GLACIAL CYCLES' TOWARD A NEW PARADIGM Didier Paillard Laboratoire des Sciences du Climat et de I'Environnement, CEA-CNRS Centred'Etudes de Saclay,Orme desMerisiers Gif-sur-Yvette, France Abstract. The largest environmentalchanges in the sheets,whereas recent high-resolution data from ice and recent geologicalhistory of the Earth are undoubtedly marine sediment cores do not support such a gradual the successionsof glacial and interglacialtimes. It has scenario.Most of the temperature rise at the last termi- been clearly demonstratedthat changesin the orbital nation occurred over a few decades in the Northern parametersof our planet have a crucial role in these Hemisphere,indicating a major and abrupt reorganiza- cycles.Nevertheless, several problems in classicalastro- tion of the ocean-atmospheresystem. Similarly, huge nomical theory of paleoclimatehave indeed been iden- icebergdischarges during glacial times, known as Hein- tified: (1) The main cyclicityin the paleoclimaticrecord rich events,clearly demonstratethat ice sheet changes is close to 100,000 years, but there is no significant may also be sometimesquite abrupt. In light of these orbitally inducedchanges in the radiativeforcing of the recent paleoclimaticdata the Earth climate systemap- Earth in this frequencyrange (the "100-kyrproblem"); pears much more unstableand seemsto jump abruptly (2) the most prominent glacial-interglacialtransition betweendifferent quasisteady states. Using the concept occursat a time of minimalorbital variations (the "stage of thresholds,this new paradigmcan be easilyintegrated 11 problem);and (3) at ---0.8Ma a changefrom a 41-kyr into classicalastronomical theory and compared with dominant periodicity to a 100-kyr periodicity occurred recent observationalevidence. If the ice sheet changes withoutmajor changesin orbitalforcing or in the Earth's are, by definition, the central phenomenonof glacial- configuration(the "late Pleistocenetransition prob- interglacialcycles, other componentsof the climate sys- lem"). Additionally,the traditionalview statesthat the tem (atmosphericCO2 concentration,Southern Ocean climatesystem changes slowly and continuouslytogether productivity,or globaldeep-ocean circulation) may play with the slow evolution of the large continental ice an even more fundamentalrole in these climatic cycles. 1. INTRODUCTION andso forth. The wholeEarth participatesin the dynamics of ice agesin a complexfashion, and its componentsare The first astronomicaltheory of paleoclimatesis al- tied togetherthrough a densenetwork of feedbacks.In readymore than 150years old (a detailedaccount of the particular,the atmosphericconcentration of CO2 and the historyof this scientificadventure is givenby Imbrie and existenceof climaticthresholds appear to have a funda- Imbrie [1979]). The astronomicalforcing is now well mentalrole in the glacial-interglacialcycles. In the current known, at least for the late Pleistocene. Recent advances contextof anthropogenicglobal warming, the understand- in geochemistryhelped to quantifythe geologicalrecord, ing of the dynamicsof ice ages,the largestrecent changes and it is now evident that climatic cycleshave frequen- in the climatesystem, is becominga key scientificissue. cies nearly identical to the Earth's orbital frequencies. In section 2 I will briefly mention some important However, the story is not finished,since we still do not historicmilestones in the discoveryof ice agesand then understandhow the climatesystem works and how small present, in more detail, the astronomicaltheory and changesin the insolationat the top of the atmospherecan some simple conceptualmodels. In section 2.4 I will be amplifiedby the Earth systemto createthe large cli- discusshow new observationaland conceptualadvances matic changesassociated with glacial-interglacialcycles. may help define a new paradigmfor glacial cyclesthat Traditionally,ice age modelshave concentratedon the could solve the traditional difficulties associated with behaviorof the largeNorthern Hemisphere ice sheets,the classicalastronomical theory. Laurentideand the Fennoscandian.In light of recentpa- leoclimaticdata this approachnow appearsinsufficient. 