The Greenland Ice Sheet Project 2 Depth-Age Scale: Methods and Results D

The University of Maine DigitalCommons@UMaine

Earth Science Faculty Scholarship Earth Sciences

11-30-1997 The Greenland Ice Sheet Project 2 Depth-age Scale: Methods and Results D. A. Meese

A. J. Gow

R. B. Alley

G. A. Zielinski

P. M. Grootes

See next page for additional authors

Follow this and additional works at: https://digitalcommons.library.umaine.edu/ers_facpub Part of the Glaciology Commons, and the Hydrology Commons

Repository Citation Meese, D. A.; Gow, A. J.; Alley, R. B.; Zielinski, G. A.; Grootes, P. M.; Ram, M.; Taylor, K. C.; Mayewski, Paul Andrew; and Bolzan, J. F., "The Greenland Ice Sheet Project 2 Depth-age Scale: Methods and Results" (1997). Earth Science Faculty Scholarship. 260. https://digitalcommons.library.umaine.edu/ers_facpub/260

This Article is brought to you for free and open access by DigitalCommons@UMaine. It has been accepted for inclusion in Earth Science Faculty Scholarship by an authorized administrator of DigitalCommons@UMaine. For more information, please contact [email protected]. Authors D. A. Meese, A. J. Gow, R. B. Alley, G. A. Zielinski, P. M. Grootes, M. Ram, K. C. Taylor, Paul Andrew Mayewski, and J. F. Bolzan

This article is available at DigitalCommons@UMaine: https://digitalcommons.library.umaine.edu/ers_facpub/260 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. C12, PAGES 26,411-26,423, NOVEMBER 30, 1997

The Greenland Ice Sheet Project 2 depth-age scale: Methods and results

D. A. Meese,• A. J. Gow,• R. B. Alley,2 G. A. Zielinski,3 P.M. Grootes,4 M. Ram,s K. C. Taylor,6 P. A. Mayewski,3 and J. F. Bolzan7

Abstract. The GreenlandIce SheetProject 2 (GISP2) depth-agescale is presentedbased on a multiparametercontinuous count approach,to a depth of 2800 m, using a systematic combinationof parametersthat have never been used to this extent before. The ice at 2800 m is dated at 110,000years B.P. with an estimatederror rangingfrom 1 to 10% in the top 2500 m of the core'andaveraging 20% between2500 and 2800 m. Parameters usedto date the core includevisual stratigraphy, oxygen isotopic ratios of the ice, electricalconductivity measurements, laser-light scattering from dust,volcanic signals, and major ion chemistry.GISP2 agesfor major climatic eventsagree with independentages basedon varve chronologies,calibrated radiocarbon dates, and other techniqueswithin the combineduncertainties. Good agreementalso is obtainedwith Greenland Ice Core Project ice core dates and with the SPECMAP marine timescaleafter correlation through the •80 of 02. Althoughthe core is deformedbelow 2800 m andthe continuity of the record is unclear,we attemptedto date this sectionof the core on the basisof the laserlight scatteringof dust in the ice.

Introduction missing years due to deflation by wind or sublimation are unlikely. The stratigraphicrecord at Summit, Greenland, oc- In the study of geologicmaterials containing paleoclimatic casionally reveals periods of very minor melt [Alley and records,it is imperative that there be a means of dating the Anandakrishnan,1995] but no more than about onceper cen- material. In most cases,some type of radioisotopicdating has tury. The GISP2 core may thus be expectedto yield a long, beenused, such as •4C or U/Th. The datingof somematerials high-quality timescale. has been obtainedusing chemical dating techniquescombined We haveidentified several parameters exhibiting annual sig- with counting of annual layers, i.e., tree rings, corals, and nals and used these in combinationto determine a chronology lake/ocean sediment varves. While some terrestrial deposits for the GISP2 ice core. The nature of these parameters,how can representaccumulations over long periods of time, very they compareto each other, and the strengthsand weaknesses few contain identifiable annual signalsthat allow for accurate of each as a dating tool are describedin detail below. Addi- dating of these deposits. tionally,means of verificationof this depth-agescale and com- Glaciers and ice sheets receive an annual net accumulation parisonswith published dates of major climatic events are that is a combinationof accumulationfrom precipitationand discussed. wind-blowndeposition and lossesdue to sublimation,melt, and erosion by wind. Contained within this accumulation Methods record are chemicalsignals and the manifestationof ongoing physicalprocesses that may provide annual or seasonalindi- Age dating of the GISP2 ice core was accomplishedby cators. On most glaciers and ice sheets,annual or seasonal identifyingand countingannual layers using a number of phys- signalscan be altered by melt, missingyears due to lack of or ical and chemicalparameters that includedmeasurements of low accumulation rates, sublimation, wind, or a combination of visualstratigraphy, electrical conductivity method (ECM), la- these. At the Greenland Ice Sheet Project 2 (GISP2) site, ser-lightscattering from dust (LLS), oxygenisotopic ratios of where accumulationis high (approximately0.24 m ice/yr), the ice (/3•80),major ion chemistry,and the analysisof glass shardsand ashfrom volcaniceruptions. Each of theseparam- eters (with the exceptionof volcanics)exhibits a distinctsea- 1U.S.Army Cold RegionsResearch and EngineeringLaboratory, Hanover, New Hampshire. sonal signal. 2EarthSystem Science Center, Pennsylvania State University, Uni- The definitivesummer stratigraphic signal at the GISP2 site versityPark. occursin the form of coarse-graineddepth-hoar layers formed 3ClimateChange Research Center, Institute for the Studyof Earth, by summer insolation [Alleyet al., 1990]. In the region around Oceansand Space,University of New Hampshire,Durham. 4LeibnizLaboratory Christian Albrechts University Kiel, Kiel, Ger- the GISP2 site the relief of the snowsurface is remarkablyflat. many. Sastrugi several centimeters in height may be produced by SDepartmentof Physics,State University of NewYork at Buffalo. storms,but subsequentdeposition, sublimation, and densifica- 6DesertResearch Institute, University and Community College Sys- tion tend to level the surface. Depth-hoar sequenceswere tem of Nevada, Reno. readilyrecognized in snowpits dug near the GISP2 drilling site 7ByrdPolar Research Center, Ohio StateUniversity, Columbus. [e.g.,Shutnan et al., 1995] (Figure 1). Copyright1997 by the American GeophysicalUnion. Recognition of the summer and winter stratigraphicse- Paper number 97JC00269. quencesin the snow pits provided important information in 0148-0227/97/97JC-00269509.00 identifyingstratigraphy in the ice core. In the core-processing

