NASA Technical Reports Server (NTRS) 20140017656: Evaluating the Dominant Components of Warming in Pliocene Climate Simulations

1.4 MB·English

by  NASA Technical Reports Server (NTRS)

Clim.Past,10,79–90,2014 O
www.clim-past.net/10/79/2014/ Climate pe
n
doi:10.5194/cp-10-79-2014 A
of the Past c
c
©Author(s)2014.CCAttribution3.0License. e
s
s
Evaluating the dominant components of warming in Pliocene
climate simulations
D.J.Hill1,2,A.M.Haywood1,D.J.Lunt3,S.J.Hunter1,F.J.Bragg3,C.Contoux4,5,C.Stepanek6,L.Sohl7,
N.A.Rosenbloom8,W.-L.Chan9,Y.Kamae10,Z.Zhang11,12,A.Abe-Ouchi9,13,M.A.Chandler7,A.Jost5,
G.Lohmann6,B.L.Otto-Bliesner8,G.Ramstein4,andH.Ueda10
1SchoolofEarthandEnvironment,UniversityofLeeds,Leeds,UK
2BritishGeologicalSurvey,Keyworth,Nottingham,UK
3SchoolofGeographicalSciences,UniversityofBristol,Bristol,UK
4LaboratoiredesSciencesduClimatetdel’Environnement,Saclay,France
5Sisyphe,CNRS/UPMCUniv.Paris06,Paris,France
6AlfredWegenerInstituteHelmholtzCentreforPolarandMarineResearch,Bremerhaven,Germany
7ColumbiaUniversity–NASA/GISS,NewYork,NY,USA
8NationalCenterforAtmosphericResearch,Boulder,Colorado,USA
9AtmosphereandOceanResearchInstitute,UniversityofTokyo,Kashiwa,Japan
10GraduateSchoolofLifeandEnvironmentalSciences,UniversityofTsukuba,Tsukuba,Japan
11UniResearchandBjerknesCentreforClimateResearch,Bergen,Norway
12Nansen-zhuInternationalResearchCentre,InstituteofAtmosphericPhysics,ChineseAcademyofSciences,Beijing,China
13JapanAgencyforMarine-EarthScienceandTechnology,Yokohama,Japan
Correspondenceto:D.J.Hill([email protected])
Received:28February2013–PublishedinClim.PastDiscuss.:26March2013
Revised:29November2013–Accepted:2December2013–Published:15January2014
Abstract. The Pliocene Model Intercomparison Project important,butthedominantwarminginfluencecomesfrom
(PlioMIP)isthefirstcoordinatedclimatemodelcomparison theclearskyalbedo,onlypartiallyoffsetbytheincreasesin
for a warmer palaeoclimate with atmospheric CO signifi- the cooling impact of cloud albedo. This demonstrates the
2
cantlyhigherthanpre-industrialconcentrations.Thesimula- importance of specified ice sheet and high latitude vegeta-
tionsofthemid-Pliocenewarmperiodshowglobalwarming tion boundary conditions and simulated sea ice and snow
◦
of between 1.8 and 3.6 C above pre-industrial surface air albedofeedbacks.Thelargestcomponentsintheoverallun-
temperatures, with significant polar amplification. Here we certainty are associated with clouds in the tropics and polar
performenergybalancecalculationsonalleightofthecou- clearskyalbedo,particularlyinseaiceregions.Thesesimu-
pledocean–atmospheresimulationswithinPlioMIPExperi- lationsshowthatalbedofeedbacks,particularlythoseofsea
ment 2 to evaluate the causes of the increased temperatures iceandicesheets,providethemostsignificantenhancements
anddifferencesbetweenthemodels.Inthetropicssimulated tohighlatitudewarminginthePliocene.
warmingisdominatedbygreenhousegasincreases,withthe
cloud component of planetary albedo enhancing the warm-
ing in most of the models, but by widely varying amounts.
Theresponsestomid-PlioceneclimateforcingintheNorth- 1 Introduction
ern Hemisphere midlatitudes are substantially different be-
tweentheclimatemodels,withtheonlyconsistentresponse Atmosphericcarbondioxideconcentrationscontinuetorise
being a warming due to increased greenhouse gases. In the due to anthropogenic emissions. The latest measurements
high latitudes all the energy balance components become show that annual mean concentrations have risen beyond
390 parts per million (Conway et al., 2012). The Pliocene
PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion.
80 D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations
Table1.KeymodelandexperimentaldesignparametersforeachoftheeightPlioMIPExperiment2simulations.
GCM Atmospheric Oceanresolution Boundary Ocean Reference
resolution (◦lat×◦long×levels) conditions initialization
(◦lat×◦long×levels) employed
CCSM4 0.9×1.25×26 1×1×60 Alternate PRISM3 Rosenbloomet
(anomaly) al.(2013)
COSMOS 3.75×3.75×19 3×1.8×40 Preferred PRISM3 Stepanekand
(anomaly) Lohmann(2012)
GISS-E2-R 2×2.5×40 1×1.25×32 Preferred PRISM3 Chandleretal.
