modelingice sheets and sea level with the community earth ... · modelingice sheets and sea level...

Post on 15-Oct-2019

4 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Modeling Ice Sheets and Sea Level with the Community Earth System Model

William LipscombCESM Tutorial17 August 2017

Outline• Introduction

• IcesheetsinCESM1

• IcesheetsintheCESM2release

• IcesheetsbeyondCESM2

Sea-level rise over the past 25 years

Globalmeansea-levelrisefromsatellitealtimetrySLRrate=3.3± 0.4mm/yr,1993–2015

Nerem 2016

Growing ice sheet contribution to sea level

~1mm/yr fromGreenlandandAntarcticicesheets(~65%Greenland)~1mm/yr fromglaciersandicecaps~1mm/yr fromoceanthermalexpansion

Credit:ESA/NASA/PlanetaryVisions

AntarcticIceSheet• 60msea-levelequivalent

– 5minmarine-basedpartsofWestAntarctica

– 20minmarine-basedpartsofE.Antarctica

• Accumulationbalancedbyflowintofloatingiceshelves;littlesurfacemelting

• MasslossinrecentyearsfromWestAntarcticaandtheAntarcticPeninsula,triggeredbywarmoceanwaterreachingthebaseoficeshelves

Antarctic ice flow speed (Rignot et al. 2011)

GreenlandIceSheet

• 7msea-levelequivalent• Accumulationbalancedby

surfacerunoffandicebergcalving

• Growingmasslossinrecentyearsfromincreasedsurfacemeltingandrunoff,andfromthinningandaccelerationofoutletglaciers

Greenland surface ice speed (Greenland Ice Mapping Project)

IcesheetsinIPCCAR5• “UnderallRCPscenariostherateofsealevelrisewillvery

likelyexceedthatobservedduring1971–2010duetoincreasedoceanwarmingandincreasedlossofmassfromglaciersandicesheets.”

• Likelyrangeof21st centuryglobalmeansealevelrise:• 0.32to0.63m(RCP4.5,2081-2100)• 0.45to0.82m(RCP8.5,2081-2100)

• “Onlythecollapseofmarine-basedsectorsoftheAntarcticicesheet,ifinitiated,couldcauseglobalmeansealeveltorisesubstantiallyabovethelikelyrangeduringthe21stcentury….”

Marineicesheetinstability(MISI)

• IceinpartsofAntarctica,especiallytheAmundsen Searegion,isvulnerabletointrusionsofwarmCircumpolarDeepWater(possiblydrivenbychangesinwindforcing).

• Unbuttressed marineiceonareverse-slopingbedisunstable.• MISImayalreadybeactiveforPineIslandandThwaites glaciers.

Schematic of warm CDW reaching the grounding line (Jenkins et al. 2016)

Antarcticicesheetsensitivity

• LastInterglacial(125Kyearsago,CO2 =280ppm)– Globalmeansealevel6–9mhigherthantoday– Only~2mfromGreenland,0.4mfromoceanthermalexpansion,

henceanAntarcticcontributionof~5m• Pliocene (3Myearsago,CO2 =400ppm)

– Globalmeansealevel10–30mhigherthantoday– Maxof~7mfromGreenland,~5mfromW.Antarctica,so

probablyanE.Antarcticcontribution• EvenwithMISIincluded,itisdifficulttosimulatethismuch

retreatwithcurrentmodels.

– Arecriticalmechanismsmissing?

MechanismsforlargeAntarcticSLR?DeConto &Pollard(2016)suggestednewmechanisms:• Marineicecliffinstability(MICI)

– Icecliffsattheedgeofoutletglacierscanbe~1kmthick– Whenmorethan~90moficesitsoutsidethewater,atypicalicecliffwillcollapse(stress>1Mpa)

– Cliffretreatonareverse-slopingbedisdynamicallyunstable(similartoMISI).

• Hydrofracture– Increasedmeltingoniceshelvescouldgrowcrevassesandincreasecalving

– Loseofbuttressingshelvescouldtriggercliffinstability(e.g.,Jakobshavn inGreenland)

SimulatedAntarcticretreatDeConto &Pollard(2016)

• RCP8.5:77cmby2100,12mby2500– WAIScollapseby2250– MajorretreatinWilkes

andAuroraBasins• Atmosphericwarmingis

themaindriver,butoceanthermalmemoryinhibitsrecovery.