26,411 26,412 MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE

cm August 1990 acidpeak from nitricacid production in the stratosphere[Nef- o tel et al., 1985].Although ECM is an excellentseasonal indicator, as stated above,nonseasonal inputs from other sources maycause additional peaks which could be confusedwith the 25 annualsummer signal. In additionto being an annualindicator, ECM is alsoused for rapid identificationof major climatic Summer changes[Taylor et al., 1993a,b] and hasproved very useful in 5o the identificationof volcanicsignals. Duringthe late springand summerin Greenland,there is an winter influxof dust[Hamilton and Langway,1967]. This dustpeak is in part a resultof duststorms that occurin both hemispheres 75 duringthe spring/summerperiod [Ram and !lling, !994] and Summer atmosphericcirculation changeswhich enable the strato- sphericload to reachthe troposphere.Dust particlesin solid lOO ice and ice meltwaterscatter incident light, and the intensityof light scatteredat 90ø to the incidentlight directionis propor- winter tional to the massof suspendedparticulates [Hammer, 1977; 125 Ram and Illing, 1994;Ram et al., 1995].Liquid and solidLLS measurementswere comparedin severalmeters of core to a depthof 1812m andmatched very closely. These comparisons indicate that the solid LLS method can be used as an annual 15o Summer layerindicator in the Holoceneand Wisconsinan even though the signalchanges from one of depth hoar to layersof in- creased dust concentration. The solid measurements are also Figure 1. Stratigraphiclayering of a !.5-m snow pit exca- consistentwith both ECM and visible stratigraphyin this re- vated near the GISP2 drilling site in August1990. The summer and winter stratigraphicsequences are readily delineated. gionof the core.LLS wasa veryvaluable dating tool through- Summer stratigraphyis dominatedby layers of depth hoar out almost the entire length of the core, particularlyin the whichwere utilized as the primary marker for delineatingan- deeperice at GISP2,where the othertechniques either fail or nual layersin datingthe top 50% of the ice sheetat GISP2. becomeincreasingly unreliable. However, an increasedpartic- ulate concentrationmay not be restrictedto the spring or summerand additionalinfluxes of dust may occur during any line at the GISP2 Summit site, sectionsof ice were cut from the part of the year, creatingadditional peaks of a nonannual core lengthwisefor chemicaland isotopicanalysis. The cut nature. The LLS signalcan also be used as an indicator of surface on the remainder of the core was microtomed for ECM major climatechanges and of somevolcanic events. analysis.This microtomingalso provided a smooth,uniform Additionally,8•80 values(the relativedifference between surfacefor stratigraphicanalysis. Annual layeringin the upper the •80/•60 abundanceratios of the ice and Vienna standard portionsof the coreis easilyrecognized in transmittedlight. In mean oceanwater (V-SMOW) expressedin per mil (%0) of the eventof scoringon the outsideof the coreincurred during the ice) [Epsteinand Sharp,1959; Dansgaard, 1964; Grootes et drillingor if the layeringbecame faint, small"windows" were al., 1993]were usedto identify seasonalcycles between the cut on the bottom side of the ice (parallelto the microtomed surfaceand 300-m depthin conjunctionwith the stratigraphic, surface)to enhancevisibility. Stratigraphy in the form of ECM, and LLS techniques.While continuousseasonal isotope depth-hoarlayer sequences remained continuous through the samplingremained a viabledating technique in the top 300 m Holocene and into the glacial transition.During the Wiscon- of the core, the effects of diffusion rapidly obliterated the sinanglacial the stratigraphicsignal was not the characteristic seasonalsignal in deeper ice. depthhoar sequence of the Holocenebut rathercloudy bands Many volcanic eruption signals,including both volcanic resultingfrom significantchanges in seasonaldust concentra- aerosols(primarily identified as H2SO4) and tephra, were iden- tion in the ice. Overlapof the two typesof annuallayer signal tifiedthroughout the core[e.g., Zielinski et al., 1994,this issue], and successful calibration between the two in Preboreal ice is therebyproviding definitive tie pointsto whichthe annuallayer describedby in Alley et al. [thisissue (a)]. countingcould be compared(Table 1). Of particularimpor- The ECM providesa continuoushigh-resolution record of tancewere volcanicsignals found duringthe period of histor- low-frequencyelectrical conductivity of glacialice, which is ically dated eruptions(i.e., over the last 2000 yearswith the relatedto the acidityof the ice [Hammer,1980; Taylor et al., A.D. 79 eruption of Vesuviusbeing the oldest and largest 1992].The measurementis basedon the determinationof the explosiveeruption dated based on historicalrecords [Zielinski, currentflowing between two movingelectrodes with a poten- 1995]).There is somelag between the time of the eruptionand tial differenceof a few thousandvolts. Stronginorganic acids depositionat the sitewhich may introducea 1- to 3-yearerror suchas sulfuricacid from volcanicactivity and nitric acid con- in the datingof the volcanicsignals [Stuiver et al., 1995].As the trolled by atmosphericchemistry cause an increasein current. lag is not consistent,a specifictime framecannot be generally Conversely,when the acidsare neutralizeddue to alkalinedust assignedto accountfor this.Deposition from the 1912Katmai from continental sources or from ammonia due to biomass eruptiondid arrivein 1912.However, Laki lags! year andthe burning,the currentis reduced[Taylor et al., 1992].As such, signalthat is probablyrelated to the 1600eruption of Huayna- resultsfrom ECM can be used for a number of different types putinalags 3-4 years[Zielinski, 1995]. Although we are ableto of interpretations.The most importantfeature of the ECM link manyof the volcanicaerosol signals (SO42- and ECM data in relation to the depth-agescale is the spring/summer [Zielinskiet al., thisissue]) to a specificeruption with a fairly MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE 26,413 high degreeof confidence,it is locatingand identifyingvolca- Table 2. Error Estimatesof the GISP2 Depth-Age Scale nic glassfrom that same eruption that verifies the volcanic Depth, Age, Estimated signal and thus the absoluteage of the particular ice layer. m years B.P. Parameters Error, % Tephra hasbeen found in the GISP2 core with a composition closelymatching that of the candidateeruption for five of the 0-300 1,133 S, D, E, V, some L 1 tie pointslisted in Table 1. Althoughthere was an absenceof 300-719 3,289 S, E, L 1 719-1371 8,021 S, some E and L 2 glassshards from the Vesuviuseruption in the GISP2 core,the 1371-1510 9,374 S, E, L 2 known record of volcanismfails to provide another likely can- 1510-2250 39,852 S, some E and L 2 didate.Another very largevolcanic signal (third largestof the 2250-2340 44,583 S, some E and L 5-10 last 5000 years) that has been dated at 53 B.C. in the GISP2 2340-2500 56,931 L, some S and E 10 2500-2800 110,694 L, some S 20 core was also observedand subsequentlydated as 50 B.C. in 2800-3030 161,313 L, some S 9 the Camp Centurycore [Hammeret al., 1980],further validat- ing the GISP2 chronology.In the caseof the GISP2 core, layer S, stratigraphy;D, isotopes;E, ECM; L, LLS; V, volcanics. countsclosely matched the datesof all the historicaleruptions givenin Table 1 with the exceptionof Eldgja. In this instance, layer countingof the GISP2 core gives an age of A.D. 938 rameters and the effect on the dating are describedin more [Zielinskiet al., 1995] as opposedto A.D. 934 reportedin the detail below. Our ability to intercalibratemultiple annual in- literature [Hammeret al., 1980;Hammer, 1984],which is within dicatorsin the upper part of the core and learn their signal the calculatederror of 1% at this depth.It is importantto note characteristics allowed us to retain considerable confidence in that the timing of the Eldgja eruption event at A.D. 934 is our interpretationseven after some indicatorswere lost with based on other ice-core records from Greenland and is not an increasingdepth. actual historicaleruption date. Similarly,the A.D. 1259 event The greatestnumber of the parametersexhibiting an annual (sourcevolcano unknown) has been observed in ice coresfrom signalremained viable in the upperportion of the core (0-600 both polar regions[Langway et al., 1988;Palais et al., 1992]. m) prior to the onsetof the brittle ice zone and particularlyin Tephra was found in the GISP2 core for five of the eruptions thetop 300m where•80 alsoretained an identifiableannual listedin Table 1 [Palaiset al., 1991, 1992;Fiacco et al., 1994; signal.As each of the signalscarries its own structureor pat- Zielinskiet al., 1995]; thosenot found includeHuaynaputina, terns of change,this providedan opportunityto determinethe Hekla, and Vesuvius.The Huaynaputina and Hekla signals precisecharacteristics of eachsignal and the way in whichthey have been identified in other ice coresfrom Greenland [e.g., intercompared.Because this structureor pattern representing Hammer et al., 1980], and the Huaynaputinaeruption has also annual signalswas determinedfor each parameter,the count- been identifiedin an Antarcticcore [e.g.,Moore et al., 1991]. ing of layerscould be continueddown the lengthof the core. It In certain sectionsof the GISP2 core, changesin the prop- is important to note that each parameter is affectedby extra- erties and mechanical condition of the core, in addition to neousor miscellaneousevents, such as volcanic eruptions, for- changesin climate, have resulted in variations in the time est fires,etc., that may influencethe timing and intensityof the stratigraphicparameters to the extentthat someproved more signal.With time and experiencethe effectsof theseevents on valuablefor annual layer dating than others.In the region of the annual signalswere determined and taken into account. the core designatedthe brittle ice zone [Gow et al., this issue] This was particularly important when fewer parameterswere that occurred between 600 and 1400 m, relaxation stresses availableand was the primary reasonwhy it was so critical to exceedthe tensile strengthof the ice, leading to widespread use more than one parameter wherever possiblein obtaining fracturingof cores[Gow, 1971;Gow et al., this issue].Freshly an accurate depth-age scale. The visual stratigraphywas a cored ice in this zone is very brittle and must be allowed to consistentparameter throughoutmost of the core. ECM and "relax" for severalmonths before it can be processed.The LLS were more valuable in some sectionsthan othersdepend- fractured ice in this zone often made it difficult to apply the ing on the atmosphericchemistry and climate at that time ECM and LLS methods.The glacial/interglacialand stadial/ (Table 2). Continuousexamination also was essential to follow interstadialtransitions are anotherexample of sectionswhere changesin annualsignals as well as to characterizethe climate all propertiesexhibit large changes.A third exampleis the variabilityfully and to avoid interpolationerrors [cf. Ram and occurrence of deformed ice in the bottom 250-350 m of the Koenig,this issue]. core. The nature of the changesin the time-stratigraphicpa- Visual stratigraphywas examined and substantiallycom- pleted in the field. This stratigraphicrecord providedan excellent preliminarydepth-age scale and an immediateindica- Table 1. List of Volcanoes Used as Tie Points to Verify tion of accumulation rate changes. A number of people the Dating of the GISP2 Core Throughout contributedto this aspectof the field program[Alley et al., this Recorded History issue(a)]. A summerstratigraphic horizon was definedas the Year of midpoint of the summerdepth-hoar sequence and was chosen Volcano Eruption (A.D.) as the definitive annual layer marker. All depth-hoar sequencesand other stratigraphicfeatures (wind crusts,melt Katmai 1912 Laki 1783 layers,etc.) were recordedon meter-longpaper at a 1:1 scale. Huaynaputina 1600 During each field season,comparisons were continuallybeing Oraefajokull 1362 made between ECM and stratigraphyand with LLS when 1259 event 1259 available.These comparisonsprovided immediate information Hekla 1104 on variations in accumulationand ice-core chemistry.Addi- Eldgja 934 Vesuvius 79 tionally, they confirmedthat all parameterswere varying syn- chronouslyand could continueto be used in the development 26,414 MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE

• -35

! >•, -30

=- 40

Lu .. 20 ._o E 0 •, 1.0

-• 0.5

• 0 I I I I I I I I loo lo2 lo4 Depth (m) Figure 2. Example of multiparametercomparisons between 100 and 104 m. Each parameter in this se- quenceyielded 16 yearsof accumulation.Lags and leadsbetween the parametersare evidentand expectedas timing of the seasonalinputs varies slightly between years and betweenparameters.

of the depth-agescale. After each field seasonthe summer picksfor that parameter.The black lines in the upper portion (annual layer) signaturesfor each of the other parameters of each box representthe final summerpicks based on com- were identified and recordedfrom peaks (or the midpoint of parisonand analysisof all parameters.In eachexample, some groupsof peaks) on data plots. The summerpicks for each lag or lead betweensome of the individualpicks and the final parameterwere chosenindependently of other parametersin picks is evident;however, there is an excellentoverall corre- order to minimize subjectivedeterminations. The initial picks spondencebetween the parameters.These lags and leads are for all parameterswere placedin a spreadsheetand compared attributedto variationsin timing of the "summer"inputs that line by line or year by year. Slightlags or leadsoccurred due to may encompass2 or 3 monthsof the summerperiod [Tayloret differencesin the timing of the signalof eachparameter. When al., 1992]. For example,in the section100-104 m (Figure 2), extraneouspeaks occurredwithin one parameter that could some lag or lead in the parametersis evident; however,the not be correlated with another parameter, they were dis- total numbersof years determinedfor each parameteris 16. counted.If two or more parametersshowed a strongpeak at From 270 to 275 m (Figure3), 20 stratigraphic(depth hoar) essentiallythe same depth that could not be discountedby sequencesand/5•80 peaks were identified in conjunctionwith eventssuch as volcanicactivity or an additionalinflux of dust, 22 ECM and 23 LLS peaks.At one location in this sequence, etc., it was countedas a year even in the infrequentabsence of /5•80, ECM, and LLS exhibitedstrong peaks that corre- a strongstratigraphic signal. Informed decisionswere made on spondedwith a wind crust,but the normallydiagnostic depth- a year-by-yearbasis by the senior author, basedon the weight hoar sequencewas not identified.This wascounted as 1 year in of the available evidence. Most were either rechecked or the annuallayer count.The value of multiparametercounting pickedindependently by A.J.G., and for somesections, partic- is evident in this instance and in other sections of core where ularly the upper 300 m, a "round-robin"discussion involving extra ECM and LLS peaks occurred that probably are not severalauthors was used (particularly R.B.A. andK.C.T.). This annual because these records are affected by volcanic and was done to ensure that subjectivedeterminations or biases other eventswhich can result in extraneouspeaks. The poten- were kept to a minimum.Additionally, much of the stratigra- tial for additional LLS signalsis not surprisingas elevated phy was reexaminedindependently by three of the authors inputsof dustcan occurat other times of the year, thoughthe (D.A.M., A.J.G., and R.B.A.) at the National Ice Core Labo- late spring/summerinput constitutesthe dominantLLS signal. ratory (NICL) in Denver, Colorado.Since our primary strati- The final countfor the 270- to 275-m sectionis 21 years(+_ 1 graphicsignal based on depth-hoarformation can only occur year), basedon the relativestrength of eachparameter. duringthe summer,the midpointof the stratigraphiclayer was An examplefrom just abovethe brittle ice zone at 501-503 used as the marker depth for each summeror annual signal. m is given in Figure 4. At this depth, isotopesare no longer Comparisonof the various signaturesshows a remarkable viable as an annualindicator, so only three parameters(stra- correspondenceof peaksthroughout the upper 2300 m of the tigraphy,ECM, and LLS) are used.In this instance,11 strati- core. Examples of this are given in Figures 2-8, where the graphicmarkers, 12 ECM peaks,and 11 LLS peakswere iden- picks for each parameter used in the layer countingfor that tified. The final count for this 2-m section is 11. The reason for particularsection are shown.In eachfigure the arrowsin the thisis describedfully in the captionfor Figure4. It canbe seen top bar indicate midsummerstratigraphic picks. The lower that although the actual depths of the peaks do not align plotsinclude some or all of the following:/5•80,ECM, and/or exactly,they typicallyrecord approximatelythe samenumber LLS. The white linesin the shadedareas represent the summer of years.As the depth for each parameterwas recordedon a MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE 26,415

