The Younger Dryas Termination and North Atlantic Deep Water

PALEOCEANOGRAPHY, VOL. 12, NO. 1, PAGES 23-38, FEBRUARY 1997

The Younger Dryas termination and North Atlantic Deep Water formation: Insights from climate model simulations

and

Greenland

ice cores

Peter J. Fawcett • Department of Physics, University of Toronto, Toronto, Ontario, Canada

AnnaMaria•gfistsddttir andRichard B. Alley Department of Geosciencesand Earth System ScienceCenter, PennsylvaniaState University University Park

Christopher A. Shuman Oceansand Ice Branch, NASA Goddard SpaceFlight Center, Greenbelt, Maryland Abstract.

Results from the GISP2

and GRIP

ice cores show that the termination

of the YoungerDryas (YD) climate event in Greenlandwas a large and extremely fast climate change.A reinitiation of North Atlantic Deep Water formationfollowing a shutdown, and its associatedwinter releaseof heat to the atmosphere,has been suggestedas the most likely causeof this climate transition. To test this idea, two general circulationmodel experimentsusing GENESIS have been completedfor

YD time (12,000calendaryearsago): onewith low heat flux in the NordicSeas

(10 W/m2, deepwatershutdown) andonewith highNordicSeaheatflux (300 w/ m• , activedeepwater formation). Comparisonof Greenlandclimatedifferences between these experimentswith the ice core recordsshowsthat when deep water is turned on, much of the YD termination warming is achieved. The increase in precipitation is underestimatedbecauseof a model tendency to overestimate summertime precipitation, which obscuresthe dominantly wintertime responseto the specifiedforcing. The winter storm track shift toward Greenland contributes much of the climate changeat the YD termination.

Introduction

The YoungerDryas (YD) climateeventis the bestknownin a seriesof abrupt returnsto almostfull glacial conditionsduring the last deglaciation. New Greenland ice core data show that the termination

of this event was

a large and extremelyfast climate change,occurringin decadesor less,and possiblyin as little as I to 3 years [Dansgaardet al., 1989; Alley et al., 1993; Mayewski et al., 1993; Taylor et al., 1993]. Both the magnitude and rapidity of the YoungerDryas - Preboreal(PB) climate transition have broad implications for our un-

derstanding of variability in the climate system, especially in view of current concernover anthropogeniccli-

mate change[e.g.,Broecker,1995].Much attentionhas therefore been focussedon identifying possibleclimate

forcing mechanismsthat operate on timescalesmuch shorter

than

orbital

variations.

Rapid climate oscillationssuchas the YoungerDryas have been attributed to sudden changesin the North

Atlantic thermohalinecirculationand associated poleward advectionof warm surfacewaters[e.g., Broecker and Denton, 1989; Broeckeret al., 1985, 1990]. Previous climate model studies[e.g., Bryan, 1986; Manabe and $touffer, 1988; Birchfieldand Broecker,1990; Wright and $tocker,1993]have shownthat the thermohaline overturn

in the North

Atlantic

can oscillate

•Now at Departmentof Earth and PlanetarySciences, betweentwo modes,"on" and "off," as a function of sur-

University of New Mexico, Albuquerque.

facesalinities.Broeckerand Denton[1989]attributed

Copyright1997 by the AmericanGeophysicalUnion.

the onset of Younger Dryas coolingto a shutdownof North Atlantic Deep Water (NADW) formationwhen

Paper number 96PA02711.

Laurentide

0883-8305 / 97/ 96PA-02711$12.00

Mississippito the St. Lawrence drainage and caused 23

ice sheet meltwater

was diverted from the

24

FAWCETT ET AL.: YOUNGER DRYAS TERMINATION

a rapid fresheningof North Atlantic surfacewaters. In this model, the YoungerDryas endedwhen the Atlantic salt conveyorreestablisheditself and warm surfacewaters were again advectedto high latitudes in the North Atlantic. In addition to the drainagediversionthere is substantialevidencefor a large icebergdischargefrom the Hudson Strait

into the North

Atlantic

at this time

(Heinrichevent, H-0) and thesein combinationcould have generated the low salinity cap necessaryto shut

imum ( 0.33 per mil per øC [Cuffeyet al., 1995]). No direcct calibration was obtained for the Younger Dryas becauseits borehole-temperaturesignal has largely diffused away, although the isotopicdata do contain tem-

peratureinformation[Cuffeyet al., 1994, 1995].These calibrations yield warmings of 7ø-8øC to 10ø-12øCat the Younger Dryas termination. Changesin moisture-

sourcetemperature [Dansgaardet al., 1989] and perhapslocation[Charleset al., 1994]may haveenhanced

In this study we use an atmosphericgeneral circulation model (AGCM) which includesan explicit param-

the isotopicchange,causingan overestimateof the temperature change. Analysis of the isotopic composition of trapped gasesapparently has detected the thermal diffusionsignal of a large, abrupt warming at the end

eterization for ocean heat transport to test the hypoth-

of the YoungerDryas [Severinghaus et al., 1996]. Pre-

esisthat the YoungerDryas to Preboreal(PB) climate

liminary calculationsbasedon the inferred gas age-ice age differenceat the Younger Dryas termination, the reconstructedaccumulationrates, and an empirical firn densificationmodel that yieldsthe gas age-iceage difference as a function of temperature and accumulation are consistent with a moderate to large warming, per-

downdeepconvection[Miller andKauj•man,1990;Bond et al., 1993;Andrewset al., 1994].

transition

resulted from a restart

of NADW

formation.

Two experimentsfor YoungerDryas-ageboundaryconditions were carried out: one with low ocean heat con-

vergencein the Nordic Seas(Norwegianand Greenland Seas)representingthe NADW shutdownphase,and one with high ocean heat convergencein the Nordic Seas representingNADW formation and advectionof warm surface waters into this region. The differencein climate results between the two experiments is compared with the high-resolutionclimate recordfrom the GISP2 ice core.

haps7ø to 8øC [Severinghaus et al., 1996]. At the Younger Dryas termination, ice accumulation rates doubled in a matter of decades,and much of this

changeoccurredin as little as I to 3 years[Alley et al., 1993]. The changein accumulationrates at this transition (10%/øC for a 7øC warming,and no lessthan 6%/øC assumingan upper limit warmingof 12øC) is

Greenland

muchhigher than would be expectedfrom a purely thermodynamiceffect (the ability of warmer air to carry

The Younger Dryas to Preboreal climate changeat Summit, Greenland, is well documentedin the various

requires dynamic changesaffecting accumulationwith the warming and most likely a storm track shift toward

proxy climate records obtained from the new GRIP and

Greenland[Kapsneret al., 1995]. Suchshiftsin storm tracks are also arguedfor by Mayewskiet al., [1993]

The Younger Dryas Termination

in

GISP2 ice cores. The upper two thirds of the GISP2 corewasdated by countingannuallayers[Alley et al., 1993;Meeseet al., 1994]whichallowsan assignment of calendar-yearagesto the climate records,and the high resolutionof the ice core allows for detailed inspection of the climate transitions. The YoungerDryas began -•12,700 J:200 (calendar)yearsago and ended-•11,500