• Mechanismsareplausible,butratesarehighlyuncertain.o Coarse-resolutionicesheetmodel(10km)with

parameterizedfluxesatthegroundinglineo Prescribedoceanmeltrateswithnocoupling

DeConto &Pollard2016

Questions• Whatrangeofsea-levelriseshouldplannersand

policymakersassume?o Canwebeconfidentthat21st centurysea-levelrisewillnotexceed1m?

• CanCESMandotherEarthsystemmodelsprovideactionablesea-levelscience?

o SurfacemassbalanceprojectionsfromEarthsystemmodelsareincreasinglycredible.

o Butitremainsdifficulttoputanupperboundonthesea-levelcontributionfromAntarctica.

IcesheetsinCESM1• Formanyyears,globalclimatemodelsdidnotincludedynamic

icesheets.Icesheetsweretreatedasbigbrightrocks.o Thetraditionalviewwasthaticesheetsevolvedonmulti-

centuryandlongertimescales.Thisviewchangedwithobservationsoficesheetmasslossinthe1990sand2000s.

• In2009,theCESMLandIceWorkingGroupformedwiththegoalsof1) integratingawellvalidated,fullydynamicalicesheetmodel

inCESM2) determiningthelikelyrangeofdecade-to-century-scale

sea-levelriseassociatedwiththelossoflandice• CESMv1.0wasreleasedin2010withapreliminary

implementationofdynamicicesheets.

IcesheetsinCESM1CESM1includedtheGlimmerCommunityIceSheetModel(Glimmer-CISM).

• DynamicGreenlandicesheetonastructured5kmgrid

• Serialcode;Glide shallow-icedynamics(validforslowinteriorflows,butnotfastflowinicestreamsandiceshelves)

CESM1alsoaddedasurface-mass-balanceschemeforlandice.• Thesurfacemassbalance(SMB)forglaciatedregionsiscomputedbytheCommunityLandModelinmultipleelevationclasses,thensenttothecoupleranddownscaledtothelocalicesheetgrid.

• AdvantagestocomputingSMBinthelandmodel:o Coupleicealbedotoatmosphereonhourlytimescaleso Avoidduplicationofsnowphysicso Computationalsavings(landgridcoarserthanicesheetgrid)

IcesheetsinCESM1Land -> Ice sheet (10 classes)n Surface mass balancen Surface elevationn Surface temperature

Coupler

Atmosphere

Ocean

Sea Ice

Land surface(Ice sheet surface

mass balance)

Ice sheet(Dynamics)

Ice sheet -> Land

n No fields passed; placeholders only

GreenlandSMB:ComparisonwithRACMO2CESM RACMO2

SMB(Gt/yr) 359± 120 376 ± 117

• Goodmatchinablationzones• Accumulationisoverestimatedintheinteriorandunderestimatedin

thesoutheast(smootherorographyinCESM)

Courtesy of M. Vizcaíno!"#$%&'&()& *+!$,-%&.//&()&

Greenland ice sheet SMB (kg m-2 yr-1), 1960-2005

Red = net accumulationPurple = net melting

GreenlandSMB:20th v.21st century20th-century (1980-1999) RCP8.5 (2080-2099)

SMB (Gt/yr) 372± 100 -78± 143

Greenland ice sheet SMB (kg m-2 yr-1)

• ForRCP8.5,precipitationincreases,butmeltandrunoffincreasemore.• Warmingisgreatestinnorth(lessseaice),leastinsoutheast(weakerMOC).• AverageSMBisnegativeby2100,implyinglong-termdecayoficesheet.

Red = net accumulationPurple = net melting

Courtesy of M. Vizcaíno !"#$%""& '$#$%""&

Glimmer-CISMinCESM1• CapturesbroadpatternsofGreenlandiceflow,butoutlet

glacierstendtobeslowanddiffuse• Icesheettooextensiveinnorthandeast(positiveSMBbias)

(a) Greenland balance velocities based on observed thickness and the SMB of Ettema et al. (2009)

(b) Vertically averaged Greenland velocities simulated by Glimmer-CISM in CESM (top-ranking ensemble member, Lipscomb et al. 2013).