(2013)
HadCM3 2.5×3.75×19 1.25×1.25×20 Alternate PRISM2 Braggetal.
mPWPcontrol (2012)
IPSLCM5A 3.75×1.9×39 0.5–2×2×31 Alternate Pre-industrial Contouxetal.
control (2012)
MIROC4m 2.8×2.8×20 0.5–1.4×1.4×43 Preferred PRISM3 Chanetal.
(2011)
MRI-CGCM2.3 2.8×2.8×30 0.5–2×2.5×23 Alternate PRISM3 Kamaeand
(anomaly) Ueda(2012)
NorESM-L 3.75×3.75×26 3×3×30 Alternate Levitus Zhangetal.
(2012)
wasthelastperiodofEarthhistorywithsimilartomodernat- betweenthesimulationshavenotbeenextensivelyexplored
mosphericCO concentrations(Kürschneretal.,1996;Seki prior to this study. In this paper the energy balance of the
2
et al., 2010; Pagani et al., 2010; Bartoli et al., 2011). These PlioMIP Experiment 2 simulations are analysed in order to
wereassociatedwithelevatedglobaltemperaturesinboththe understandthecausesofPlioceneatmosphericwarmingand
ocean (Dowsett et al., 2012) and on land (Salzmann et al., the latitudinal distribution of increased surface air tempera-
2013).Asthelastperiodofglobalwarmthbeforetheclimate tures. This analysis allows us to analyse the causes of the
transition into the bipolar ice age cycles of the Pleistocene, warming,bothdirectlythroughthesimulatedenergybalance
the mid-Pliocene warm period (mPWP) has been a target componentsandthroughexaminationofEarthsystemcom-
for both palaeoenvironmental data acquisition and palaeo- ponent changes that are driving these. It is also important
climate modelling over a number of years (Dowsett et al., when discrepancies with available proxy reconstructions of
1992;Chandleretal.,1994;Haywoodetal.,2009;Dowsett Pliocenewarmingareconsidered(Salzmannetal.,2013).
etal.,2010).Althoughanumberofdifferentgeneralcircula-
tionmodels(GCMs)havebeenusedtosimulatePliocenecli-
2 Participatingmodels
mates(Chandleretal.,1994;Sloanetal.,1996;Haywoodet
al.,2000,2009;Dowsettetal.,2011),itisonlyrecentlythata
Eightdifferentmodellinggroupshavesubmittedsimulations
coordinatedmulti-modelexperimenthasbeeninitiated,with
to PlioMIP Experiment 2. All of these models are coupled
standardizeddesignformid-Pliocenesimulations(Haywood
ocean–atmosphereGCMs,butrangeincomplexityandspa-
etal.,2010,2011).
tial resolution. Table 1 contains the details of each of the
The Pliocene Model Intercomparison Project (PlioMIP)
models’simulation,includingtheresolutionatwhichitwas
represents the first coordinated multi-model experiment to
run,theboundaryconditionsemployedandthemodelinitial-
simulateawarmerthanmodernpalaeoclimate,withhighat-
ization.Eachofthesimulationsisdocumentedinmuchmore
mospheric CO concentrations (405ppmv). It has recently
2 detailinaseparatepaperwithinaspecialissueofGeoscien-
been added to the Paleoclimate Model Intercomparison
tificModelDevelopment,referencedinTable1.Thegeneral
Project (PMIP; Hill et al., 2012) and the first phase, in-
climatesensitivityofthemodelandtheannualmeanglobal
corporatingtwo simulations,completed.This paperfocuses
warmingproducedinitsPlioMIPExperiment2simulationis
on PlioMIP Experiment 2, designed for coupled ocean–
detailedinTable2.Furtherdetailsaboutthemodelscanalso
atmosphereGCMs(Haywoodetal.,2011).Although,many
befoundinHaywoodetal.(2013)andthereferencestherein.