+200 yearsago [Alley et al., 1993]. Electrical conductivity measurementsof the ice core

showan abrupt decrease(10 yearsor less)in alkaline dust loadingat the YoungerDryas termination[Taylor et al., 1993],and chemicalanalysesof the ice also showa sharp decreasein atmosphericloadingof crustal material and sea salt, as well as large fluctuationsin

more moisture,4%/øC [Johnsenet al., 1989]). This

basedon rapid changesin the loadingof dust and other particulates in the ice. Climatic

oscillations

recorded in North

Atlantic

mar-

ine sediment cores support the hypothesis of a link between

the ocean's thermohaline

circulation

and the

abrupt climate changesof the last deglaciation. Finescalecolor changesrelated to abundancesof Neoglobo-

quadrinapachyderma(s) in Deep Sea Drilling Project core 609 in the central Atlantic resemble the 5•sO

record in the Camp Century core (south Greenland), whichsuggestsa commonclimaticorigin [Broeckeret al., 1990]. The high-resolution Troll 3.1 corefrom the Norwegiantrenchrecordsseasurfacewarmingof 5øCor

ammonium[Mayewskiet al., 1993;Alley et al., 1995]. more in fewer than 40 years at the end of the Younger The size and rapidity of these changesprobably indicate a shift in atmosphericcirculation patterns rather than changesin the dust sourceareas.

Dryas due to increased inflow of warm Atlantic sur-

facewater to the NorwegianSea[LehmanandKeigwin,

The 51sO(permil relativeto SMOW) of iceshows a 3.5 to 4 per mil increaseover 50 years at the Younger

Dryas termination [Dansgaardet al., 1989; Johnsenet al., 1992; Grooteset al., 1993].The isotopicthermome-

Model and Boundary Conditions The GENESIS (version1.02) generalcirculationmo-

ter has been calibrated against boreholetemperatures

del [Thompsonand Pollard,1995]usedin this study

for the warmingsfrom the Little Ice Age ( 0.55 per mil per øC [Cuffeyet al., 1994) and from the glacialmax-

is an extensivelymodified versionof the National Center for AtmosphericResearch(NCAR) CommunityCli-

FAWCETT ET AL.: YOUNGER DRYAS TERMINATION

2:5

of the cool climate episode.Paleogeography (land-sea distribution, continentalice sheetdistribution,and paleotopography)for 12,000 years ago was taken from Peltier[1994]andis shownin Figure1. The present-day

mate Model CCM1 [Williamsonet al., 1987]. The atmosphericmodel is coupledto a 50-m slab mixed layer ocean in which poleward ocean heat transport is includedas a zonally symmetricfunctionof latitude based on present-dayobservationsusing 0.3 times the "0.5 X

solarconstantof 1370W/m 2 and the Earth'sorbital

OCNFLX" caseof Coveyand Thompson[1989]. This

configurationfor 12,000 years ago determinedthe lat-

reductiongivesa best fit to presentzonal mean seasur-

itudinal

face temperatures[Thompsonand Pollard, 1995]. A

ing Berger[1978]. AtmosphericCO• is set at 250 ppm from the Vostokice corerecord[Barnolaet al., 1987] (presentcontrolvalueis 340 ppm). A globalintermediate vegetationtype (savannah)wasselectedalongwith

regionof enhancedoceanheat flux during winter in the Nordic Seaswas added to prevent unrealisticbuildup of sea ice. This flux warms the mixed layer wheneversurface water temperature falls below 1.04øC in a rectangular region between 66ø and 78øN and-10 ø and 56øE,

and seasonal distribution

intermediate

of insolation follow-

values for soil texture

and color.

The basisfor the oceanheat transportsensitivitytest and increases linearlyto 500 W/m 2 if the oceancools is the enhancedNordic Sea heat convergencein winter. to its freezingpoint (-1.96øC) [Thompsonand Pollard, In the first experiment the additional Nordic Sea heat 1995]. This simulatesthe bufferingeffectof the deep- flux is turnedoff (actuallyreduced to 10 W/m •) in the ening mixed layer in winter, and advection of heat by model to simulate the NADW shutdown and the cold warmoceancurrents[e.g.,Hibler andBryan, 1987].At phaseof the YoungerDryas. This experimentis termed eachtime step after the Nordic Sea adjustment is made, "Nordic Sea Heat Flux Off" (Heat Off). The second an additive global adjustment is made between 55øN experiment has the Nordic Sea heat flux turned on to and 55øSto keep the globalintegral of oceanheat con- simulate the surfaceeffectsof active NADW formation, and the warmer climate followingthe Younger Dryas vergence at zero. The present-dayperformanceof the model is compa- (Preboreal). This experimentis termed "Nordic Sea rable to that of previous coarse-gridmodels with pre- Heat Flux On" (Heat On). dicted sea surfacetemperatures[Thompsonand PolResults lard, 1995]. The primary errors are too warm high- Model latitude

land areas in summer

associated with too few

For both experiments, the model was executed with

clouds, and too large a global precipitation value due to too active penetrative convectionin the lower troposphereand an overly large value of the roughnesslength

an AGCM spectralresolutionof R15 (4.5ø latitude by 7.5ø longitude)and a 2ø by 2ø surfacemodelresolution. Unless otherwise noted, all fields are means averaged

overopenocean(whichaffectsevaporationrates). The

over 10 consecutiveyears of an equilibrated run.

model does predict reasonablevaluesfor surfacetemperature, diurnal ranges of temperature, atmospheric energyfluxes,strength of jet stream maxima, and locations of precipitation maxima. The Younger Dryas experiments were designedfor 12,000 calendar years ago, the approximate midpoint

Surface Temperature

The specificationof Younger Dryas boundary conditions producessubstantialchangein global climate relative to the present day. Figure 2 showszonally aver-

90N

60N

30N

30S

90S 180

•

90W

0

90E

180

Figure 1. Paleogeographic map for 12,000calendaryearsago (adaptedfrom Peltier [1994])at the surface model resolution of 2ø by 2ø. Heavy lines show the extent of ice sheets;thin lines showtopography with a 1000-m contourinterval. G marks the GISP2 site, central Greenland, T is Troll 3.1 core, NorwegianSea, and D is DSDP Site 609, North Atlantic.

26

FAWCETT

ET AL.' YOUNGER

DRYAS TERMINATION

agedannualtemperatureprofilesfor threesimulations, the two YoungerDryasexperiments and a present-day

4O

control experiment. The ¾D experimentsare about 2ø-

20

3øCcoolerin the tropicsthan the presentdayandshow temperatures increasinglycoolerthan presentonestoward the poles. The northern hemispherecooledmore

/!

0

than the southernhemisphererelativeto the present day in both ¾D experimentsdue to the specification of

/,' -20

the Laurentide and Fennoscandian ice sheets.

Temperaturedifferences betweenthe YoungerDryas Heat Off and Heat On experimentsare largestbetween 60ø and 80øN.In the southernhemisphere, temperature differences are approximatelya degreeor less.The Heat On experimentis warmerthan the Heat Off experiment

-40

I

90

' o

60

,

,

•

,

,

1

in the northern hemisphereand cooler in the southern hemisphere. Figures 3 and 4 show the predicted December-Jan-

Latitude

Figure 2. Zonallyaveragedsurfacetemperaturefor three GENESIS experiments:the long-dashed line is

uary-February (DJF) andJune-July-August (JJA) surface temperatures for the ¾D Heat Off and Heat On experiments, respectively. The seasonaltemperature structure is very similar in both experimentsfor much

the present-daycontrol,the solidline is the YD Nordic Sea Heat On experiment,and the short-dashedline is the ¾D Nordic Sea Heat Off experiment. A.