-34

-37

1.0

0.5

0.0 I ' 270 271 272 273 274 275 Depth (m) Figure3. Multiparametersequence between 270 and275 m. Twentystratigraphic layers and •80 peaks werecounted in conjunctionwith 22 ECM and23 LLSpeaks. Annual layer markers for LLS,ECM, and•80 are shownas vertical white lines.These correspondto the spring/summerinputs for each parameter.Details of the interpretationof this record are given in the text. computerizedspreadsheet, depths were viewed continuously section and constitutedthe primary annual layer indicator. and peaks occurringvery near meter breaks could be more However, scattered ECM and/or LLS data from the brittle ice easily aligned in comparisonto correlationsbased on paper zone were available,and they provideduseful checkson strati- plots as in the examplesgiven here. graphic layer counts.A comparisonof the stratigraphicand The brittle ice zone (719-1371 m) was a region of some LLS recordsis shownin Figure 5 between 871 and 873 m. Both concerndue to the large numberof fracturesin the core [Gow parametersexhibit 12 peaksin this section,but becauseof lags et al., thisissue]. While thisfracturing of the ice coreinterfered or leadsthe final picksdo not alwayscoincide. It is evidentthat particularlywith the ECM and LLS measurements,visual stra- although the mechanical condition of the ice deteriorated tigraphyremained very clear and readily decipherablein this somewhatin this region, the parametersused for layer identi-

A 500 • 400 o• • 300 _• = 200 o z: 100 • 0 501 502 503 Depth (m) Figure 4. In this 2-m sectionfrom 501 to 503 m, 11 stratigraphicsummers were identified in conjunction with 12 ECM peaksand 11 LLS peaks.A final count of 11 annual layerswas determinedfor this 2-m section of ice. The stratigraphicsequence corresponding to the ECM and LLS peaksat 501 m is actuallyin the 500-m sectionand was countedthere. At approximately502.72 m a stratigraphicsequence is present as well as a strongECM signal;however, the LLS signalis very weak and was not counted,resulting in an undercountin the LLS record. However, this did not affect the final count. Also, at approximately502.9 m a stratigraphic sequencewas observed, but there is not a correspondingstrong ECM or LLS peak and this sequencewas thus droppedfrom the count.Again, this sectionof ice showsthat althoughthe actual depthsof the peaks are slightlyoffset, approximatelythe samenumber of yearsis revealed for each parameter. 26,416 MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE

• • 400 .-J ,a 300 o ,- 200 •v 100I•i :¾".:•:'.'.'.': •':'..•-: ' '.'%:':' .,. ,':'' 871 872 873 Depth (m) Figure 5. This sectionfrom 871 to 873 m is from the brittle ice zone. In this zone, stratigraphywas the primary annual signalindicator and ECM and LLS were usedwhen available.In this section,12 annual layer markers are observedin both the stratigraphyand the LLS, indicatingthat countingis still viable within the brittle ice zone.

ficationremained sufficiently consistent to sustainannual layer 2247.5 and 2248.0 m is shown in Figure 7. In this core, 42 counting,essentially without a break,through the brittle ice zone. stratigraphicand 39 LLS annual peaks were identified. The Between 1370 and 1678 m, in the Preborealpreceeding the differencein countsusing these two techniquesin this section onsetof the Younger Dryas, the characterof the visualstrati- of the record is clearlyhigher than in the Holocenebut is still graphic signal changeddrastically from that of the character- less than 10%. istic depth-hoar sequencesto one consistingof alternating Beginning around 2400 m, visual stratigraphyin extended clear and cloudybands with the cloudybands containing much sectionsof the core became so faint that the annual layer higher concentrationsof dust than were present in younger structurebecame very difficultto decipher.This was especially Holoceneice. The regionbetween 1680 and 1800 m consistsof true of those sections of ice associatedwith Dansgaard- the Younger Dryas, the B011ing/Aller0d,and the transition to Oeschgerevents which contained greatly reduced dust levels. full glacialconditions. Once into the full glacial(Wisconsinan) Becauseof the low dust levels, resultingin weak to nondeci- period, below 1800 m, dust concentrationsincreased signifi- pherablestratigraphy, significant undercounting of annual lay- cantly and the magnitudeof changesseen in the ECM record ers occurredin the regionof the GISP2 corebetween 2500 and decreased.As a result of this and perhapsbecause of limita- 2800 m. This forced greater reliance on the LLS record in tions imposedby the spatialresolution of the ECM technique, which annual layer peaks could still be readily observed.How- it became increasinglydifficult to use the ECM record for ever, many sectionsof core between2500- and 2800-m depth annuallayer signalidentification in deeperice-age ice. A pho- retained identifiable stratigraphy,especially those sectionsof tographof a 19-cm-longsection of core in Figure 6 demon- ice containingelevated dust levels associatedwith colder cli- strates the very distinct annual layering associatedwith 12 matic conditions.Such sectionsexhibited readily decipherable summer peaks, based on the bright white light bands corre- layer structure that correspondedclosely with annual layer spondingto spring/summerdust rich layers in the core. A peaks derived from the LLS record. 0.5-m sectionof stratigraphicand LLS recordsfrom between As records were obtained, the actual depth recorded for I I

Figure 6. Photographof a 19-cm-longsection of ice from 1855 m showingannual stratigraphiclayers in the Wisconsinan.This sectioncontains 12 summerlayers (arrowed) sandwiched between darker winter layers.The lighter layersare a result of increasedscattering of light due to higher concentrationsof dust. MEESE ET AL.' THE GISP2 DEPTH-AGE SCALE 26,417

1o.6

o i 2247.5 2247.6 2247.7 2247.8 2247.9 2248.0 Depth (m) Figure 7. This sectionbetween 2247.5 and 2248.0 m showsthe comparisonbetween stratigraphyand LLS much deeper in the core. It can be seen that the methodologyused from the top of the core still holds and the comparisonsare excellent.In this case, there were 42 stratigraphicand 39 LLS annual layer markers identified.

eachsummer (or annuallayer) was based on the positionof the al., 1995;Gow et al., this issue].Sowers and Bender developed visual stratigraphicmarker. The stratigraphicrecord was cho- a correlatedtimescale [Bender et al., 1994] in which an inverse sen becausethe timing of the annual signal is more clearly correlationtechnique was used to mapthe recordof •80 in defined than other parametersand is typicallythe most con- atmospheric02 ((•18Oatm)from GISP2 into the Vostok sistent.In the event of weak or absentdiagnostic stratigraphy, (•18Oatm that has been tied to the SPECMAPocean timescale other parameterswere used in the following order: isotopes [Sowerset al., 1993]. They predicted the age of the ice at (0-300 m), ECM, and LLS. In instanceswhere core lossoc- 2800 m to be about 110,000years, 25,000 years older than had curred [Gow et al., this issue;Greenland Ice SheetProject 2 been originally counted on the basis of visual stratigraphy ScienceManagement Office, 1993], approximateannual depths [Meeseet al., 1994]. The preliminary depth-agescale derived were interpolatedby usingthe averagevalue of the number of from the analysisof the visual stratigraphybelow 2200 m annuallayers in an equivalentamount of core aboveand below [Meeseet al., 1994] and the Sowers-Bendercorrelated time- the area of loss.Although this resultsin loss of detail of the scale started to deviate at 2341 m. annual variation in the accumulationrecord throughoutthese Becauseof the abovediscrepancy the seniorauthor returned areas,we believe that the overall chronologyis not seriously to the National Ice Core Laboratory and recheckedthe visible affected. stratigraphy.No significantchanges from the original counts It is believed that viable dating is retained to 2800 m below were observed.Study of the LLS record, now available in a whichflow disturbancein the ice becomesquite severe[Alley et more detailed format using a 1-mm beam width rather than 8

500 { 400

•00

200

•00

0 2552.60 2552.64 2552.68 2552.72 2552.76 2552.80 Depth (m) Figure 8. A thin sectionphotograph taken between crossedpolarizers showingstratigraphy in a section between 2552.60 and 2552.80 comparedto the LLS record. Most of the bright white layers (containing elevatedlevels of dust) compareclosely to the peaksin the LLS record. 26,418 MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE

-268[••J

'2808m i MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE 26,419 mm [Ram and Koening,this issue],showed there was more a definitivetrend, below 2500 m the number of LLS layerswas structureindicating a greater number of annual layers,leading consistentlyhigher than the number identified by visible stra- to a more expandedtimescale than had been interpretedfrom tigraphy, based on the peaking characteristicsof the LLS the stratigraphicrecord. Therefore the entire high-resolution record. A comparisonof countsin yearsper meter showsthat LLS record between 2300 and 2800 m was reexamined inde- the overall structureof the LLS record is very similar as de- pendentlyby two of the authors(D.A.M. and A.J.G.). Dating termined by both individuals,but with one personconsistently of the GISP2 core by LLS is ultimately limited by the spatial countingon average20% higher than the other. To obtain a resolutionof the measurementtechnique (currently 1 mm). final GISP2 depth-agescale, the two setsof countswere then Severalpoints are requiredto identifya definitivepeak; for the averaged.This average was then compared to the Sowers- LLS method the minimum thicknessnecessary for successful Bender correlated timescale and showed a maximum differ- identificationof an annuallayer is 3 or 4 mm (requiringfour or ence of 1.1% with an age of approximately111,000 years B.P. five pointsto identify the peak). Layersless than 3 mm thick Comparisonswere alsomade betweenthe counts,the ECM could be missed,however, leading to undercounting.The po- record,and the 8•80 record.A goodcorrespondence exists tential for undercountingannual layers in the stratigraphic betweenthe layer counts,the ECM, and the 8•80 record, record is known to exist in low accumulation areas such as the indicatingthat the LLS record in this region of the core con- southpole [Gow, 1965]. However, suchundercounting due to tinues to track the climatic signal and that the annual layer extremelow accumulationyears or hiatusyears is estimatedby record still remains intact. It is clear that the original exami- Gow not to exceed 10%. The solid ice LLS instrument used on nation of the visual stratigraphyled to undercountingbelow the GISP2 core [Ram and Koenig,this issue]was configuredto 2400 m. Reexamination of a number of core sections indicates produce1000 readingsper meter for most of the deeper core; that undercountingvery likely resultsfrom either the thinning hence the theoretical maximum number of annual layers that and mergingof layersor the fact that the eye could no longer could be identified in a meter of core is somewhat below 500. distinguishindividual layers, especially in coresexhibiting very A photographof a thin verticalsection of ice between2552.6 faint stratigraphyrelated to low dust levels. Complications and 2552.8 m and its relationshipto the LLS record is shown arising from layer deformation are another possibility.Very in Figure 8. In this section of ice, photographedbetween small folds and waves in the layerswere observedas high as crossedpolarizers to revealdifferences in the dust-relatedsizes 2487 m, but these are not likely to have causeddisruption of of crystalsin the summer and winter layers, 21 stratigraphic the orderlystratigraphic record. However, in the eventof more sequenceswere identifiedand 23 LLS peakswere counted.As severedeformation, stratigraphic succession could have been opposedto the regionsof the core above 2500 m, where the disruptedto the point where it becomesimpossible to distin- number of layerscounted varied betweenparameters without guish and/or count individual layers. As statedearlier, variouscombinations of parametersbased on the chemicaland physicalproperties of the ice were used throughoutthe length of the core to evaluate the depth-age Figure 9. (opposite)Examples of annuallayer structureseen scale. Table 2 lists the depth and age ranges for different in GISP2 ice coresilluminated from belowby a fiber optic light combinationsof layer-counting parameters and error esti- source.Examples of both undisturbedand disturbedstratigra- mates. phy are shown.(a) 2280 m. Distinctiveannual layer structure. It has been reported elsewhere[Alley et al., this issue (a); (b) 2304 m. Lessdistinctive annual layering due to decreasing Ram and Koenig,this issue]that countinglimited to only one dustlevels in the ice. (c) 2379 m. Even lessdistinctive layering due to the decreaseddust content of the ice; layers are more parameter or to the countingof 1 m in every 5 (or 20%) widely spacedindicating higher annual snowfall.(d) 2455 m. providesnearly as accuratea timescaleas the multiparameter, Return to more distinctiveannual layering. (e) 2465 m. The continuouscount method. Stratigraphiclayer countingyields a fewer layers in this core indicate a higher accumulationrate timescalethat is within 1% of the multiparametermethod in associatedwith lower dust content, indicating a warmer cli- the Holocene and even deeper. However, deeper in the core mate than at 2455 m. (f) 2518 m. There are more layersin this the differencebetween the one parameterand multiparameter section,indicating lower accumulationand a cooler climate. methods tends to increase and much of the fine detail that can Layersare still fairly horizontalwith someminor waviness.(g) only be obtained by continuouscounting methods is likely to 2537 m. A minor disturbance is evident with a small z-fold near be missed.Variations in annual layer thicknesscan be readily the bottomof the photograph.This fold is smallenough that it seen on an annual basis when continuouscounting is com- did not disturb the climatic record or the layer counts.(h) 2541 m. Prominent folding of annual layers in the ice. How- pleted.For example,when layer countingis limited to only one ever, layer countsare still decipherable.(i) 2558. Return to method, i.e., stratigraphyin the upper 2300 m of the core, a undisturbedannual layering. (j) 2573.Distinctive annual layer more subjectiveapproach is taken in rejecting the extreme structure;some minor wavinessof layersevident. (k) 2622 m. years from the record. Not only do these extreme years exist, Some minor wavinessor bending of the layers in evidencein they may indicate short-term changes,particularly if these this core. (1) 2685 m. Annual layers becomingvery closely trendspersist for more than a number of years.A comparison spacedin thiscore. (m) 2808m. Layerstructure still discernible of all parameters sometimesreveals peaks of a nonannual but difficult to count. (n) 2916 m. Layer structurevery dis- nature, for example,peaks related to volcanicevents, which if turbed and discontinuous;annual layering is no longer discern- includedin the count would result in overcounting.In addition ible. (o) 2923 m. Layer structuredisturbed and discontinuous; to identifyingshort-term trends, the multiparameterapproach no discernibleannual layering.(p) 2957 m. No definitivean- nual layering present; sectionsof cloudy ice alternate with identifies variations which may be occurring in only one pa- clear ice, and layersare curvedand discontinuous.(q) 2973 m. rameter. As indicatedearlier, a prime exampleis the problem Highly compressedlayer structure;layers do not appear to be of undercountingof layers associatedwith large sectionsof deformed.(r) 3028 m. Intermixingof cloudyand clear ice; no core exhibitingmerged layers or very faint visual stratigraphy discerniblelayering present. below 2500 m. The resultantlarge differencein the timescale, 26,420 MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE

200

960

•'720

480

240

2808.32 2808.36 2808.40 2808.44 2808.48 2808.52 Depth (m) Figure 10. Exampleof an LLS recordfrom 2808 m. Somefaint stratigraphiclayering is decipherablein this meter of core and appearsto be disturbed.Even though the stratigraphicrecord is disturbedin this section of core, the LLS record appearsto be intact, and its overall characteris identical to that observedin undisturbedice at higher levelsin the ice sheet. basedon the original visible layer counts,compared with the The diagnosticfeatures of the LLS record (on which we Sowers-Bendercorrelated scale provided the impetusto reex- relied heavily to evaluate annual layer countingbetween 2500 amine the visual stratigraphyin greater detail in conjunction and2800 m) appearto be retainedin the icebelow 2800 m, and with the LLS record. This reanalysishas resultedin what we for this reasonalone, layer countingwas continuedto a depth believeis a much more accuratescale than we previouslyhad of 3030 m. Given the widespreadevidence of disturbedstruc- obtained. ture in the ice, it is clear that the layer count is very unlikely to The region below 2800 m has been a subjectof much dis- be continuous.Yet, it providesinformation on the sectionsof cussion based on the conclusion from the Greenland Ice Core apparentlylittle disturbedice in the deformed region of the Project (GRIP) core that the interglacialclimate during the core, includingthe number of yearscontained in each of those Eemian had been unstable.This was put into doubt when it sections. was realizedthat the profilesof isotopesand other core prop- An example of the clarity of the LLS record obtained at erties from the GISP2 and GRIP cores no longer agreed.At 2808 m is shown in Figure 10. Faint banding indicative of depthsbelow 2800 m, two major complications,loss of a con- annuallayering was still discernibleat this depth. Examplesof sistentvisible stratigraphic signal and possiblelack of sufficient the LLS record shownin Figures 11 and 12 are of sectionsof spatialresolution to resolvethe annuallayering, limit our abil- ice exhibitingsignificantly different stratigraphicstructure. At ity to identify annuallayers. The stratigraphicrecord revealed 2840 m (Figure 11) the ice wasvery clear and coarsegrained evidenceof appreciabledeformation in this regionof the core. with very little stratigraphicstructure visible. At 2850 m (Fig- Examplesof overturnedfolds, z folds, highly inclined layers, ure 12), where identifiablelayering briefly reappeared, the ice and other structural features indicating disruption of strati- was much finer grained.The reappearanceof visualstratigra- graphicorder are abundant(Figure 9). These deformational phy in conjunctionwith the formation of fine-grainedice is features include reversalsin the direction of inclined layering entirely compatiblewith the increaseddust concentrationsev- in sectionsof core up to 1 m long, which couldindicate over- ident in the vertical scale of the LLS record. To determine the turned folds. Some interpretation of the structuresin this re- number of layersbetween 2800 and 3030 m, the seniorauthor gionhas been presented[Alley et al., this issue(b); Gow et al., (D.A.M.) countedcontinuously to 3030m, approximately10 m this issue],but much more work is needed to determine the abovethe siltyice contact.Another one of the authors(A.J.G.) precisenature and extent of the deformation.It is likely that countedcontinuously to 2850 m and then 1 m in every 5 m to this area of the core also contains structural discontinuities. 3030 m. Interpolationswere then made in this latter record in Suchdiscontinuities could arise from the depositionof one ice order to compare the countsof both observerson a relative sheet on top of remnantsof previousice sheets,leading to basis.More structureis evidentin the countsmade by D.A.M., significanttime gapsin the stratigraphicrecord. In theoryit is showingthe importanceof counting continuously.Nonethe- possibleto identify ice from a singleage that has been folded less,the averagevariation in the two setsof countswas 20% on sufficientlyto be duplicatedin the core by searchingfor ice at a layer per meter basisbetween 2500 and 2800 m and 24% two depths that have the same chemical, isotopic, and gas between 2800 and 3000 m. At 2800 m the age of the ice is characteristics.In practice,however, depositional spatial vari- determinedto be 110,693years B.P. _+10%. The number of ability, diffusion, and measurementlimitations are likely to layers obtainedby averagingthe countsof the two observers make identification of repeated sectionsdifficult to resolve. between 2800 and 3030 m is 50,600 with an unknown error. We have not identified any large repeated sections,but we Because of deformation and likelihood of discontinuities,it is considerit possiblethat duplicatedsections exist. not possibleto assigna timescaleor agesto this region of the MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE 26,421

lOO

2840.00 2840.04 2840.08 2840.12 2840.16 2840.20 Depth (rn) Figure 11. An exampleof the LLS record from ice that is very clear and coarsegrained with little visible stratigraphy. core. If future work allows us to deconvolve this area of the original layer counting,the volcanicsignals were examinedin core, we may be able to identify climaticevents and place age relation to the age scalebefore any reevaluationof the years limits on them. was completed.The worst casewas for Vesuvius(A.D. 79). Analysisof the GISP2 core gave a date which was 12 years Error and Verification older than the historicaleruption date. This differenceresults in a 0.63% error, which is lessthan the 1% stated for this work. Putting error estimateson the depth-agescale is difficult. Assessingthe accuracyof our depth-agescale based on the The accuracyof the depth-agescale can be checkedagainst historicallydated volcanic events over the last 2000years (i.e., multiparameterapproach in deeper ice becomesincreasingly tie points;Table 1). If a discrepancyexisted between the date more difficultas there are no independentmeans of datingthe of a knownvolcanic eruption and the correspondingdate as- ice by radiometrictechniques. Possible comparisons might be signedon the basisof elevated ECM and sulfate peaks, the made with eventsthat have been dated in corals and deep-sea strengthsof the variousannual signalswere then reevaluated cores, but these events likely have larger errors than those to determine if perhapsa more accuratedate could be ob- applyingto the GISP2 timescale.Additionally, a depth-age tained. Most of our final datesof the major historicalvolcanic scaleexists for the EuropeanGRIP core(largely based on flow eruptionsidentified in the corematch stratigraphically defined modelingand noncontinuousmethods) which can alsobe used datesprecisely. To obtain an estimateof error basedon the for comparison.The errorsthat are listedin Table 2 are based

200

160

e 120

o .,-, 80

o I 2850.00 2850.04 2850.08 285o.12 285o.16 2850.20 Depth (m) Figure 12. An exampleof the LLS record from ice that containsdecipherable stratigraphic layering and is much finer grainedthan the sectionof ice presentedin Figure 11. 26,422 MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE

on variations noted between methods and between analysts. Summary All of these estimates are conservative and should be consid- The GISP2 ice core has been dated continuouslyfrom 0 to ered to err on the side of maximum error. 2800 m with considerable precision. The age obtained at The YoungerDryas (YD)-Bolling/Allerod-Oldest Dryas se- 2800 m was 111,000years B.P. with an estimatederror ranging quenceprovides one tie point, particularlythe YD termination from i to 10% and up to 20% between2500 and 2800 m. Layer at 1678 m where a dramatic changeoccurs in layer thickness counting was extended to 3030 m on the basis of the LLS (resultingin an estimatedtwofold to threefoldchange in ac- record mainly to obtain a possibleage limit of the ice near its cumulationrate), correspondingto an ageof 11,640years B.P. contactwith the bed, assumingminimal disturbanceby defor- This agreesvery closelywith most publisheddates of around mation of the stratigraphicrecord. Layer countsin excessof 11,500years B.P. (a variationof slightlymore than 1%) for the 300/m were measuredin the deepestice, and an estimatedage YD termination[Alley et al., this issue(a)]. of 161,000years B.P. was obtainedat 3030 m with an unknown Another tie point that canbe usedis HeinrichEvent 3 (H3) error. This age estimateis significantlyless than that obtained betweenInterstadial 4 and 5 [Meeseet al., 1993]. On the basis at GRIP, largelyon the basisof ice flow modeling.However, it of comparisonsof high-resolutiondeep-sea sediment records is likely that structuraldisturbance observed in the basal250 m from the North Atlantica •4C date of 26,500was obtained for of ice at both locations has affected both the timescales to an core 2381 and a date of 27,000was obtainedfor Orphan Knoll unknown degree. (G. Bond, personalcommunication, 1996). These dates are Comparisonof the GISP2 depth-agescale with published approximatelyequivalent to a calendardate of 30,000-31,000 datesfor the termination of the Younger Dryas showscorre- years B.P. [Bond and Lotti, 1995] based on the U/Th calibra- spondencewithin the combinederrors. Another tie point exists tion of Bard et al. [1990]. The date obtainedfrom continuous deeper in the core for Heinrich Event 3 at approximately counting of annual layers in the GISP2 core for H3 is 31,200 30,000years B.P., againwith excellentcorrespondence to pub- years B.P. This again providesgood agreement. lished records. Below this event, correlations made with a A comparisoncan also be made between the correlated modelbased on (5•80of 02 reveala closecorrespondence. timescalepresented by Benderet al. [1994] and the depth-age scale.The modelis basedon (•18Oatm of 0 2 measurementsof the GISP2 ice core and comparedto the Vostok ice core and Acknowledgments. The authors would like to thank the National SPECMAP for the oceans.At 2500 m, annuallayer countsgive ScienceFoundation Office of Polar Programs for funding of this an age equivalentto 56,930years B.P., and the correlatedscale project and the GISP2 ScienceManagement Office, the Polar Ice Coring Office, the 109th Air National Guard, the National Ice Core of Benderet al. [1994] gives57,600 years B.P. This differenceis Laboratory,and numerouscolleagues for their assistanceduring the slightlyhigher than 1%. This is somewhatremarkable, since drilling and assistancein obtainingthe data necessaryfor the depth- the error for both methodsis estimatedto be muchhigher than age scale.D.A.M. and A.J.G. also thank the USA Cold RegionsRe- this. The same trend holds at depth, where layer countingat searchand EngineeringLaboratory for supportand Army basicre- searchprojects for partial funding. 2700 m gives 89,084 years B.P. comparedto a model age of 89,244 years B.P. (now less than 1% variation in the two scales).Over the depthrange, 2700-2800 m, an excellentcor- References relationalso exists between the (•18Oatm in the Vostokand GISP2 cores[Bender et al., 1994].Bender et al. [1994] suggest Alley, R. B., and S. Anandakrishnan,Variations in melt-layer fre- that the ice below 2850 m may largelybe from the substages5e quencyin the GISP2 ice core: Implicationsfor Holocene summer temperaturesin central Greenland,Ann. Glaciol.,21, 64-70, 1995. and 5d, approximately128,000-140,000 yearsB.P. If the layer Alley, R. B., E. S. Saltzman, K. M. Cuffey, and J. J. Fitzpatrick, countsbased on the LLS technique are used as dates in this Summertimeformation of depth hoar in central Greenland, Geo- sectionof the core, the estimatedage would be 121,200years phys.Res. Lett., 17, 2393-2396, 1990. B.P. at 2850 m, lessthan a 5% variation at 2850 m as suggested Alley, R. B., A. J. Gow, S. J. Johnsen,J. Kipfstuhl,D. A. Meese, and T. H. Thorsteinsson,Comparison of deep ice cores,Nature, 373, above, indicatingthat this ice may in fact be part of substages 393-394, 1995. 5e and 5d. However, as stated above, caution must be exercised Alley, R. B., et al., Visual-stratigraphicdating of the Greenland Ice as the full extent of the deformation in this region has not yet Sheet Project 2 ice core: Basis,reproducibility, and application,J. been completelydefined. Geophys.Res., this issue(a). A final comparisonis made betweenthe GISP2 depth-age Alley, R. B., A. J. Gow, D. A. Meese, J. J. Fitzpatrick,E. D. Wad- dington,and J. F. Bolzan, Grain-scaleprocesses, folding, and strati- scalebased on annual layer countsand the GRIP depth-age graphicdisturbance in the GreenlandIce SheetProject 2 ice core,J. scale [Dansgaardet al., 1993;Johnsen et al., 1995]. The GRIP Geophys.Res., this issue(b). timescaleback to 14,500 years B.P. was derived from annual Bard, E., B. Hamelin, R. G. Fairbanks, and A. Zindler, Calibration of layer countsfrom the surface.Dating below 14,500years B.P. the •4Ctimescale over the past30,000 years using mass spectromet- ric U-Th agesfrom Barbadoscorals, Nature, 345, 405-410, 1990. was calculated on the basis of flow modeling and chemical Bender, M., T. Sowers, M. Dickson, J. Orchardo, P. Grootes, techniquesincluding Ca 2+, NH•-, andmicroparticles [Johnsen P. Mayewski,and D. Meese, Climate correlationsbetween Green- et al., 1992]. There are potentialproblems with both the GISP2 land and Antarctica during the past 100,000 years, Nature, 372, and GRIP timescales.Most ice flow modelspredict extremely 663-666, 1994. rapid thinning near the bottom of an ice sheet.Annual layer Bond, G., and R. Lotti, Icebergdischarges into the North Atlantic on millennial time scalesduring the last glaciation,Science, 267, 1005- countingbased on LLS results at GISP2 showsthat thinning 1010, 1995. does not appear to occur at the rate predictedby the models Dahl-Jensen, D., S. J. Johnsen, C. U. Hammer, H. B. Clausen, and [SchOttet al., 1992;Dahl-Jensen et al., 1993].However, both the J. Jouzel, Past accumulation rates derived from observed annual GISP2 and GRIP cores are known to exhibit evidence of se- layersin the GRIP ice core from Summit,central Greenland,in Ice in the ClimateSystem, NATO ASI Ser. I, vol. 12, edited by W. R. vere structural disturbance in the basal 200-300 m, and this Peltier, pp. 517-532, Springer-Verlag,New York, 1993. has likely led to discontinuitiesor breaks in the stratigraphic Dansgaard,W., Stableisotopes in precipitation,Tellus, 16(4), 436- sequence. 468, 1964. MEESE ET AL.: THE GISP2 DEPTH-AGE SCALE 26,423