ofthelarge-scalefeaturesofthesimulatedPlioceneclimate
havebeenwelldocumented(Dowsettetal.,2012;Haywood
et al., 2013; Zhang et al., 2013, 2013a,b; Salzmann et al.,
2013), the causes of the simulated changes and differences
Clim.Past,10,79–90,2014 www.clim-past.net/10/79/2014/
D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations 81
Table 2. The climate sensitivity and global mean annual surface 4 PlioMIPExperiment2globalwarming
air temperature warming of each of the models with simulations
in the PlioMIP Experiment 2 ensemble. Climate sensitivity is the Overall the PlioMIP models simulate mPWP annual mean
◦
equilibriumglobalwarmingforadoublingofatmosphericCO2 – globalsurfaceairtemperature(SAT)increasesof1.8–3.6 C
thevaluesquotedhereareageneralvalueforeachmodelanddo (Table 2). Tropical temperatures increased by only 1.0–
notrefertotheparticularset-upandinitializationproceduresused 3.1◦C,whileintheArcticsurfaceairtemperaturesincreased
forPlioMIP. ◦
by3.5–13.2 C(Fig.1).Seasurfacetemperatures(SSTs)fol-
lowasimilarpattern,butwithareducedmagnitudeofglobal
GCM Climate Meanannual
warmingandsignificantlygreaterwarmingintheNorthPa-
sensitivity mPWPSAT
◦ ◦ cific(Fig.1d).Thepatternsofwarminginthenorthernmid-
( C) warming( C)
latitudesandsouthernhighlatitudesaremuchmorevariable
CCSM4 3.2 1.86 betweenthedifferentmodels.Relativevariationbetweenthe
COSMOS 4.1 3.60
modelsislargestintheNorthAtlantic,midlatitudemountain
GISS-E2-R 2.7 2.24
regionsandcentralAntarcticaforSATs(Fig.1c)andinthe
HadCM3 3.1 3.27
NorthAtlantic,NorthPacificandseaiceareasoftheArctic
IPSLCM5A 3.4 2.03
andSouthernoceansforSSTs(Fig.1f).
MIROC4m 4.1 3.46
ThewarmingofthePlioMIPsimulationsisaccompanied
MRI-CGCM2.3 3.2 1.84
NorESM-L 3.1 3.27 byincreasedprecipitation(Haywoodetal.,2013)andmon-
soonal activity (Zhang et al., 2013) and reductions in sea
ice (Clark et al., 2013), although the Atlantic Meridional
Overturning Circulation shows little response (Zhang et al.,
3 PlioMIPExperiment2
2013b).Globalmeantemperatureresponse(Table2),aswell
PlioMIPusesthelatestiterationofthePRISM(PlioceneRe- as polar amplification (Salzmann et al., 2013), do not show
search, Interpretation and Synoptic Mapping) mid-Pliocene a strong correlation to either the use of preferred or alter-
palaeoenvironmental reconstruction, PRISM3 (Dowsett et nate boundary conditions or to the initial conditions of the
al., 2010), as the basis for the imposed model boundary ocean. Although land-sea masks vary between the different
conditions.Thisreconstructionrepresentsthepeakaveraged models in the Hudson Bay and West Antarctic region, they
(DowsettandPoore,1991)warmclimateofthemid-Pliocene donotshowthelargestrelativevariance,suggestingthatthe
warmperiod(mPWP;3.246–3.025Ma;Dowsettetal.,2010) alternativesusedinPlioMIPExperiment2donotintroduce
in the middle of the Piacenzian Stage. It incorporates sea significantbiases.
surface temperatures, bottom water temperatures (Dowsett
et al., 2009), vegetation (Salzmann et al., 2008), ice sheets
5 Energybalanceapproach
(Hilletal.,2007,2010),orography(Sohletal.,2009)anda
globalland-seamaskequivalentto25mofsealevelrise.The Energybalanceanalyseshavebeenusedinmanypalaeocli-
vegetation,icesheetsandorographicreconstructionsareall matesimulationsandensemblestounderstandthesimulated
requiredasboundaryconditionswithinthemodels,although temperaturechanges(e.g.Donnadieuetal.,2006;Murakami
they must be translated onto the resolution of each individ- etal.,2008).TheresultsfromeachoftheGCMscanbebro-
ualmodel.VegetationwasreconstructedusingtheBIOME4 kendownintothevariouscomponentsintheenergybalance
classification scheme (Kaplan, 2001) and must therefore be ofeachindividualsimulation.Theapproachtakenbuildson
translatedontothevegetationschemeusedbyeachmodel. the energy balance modelling of Heinemann et al. (2009)
Although as part of PlioMIP a standard experimental de- andLuntetal.(2012),wheregloballyaveragedtemperatures
signwasimplemented,itwasappreciatedthatnotallofthe areapproximatedusingplanetaryalbedoα andtheeffective
modellinggroupswouldbeabletoperformtheidealmPWP longwaveemissivityε.
experiment. As such, alternate boundary conditions were
S
specified for those models that could not effectively change 0(1−α) = εσT4, (1)
4
the land-sea mask from the present-day configuration. This
meant that the ocean advance specified in low-lying coastal where S is the total solar irradiance (1367Wm−2) and σ
0
regionsandWestAntarcticaaswellasthefillingofHudson istheStefan–Boltzmannconstant(5.67×10−8Wm−2K−4).