I

90N I •--.. ,

I-,

I

I

I

60N

30N

30S

60S

90S 180

90W

0

90E

180

B. 90N

,

I

I

I

I

I

60N

30N

I

I

I

I

I

20

20

20 %

30S

2O

60S :'I":: :•20 -_-...... 180

Figure 3. Model-predicted surface temperature (10øCcontour interval)for the YD HeatOff experiment' (a) December-January-February (DJF)and(b) June-July-August (JJA).

FAWCETT

90N

ET AL.-

YOUNGER

DRYAS

TERMINATION

27

.._

I

I

I

I

I

I

60N

30N

20

".

3os

20

60S

90S 180

.........

I

90W

'I

I

I

0

I-"

I

90E

180

90N

I

•

I

I

I

I

I

60N

30N

30S

_

0

ø

6os i• :-:-:--_20 :•_.•._ •,0:--:•:---' '-----:::::::::::::=: .............. 90S 180

[-----1' ..... r .... 3---:-] ..... 90W

;--" 0

•

• 90E

180

Figure 4. Modelpredicted surface temperature (10øCcontourinterval)for the YD Heat On

experiment:(a) DJF and (b) JJA.

of the globewith the exceptionof the North Atlantic region. In the Heat Off case,the DJF 0øC isothermruns almost east-west,from the U.S. mid-Atlantic coast to

North Americaa very sharptemperaturecontrastruns along the Laurentide ice sheet margin with subfreezing temperatures over the ice itself and temperatures the southernBritish Isles. A sharptemperaturegra- of 10øC and higher south of the ice margin. Central dient existsoverthe North Atlantic and temperatures Greenlandagainhas the coldestnorthernhemisphere of-40 ø to-60øC

are found in northern North America

and Greenland.The summertemperaturepattern over North Americais complex,with subfreezing temperaturesoverthe specifiedice sheetsand very sharpgradientsalongthe icemargins.CentralGreenlandtemperatures are the coldestin the northernhemisphere.In the

summer temperatures.

The differencein annualaveragesurfacetemperature betweenthe two ¾D experimentsis shownin Figure 5; values from the Heat Off run are subtracted from

the Heat On run values.There is a largepositivedifferencein temperaturein the circum-Atlanticregionwhich Heat On experiment,the DJF 0øC isothermruns north- reachesa maximumof 25øCin the BarentsSearegion. east from the mid-Atlantic coast to north of the British Much smaller but hemisphere-widedifferencesare also Isles. Largeportionsof the Norwegianand BarentsSeas apparent,with the northernhemisphereslightlywarmer are at, or just slightlybelow,0øC.The sharpesttemper- and the southernhemisphereslightlycoolerthan in the ature gradientsoccurin easternNorth Americaalong Heat On experiment(seealsoFigure2). Only a small the SE marginof the Laurentideice sheet,and along portion of western North America is warmer in the Heat the SE Greenland coast. North American and GreenOff simulation.The largesttemperaturedifferences are land interior temperaturesrangefrom-40 ø to-50øC. In centeredin the region of the ocean heat flux forcing

28

FAWCETT

ET AL.: YOUNGER

DRYAS

TERMINATION

..........

.......

o

..... 90S 180

............................ 90W

_-•.. •

":::-'-•.• ,__, • 0

•

•

_ • 90E

180

Figure 5. Annualaverage surface temperature differences (HeatOn minusHeatOff). Positive differences are shaded,contourintervalis 2ø. Negativedifferencecontoursare dashed.

and are easily understoodin this context; the smaller but more widespreaddifferencesare more complex and will be discussed below.

Temperatures

at the GISP2

(Heat On versusHeat Off experiments)for the GISP2

Site

The predicted surface temperatures for the GISP2

sitein centralGreenland(76.2øN,38.5øW)for both YD experiments are averagesof the values from the four surfacegrid cells surroundingthe site. Table i shows the actual model grid point temperatures and temperatures that

are corrected

for a lowered Greenland

by 10øC and precipitation is reduced to 66.5% of the original rate. The annual average surface temperature difference

sum-

mit. GCMs that use a spectral method of solutionmust first smooth the surfacetopography which usually resultsin a small adjustment to the surfaceelevation. The surface of central Greenland, however, is lowered a full

site is 2.8øC. This differenceis strongly seasonal,with much of it occurring in winter and virtually identical summer(JJA) temperaturesin both experiments(Figure 6). Sea

Ice

In the GENESIS climate model, sea ice forms whenever the ocean surfacetemperature drops below -1.96øC.

The large differencesin North Atlantic ocean surface temperatures betweenthe two YD experimentsdo generate significant differencesin the sea ice. In the Heat kilometer(from3 km to 2 km) becauseof the steepslope Off case,the DJF ice margin runs from northern Newof the ice sheet margins. We apply a correctionusing foundlandto the northern tip of the British Isles(Figure the dry adiabaticlapserate of 10øC/km to the predicted 7) and the JJA margin lies slightlynorth of its winter surface temperature values, and reduce predicted pre- position. The Heat On experiment has a DJF sea ice cipitationvaluesby 4%/øC (seeTable 2). With a 1-km margin that runs from Newfoundland to Iceland and elevation difference, surface temperatures are reduced then north acrossthe northern part of the Barents Sea

Table 1. GENESIS-predicted surfacetemperatures for the Greenland summit Model Predicted Surface Temperatures, øC Younger Dryas Experiment

DJF

JJA

Annual

Heat

Flux

Off

-55.53 (-65.53)

-14.93 (-24.93)

-38.94 (-48.94)

Heat

Flux

On

-49.88 (-59.88)

-14.27 (-24.27)

-36.17 (-46.17)

Temperature difference

5.65

0.66

2.77

Surface temperature values correctedfor elevation are given in parentheses. DJF, December-January-February;JJA, June-July-August.

FAWCETT ET AL.' YOUNGER DRYAS TERMINATION

29

Table 2. GENESIS-predicted precipitation valuesfor the Greenland summit Model Predicted Precipitation, cm/month DJF

Younger Dryas Experiment

JJA

Annual

Heat Flux Off

0.62 (0.41)

3.34 (2.22)

18.48 (12.29)

Heat Flux On

1.07 (0.71)

3.52 (2.34)

21.08 (14.02)

Precipitation difference

0.45 (0.30)

0.18 (0.12)

2.70 (1.79)

(+73%)

(+5%)

(+15%)

Increase

Precipitation values correctedfor elevation are given in parentheses. DJF, December-January-February;JJA, June-July-August.

(Figure 7). The JJA marginalsoliesjust northwardof

period, this contrast is enhanced by the presenceof the

its winter position. The major differencebetween the

Laurentide

two simulations

is in the much wider extent of sea ice

Winter

Tracks

ice sheet.