Dansgaard,W., et al., Evidencefor generalinstability of past climate Ram, M., and G. Koenig, Continuousdust concentrationprofile of from a 250-kyr ice-corerecord, Nature, 364, 218-220, 1993. pre-Holocene ice from the Greenland Ice Sheet Project 2: Dust Epstein,S., and R. P. Sharp,Oxygen isotope studies, Eos Trans.AGU, stadials,interstadials, and the Eemian, J. Geophys.Res., this issue. 40(1), 81-84, 1959. Ram, M., M. Illing, P. Weber, G. Koening,and M. Kaplan, Polar ice Fiacco, R. J., Jr., T. Thordarson, M. S. Germani, S. Self, J. M. Palais, stratigraphyfrom laserlight scattering:Scattering from ice, Geophys. S. Whitlow, and P. Grootes,Atmospheric aerosol loading and trans- Res. Lett., 22, 3525-3527, 1995. port due to the 1783-84 Laki eruptionin Iceland, interpretedfrom Sch0tt, C., E. D. Waddington, and C. F. Raymond, Predicted time- ash particles and acidity in the GISP2 ice core, Quat. Res., 42, scales for GISP2 and GRIP boreholes at Summit, Greenland, J. 231-240, 1994. Glaciol., 38, 162-168, 1992. Gow, A. J., On the accumulation and seasonalstratification of snow at Shuman,C. A., R. B. Alley, S. Anandakrishnan,and C. R. Stearns,An the southpole, J. Glaciol.,5, 467-477, 1965. empirical technique for estimating near-surfaceair temperature Gow, A. J., Relaxationof ice in deep drill coresfrom Antarctica,J. trends in central Greenland from SSM/I brightnesstemperatures, Geophys.Res., 76, 2533-2541, 1971. Remote Sens. Environ., 51, 245-252, 1995. Gow, A. J., D. A. Meese,R. B. Alley, J. J. Fitzpatrick,S. Anandakrish- Sowers, T., M. Bender, L. Labeyrie, D. Martinson, J. Jouzel, nan, G. A. Woods, and B.C. Elder, Physicaland structuralproper- D. Raynaud, J. J. Pichon, and Y. S. Korotkevich,A 135,000-year ties of the Greenland Ice Sheet Project 2 ice core: A review,J. Vostok-SPECMAP common temporal framework, Paleoceanogra- Geophys.Res., this issue. phy, 8(6), 737-766, 1993. Greenland Ice Sheet Project 2 ScienceManagement Office, GISP2 Stuiver,M., P.M. Grootes,and T. Braziunas,The GISP2 (5•80climate core data book, 114 pp., Univ. of N.H., Durham, 1993. record of the past 16,500 years and the role of the sun, ocean, and Grootes, P.M., M. Stuiver, J. W. C. White, S. Johnsen, and J. Jouzel, volcanoes,Quat. Res., 44, 341-354, 1995. Comparisonof oxygenisotope records from the GISP2 and GRIP Taylor, K., R. Alley, J. Fiacco, P. Grootes, G. Lamorey, P. Mayewski, Greenland ice cores, Nature, 366, 552-554, 1993. and M. J. Spencer,Ice-core dating and chemistryby direct-current Hamilton, W. L., and C. C. LangwayJr., A correlationof microparticle electrical conductivity,J. Glaciol., 38, 325-332, 1992. concentrationswith oxygenisotope ratios in 700 year old Greenland Taylor, K. C., G. W. Lamorey, G. A. Doyle, R. B. Alley, P.M. Grootes, ice, Earth Planet. Sci. Lett., 3, 363-366, 1967. P. A. Mayewski,J. W. C. White, and L. K. Barlow, "The flickering Hammer, C. U., Past volcanism revealed by Greenland Ice Sheet switch" of late Pleistoceneclimate change,Nature, 361, 432-436, impurities,Nature, 270, 482-486, 1977. 1993a. Hammer, C. U., Acidity of polar ice cores in relation to absolute Taylor, K. C., C. U. Hammer, R. B. Alley, H. B. Clausen, D. Dahl- dating, past volcanism,and radio-echos,J. Glaciol., 25, 359-372, Jensen,A. J. Gow, N. S. Gundestrup,J. Kipfstuhl,J. C. Moore, and 1980. E. D. Waddington, Electrical conductivitymeasurements from the Hammer, C. U., Traces of Icelandic eruptions in the Greenland Ice GISP2 and GRIP Greenland ice cores, Nature, 366, 549-552, 1993b. Sheet, Joekull, 34, 51-54, 1984. Zielinski, G. A., Stratosphericloading and optical depth estimatesof Hammer, C. U., H. B. Clausen, and W. Dansgaard, Greenland ice explosivevolcanism over the last 2100 yearsderived from the GISP2 sheet evidence of post-glacialvolcanism and its climatic impact, Greenland ice core, J. Geophys.Res., 100, 20,937-20,955, 1995. Nature, 288, 230-235, 1980. Zielinski, G. A., P. A. Mayewski, L. D. Meeker, S. Whitlow, M. S. Johnsen,S. J., H. B. Clausen, W. Dansgaard,K. Fuhrer, N. Gun- Twickler, M. Morrison, D. Meese, R. B. Alley, and A. J. Gow, destrup,C. U. Hammer, P. Iverson,J. Jouzel,B. Stauffer, and J.P. Record of volcanism since 7000 B.C. from the GISP2 Greenland ice Steffensen,Irregular glacialinterstadials recorded in a new Green- core and implicationsfor the volcano-climatesystem, Science, 264, land ice core, Nature, 359, 311-313, 1992. 948-952, 1994. Johnsen,S. J., H. B. Clausen,W. Dansgaard,N. Gundestrup,C. Ham- Zielinski, G. A., M. S. Germani, G. Larsen, M. G. L. Baillie, S. Whit- mer, and H. Tauber, The Eem stableisotope record along the GRIP low, M. S. Twickler,and K. Taylor, Evidenceof the Eldgja (Iceland) ice core and its interpretation,Quat. Res.,43, 117-124, 1995. eruptionin the GISP2 Greenlandice core:Relationship to eruption Langway,C. C., Jr., H. B. Clausen, and C. U. Hammer, An inter- processesand climatic conditionsin the tenth century,Holocene, 5, hemisphericvolcanic time-marker in ice coresfrom Greenlandand 129-140, 1995. Antarctica, Ann. Glaciol., 10, 102-108, 1988. Zielinski, G. A., P. A. Mayewski, L. D. Meeker, K. Gr6nvold, M. S. Meese, D. A., A. J. Gow, R. B. Alley, P.M. Grootes,P. A. Mayewski, Germani, S. Whitlow, M. S. Twickler, and K. Taylor, Volcanic aero- M. Ram, K. C. Taylor, E. D. Waddington,G. A. Zielinski, and G. C. sol records and tephrochronologyof the Summit, Greenland, ice Bond, Counting down... the GISP2 depth/age scale,Eos Trans. cores,J. Geophys.Res., this issue. AGU, 74(43), Fall Meet. Suppl.,83, 1993. Meese, D. A., R. B. Alley, A. J. Gow, P. Grootes,P. A. Mayewski,M. R. B. Alley, Earth SystemScience Center, PennsylvaniaState Uni- Ram, K. C. Taylor, E. D. Waddington,and G. Zielinski, Preliminary versity,University Park, PA 16802. depth-agescale of the GISP2 ice core, CRREL Spec.Rep. 94-1, 66 J. F. Bolzan, Byrd Polar ResearchCenter, Ohio State University, pp., Cold Reg. Res. and Eng. Lab., Hanover, N.H., 1994. Columbus, OH 43210. Moore, J. C., H. Narita, and N. Maeno, A continuous770-year record A. J. Gow and D. A. Meese, U.S. Army Cold RegionsResearch and of volcanic activity from East Antarctica, J. Geophys.Res., 96, Engineering Laboratory, Hanover, NH 03755. (e-mail: dmeese@ 17,353-17,359, 1991. hanøver-crrel'army'mil) Neftel, A., M. Andree, J. Schwander,B. Stauffer, and C. U. Hammer, P.M. Grootes, Leibniz-Labor, Christian-Albrechts-Universit•it Kiel, Measurementsof a kind of DC-conductivityon coresfrom Dye 3, in Max-Eyth-Strasse11, 24118 Kiel, Germany. GreenlandIce Core:Geophysics, Geochemistry, and theEnvironment, P. A. Mayewski and G. A. Zielinski, Climate Change Research Geophys.Monogr. Ser., vol. 33, edited by C. C. Langway Jr., H. Center, Institute for the Studyof Earth, Oceansand Space,University Oeschger,and W. Dansgaard,pp. 32-38, AGU, Washington,D.C., of New Hampshire,Durham, NH 03824. 1985. M. Ram, Department of Physics,State Universityof New York at Palais,J. M., K. Taylor, P. A. Mayewski,and P. Grootes,Volcanic ash Buffalo, Buffalo, NY 14260. from the 1362 A.D. Oraefojokulleruption (Iceland) in the Green- K. C. Taylor, Desert ResearchInstitute, Universityand Community land Ice Sheet, Geophys.Res. Lett., 18, 1241-1244, 1991. College Systemof Nevada, Reno, NV 89506. Palais,J. M., M. S. Germani, and G. A. Zielinski, Inter-hemispheric transportof volcanicash from a 1259A.D. volcaniceruption to the Greenland and Antarctic Ice Sheets,Geophys. Res. Lett., 19, 801- 804, 1992. Ram, M., and M. Illing, Polar ice stratigraphyfrom laser-lightscatter- (ReceivedFebruary 3, 1996; revisedDecember 22, 1996; ing: Scatteringfrom meltwater,J. Glaciol., 40, 504-508, 1994. acceptedJanuary 16, 1997.)