Baywerenotincludedinsomeofthesimulations(Table1). Planetaryalbedoistheratioofoutgoing(↑)toincoming(↓)
Furthermore a choice was given concerning the initial state shortwaveradiationatthetopoftheatmosphere(TOA)and
of the ocean between a specification of the PRISM3 three- effective longwave emissivity the ratio of TOA to surface
dimensional ocean temperatures (Dowsett et al., 2009) and (SURF)upwardlongwaveradiation,
initialization with the same ocean temperatures as the pre-
↑ ↑
SW LW
industrialcontrolsimulation(Haywoodetal.,2011). α = TOA, ε = TOA . (2)
↓ ↑
SW LW
TOA SURF
www.clim-past.net/10/79/2014/ Clim.Past,10,79–90,2014
82 D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations
a b c
d e f
°C
Fig.1.Multi-modelmeanPlioMIPExperiment2warmingbetweenmid-Plioceneandpre-industrialsimulations.(a)Annualmeansurface
air temperature (SAT) warming, (b) zonal mean SAT warming (solid line), with shading showing the range of model simulations, and
(c)relativevariancebetweenthePlioMIPExperiment2simulations(σ/(cid:5)SAT).(d)Annualmeanseasurfacetemperature(SST)warming,
(e)zonalmeanSSTwarmingand(f)relativevarianceofSSTs.
This can be expanded to approximate the one dimensional, further broken down into the changes in the impact of at-
zonallyaveragedtemperaturesateachlatitudeofthemodel mospheric greenhouse gases ((cid:5)Tggε), clouds (via impacts
gridbyincludingacomponentfortheimpliednetmeridional on both emissivity, (cid:5)Tcε, and albedo, (cid:5)Tcα; see Sect. 6
heattransportdivergence(H). fordiscussion)andclearskyalbedo((cid:5)Tcsα;generallydom-
inated by changes in surface albedo, but including atmo-
SW↓ (1−α)+H = εσT4, (3) spheric absorption and scattering components). In experi-
TOA
ments and latitudes where changes in topography occur be-
where
tween the Pliocene and pre-industrial times, the impact of
(cid:2)(cid:2) (cid:3) (cid:3)
H = − SW↓ −SW↑ −LW↑ . (4) these changes in surface altitude ((cid:5)Ttopo) must also be ac-
TOA TOA TOA countedfor.
ThusthetemperatureateachlatitudeinaGCMexperiment
isgivenby (cid:5)T ≈ (cid:5)Tggε+(cid:5)Tcε+(cid:5)Tcα+(cid:5)Tcsα+(cid:5)TH+(cid:5)Ttopo (7)
(cid:4) (cid:5)
SW↓ (1−α)−H 1/4 Each of these components can be calculated from various
T = TOA ≡ T(ε,α,H). (5) combinations of Pliocene and pre-industrial albedos, emis-
εσ
sivitiesandimpliedheattransports,althoughtodifferentiate
betweencloudandclearskycomponentssomemustbecal-
ByapplyingthenotationofLuntetal.(2012)todenotethe
culatedintheclearskycase(denotedwithasubscriptcs).
pre-industrialcontrolexperimentasasecondexperimentrep-
resentedbyanapostrophe,thePliocenesurfaceairtempera- (cid:5)Tggε = T(εcs,α,H)−T(εc(cid:6)s,α,H)−(cid:5)Ttopo (8)
turewarming((cid:5)T)canbecalculatedby (cid:5)T = (T(ε,α,H)−T(ε ,α,H))
cε (cid:6) (cid:6)cs (cid:7)(cid:7)
(cid:5)T = T(ε,α,H)−T(ε(cid:6),α(cid:6),H(cid:6)). (6) − T(ε(cid:6),α,H)−T εc(cid:6)s,α,H (9)
(cid:5)T = (T(ε,α,H)−T(ε,α ,H))
cα (cid:6) (cid:6) cs (cid:7)(cid:7)
Due to their small changes relative to their absolute values, − T(ε,α(cid:6),H)−T ε,α(cid:6) ,H (10)
Pliocene warming can be approximated by a linear combi- (cid:6) cs (cid:7)
nation of changes in emissivity ((cid:5)Tε), albedo ((cid:5)Tα) and (cid:5)Tcsα = T(ε,αcs,H)−T ε,αc(cid:6)s,H (11)
heat transport ((cid:5)T ). However, these components can be (cid:5)T = T(ε,α,H)−T(ε,α,H(cid:6)) (12)
H H
Clim.Past,10,79–90,2014 www.clim-past.net/10/79/2014/
D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations 83
Changing the topography within a climate model can have
many effects on the energy balance of the model, including
changing circulation patterns, heat transport, surface condi-
tions, cloud formation, storm generation, etc. Most of these
features cannot be properly quantified without performing
further simulations. However, a simple lapse rate correc-
tion will remove the direct impact on surface temperatures
of changing the height of the surface. Although lapse rates
varyovertimeandspace,theimpactofchangingthetopog-
raphy in the Pliocene simulations ((cid:5)T ) can be approxi-
topo
mated by multiplying the change in topography ((cid:5)h) by a
constant atmospheric lapse rate (γ ≈5.5Kkm−1; Yang and
Smith,1985).