LandicegoalsforCESM2

• Parallelicesheetmodelwith“higher-order”flow;validinallpartsoftheicesheet

• Land-iceSMBcalculationssupportedbydefault;notjustforcustomLIWGsimulations

• Improvedtreatmentofsnowphysics(deepfirnlayers,insteadofsnowdepthcappedat1m)

• Supportfortwo-waycoupling:Changesinicesheetelevationandextentcanfeedbackontheclimate

HierarchyofStokesapproximations

• Previousgenerationoficesheetmodelsmostlyusedshallow-iceorshallow-shelfapproximations

• Newermodels(BISICLES,Elmer-Ice,ISSM,PISM,PSU,etc.)haveoneormorehigher-ordervelocitysolvers

• PrimarygoalforCISM2wasarobust,parallelhigher-ordersolver

Stokes3Dsolveforu,v,w,p

Higher-order(Blatter-Pattyn)3Dsolveforu,v

Depth-integratedhigher-order2Dsolveforu,v

Shallowiceapproximation

(Verticalshearstresses)

Shallowshelfapproximation(Membranestresses)

CISM2CISM2 wasreleasedinOct.2014:

• Codeathttp://oceans11.lanl.gov/cism/;git repoat https://github.com/cism

• Paralleldynamicalcore(Glissade)withsuiteofhigher-ordervelocitysolvers

• Incrementalremappingformassandtemperaturetransport

• Newtestcases(shallow-ice,higher-order)withPythontools

• Improvedcouplinginterface• PortedtoCESMbyBillSacks

SimulatedCISM2velocities.Top:GreenlandicesheetBottom:RossIceShelf

CISM2: ISMIP-HOM tests• Compared higher-order model results to community benchmarks

(Pattyn et al. 2008) for problems with small-scale variations in topography and basal traction

• Glissade’s higher-order solvers agree well with benchmarks

ISMIP-HOM Test A: Sinusoidal pattern in basal topography at 6 grid scales (Glissade output shown by black lines)

CISM2:Greenlandthickness

Observedicethickness(m)

CISMicethickness(m)after50ka spin-up

Modelminusobservedthickness(m)

• CISM2isrobustandefficientforlongGreenlandspin-ups(~1200yr/wallclockhouron4kmgrid).

• Aftera50kyr spin-upwithSMBforcingfromRACMO2,modelthicknessisclosetoobservations(abitthininnorthandwest,thickinsoutheast).

CISM2: Greenland velocities• Usingadepth-integratedHOsolver(DIVA)andpseudo-plasticsliding

law,CISMvelocitiesareingoodagreementwithobservations.

Observedsurfacespeed(m/yr,logscale)

ModeledsurfacespeedinCISM(m/yr,logscale)

CISM2: Greenland basal state• CISM’sdistributionoffrozenandthawedregionsissimilartoestimates

basedonobservationsandothermodels.

SynthesisofGreenland’sbasalthermalstatefromMacGregoretal.(2016)

Basalwaterdepth(m)inCISM;purple =frozen(nobasalwater),red =thawed(waterpresent).

CESM1.0 CESM2.0

Serial,shallowiceapproximation Parallel,higher-order approximation

One-waycoupling(CLMà CISM) Two-waycoupling(CLMßà CISM)with dynamic landunits

DownscalinginCISM/Glint,withSMBnotconserved

Downscaling inthecoupler,withSMBconserved

1-msnowpackinCLM Firn modelinCLM(10-msnowpack,improvedsnowdensity)

SMB computedonlyinrunsdonebyLIWG SMB computedinallruns

LandiceprogressinCESM

IcesheetsinCESM1Land -> Ice sheet (10 classes)n Surface mass balancen Surface elevationn Surface temperature

Coupler

Atmosphere

Ocean

Sea Ice

Land surface(Ice sheet surface

mass balance)

Ice sheet(Dynamics)

Ice sheet -> Land

n No fields passed; placeholders only

IcesheetsinCESM2Land -> Ice sheet (10 classes + bare land )n Surface mass balancen Surface elevationn Surface temperature

Coupler

Atmosphere

Ocean

Sea Ice

Land surface(Ice sheet surface

mass balance)

Ice sheet(Dynamics)

Ice sheet -> Landn Ice extentn Ice surface elevationn SMB mask

Ice sheet -> Oceann Solid and liquid fluxes

Ice sheet -> Atmosphere (offline)n Surface topography

Greenland surface mass balance in CESM2Overall Greenland climate is very good.