We determine the position of the winter storm tracks in the Norwegianand Barents Seasin the colderexper- in the Younger Dryas simulationsby taking the staniment (Heat Off) and is a direct resultof the specified dard deviation of the 500-mbar geopotentialheight field, ocean heat flux forcing. The distribution of sea ice in followingBlackmonet al. [1976]. For mid latitude transient events, a time filter of 2.5 to 6 days is northesetwo experimentsstronglyaffectsthe nature of the atmosphericcirculation in the North Atlantic. mally applied to give a good measureof regionswhere strong high and low pressuresystemspass frequently. Storm

Synoptic-scalecyclonesare the dominant contributors of precipitationto Greenland,accountingfor 90% of the total moistureconvergence [Robasky and Bromwich, 1994]. Winter stormsare closelytied to the jet stream, which occurs where the average tropospheric meridional temperature gradient is greatest. The strongest temperature contrastsand highestwind speedsin winter occur at the eastern edgesof continents,and as we have noted in the case of the Younger Dryas time

At higher latitude regionssuchas Greenland,however, cyclonicdisturbancescan persistfor periodslongerthan 6 days, and so we apply a lower-frequencyfilter of 2 to 20 days for better resolution of the storm tracks. Variancesin the geopotentialheight of 80 m and greaterare used to define the position of the storm track. The northern-hemispherewinter storm track for the YD Heat Off experiment is a prominent feature which extends

from

ern North

eastern Siberia across much of northAmerica and across the North Atlantic into

northernEurope(Figure8). The regionof highestbaroclinic activity is anchored on the southeasternmargin of the Laurentide ice sheet, and this extends acrossthe

Atlantic towardsIceland. Central Greenlandhas significant storm activity, but the major axis of the Atlantic storm track lies to the southeast, just poleward of the DJF sea ice margin. The winter storm track for the Heat On experiment (Figure8) is reducedboth in extentand intensitycompared with that of the Heat Off experiment, yet it remains a very prominent feature. It also begins in east-

-2O

I-- -40

ern Siberia -60

'-..

.-" YD HeatOFF

't _

-8O

J

F

M

A

M

J

J

A

S

O

N

D

Figure 6. Model predicted annual surface temperature cycle at the GISP2 site, Greenland, for the YD

Heat On (solidline) andthe YD Heat Off (dashedline) experiments.

and crosses into northern

North

America

but endsin the North Atlantic. Again, the most intense portion of the storm track is anchoredon the southeast margin of the Laurentide ice sheet,althoughit doesnot showquite as much variability in the height field as in the colder simulation.

The axis of the North

Atlantic

storm track extends from this position to the northeast toward Greenland, roughly followingthe DJF sea

ice margin. Central Greenlandshowsmore variancein geopotentialheight in the Heat On simulationthan in the Heat Off run, consistentwith a higher frequencyof

30

FAWCETT ET AL.- YOUNGER DRYAS TERMINATION 90N

I

I

I

I

I

I

I

I

[

I

I

I

[

I

I

I

I

I

I

[

60N

30N

30S

6øs i

90S I 180

I

I

90W

0

90E

180

Figure 7. Model predictedDJF sea ice distribution (50-cm thickness)for the YD Heat Off (dashedline) and the YD Heat On (solidline) experiments.

90N

60N

30N

30S

60S

90S 180

90W

0

90E

180

90N

60N

30N

30S

60S

90S 180

Figure 8. Winter (DJF) stormtracks(standarddeviationof 2 to 20 day time-filteredgeopotential heightfield) for 5 yearsof the (a) YD Heat Off and (b) YD Heat On experiments.Contours start

at 80 m with

a 10-m interval.

FAWCETT

ET AL.: YOUNGER

and central

Greenland.

In the Heat

31

cipitation is low over the northern hemisphere continents with the exception of northwestern North America. The precipitation maxima over the North Atlantic

cyclonesreaching the interior. A secondstorm track affecting Greenland lies farther to the north, running from the north central part of the Laurentide ice sheet acrosscentral Greenland. A relative minimum in height field variance lies between these two axes, over the Labrador Sea. The explanation for this may be a northward propagationof cyclonesthrough the Nordic Seas to the higher latitudes and then a westwarddrift across northern

DRYAS TERMINATION

follow their respectivewinter storm tracks (southof the main axis of the tracks) with Greenlandreceivingmore precipitation in the YD Heat On experiment. In both experiments, a JJA North American precipitation belt followsthe southern margin of the Laurentide ice sheet, and the Middle East receivessummer monsoonalprecipitation (a result similar to that of COHMAP Members [1988]). Annual precipitation differencesbetween the two YD experiments are shown in Figure 10; values from the

Off case

the more zonal temperature pattern apparently blocks the northward propagation of cyclonesthrough the icecoveredNordic Seas. There are significantdifferencesin the path of winter storms between the two YD simulations, with more storm activity over central Greenland

Heat Off run are subtracted

from the Heat On run. The

in the Heat On case. The maximum intensity of northern hemispherewinter storm activity is greater in the

two regions of greatest difference in precipitation are

Heat

large changesin the tropics are consistentwith a north-

overthe Norwegian/BarentsSeasand the tropics. The

Off simulation.

ward shift of the intertropicalconvergence zone(ITCZ)

Precipitation

in the Heat On experiment. Northwestern North America receivesmore precipitation in the Heat Off casethan in the Heat On; this is one of the few locations in the

The predicted D JF precipitation fields for both experiments are shown in Figure 9. In both cases,pre-

B 90N

60N

30N

30S

90S

180

I

I 90W

I

I 0

I

I

I 90E

Figure 9. ModelpredictedDJF precipitation rates(millimetersper day,variablecontourinterval) for the (a) YD Heat Off and (b) YD Heat On experiments.

32

FAWCETT

ET AL.: YOUNGER

0

2

5

DRYAS TERMINATION

10 50 100

cm/yr

Figure 10. Annualaverage precipitation rate differences (centimeters peryear,variablecontour interval),YD HeatOnminusYD HeatOffexperiments. Positive differences areshaded.Negative contoursare-5,-50 and -100 cm/yr.

northern hemispherewhere this occurs. Precipitation can place confidencein these results, we need to assess rates are higher over central and northern Greenland how well the GENESIS model performsfor the present and over the Arctic Islandsin the Heat On experiment day at the summit of Greenland. and are about the samein both experimentsin southern Performance GISP2 Site

Greenland.

Precipitation

at the GISP2

Site

Predicted values for precipitation at the GISP2 site with the elevationcorrectionare givenin Table 2. On an annual averagebasis, the changein precipitation from the Heat Off to Heat On experiment is small with an

of GENESIS

for the

Modern

The surface temperature record of the GISP2 site has been obtained from long-term satellite passivemicrowavebrightnesstemperature trends and short-term

automaticweatherstationdata [Shumanet al., 1995]. 4

increaseof 1.5 cm/yr, or 15%. There is a very strong seasonalcomponentto this annual average,with most of the increaseoccurringin winter (Figure 11).

3

Discussion

We haveconductedtwo experimentsrepresentingthe Younger Dryas to Preboreal climate transition, in which the amount of oceanheat convergence specifiedin the Nordic Sea is varied. This isolatesthe predicted re-

•

2

predictedclimatechange withthe actualclimatechange reconstructed from the GISP2 ice core should allow an

assessment of howmuchof the changecanbe attributed to the turn-on of NADW formation. However,beforewe

-

I

sponse of Greenland's climate to a sudden reinitiation of North Atlantic Deep Water formation and associated

changesin ocean surfaceheat flux. Comparisonof the

YDHeatON

'"'-... .............. '"';D Heat OFF 0

•

I

,

I

J

F

M

A

I

M

J

"" .....