(cid:5)T = (cid:5)h·γ (13) Fig.2.ComparisonoftheimpactonPliocenewarmingofthecom-
topo
ponents of planetary albedo under different formulations for the
HadCM3 model. The grey lines are the formulations presented in
6 Treatmentofcloudswithintheenergybalance this Sect. 5, while the black lines represent an altered calculation
calculations ofthealbedocomponentsapproximatingtheimpactofcloudsus-
ingthedifferencebetweentheincomingradiationatthetopofthe
Theenergybalancecalculationspresentedheresplittheplan- atmosphereandthemodelleddownwardshortwaveradiationatthe
etary albedo and emissivity impacts into a component due surface.
to clouds and a clear sky one. In order to do this the clear
skyradiationfluxes,whicharetheradiationfluxesthatwould
haveoccurredhadtherenotbeenanyclouds,areusedwithin methodologyoutlinedinSect.5.Althoughusingtheoutlined
the calculations. The cloud components are then the global energybalanceapproachproducesgreatermagnitudesofim-
temperaturechangeduetotheimpactofcloudsonplanetary pactsfromclearskyandcloudalbedo,aswouldbeexpected,
albedo and emissivity. If all else in the climate system re- theoverallstructureandtheconclusionsthatwouldbedrawn
mainedthesameexceptforthesurfacealbedothentheclear fromthesecalculationsremain.
skyalbedowouldshowachangeintemperature,butsoalso
wouldthecloudalbedo.Thisisbecausetheimpactofiden-
tical clouds on the total planetary albedo have changed due 7 Energybalanceresultsforindividualsimulations
tothesurfacewhichtheyarecoveringbeingdifferent.Simi-
larlyanincreaseinhighlyreflectivecloudtypeswouldhave The energy balance calculations for each of the individual
a greater impact on temperature over the dark oceans than simulations within the PlioMIP Experiment 2 ensemble are
overahighalbedolandsurface. showninFig.3.Theoverallstructureoftheenergybalance
A more thorough methodology for calculating the cloud componentsislargelythesamebetweenallthesimulations,
and clear sky components would be to incorporate cloudi- although there are differences in the responses in each of
nessintothecalculationsandtherebyremovetherelianceof the models and their magnitudes. In the tropics the warm-
cloudimpactonthesurfaceconditions.Thisrequiresanun- ingisdominatedbythegreenhousegasemissivity,withthe
derstanding of the impact of different modelled cloud types othercomponentshavingasmallimpactonwarming.Those
on the energy balance and, as we are considering a multi- simulations showing greater tropical warming tend to also
modelensemble,theconsistentapplicationofthisacrossall have a significant warming component from cloud impact
the simulations within the ensemble. Total cloud fractions on albedo. The midlatitudes are the region where the mod-
areproducedbyeachmodel,buttheseusedifferentoverlap els show the least consistency, especially in the Northern
functions,allofwhicharesimplificationsoftheobservedde- Hemisphere.Theonlyconsistentcomponentofthewarming
pendenciesofcloudoverlaponatmosphericdynamics(Wang comes from greenhouse gas emissivity, although the cloud
and Dessler, 2006; Naud et al., 2008). Therefore, the total andclearskyalbedosalsotendtohaveawarmingimpact.In
cloudfractionsofthemodelsarenotonlyincomparable,but thehighlatitudesmuchofthewarmingcomesfromchanges
mayalsonotbethebestrepresentationofcloudimpactson in the clear sky albedo, which are only partially offset by
energybalance. changesintheoverallcloudyskyimpactonplanetaryalbedo.
The impact of using the pragmatic methodology outlined There are also slightly enhanced greenhouse gas warming
above can be approximated using the surface downward and a warming impact of cloud emissivity. In the Southern
shortwave radiation to approximate the impact of clouds. Hemisphere the clear sky albedo warming often has a dou-
◦
Figure2showsthealbedocomponentsofPliocenewarming blepeak,thefirstaround60 S,representingchangesinthe
intheHadCM3model,usingboththisapproximationandthe simulated Southern Ocean sea ice and the second over the
www.clim-past.net/10/79/2014/ Clim.Past,10,79–90,2014
84 D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations
Fig.3.EnergybalanceanalysisforeachoftheeightPlioMIPExperiment2simulations,from(a)CCSM4,(b)COSMOS,(c)GISS-E2-R,
(d)HadCM3,(e)IPSLCM5A,(f)MIROC4m,(g)MRI-CGCM2.3and(h)NorESM-L.Plotsshowthezonalmeanwarming,ateachlatitude
inthemodel,fromeachoftheenergybalancecomponents.Thesolidblacklineisthezonalmeansurfaceairtemperatureincreasefromthe
GCMsimulation,whilethedashedgreylineisthePliocenewarmingapproximatedbytheenergybalancecalculations.