• SW and LW biases reduced compared to CESM1

• Recent tuning has improved winds and precipitation

• Model captures the western ablation zone

• Too little ablation in north; cold climate bias

GreenlandSMBinCESM2onCLMgrid(left)anddownscaledtoCISMgrid(right),comparedtoRACMO2(center).Red =accumulation,blue =ablation

Antarctic surface mass balance in CESM2

• Excellent Antarctic climate and SMB

• Improvements in snow physics have reduced spurious melting

• Good baseline for warmer climates with increased ablation AntarcticsurfacemassbalanceinrecentCESM2

tuningruncomparedtoRACMO2.Red =accumulation, blue =ablation

LandicegoalsforCESM2+• Supportmultipleicesheets(notjustGreenland)withappropriate

physicsandnumericso Antarcticao Paleo icesheets(e.g.,Laurentide)

• Worktowarddynamiccouplingofmarineicesheets

o Long-term:Interactivecouplingwithoceancirculationinsub-shelfcavitiesandadynamicice/oceaninterface

o Nearterm:UsesimplemodelsandparameterizationstolinkoceanGCMoutput(atvariousresolutions,withorwithoutcavities)tosub-shelfmelting

• ParticipateintheIceSheetModelIntercomparisonProjectforCMIP6

o ExperimentsforstandaloneISMsandcoupledAOGCM/ISMs

CurrentstateofCISM• User-friendly• Portable• Maintainableandextensible• Robust• Verified• Versatile

• SuiteofStokesparameterizations(BP,DIVA,SSA,SIA)• Butsomephysicsparameterizationsneedmoreresearch

• Efficient• NativeFortransolverisfastformoderateresolution

Greenland(~4km),adequateforfineresolution(~1–2km)• Solverimprovementswouldfacilitatewhole-icesheet

simulationsforAntarctica

NewCISMphysicsandnumerics• Newphysicsformarineicesheets

o Grounding-lineparameterization(complete)o Damage-basedcalving(inprogress)o Sub-shelfplumemodel(inprogress)o Marineicecliffinstabilityandhydrofracture

• Otherphysicsimprovementso Evolutionarybasalhydrologyo Fullyparallelisostasymodel

• Improvedefficiency(for1-2kmresolution)o Fastermatrixsolver(betterpreconditioningforiceshelves)o Betterloadbalancing(removeice-freeoceancells)

CISMAntarcticicesheetsimulations• CISM’s higher-order velocity solvers can handle Antarctica

out of the box at resolutions down to 1 km.• Challenge is to simulate realistic long-term evolution.

ObservedAntarcticsurfacevelocities(m/yr,logscale)

ModeledAntarcticsurfacevelocities(m/yr,logscale)

MarineIceSheetModelIntercomparisonProject

• MISMIP3d(Pattyn etal.2013)• Perturbedbasalslidingparametersgivelateralvariationandbuttressing,withcurvedgroundinglines

• CISMusesagroundinglineparameterization(GLP)toresolvesubgrid variationsinbasaltractions

• Thisallowsustomodelgroundinglinesaccuratelyatpracticalresolutions(~1–2km)

• MISMIPconsistsofidealizedexperimentsthattestamodel’sabilitytotrackgrounding-lineadvanceandretreat.DoestheGLreturntoitsstablestartingposition?

MISMIP3d: Applied basal perturbation

SSA withGLP(1km):GLreturnstostartposition(598km),closetoanalyticsolution(612km)

Black =startingposition; red =advance;lt.blue=return

592 594 596 598 600 602 604 606 608 6100

10

20

30

40

50

60

70

80

90

100

Cha

nnel

wid

th (k

m)

Channel length (km)

GL position throughout experiments

StndP75SP75R

500 505 510 515 520 525 5300

10

20

30

40

50

60

70

80

90

100

Chan

nel w

idth

(km

)

Channel length (km)

GL position throughout experiments

StndP75SP75R

SSA withoutGLP(1km):GLtoofarretreatedatstart(504km)

andfailstoreturn

Startfromsteadystate;runfor100yearswithbasalperturbation;turnoffperturbationandreturntosteadystate.

