ancient DNA

Sedimentary ancient DNA (sedaDNA, Willerslev-Hansen-Binladen-Brand-Gilbert-Shapiro-Bunce-Wiuf-Gilichinsky-Cooper 2003 Science 300; Pedersen-Overballe-Petersen-Ermini-Sarkissian-Haile-Hellstrom-Spens-Thomsen-Bohmann-Cappellini-Schnell-Wales-Caroe-Campos-Schmidt-Gilbert-Hansen-Orlando-Willerslev 2015 Philosophical Transactions of the Royal Society B 370) recovers DNA from permafrost and lake sediments back ~2 Ma (Kjær-Winther Pedersen-De Sanctis-De Cahsan-Korneliussen-Michelsen-Sand-Jelavić-Ruter-Schmidt-Kjeldsen-Tesakov-Snowball-Gosse-Alsos-Wang-Dockter-Rasmussen-Jørgensen-Skadhauge-Prohaska-Kristensen-Bjerager-Allentoft-Capellini-Gilbert-Mortensen-Markussen-Bocherens-Hofreiter-Dalén-Stafford-Stenderup-Sikora-Lipotinen-Margaryan-Wilson-Bramanti-Carmagnini-Pinhasi-Foley-Lindqvist-Lewallen-Bennett-Bristow-Brusgaard-Lorenzen-Heintzman-Korotin-Schaltout-Mehler-Bocherens-Sukhova-Steinmetz-Hofreiter-Dalén-Friebe-Stenderup-Sikora-Prohaska-Schaal-Wickström-Holdaway-Beckmann-Sklenar-Andreasen-Ascough-Nye-Stafford-Bunce-Møldrup-Hansen-Borchhardt-Margaryan-Sjögren-Allentoft-Sikora-Stafford-Møldrup-Stenderup-Sikora-Christophersen-Nielsen-Brock-Schöne-Salonen-Sand-Mortensen-Wilson-Møldrup-Bjerager-Korneliussen-Lokugamage-Bjerager-Friebe-Dalén-Christensen-Mortensen-Vinner-Allentoft-Sikora-Boertien-Margaryan-Stafford-Bocherens-Stenderup-Mehler-Schaltout-Bocherens-Bjerager-Korneliussen-Brusgaard-Brusgaard-Bracalini 2022 Nature 612 — Kap Kobenhavn Formation, 2 Ma, oldest sedaDNA so far).

13.4 Tephrochronology

Volcanic-ash horizons provide isochrones traced across thousands of km. Useful synchronisation tools include the Toba 74 ka tephra (visible in Indian Ocean cores), the Laacher See 12.9 ka tephra (European reference horizon coincident with Younger Dryas onset), Vedde Ash (12.0 ka), Saksunarvatn Ash (10.3 ka). The INTIMATE working group coordinates tephrochronology across NH paleoclimate records.

13.5 Paleomagnetism and magnetic stratigraphy

Geomagnetic polarity reversals provide a chronologic framework for marine sediments back >150 Ma. The Cande-Kent 1995 Journal of Geophysical Research 100 polarity timescale, refined by Gradstein 2020 GTS, ties to absolute U-Pb ages. The Laschamp event (~41 ka) is the youngest geomagnetic excursion; the Blake event (~120 ka); Iceland Basin (~190 ka). Paleomagnetic intensity (relative paleo-intensity, RPI) records (Channell-Hodell-Singer-Xuan 2009 Earth and Planetary Science Letters 282) provide independent chronometry.

14. Connection to projection and policy

Paleoclimate evidence enters policy and projection in three ways. First, paleo data anchor the high-end of climate sensitivity distributions used in the IPCC ECS assessment, with the Sherwood et al. 2020 synthesis explicitly weighting paleo lines of evidence comparably to process-based and historical. Second, mid-Pliocene and Last Interglacial sea levels (17 m, 6–9 m respectively) provide empirical benchmarks for the multi-millennial sea-level commitment of sustained warming above pre-industrial — relevant to the framing of “long-term temperature goals” in the Paris Agreement. Third, abrupt-event records (DO, Younger Dryas, 8.2 ka) are the empirical basis for the “tipping point” discourse (Lenton-Held-Kriegler-Hall-Lucht-Rahmstorf-Schellnhuber 2008 PNAS 105; Armstrong McKay-Staal-Abrams-Winkelmann-Sakschewski-Loriani-Fetzer-Cornell-Rockström-Lenton 2022 Science 377) feeding the “Hothouse Earth” framing (Steffen-Rockström-Richardson-Lenton-Folke-Liverman-Summerhayes-Barnosky-Cornell-Crucifix-Donges-Fetzer-Lade-Scheffer-Winkelmann-Schellnhuber 2018 PNAS 115).

Further reading

  • Bradley, R. S. 2015. Paleoclimatology: Reconstructing Climates of the Quaternary (3rd ed.).
  • Cronin, T. M. 2010. Paleoclimates: Understanding Climate Change Past and Present.
  • Ruddiman, W. F. 2014. Earth’s Climate: Past and Future (3rd ed.).
  • Bender, M. L. 2013. Paleoclimate.
  • Imbrie, J. and K. P. Imbrie 1979. Ice Ages: Solving the Mystery.
  • Hoffman, P. F. et al. 2017. “Snowball Earth climate dynamics and Cryogenian geology-geobiology.” Science Advances 3.
  • Zachos, J. et al. 2001. “Trends, rhythms, and aberrations in global climate 65 Ma to present.” Science 292.
  • Rohling, E. J. et al. (PALAEOSENS) 2012. “Making sense of palaeoclimate sensitivity.” Nature 491.
  • Tierney, J. E. et al. 2020. “Past climates inform our future.” Science 370.
  • IPCC AR6 WG1 Chapter 2 (Gulev et al. 2021) — “Changing State of the Climate System.”
  • IPCC AR6 WG1 Chapter 7 (Forster et al. 2021) — “The Earth’s Energy Budget, Climate Feedbacks, and Climate Sensitivity.”
  • Sherwood, S. C. et al. 2020. “An assessment of Earth’s climate sensitivity using multiple lines of evidence.” Reviews of Geophysics 58.
  • PAGES 2k Consortium 2019. “Consistent multidecadal variability in global temperature reconstructions and simulations over the Common Era.” Nature Geoscience 12.
  • Hönisch, B. et al. (CenCO2PIP) 2023. “Toward a Cenozoic history of atmospheric CO2.” Science 382.
  • Haywood, A. M. et al. 2020. “The Pliocene Model Intercomparison Project Phase 2.” Climate of the Past 16.
  • Lunt, D. J. et al. 2021. “DeepMIP: model intercomparison of early Eocene climatic optimum experiments.” Climate of the Past 17.

Adjacent

Graph View