Clim.Past,10,79–90,2014 www.clim-past.net/10/79/2014/
D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations 85
Antarcticcontinents,fromprescribedchangestotheAntarc-
ticicesheets.
8 PlioMIPExperiment2energybalance
In order to evaluate the simulation of warm climates of
the Pliocene in general, a simple mean of the energy bal-
ance components from each of the individual simulations
within the PlioMIP Experiment ensemble has been per-
formed.Whencombinedwiththerangeofvalueswithinthe
ensemble this allows an assessment of the general cause of
warmingwithinthePlioMIPsimulationsandtherobustness
of any conclusion that can be drawn. Figure 4 shows the
ensemble mean of the various energy balance components
Fig.4.SummaryofthePlioMIPExperiment2energybalanceanal-
alongwiththerangefromtheeightsimulations,whileFig.5
ysis.Solidlineshowsthemulti-modelmeanwarmingforeachcom-
showstheindividualenergybalancecomponentsforeachof ponent,withtheassociatedshadingrepresentingtherange.These
thePlioMIPsimulations. valueshavebeeninterpolatedontoa1◦ latitudegridforcompari-
Clear sky albedo includes contributions from surface sonpurposes.
albedo changes and atmospheric absorption and scattering.
The latter could become important, even in models with
no mechanisms for changing atmospheric transparency, as especially in the Arctic (Fig. 5b). This is consistent with
atmospheric thickness can increase due to changes in sur- the prescribed increases in CO (at 405ppm for the mid-
2
face altitude. In the PlioMIP simulations clear sky albedo Pliocene, as opposed to 280ppm in the pre-industrial sim-
showslittlecontributiontowarminginthetropicsandSouth- ulations). The amplified high-latitude response is due to in-
ern Hemisphere midlatitudes. In the Northern Hemisphere creases in the atmospheric water vapour predicted by the
midlatitudes most models show a warming due to clear models. Differences in the simulation of this water vapour
sky albedo, apart from the MRI-CGCM 2.3 simulation that increase between different models explain why the range
showsacooling(Fig.5a).Inthepolarregions,allthesimu- of temperature increases due to greenhouse gas warming is
lationsshowastrongwarmingsignalfromclearskyalbedo, muchhigherinthepolarregions.
although the range in the magnitude of this warming is The changes in the impact of clouds on planetary albedo
large.Changesinclearskyalbedomostlyreflectchangeson are small in the tropics and midlatitudes. Different models
Earth’ssurface.Vegetation,snowandice(bothterrestrialice seemtoproducesignificantlydifferentresponsesmakingthe
masses and sea ice) are generally the main contributors to signalparticularlynoisy(Fig.5c).However,themulti-model
these changes. The warming found in the Northern Hemi- mean cloud albedo impact on warming appears to reflect
◦
sphere,from15–60 Nislargelybeingdrivenbychangesin some of the large-scale features of the PlioMIP simulations
thevegetationboundaryconditions,particularlyovertheSa- (Haywoodetal.,2013).BetweentheEquatorand∼45◦there
hara, Arabia and central Asia (Fig. 6). In the Arctic, warm- isageneralwarmingduetoareductionintheimpactofcloud
ing due to clear sky albedo is primary driven by changes in albedo,interruptedbyacoolingintheNorthernHemisphere
icesheetboundaryconditions(reducedGreenlandIceSheet) tropics.Thiscoolingisduetoanincreaseincloudcoverre-
and changes in the predicted sea ice, but also by the pole- sultingfromanorthwardshiftoftheInter-TropicalConver-
ward shift of the Arctic tree line (Salzmann et al., 2008). gence Zone (Kamae et al., 2011). In the high latitudes the
In the Southern Ocean and Antarctica the warming due to increasedimpactofcloudsonplanetaryalbedo,partlydueto
clear sky albedo has a double peak in most models, reflect- changes in the underlying surface albedo, leads to a signif-
◦
ingareductioninthesimulatedSouthernOceanseaiceand icant cooling, peaking at between 3 and 6 C in both hemi-
a reduction in the prescribed Antarctic Ice Sheet (Fig. 7). spheres.Cloudemissivityshowsasimilarpatternofimpacts,
Mostmodelsshowreductionsinseaicecoverageacrossthe but in the opposite direction. However, the response is gen-
wholerangeoflatitudesinwhichseaiceismodelled,except erallyofasmallermagnitude(Fig.5d).