2. ASTRONOMICAL THEORY OF PALEOCLIMATES Indeed,ice agecycles involve a reorganizationnot onlyof the ice sheetsbut alsoof the ocean-atmospheresystem, the 2.1. FromGeology, to Astronomy,to Geochemistry deepocean and its sedimentaryinterface, ocean chemistry, The idea that the Earth experiencedsevere glacia- the carboncycle, the terrestrialand marine ecosystems, tions in the past originates at the beginning of the Copyright2001 by the AmericanGeophysical Union. Reviewsof Geophysics,39, 3 / August2001 pages325-346 8755-1209/01/2000RG000091 $15.00 Papernumber 2000RG000091 ß 325 ß 326 ß Paillard: GLACIAL CYCLES 39, 3 / REVIEWSOF GEOPHYSICS nineteenthcentury. Agassiz [1838] was among the first to changes,not by temperature changes.With the studyof recognizethat glaciationwas the most natural explana- foraminiferal assemblagesin marine sedimentcores, Im- tion for the erratic boulders, moraines, and deeply brie and Kipp [1971] could confirm these results and scratchedbedrocks that could be found in many places provide quantitativeestimates of the glacial-interglacial in the Alps, Scotland,and North America. Though sev- temperature changes.Dating methodsbased on the ra- eral othershad suggestedmajor glacialadvances before, dioisotopesof uranium, thorium, and potassiumwere Agassiz widely promoted the idea of an ice age and alsodeveloped at this time. When appliedto fossilcoral starteda scientificdebate that lastedmore than 30 years. reefs [Broecker,1966; Broecker et al., 1968], as well as to A few years after Agassiz,Adh•mar [1842] suggested magneticreversals [Cox et al., 1963, 1964], a cyclicityof that the orbital variationsof the Earth could be respon- 100 kyr was clearly demonstrated [Broeckerand van sible for climatic changes.Adh6mar's theory was based Donk, 1970;Hays et al., 1969;Kukla, 1975].However, the on the known precessionof the equinoxand suggested Milankovitch theory stated that the main cyclesshould that glaciationswere causedby the changein the lengths be at 23 and 41 kyr. Careful spectralanalysis of marine of the seasons.Glaciations, causedby longer winters, sediment cores led to the clear confirmation of the would thus occur every 23 kyr in the Northern Hemi- astronomicaltheory. Indeed, besidesthe 100-kyr cycle, sphere,with ice agesoccurring in the SouthernHemi- three other cyclescould be identified:41, 23, and 19 kyr spherein oppositephase, as shownby the current pres- [Hayset al., 1976].The computationsof the astronomical enceof the Antarctic Ice Sheet.This theorywas rapidly time series were now made easier with the help of proved to be incorrect since the annual mean solar computers,and Vemekar[1972] and Berger[1977, 1978] heating at the top of the atmospheredoes not change showed that the precessionfrequency is split into a with precession.Nevertheless, the idea of cyclicglacia- 23-kyr and a 19-kyr cycle. Ice age cyclesare therefore tions forced by the Earth's orbital changeswas taken undoubtedlylinked in some fashion to Earth's orbital further by Croll [1875], who elaboratedthe first astro- variations. nomicaltheory of paleoclimate.Croll hypothesizedthat precessionalforcing, though only seasonal,might be 2.2. Orbital Forcing crucial and that winter insolation might be critical. Accordingto Kepler's laws, the Earth's orbit around Colder winterswould producelarger areascovered with the Sunis an ellipse,and one of its two foci is roughlythe snow, which could lead to glacial age becauseof the Sun. This orbit is also influencedby the motion of the snow albedo feedback.He also showedthe importance other planetsof the solar system.Among the geometri- of the modulationby the 100-kyreccentricity changes for cal characteristicsof the Earth's orbit, only two have an this precessionalforcing. He further hypothesizedthat influenceon the solarheating receivedby the Earth: the the changingtilt of the Earth should play a role, and ellipsesemimajor axis a, which measuresthe size of the being aware that the astronomicalforcing is small, he ellipse,and its eccentricitye (definedas e = c/a, where tried to find someinternal amplifyingmechanisms in the c is the distancefrom focusto center),which measures ocean circulation. its elongation(Figure 2). There is no theoreticalor Interest in an astronomicaltheory of glacial cycles experimentalreason to think that a has changedin the was renewed with the work of M. Milankovitch between past. In contrast,the eccentricitye changessignificantly 1920 and 1941 [Milankovitch,1941], in which he com- with 100-kyr,400-kyr, and 2-Myr periodicities.When the puted the solar radiation at the top of the atmosphere eccentricityis large, the ellipse is more elongated,the for different latitudes,taking into accountthe changesin annual mean Earth-Sun distanceis slightlysmaller, and eccentricity,precession, and tilt of the Earth. In contrast the energy captured by the Earth is increased.The to Croll's theory, Milankovitch argued that the summer global mean annual solar radiation receivedby Earth is seasonwas critical. Colder summersenable the persis- tence throughout the year of snowfieldsin some high- S So latitude regions,leading to a net accumulationof ice and W'•=4X/1 - e2- 4' to the buildingof ice sheets(Figure 1). The Milanko- vitch theory predicted that the climatic cyclicityshould where S is the solar radiation received at a distance a be mainly at 23 kyr, becauseof precession,and