I

J

A

I

S

O

N

D

Figure 11. Model-predicted annualprecipitationcycle at the GISP2 site,Greenland for the YD HeatOn (solid line) and YD Heat Off (dashedline) experiments.

FAWCETT

ET AL.' YOUNGER

An interestingfeatureshownby theserecordsis that the summit area experiencesprominentsecondarywarm periodsin late fallSearlywinterin additionto the primary summer warm period. The annual averagetemperature at GISP2 is -31øC and the annual temperature cycle

DRYAS TERMINATION

Younger Dryas-

33

Preboreal

Climate

Differences

The Nordic Sea heat flux experiments predict a substantial climate change in Greenland as a consequence of switchingon North Atlantic Deep Water formation. The annual averagetemperatureincreasefrom the Heat Off to Heat On experiments at the GISP2 site is 2.8øC

rangesfrom -46øC in winter to about -8øC in summer - a rangeof 38øC (Figure12a;data from $humanet al. (Table 1), which is approximatelyhalf of the temper[1996]). ature increaseindicatedby the ice core(•lSO profile. A present-dayGENESIS controlexperimentwith the Nordic Sea Heat Flux On, a 2ø by 2ø surface resolution, and a 4.5ø by 7.5ø atmosphericresolution has a

predicted annual averagesurfacetemperature for the summit of Greenland of-16.0øC. The topography of the Greenland ice sheet is also truncated by a full kilometer in this experiment, so with the elevation correction the mean annual temperature is-26.0øC. This is 5øC warmer than is observed, which is consistentwith

the Thompson andPollard[1995]observation that highlatitude areas are in generalpredictedto be too warm in GENESIS. The annual range in temperature is quite well simulated, however, ranging from -46øC in win-

ter to-11øC in summer(Figure 12b), a 35øC annual range. The late fall/early winter secondarywarm intervals found in the Greenland temperature records are

well predictedby the model (Figure 12) with similar timing and magnitudes.Thus we concludethat GEN-

The actual increasein accumulationrate is 100%,much larger than the few tens of percentpredicted. The predicted temperature and precipitation differencesare very seasonalin character(Tables I and 2; Figures6 and 11). There is a largetemperatureincrease in DJF and a small increasein JJA. Seasonalprecipitation differencesare even more pronounced, with the largest increasein DJF. The annual surfacetemperature rangesare larger for the Younger Dryas than for the present day becauseof the increasedhigh-latitude

summerinsolation12,000yearsago [Berger,1978]so we expect that there will be enhanced precipitation seasonality for the ¾D also. In fact, there is a very pronouncedseasonalcycle in precipitation for both ¾D experiments(Figure 11). However,the overprediction of summer precipitation in the present-daycontrol experiment suggeststhat the predicted YD summer precipitation values are probably also too high and that

ESIS is able to simulate the temperature structure for central Greenland including both the annual tempera-

ture range and the secondarywarm interval, to within a few degreesof error. The

modern

annual

accumulation

rate at GISP2

,

,

GISP2 Site Greenland, Observed

is

24.7 cm/yr [Alley et al., 1993]. Accumulationrates have a low seasonalitywith winter averagesof 1.5 to 2

•-20

cm/monthand summeraveragesof approximately2.5 cm/month[Bromwichet al., 1993;$humanet al., 1995]. The GENESIS control experiment predicts annual pre-

cipitation rates of 35.1 cm/yr for the Greenlandsummit, which is about 40% higherthan observed.There is a strong annual cyclein predictedprecipitationwith winter valuesof 2 cm/month (closeto observed)and

-40

Dec, 1987

summervaluesof 5 to 7 cm/month(muchhigherthan

Dec, 88

Dec, 89

Dec, 90

GISP2 Site Greenland, GENESI

Dec,91

Model

observed).The overpredictionin summerprecipitation is a result of the model overestimation of evaporation

rates over the openocean(the largeroughness length discussed above). The errorin evaporationratesis more pronouncedin summer, when surfacetemperatures are high over the moisture sourceregionsfor Greenland. In winter, when the source temperatures are cooler, less evaporationoccursand the error is diminished. Thus we concludethat while the geographicpatterns of precip-

itation are largelywell simulated[Thompsonand Pollard, 1995],the seasonalamplitudeof precipitationpredicted for the summit of Greenland is too large, with winter values closeto, and summer valueshigher than observed. We expect that a similar summer overprediction will occur for other experimentsusing this model version.

(Correctedfor elevation) -60

'

I Dec

,

• Dec

,

I

I

Dec

Dec

Dec

Figure 12. (a) Recent near-surfaceair temperature for the GISP2 site, Greenland(data from $human et al. [1996]) and (b) 5 years of model-predictedsurface temperatures for the GISP2 site, Greenland, for the present-daycontrolexperiment.(Both are monthly averages).

34

FAWCETT

ET AL.:

YOUNGER

the seasonalrange in precipitation should be lower for both experiments. The relative contributions from the winter seasonsto the annual precipitation totals should then be higher. As most of the precipitation increase from the Heat Off to the Heat On experiment occursin winter, we should also expect the annual averageprecipitation differenceto be larger and thereforecloserto the observed

value.

The predicted change in precipitation seasonalityat the Younger Dryas to Preboreal climate transition has important implications for the isotopic record of temperature change. In the cold YD, low winter precipita-

DRYAS

TERMINATION

ducedhemisphericstorminess(and reducedpeak wind speeds). The spatial patterns of increasedsurfacetemperature and precipitation(annualdifferenceplotsin Figures5 and 10) are consistentwith the shift in winter storm tracks. For both climatic variables,the largestchanges occur in northern Greenland and the magnitude of change decreasesfrom east to west. This GISP2 site actually liesin a temperaturedifferenceminimum(Figure 5). In the Heat On experiment,winter stormsdrift westward

across central to northern

Greenland

from the

open Norwegian Sea, delivering more precipitation and

tion amountsbias the annualaverage51sOvalueto- warming these regionsby advection of warmer air into wards summer when precipitationis higher (because the interior of the ice sheetand by increasedcloudcover

which traps surface-emitted longwaveradiation. The mometer). In the warmerPB, winter precipitationin- trapped IR is a significantpart of the radiation budget creasesmarkedly, and the thermometer sampleswinter at these high latitudes becauseincomingwinter insolatemperaturesmorefrequently.The changein 5•sO in tion is essentially zero. There may also be some comthe ice core will then underestimate the actual temper- ponent of increasedlatent heating of the atmosphere ature increase,all elsebeingequal[$teiget al., 1994]. associatedwith the storms, but this is probably suborThe significantdifferencesbetweenthe modern spatial dinate to the other mechanisms. 51sO- T calibration of 0.67per mil per øC [Johnsen The shift in storm tracks over Greenland explains

it is the snowfall itself that contains the isotopic ther-

et al., 1989]and the temporal5•sO- boreholeT cali- why the winter temperature(and precipitation)differ-

brationsof 0.51 and 0.33 per mil per øC [CuJfeyet al., 1994,1995]may well resultfromthis type of changein seasonality of precipitation.