forthehighestNorthernHemispherelatitudesandinCCSM Reconstruction of mid-Pliocene sea surface temperatures
closetoAntarctica(Fig.7c).Thesimulatedchangesintotal hasledtoincreasedheattransportintheNorthAtlanticbeing
cloudarerelativelysmallandinconsistentbetweenthemod- suggestedasaprimarydriverofwarminginthemid-Pliocene
els(Fig.7e),sodonotappeartointroducesystematicbiases. (Dowsettetal.,1992;Raymoetal.,1996).However,theim-
Allthesimulationsshowawarmingduetogreenhousegas pliedoverallmeridionalheattransportinthePlioMIPsimu-
◦
emissivityofaround1–2 C.Theseimpactsarelargelycon- lations,whichintegratesbothoceanicandatmospherictrans-
stant across latitudes, but with a slight polar amplification, ports,showslittlecoherentsignal.Thefactthatthereisonly
www.clim-past.net/10/79/2014/ Clim.Past,10,79–90,2014
86 D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations
Fig. 5. Breakdown of the energy balance components, (a) clear sky albedo, (b) greenhouse gas emissivity, (c) cloud albedo, (d) cloud
emissivityand(e)impliedmeridionalheattransport.Solidblacklineshowsthemulti-modelmean,rangeisshownbythegreyshadingand
eachoftheindividualmodelresultsisshownbythecolouredsolidlines.
one region where all of the simulations show a temperature 9 Conclusions
change of the same direction suggests that the only robust
conclusion that can be drawn about heat transport is a re- Themid-PliocenewasprobablythelasttimeinEarth’shis-
duction of overall transport into the Arctic (Fig. 5e). This tory when atmospheric carbon dioxide concentrations were
would be an expected result of reduced thermal gradients similar to today (Kürschner et al., 1996; Seki et al., 2010;
duetopolaramplificationintheArcticregionunderclimate Paganietal.,2010;Bartolietal.,2011).Ithasbeenthefocus
warming. These energy balance calculations support analy- of palaeoenvironmental reconstructions and palaeoclimate
sisoftheAtlanticMeridionalOverturningCirculationinthe model experiments for many years. However, the recently
PlioMIPensemble,whichshowsthatthereislittlechangein begun Pliocene Model Intercomparison Project is the first
thenorthwardheattransportintheNorthAtlantic(Zhanget time that coordinated multi-model experiments, with com-
al., 2013b). This calls into question the role of ocean heat mon boundary conditions and experimental protocols, have
transportinthegeneralwarmingofthemid-Pliocene.How- beenundertaken.ThewarmingseeninthePliocenehasbeen
ever, it may be important in the Pliocene variability of sea well documented from a wide variety of sites from across
surfacetemperatures,whichisparticularlyhighintheNorth the globe and using a number of different proxy techniques
Atlantic(Dowsettetal.,2012). (Dowsettetal.,2012;Salzmannetal.,2013).Previoussim-
ulationsofPliocenewarmthhavebeenperformedwithonly
asinglemodelandmulti-modelanalyseshavebeenseverely
Clim.Past,10,79–90,2014 www.clim-past.net/10/79/2014/
D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations 87
Fig.6.Spatialdistributionofmulti-modelmeanchangesintheclearskyalbedobetweenmid-Plioceneandpre-industrialsimulations.This
primarily shows changes due to specified vegetation and global ice sheets and modelled sea ice and snow cover, although it includes an
atmosphericcomponent.Greyscaleshowsreductionsinalbedo,generallyassociatedwithwarmingandyellowshadingindicatesincreases
inalbedothatwouldgenerallycauseareductioninsurfaceairtemperature.Increasesinalbedocloselyfollowthenorthwardexpansionof
grasslandimposedonthesimulations(Haywoodetal.,2010).
hamperedbydifferingexperimentaldesigns(Haywoodetal., insight into the levels of CO required to produce Pliocene
2
2009). For the first time a robust analysis of the causes of climates(forexampleinasimilarwaytoLuntetal.,2012,for
warminginPlioceneclimatemodelsispossible. the Eocene). Alternatively, accurate reconstructions of sur-
Energybalancecalculationsshowthatthetropicalwarm- facetemperaturesandatmosphericCO incombinationwith
2
ing seen in all the models is primarily caused by green- modellingstudiescouldrevealtheextentofchangestotrop-
housegasemissivity,withspecifiedincreasesinatmospheric icalcloudcoverinthewarmerPlioceneworld.