MISMIP3d: Applied basal perturbation

SSA withGLP:GLreturnstostartposition(598km),closetoanalyticsolution(612km)

Black =startingposition; red =advance;lightblue=return

592 594 596 598 600 602 604 606 608 6100

10

20

30

40

50

60

70

80

90

100

Cha

nnel

wid

th (k

m)

Channel length (km)

GL position throughout experiments

StndP75SP75R

DIVAwithGLP:GLreturnstostartposition(558km);iceissofterwithverticalshearstresses

554 556 558 560 562 564 566 568 570 572 5740

10

20

30

40

50

60

70

80

90

100

Cha

nnel

wid

th (k

m)

Channel length (km)

GL position throughout experiments

StndP75SP75R

• SSAgroundinglineisclosetoanalytic1Dsolution.• DIVAandBPresultsareclosetobenchmarkStokessolution.

MISMIP summary

• Veryhighresolution(~200m)doesnotseemnecessaryforaccurategrounding-lineresolution.WithaGLP,aresolutionof~1kmisadequate,evenwithasharptransitionzone.

• Depth-integratedviscosityapproximation(DIVA)appearstobeanaccurateandefficientcompromisebetweenBlatter-PattynandSSA.

• Thisisgoodnewsformodels!At1kmwithadepth-integratedsolver,itisfeasibletomodelentireicesheets,evenonCISM’sstructuredgrid.

Calving• CalvingaccountsforabouthalfofmasslossfromAntarctica

andGreenland.• CISMhasseveralsimplecalvingschemesbasedonice

geometryorlocation:o Calveallfloatingice.o Calveicewherethebedisdeeperthanacriticaldepth.o Calvefloatingicewhenthinnerthanacriticalthickness.

• Theseschemesdonotaccountforicemechanics.Icebergscalvewhentheiceissufficientlydamaged.

Schematicoffloatingicetonguewithcrevassesleadingtocalving.CourtesyofM.Whitcomb.

Damage-basedcalving• DOEcollaborationtodevelopadamage-basedcalvingmodel:U.

Michigan(J.Bassis,M.Whitcomb),LANL• Iceisdescribedbyadamagetracerrepresentingcrevassedepth.

CalvingoccurswhenD=1.• Tensilestressandbasalmeltingopencrevasses;gravitationalforces

closecrevasses.• Modelappliedtoobservedicestreamsandicetongues:Erebus,

Drygalski,PineIsland,Petermann

ModeleddamageofErebusIceTongue,comparedtoobservedthicknessprofile.CourtesyofM.Whitcomb.

Sub-shelfmelting

• Sub-shelfmeltratesaresometimesparameterizedasafunctionofdepth,failingtocapturethespatialstructureofmelting.

• ItisexpensivetorunoceanGCMsbeneathiceshelves(requiredgridresolution~2km),andnotalloceanmodelshavethiscapability.o POPgridhasaverticalwallattheshelfedge.o MOM6hasbeenrunwithoceancavities,butnotyet

operationalforglobalsimulations.

• Canweestimatesub-shelfmeltrateswithamodelofintermediatecomplexity?

Plumemodel• Holland,Jenkins&Holland(2008)modeledoceanflowinthecavity

beneathastaticiceshelf.TheysuggestedthatoceanGCMresultscanbeexplainedintermsofasteady-stateplumemodel.

• Theplumeisawell-mixed,buoyantlayerattheiceshelfbase,characterizedbythicknessD,temperature T,salinityS,andvelocityu=(u,v).

• Wearegiventhecavitygeometry(shelfbase,bedtopography) andtheambienttemperatureandsalinityTa andSa.

• TheplumemodelisbeingappliedtotheISOMIP+experimentsforsub-shelfoceanmodels(Asay-Davisetal.2016).

• T0 =-1.9oC,Tbot =1.0oC

• S0 =33.8psu,Sbot =34.7psu

Plumethicknessandvelocity

• Dominantflowisupslope(eastward),limitedbydraganddivertednorthwardbytheCoriolis force.