Predictions of large winter changesin temperature and precipitation are not surprisinggiven that the forcing factor, North Atlantic Deep Water formation, is a

seasonal(winter) phenomenon.Present-dayEuropeis considerably warmer in winter lying downwind of this heat source than it would be otherwise. Greenland, however, lies upwind of this heat source. When NADW is "switched on," the increasesin temperature and precipitation at the GISP2 site must be due to a more

complexprocessthan simpledownwind(westerly)advection

encesare so large. The Heat Off experiment, in which winter storms track away from Greenland, has very cold winters and a reduced to nonexistentsecondary warm peak (Figure 13). Winters in the Heat On experiment, where winter storms track close to Greenland, are considerablywarmer and have the secondarywarm peak noted in the present-day control experiment and in modern

climate

data.

This difference in winter tem-

perature structure shows that the shift in the North

Atlantic storm track drives regional climate changein Greenland

at the YD

termination.

The high correspondenceof the climate model results to the ice core data suggeststhat a suddenreinitiation

of heat and moisture.

The shifts in atmosphericcirculation patterns, especially the winter storm tracks, from the Younger Dryas to Preboreal appear to be the explanation. In the Heat Off experiment, the North Atlantic storm track is very intense and extends east from the southern edge of the Laurentide ice sheet out into the Atlantic (Figure 7). Its counterpart in the Heat On experiment is not as intense, does not extend as far into the Atlantic, and curls up onto central and northern Greenland from the

YD HeatOFF

GISP2SiteGreenland

YDHeatON

(Corrected forelevation)

-20

I

-40

I ¾, I/ I"1 '

NorwegianSea(Figure7). The resultingchangefor central Greenlandis increasedregionalwinter storminessin

-60

concert with hemispherically reduced storminess. Both aspectsof this prediction are well supported by analy-

sisof the ice core. Kapsneret al. [1995]demonstrated that a storm track shift onto Greenland was required to explain the extra increase in accumulation relative to temperature across the YD-PB transition, and the abrupt decreasein dust and particulate loading in the

ice acrossthis transition[Mayewskiet al., 1993; Taylor et al., 1993; Alley et al., 1995]is consistent with re-

-80

,v,i,,,,,,u/ v,/ I

•

I

I

Dec

Dec

',,

¾' -',,/ I

Dec

,

I

Dec

Dec

Figure 13. Five years of model predictedannual average surfacetemperaturefor the GISP2 site, Greenland, for the YD Heat Off (dashedline) and YD Heat On experiments

FAWCETT

ET AL.:

YOUNGER

of NADW formation after an extended period of shutdown is a very plausibleexplanationfor the large and rapid climate change in Greenland at the end of the Younger Dryas. However, the results do not explain all of the observedclimate change, and this raises three

possibilities.(1) The hypothesisis incorrect,and the apparent successof the results is merely a chance oc-

currence.(2) The hypothesisis correct,but the model is not fully sensitiveto the climate forcing. (3) The hypothesisis partially correct, but some other climate

forcingfactor(s) have also contributedto the climate change. We reject the first possibility on the basis of widespread paleoceanographicevidencefor NADW formation shutdown during the Younger Dryas, the bistable nature of deep ocean convectionin the North Atlantic in a number of ocean circulation models, and the degree to which our atmosphericmodeling results match patterns of changein the ice core record. This leads us to concludethat the model is not fully sensitiveto this

climateforcingand/or that other factorshaveactedin combination

with the Atlantic

thermohaline

circulation.

The coarse atmospheric resolution of the GENESIS experiments may have contributed to an underestimation of the actual climate change. Close examination

of Figures5 and 10 showsthat the GISP2siteactually lies in relative minima for both temperature and precipitation differences. As we have determined that these differencesare largely due to storm track migration, a higher- resolutionmodel experimentwith a more realistic topography could changeslightly the position of the Greenland

winter

storm tracks and increase the

differencesbetween the ¾D experiments. For other climate forcing factors to be consideredfor the YD termination, two important criteria must be

met. They must operate on a very rapid timescaleand have a global effect. The Younger Dryas has long been recognizedin terrestrial climate recordsof northwestern Europe, and more recently in Greenland ice coresand deep-seasedimentcoresfrom the North Atlantic basin. With improved dating techniques,cooling events contemporaneouswith the North Atlantic Younger Dryas have been recognizedin a number of widespreadlocal-

ities (seePeteet [1995]for review): easternand western North America, Alaska, the South AmericanAndes, New Zealand, and the westernPacific Ocean. In East Africa a YD-termination-contemporaneouschangefrom a more arid to a more humid climate is shown by dra-

DRYAS

TERMINATION

can affect radiation balancethroughalbedoeffectsand IR absorption). Ice core recordsfrom Greenlandand Antarctica

show increases from the ¾D to P B in both

CH4andCO2 [SowersandBender,1995],althoughthey are relatively small comparedto longer-termvariations

(CO2 increases by 10 ppmv and CH4 increases by 220 ppbv). The increasein CH4 is relativelylargerthan that for CO2 and could have contributed a small amount to

the YD end warming. The Greenland ice core record also showsa sharp decreasein atmosphericdust load-

ing from YD to PB [Taylor et al., 1993; Mayewskiet al., 1993; Alley et al., 1995], but the actual climatic responseto this change is not well understood. The Antarctic

ice core records do not show an increase in

atmosphericdust contentduringthe YD [Jouzelet al., 1995],so this climaticforcingappearsto be restricted to the northern hemisphere. The ocean circulation is another possiblecandidate

for transmittinga climatesignalaroundthe globe. Our modelresultssuggestthat the rapid climatechangesin tropical and subtropicalAfrica could have resultedfrom the small changesin tropical sea surfacetemperatures by a mechanism similar to that proposed by StreetPertoft and Pertoft [1990]. The turnoff of NADW formation in the North Atlantic produces a cooling which extendswell into the northernsubtropics(Figure 5). This is coupledto reducedprecipitation(and precipitation-evaporation)in large parts of Africa and centralAmerica (Figure 10) and is consistentwith the lower lake levels in East Africa during the YD. When North Atlantic Deep Water formation resumes,the opposite pattern occurswith higher precipitation in East Africa. Though of marginal statistical difference,this result is strengthenedby its strong agreementwith the paleoclimatic data. The atmospheric methane variations recorded

in the ice cores could have resulted

in

part from suchhydrologicchangesaffectingtropical and subtropical vegetation. The model results for southern hemispherelocalities

(Antarctica,New Zealand)are not consistentwith the regional paleoclimatic records in that no pronounced coolingduring the YD is predicted. This may be because we do not include the effects of CO2 and CH4 in the experiments, or becausethe actual climate signalinvolves ocean circulation

and cannot be simulated

with-

out a true ocean general circulationmodel. If the interactionbetweenNADW (slightlywarmerand saltier) and southern ocean water masseschangessignificantly maticchanges in lakelevelsandsalinities[$treet-Perrott when deep water formation rates changein the North and Perrott, 1990; Robertset al., 1993]. If all of these Atlantic, then it is possible that the southern hemieventsare causallyrelated, then the forcingfactor(s) sphereclimate system could be affected. This idea cannot be tested without a more detailed model of ocean affectingthe North Atlantic climate must have had a circulation. direct global impact, or at least an indirect impact via climate

feedbacks or teleconnections.