CO concentration being the most important factor. Along Particularlystrongwarminginthehighlatitudesisdriven
2
withdifferentsensitivitytotheimposedCO concentrations, by changes in albedo, especially from sea ice, ice sheets,
2
changes in warming due to the cloud impact on planetary snowcoverandvegetation,whichareonlypartiallyoffsetby
albedo drive differences between the models in the tropics. extra cooling components from increased cloud albedo im-
Atpolarlatitudesalltheenergybalancecomponentsbecome pacts.Thisistheregionwiththelargestwarmingsignaland
important, but clear sky albedo is the dominant driver of alsothelargestuncertaintiesbetweenthesimulations.There-
the high levels of warming and polar amplification. This is fore,improvementsinthereconstructionofglobalicecover
largelyduetoreductionsinthespecifiedicesheetsandsim- andArcticvegetation,alongwithimproveddatatoevaluate
ulated sea ice, but in the Northern Hemisphere also reflects the simulation of sea ice and high Arctic atmospheric and
a northward shift in the treeline. The models show a very ocean temperatures, could significantly improve the simu-
differentresponseinthemidlatitudesoftheNorthernHemi- lations and allow much better constraints on total Pliocene
sphere, with large uncertainties in the relative contributions warming.FromthePlioMIPExperiment2simulationsitap-
of the different energy balance components. This is partic- pearsthathigherCO concentrationswarmedtheplanetdur-
2
ularly true for the North Atlantic and Kuroshio Current re- ingthePlioceneanddrovelargesurfacealbedofeedbacksin
gions, where intermodel variability is highest and warming thehighlatitudesthroughchangesinseaice,vegetationand
issimulatedverydifferently(Haywoodetal.,2013).Amore icesheets.Thelattertwoofthesefactorsareimportantcom-
completepictureofthesecurrents,theirstrengthandvariabil- ponents of long-term Earth system sensitivity, further sup-
itywithinthePliocene,wouldenableamuchbetteranalysis porting a long-term response of CO greater than conven-
2
oftheskillofthemodelsinthesekeyregions. tionalclimatesensitivity(Luntetal.,2010;Haywoodetal.,
AtmosphericCO concentrationsremainasignificantun- 2013).
2
certaintyinPlioceneclimate,withdifferentproxytechniques
producing values between pre-industrial levels and double
pre-industriallevels(Kürschneretal.,1996;Sekietal.,2010;
Paganietal.,2010;Bartolietal.,2011).Astropicalwarming
islargelydrivenbythisfactor,simulationswithaccuraterep-
resentation of low latitude clouds could provide some new
www.clim-past.net/10/79/2014/ Clim.Past,10,79–90,2014
88 D.J.Hilletal.:EvaluatingthedominantcomponentsofwarminginPlioceneclimatesimulations
(a) Warming impact of clear sky albedo
(b) Surface albedo change (d) Imposed snow free albedo change
(c) Sea ice change (e) Change in modelled total cloud
Fig.7.ZonalmeanofaseriesoffactorsthatcontributetotheclearskyalbedoanditsimpactonPliocenewarming.(a)ThetotalPliocene
warmingandtheclearskyalbedocomponentofthis(blackandgreenrespectively;seeFig.4).(b)Totalzonalsurfacealbedochangeineach
ofthe8models.(c)Zonalmeanseaicepercentcoverageinthemodels,solidlineisthePliocenetotalandthedashedlineisthechange
betweenPlioceneandpre-industrialexperiments.(d)Imposedsnowfreealbedochange,ascalculatedfromtheobservedandreconstructed
Pliocenemega-biomes,whichincludevegetationandicesheetchanges(Salzmannetal.,2008;Haywoodetal.,2010).(e)Changeinmodelled
totalpercentcloudcoverforeachmodel.Thesearecalculatedbyeachofthemodelsusingtheirownalgorithmandcloudparameterizations,
soareindicativeratherthandirectlycomparable.
Acknowledgements. D.J.HillacknowledgestheLeverhulmeTrust simulations were carried out using the computational facilities of
for the award of an Early Career Fellowship and the National theAdvancedComputingResearchCentre,UniversityofBristol–
Centre for Atmospheric Science and the British Geological http://www.bris.ac.uk/acrc/.G.Lohmannreceivedfundingthrough
Survey for financial support. A. M. Haywood and S. J. Hunter the Helmholtz research programme PACES and the Helmholtz
acknowledgethattheresearchleadingtotheseresultshasreceived Climate Initiative REKLIM. C. Stepanek acknowledges financial
fundingfromtheEuropeanResearchCouncilundertheEuropean support from the Helmholtz Graduate School for Polar and
Union’s Seventh Framework Programme (FP7/2007-2013)/ERC Marine Research and from REKLIM. Funding for L. Sohl and
grant agreement no. 278636. A. M. Haywood acknowledges M. A. Chandler provided by NSF (National Science Foundation)
fundingreceivedfromtheNaturalEnvironmentResearchCouncil GrantATM0323516andNASAGrantNNX10AU63A.B.L.Otto-
(NERC Grant NE/I016287/1, and NE/G009112/1 along with BliesnerandN.A.RosenbloomrecognizethatNCARissponsored
D. J. Lunt). D. J. Lunt and F. J. Bragg acknowledge NERC grant by the US NSF and computing resources were provided by the
NE/H006273/1. D. J. Lunt acknowledges Research Councils Climate Simulation Laboratory at NCAR’s Computational and
UK for the award of an RCUK fellowship and the Leverhulme Information Systems Laboratory (CISL), sponsored by the NSF
Trust for the award of a Phillip Leverhulme Prize. The HadCM3 and other agencies. W.-L. Chan and A. Abe-Ouchi would like to
Clim.Past,10,79–90,2014 www.clim-past.net/10/79/2014/