• Theplumethicknessis5to20minmostofthedomain.• Waterpilesuponthenorthcavityedge,drivinganeastwardjet.

Basalmeltrate

Plumemodel

• PreliminaryISOMIP+resultsfromPOP2X andMPAS-Ocean,courtesyofX.Asay-Davis

• LargestmeltratestendtobeinNWandSWcornersofthedeepcavity

• PlumemodelandMPAS-OceanarelessnoisythanPOP2x(noiseassociatedwithz-gridstair-stepping)

Plumesummary

• PlumevelocitiesandmeltratesarecomparabletotheresultsofoceanGCMs.o Thereissignificantvariabilityamongoceanmodels.ForISOMIP+

theplumemodelisdifferent,butnotclearlybetterorworse.

• AlthoughlessphysicallyrealisticthanoceanGCMs,theplumemodelissimpler,cheapertorun,andeasiertocouple.o Couldbeforcedbyanoceanmodelwithoutcavities;justrequires

verticalprofilesofTa andSa neartheshelfedge

• Nextsteps:o ApplytoMISOMIP1testcase(coupledoceanandicesheet).

o TestinmorerealisticGreenlandandAntarcticageometries.

IceSheetModelIntercomparisonProjectforCMIP6(ISMIP6)

• ISMIP6isanewtargetedactivityoftheClimateandCryosphere (CliC)projectoftheWorldClimateResearchProgram.

• Primarygoal:ToestimatepastandfuturesealevelcontributionsfromtheGreenlandandAntarcticicesheets,alongwithassociateduncertainty

• Secondarygoal:Toinvestigatefeedbacksduetodynamiccouplingbetweenicesheetandclimatemodels,andimpactsoficesheetsontheEarthsystem

• TheLIWGplanstorunISMIP6standaloneexperimentswithCISM(forbothicesheets)andcoupledexperimentswithCESM(forGreenland).

ExperimentaldesignforISMIP6

1. ExistingCMIP6experimentstobeanalyzed intermsoficesheetforcing

2.StandaloneicesheetexperimentsbasedonCMIP6modeloutputtoestimatepastandfuturesealevelrise,andexploreuncertaintyduetoicesheets

3.CoupledAOGCM-ISMexperimentstoexploreimpactsandfeedbacksduetoicesheets

CMIP6exptobeusedbyISMIP6(allAOGCM)

- Pre-industrial control- AMIP- 1%peryrCO2 to4xCO2- Abrupt4xCO2- CMIP6Historical Simulation- ScenarioMIP RCP8.5/SSP5x(uptoyear2300)- LastInterglacial PMIP

Standalone ISMIP6exp (ISMonly)

- ISMcontrol- ISMfor lastfewdecades(AMIP)- ISMforthehistoricalperiod- ISMforcedby1%peryrCO2 to4xCO2- ISMfor21st /23rd century(RCP8.5/SSP5x)- ISMforLastInterglacial- ISMspecificexperimentstoexploreuncertainty

Newproposed ISMIP6exp (coupledAOGCM-ISM)

- Pre-industrial control- 1%peryrCO2 to4xCO2- ScenarioRCP8.5/SSP5x(toyear2300)

Summary• Sea-levelriseisstillawide-openscientificproblem,largely

becauseofuncertaintiesinthedynamicsofmarineicesheets.

• TheCESMcommunity—withitsinterdisciplinaryworkinggroupstructure,linkstotheacademiccommunity,andexperienceindevelopingCESM1andCESM2—iswellequippedtotacklethescience.

• CISMisaccurate,efficientandversatileenoughtosupportsimulationsoftheGreenland,Antarcticandpaleo icesheetsoverthenextseveralyears.

• CISMcanreadilyincorporatenewdynamicsandphysicstosupportemergingscience.

LandIceWorkingGroupinfo

Webpage:http://www.cesm.ucar.edu/working_groups/Land+Ice/

Liaisons:GunterLeguy (gunterl@ucar.edu) andBillSacks(sacks@ucar.edu)

Co-chairs:JanLenaerts(Jan.Lenaerts@colorado.edu) andBillLipscomb(lipscomb@ucar.edu)

Upcomingmeetings:• WinterLIWGmeeting,Boulder,10-11January2018• 23rdannualCESMworkshop,Colorado,18-21June2018

top related