At the time of the YD termination, rapidly changing factors that•,..•.might have had a global impact in-

Conclusions

clude changesin atmosphericgreenhouse gases(CH4, C02) and changesin atmosphericdust content(which

A variety of paleoclimaticindicatorsin the GISP2 ice core showthat the ¾D climate event ended abruptly in

FAWCETT

ET AL.: YOUNGER

DRYAS TERMINATION

the North Atlantic region,possiblyin as little as I to 3 years. This termination is characterizedby a 3.5 to 4.0

10. The GISP2 site climatechanges(in temperature and precipitation)are not fully explainedby the model

per mil increase in 5180of ice (whichcorresponds ap-

experiments,indicating that the model is not fully sen-

proximatelyto a 7øCtemperatureincrease),a doubling sitive to the oceanheat flux forcingand/or that some of the summit accumulationrate, and an increasein re- other climate forcing factor(s), have acted in concert gionalstorm activity but a decreasein stormactivity in with the switches in thermohaline circulation. broader areas. A reorganizationof the North Atlantic 11. Abrupt changesin subtropicalhydrologicbalthermohalinecirculationhas beensuggested to explain ancescould have been causedby reorganizationsin the both the rapidity and the size of the YD termination. The GENESIS climatemodelcontainsa provisionfor ocean heat transport which allows us to test the effects of NADW-formation-relatedexperimentsfor YD conditions, one with low and one with high oceanheat flux in the Nordic Seas(representingdeepwater shutdown and reinitiation), werecomparedto the GISP2 ice core record of climate change. These results showthe following:

1. Annual averagesurfacetemperatureincreases by 2.8øC at the Greenlandsummit when high oceanheat flux is appliedin the NordicSeas.This is approximately half of the temperature increaserecordedin the ice core. 2. Annual averageprecipitationrate increasesby a few tens of percent, as comparedto the 100% increase in accumulation

East African precipitationwhenhigh NordicSeasheat flux is applied, and cross-equatorialocean heat transport increasesto compensate,is consistentwith higher lake levelsin that region at the end of the YD.

12. The predictedchangesin subtropicalhydrologic balance should have affectedtropical and subtropical vegetationand could have contributedto the higher atmospheric CH4 levels at the close of the YD.

Together, these resultsstronglysupport the hypothesis that a change in oceanic heat flux in the Nordic

Seasis linked to the YoungerDryas but suggestthat processesor feedbacks not modeled here were also important.

rate in the ice core.

3. Both of these increaseshave a pronouncedseasonality,with most of the changeoccurringin winter; this result has implicationsfor interpretationof ice core

51sOpaleothermometry. 4. The large increasein winter precipitationis "swamped" by an overpredictionof summer precipitation. This explains, in part, the model underestimation of the annual

ocean circulation system which act to modify tropical seasurfacetemperatures.A model-predicted increasein

accumulation

rate increase.

Acknowledgments. We thank Steve Feldstein, Sukyoung Lee, and Dave Pollard for helpful discussions.Reviews by L.C. Sloan, M.L. Delaney, and an anonymousreviewer substantially improved the paper. This work was supportedby the NSF O•ce of Polar Programsand the David

and Lucile Packard

Foundation.

References

5. North Atlantic seaice extent is greatly reducedin Alley, R. B., et al., Abrupt increasein Greenlandsnowaccuthe high Nordic Sea ocean heat flux experiment. The mulation at the end of the YoungerDryas event, Nature, predicted changesin sea ice distribution roughly cor362, 527-529, 1993. respondto polar front migrationsrecordedin deep-sea Alley, R.B., R.C. Finkel, K. Nishiizumi, S. Anandakrishsediment

6.

cores.

North

Atlantic

winter

storm tracks shift toward

Greenland in the warmer climate experiment (Heat On), away from the more zonal orientationin the cold climate experiment (Heat Off). They also weakenin intensity over North America and the westernAtlantic, which is consistent

with the reduced continental

dust

and sea salt loading in the ice cores. 7. Heightened storminessover Greenland in the warm climate experiment contributesto the large increasein predicted winter precipitation rates. 8. A present-daycontrol experiment showsthat the

nan, C.A. Shuman, G.R. Mershon, G.A. Zielinski, and P.A. Mayewski, Changesin continental and sea-salt atmosphericloadingsin centralGreenlandduringthe most recent deglaciation: Model-basedestimates,J. Glaciol., • 1, 503-514, 1995.

Andrews,J.T., H. Erlenkeuser,K. Tedesco,A.E. Aksu and

A.J.T. Jull, Late Quaternary(Stage2 and 3) meltwater and Heinrichevents,northwestLabradorSea, Quat. Res., J1, 26-34, 1994.

Barnola, J.M., D. Raynaud, Y.S. Korotkevich,and C. Lorius, Vostok ice core provides 160,000-yearrecord of atmosphericCO•., Nature, 329, 408-414, 1987. Berger, A.L., Long-term variations of caloric insolation resulting from the Earth's orbital elements,Quat. Res., 9,

secondarylate fall/winter temperature maximum in 139-167, 1978. Greenland summit surfacetemperature data is caused Birchfield, G.E., and W.S. Broecker, A salt oscillator in by the passageof winter storms. The warmer temperathe glacial Atlantic?, 2, A "scale analysis" model, Patures result from heat advection into central Greenland

leoceanography, 5, 835-843, 1990.

and from increased cloud cover which traps surface- Blackmon,M.L., J.M. Wallace, N.C. Lau, and S.L. Muller, An observationalstudy of the northernhemispherewinemitted IR. 9. This winter-storm-related secondarytemperature

maximumoccursat the GISP2 sitein the (YD) warmer climate experiment, and does not in the cold climate experiment.

tertime circulation, J. Atmos. Sci., 3•, 1040-1053, 1976.

Bond, G., W.S. Broecker, S. Johnsen,J. McManus, L. Labeyrie, J. Jouzel, and G. Bonani, Correlations between climate records from North Atlantic sediments and Green-

land ice, Nature, 365, 143-147, 1993.

FAWCETT

ET AL.: YOUNGER

Broecker,W.S., Chaotic climate, Sci. Am., 273 (5), 62-68, 1995.

TERMINATION

37

coupled ocean-atmospheremodel, J. Clim., 1, 841-866, 1988.

Broecker, W.S., and G.H. Denton, The role of ocean-atmosphere reorganizations in glacial cycles, Geochim. Cosmochim. Acta, 53, 2465-2501, 1989. Broecker, W.S., D. Peteet, and D. Rind, Does the oceanatmospherehave more than one stable mode of operation, Nature, 315, 21-25, 1985. Broecker, W.S., G. Bond, and M. Klas, A salt oscillator in the glacial Atlantic?, 1, The concept, Paleoceanography, 5, 469-477, 1990. Bromwich,D.H., F.M. Robasky,R.A. Keen, and J.F. Bolzan, Modeled variations of precipitation over the (]reenland ice sheet, J. Clim., 6, 1253-1268, 1993. Bryan, F., High-latitude salinity effects and interhemispheric themohaline circulations, Nature, 323, 301-304, 1986.

Charles, C.D., D. Rind, J. Jouzel, R.D. Koster, and R.(]. Fairbanks, (]lacial-interglacial changesin moisture sources for Greenland:

DRYAS

Influences

on the ice core record of

climate, Science, 263, 508-511, 1994.

COHMAP Members, Climatic changesof the last 18,000 years: Observationsand model simulations,Science,2,il, 1043-1052, 1988.

Covey,C., and S.L. Thompson,Testing the effectsof ocean heat transport on climate, GlobalPlanet. Change,1, 331341, 1989.

Mayewski, P.A., L.D. Meeker, S. Whirlow, M.S. Twickler, M.C. Morrison, R.B. Alley, P. Bloomfield,and K. Taylor, The atmosphereduring the Younger Dryas, Science,261, 195-197, 1993.

Meese, D.A., A.J. (low, P. (]rootes, P.A. Mayewski, M. Ram, M. Stuiver, K.C. Taylor, E.D. Waddington, and (].A. Zielinkski, The accumulationrecordfrom the GISP2 core as an indicator of climate change throughout the Holocene, Science, 266, 1680-1682, 1994. Miller, G.H., and D.S. Kaufman, Rapid fluctuations of the Laurentide

ice sheet at the mouth

of Hudson

Strait:

New

evidencefor ocean/ice-sheet interactionsas controlon the Younger Dryas, Paleoceanography,5, 907-919, 1990. Peltier, W.R., Ice age paleotopography, Science, 265, 195201, 1994. Peteet, D.M., Global Younger Dryas?, Quat. Int., 28, 93104, 1995.

Robasky, F.M., and D.H. Bromwich, Greenlandprecipitation estimatesfrom the atmosphericmoisturebudget, Geophys.Res. Left., 21, 2495-2498, 1994. Roberts, N., T.M. Taieb, P. Barker, B. Damnart, M. Icole, and D. Williamson,Timing of the YoungerDryaseventin East Africa from lake-levelchanges,Nature,366, 146-148, 1993.

Severinghaus,J.P., E.J. Brook, T. Sowers, and R.B. Al-

ley,Gaseous thermaldiffusionasa gas-phase stratigraphic markerof abruptwarmingsin icecoreclimaterecords(abstract), Eos Trans. AGU, 77 (17), SpringMeet. Suppl.,

Cuffey, K.M., R.B. Alley, P.M. (]rootes, and S. Anandakrishnan, Toward using boreholetemperatures to calibrate an isotopicpaleothermometerin central (]reenland, Palaeogeogr.Palaeoclim. Palaeoecol.,98, 265-268, 1992. Cuffey, K.M., R.B. Alley, P.M. Grootes, J.F. Bolzan, and

Shuman,C.A., R.B. Alley, S. Anandakrishnan,J.W. White,

S. Anandakrishnan, Calibrationof the (•lSoisotopic pale-

P.M. (]rootes, and C.R. Stearns, Temperature and accu-

othermometerfor central Greenland, using boreholetemperatures, Y. Gla½iol.,,iO, 341-349, 1994. Cuffey, K.M., (I.D. Clow, R.B. Alley, M. Stuiver, E.D. Waddington, and R.W. Saltus, Larger Arctic temperature change at the Wisconsin- Holoceneglacial transition, Science, 270, 455-458, 1995. Dansgaard, W., J.W. White, and S.J. Johnsen, The abrupt termination of the Younger Dryas climate event, Nature, 339, 532-534, 1989. Grootes, P.M., M. Stuiver, J.W.C. White, S. Johnsen,and J. Jouzel, Comparisonof oxygenisotoperecordsfrom the

GISP2 and GRIP Greenlandice cores,Nature, 366, 552554, 1993.

Hibler, W.D., and K. Bryan, A diagnosticice ocean model, J. Phys. Oceanogr.,17, 987-1015, 1987. Johnsen, S.J., W. Dansgaard, and J.W.C. White, The origin of Arctic precipitation under presentand glacial conditions, Tellus, Set. B., •1, 452-468, 1989. Johnsen, S.J., H.B. Clausen, W. Dansgaard, K. Fuhrer,

N. (]undestrup, C.U. Hammer, P. Iverson,J. Jouzel,B. Stauffer, and J.P. Steffensen,Irregular glacial intersradials recorded in a new (]reenland ice core, Nature, 359, 311-313, 1992.

Jouzel, J., et al., The two-step shape and timing of the last deglaciationin Antarctica, Clim. Dyn., 11, 151-161, 1995. Kapsner, W.R., R.B. Alley, C.A. Shuman, S. Anandakrishnan, P.M. (]rootes, D.A. Meese,and A.J. Gow, Response of snow accumulationto temperature variations in central Greenland, Nature, 373, 52-54, 1995. Lehman, S.J., and L.D. Keigwin, Sudden changesin North Atlantic circulation during the last deglaciation,Nature, 356, 757-762, 1992. Manabe, S., and R.J. Stouffer, Two stable equilibria of a

S157, 1996.

mulationat the Greenlandsummit: Comparisonof highresolutionisotopeprofilesand satellite passivemicrowave brightnesstemperature trends, J. Geophys. Res., 100, 9165-9177, 1995.

Shuman, C.A., R.A. Bindschadler,R.B. Alley, and C.R. Stearns, Composite temperature record from the Greenland summit, 1987-1994: Synthesisof multiple automatic weather station recordsand SSM/I brightnesstemperatures, J. Clim., 9, 1421-1428, 1996. Sowers,T., and M. Bender, Climate recordscoveringthe last deglaciation, Science,269, 210-214, 1995. Steig, E.J., P.M. (]rootes, and M. Stuiver, Seasonalprecipitation timing and ice core records, Science, 266, 18851886, 1994.

Street-Perrott, F.A., and R.A. Perrott, Abrupt climate fluctuations in the tropics: The influence of Atlantic Ocean circulation, Nature, 3,i3, 607-612, 1990. Taylor, K.C., (].W. Lamorey,G.A. Doyle, R.B. Alley, P.M. (]rootes,P.A. Mayewski,J.W.C. White, and L.K. Barlow,

The 'flickeringswitch'of late Pleistoceneclimatechange, Nature, 361, 432-436, 1993. Thompson, S.L., and D. Pollard, A global climate model

(GENESIS) with a land-surfacetransferscheme(LSX). Part I: Present climate simulation, J. Clim., 8, 732-761, 1995.

Williamson, D.L., J.T. Kiehl, V. Ramanathan, R.E. Dickinson, and J.J. Hack, Description of NCAR Community Climate Model (CCM1), NCAR Tech. Note TN285, Natl. Cent. for Atmos. Res., Boulder, Colo., 1987.

Wright, D.(]., and T.F. Stocker, Younger Dryas experiments,in Ice in the ClimateSystem,NATO ASI Set., Set I, vol. 12, edited by W.R. Peltier, pp. 395-416,SpringerVerlag, New York, 1993.

FAWCETT

ET AL.: YOUNGER

DRYAS TERMINATION

C. A. Shuman, Mail Code 971, Oceans and Ice Branch,

A.M. ,igdstsddttir andR. B. Alley,Department of Geo- NASA Goddard SpaceFlight Center, Greenbelt, MD 02543.

sciencesand Earth System ScienceCenter, The Pennsylvania State University,University Park, PA 16802. (e-mail: annamari@essc. psu.edu; ralley@essc. psu.edu) P. J. Fawcett, Department of Physics, University of Toronto, 60 St. George Street, Toronto, Ontario, Canada, M5S

1A7.

(email:[email protected])

(e-mail: shuman@hardy. gsfc.nasa.gov)

(Received December 29,1995;revised September 4,1996;

acceptedSeptember6, 1996.)