1 GNET 2017 Forward: The future shape of a Greenland GNSS observation network A whitepaper produced by the participants of the NSF- supported GNET workshop, NASA GSFC, 26-27 January 2017
1
GNET2017Forward:ThefutureshapeofaGreenlandGNSS
observationnetwork
AwhitepaperproducedbytheparticipantsoftheNSF-supportedGNETworkshop,NASAGSFC,26-27January2017
2
ExecutiveCommittee:
RobertHawley,DartmouthCollege
ErikIvins,NASAJetPropulsionLaboratory
TomNeumann,NASAGoddardSpaceFlightCenter
ContributingAuthors:SurendraAdhikari,NASAJetPropulsionLaboratoryLaurenAndrews,NASAGoddardSpaceFlightCenterGregBabonis,UniversityofBuffaloMikeBevis,TheOhioStateUniversityDavidBromwich,TheOhioStateUniversityClaraChew,NASAJetPropulsionLaboratorySorenChristensen,AgencyforDataSupplyandEfficiency,DenmarkReneeCrain,NationalScienceFoundationBeaCsatho,UniversityofBuffaloReneForsberg,TechnicalUniversityofDenmarkRonniGrapenthin,NewMexicoInstituteofMiningandTechnologyRobertHawley,DartmouthCollegeErikIvins,NASAJetPropulsionLaboratoryAbbasKhan,TechnicalUniversityofDenmarkKristianKjeldsen,UniversityofCopenhagenPerKnudsen,TechnicalUniversityofDenmarkEricLarour,NASAJetPropulsionLaboratoryFinnBoMadsen,TechnicalUniversityofDenmarkGlenMattiolli,UNAVCOTomNeumann,NASAGoddardSpaceFlightCenterThomasNylen,UNAVCOKeshavDevSingh,UniversityofCalifornia,DavisTonieVanDam,UniversityofLuxembourgMikeWillis,UniversityofColorado
3
TableofContentsExecutiveSummary...................................................................................................................................4
1 Currentstateofthenetwork.........................................................................................................6
2 CurrentandfuturescienceusingGNETdata...........................................................................72.1 SolidEarthandIsostaticAdjustment................................................................................72.2 IceMassBalance......................................................................................................................92.2.1 Surfacemassbalance...........................................................................................................................92.2.1.1 Background...................................................................................................................................92.2.1.2 Theconnectionofincreasedmeltondischarge.........................................................112.2.1.3 SMBAnalysiswithGNETData...........................................................................................132.2.1.4 UseofGNETDataforFundamentalImprovementinSMBModeling...............142.2.1.5 GNETZenithTotalDelay:UseforRegionalAtmosphericModels......................152.2.1.6 ChangesinGrISdrivenbybothglacierdischargeandSMB.................................172.2.1.7 OtherpossiblesurfacemassbalancestudiesbenefitingfromGNET...............21
2.2.2 SupportforInterpretationofAltimetry...................................................................................222.2.3 SupportforGRACEandtheTotalMassBalanceTimeSeries.........................................23
2.3 Ionosphere,Troposphere...................................................................................................232.4 Largescalegeodesy...............................................................................................................252.4.1 GNSScoordinatetimeseriesandreferenceframes............................................................252.4.2 MakingITRFavailabletotheusers............................................................................................262.4.3 GNSSreflectometry...........................................................................................................................27
3 DataManagement...........................................................................................................................273.1 GeneralBackground.............................................................................................................273.2 AncillaryData.........................................................................................................................283.3 GenerationofTimeSeries..................................................................................................283.4 DataStreamandArchivingFundedbyNSF..................................................................29
4 Bestconfigurationmovingforward..........................................................................................304.1 EvaluationofGNETstations...............................................................................................304.2 Addingstationstotheexistingnetwork........................................................................324.3 Detailedsitecharacterizationofexistingsites,additionalmeasurements.......334.4 Expandedskycoverageanddataaccessforexistingsites......................................33
5 REFERENCES:....................................................................................................................................34
4
GNET2017forward:ThefutureshapeofaGreenlandGNSSobservationnetwork
AwhitepaperproducedbytheparticipantsoftheNSF-supportedGNETworkshop,NASAGSFC,26-27January2017
ExecutiveSummaryAninternationallycoordinatedresearchcampaign,theInternationalPolarYear(IPY)2007-2008,initiatedacoordinatedefforttostudythepolarregionsusingmodernobservationaltechniques,includingamajorinvestigationusinggeodeticandseismicinstrumentation.Thiseffortisformallyknownandfundedunderthename“POLENET”,orthePolarEarthObservingNetwork.InGreenland,theGNETprojectwasdevelopedtoestablishanetworkofGPSreceiversoperatingcontinuouslyandautonomouslyonstablebedrockaroundGreenland.ThemostprominentinstitutionsinvolvedinsupportingthisefforthavebeenTheOhioStateUniversity,theUniversityofLuxembourg,UNAVCO(Boulder,CO)andtheInstitutforRumsforskningogRumteknologi(DTUSpace,Copenhagen).ThepurposeoftheJanuary26-27,2017workshopwastodiscussthefutureofthisnetworkfromageodeticperspective.WewereespeciallyfocusedondocumentinghowGNETdataarebeingusednow,toprobehowthenetworkcouldevolve,andtoaskwhatscientificquestionsmotivateitsfuture.GNETdatahaveservedavarietyofusefulscientificpurposes,suchasprovidinggroundtruthforpredictivemodelsforpostglacialrise/fallofbedrockadjacenttotheicesheetthatinturn,playanessentialroleincorrectingsatellitegravityandaltimetrybasedestimatesoficemassbalanceondecadaltimescales.Adecadeofobservationsnowindicatethatverticalmotionstendtobedominatedbythemassunloadingcausedbysecularicesheetmasslosstotheocean.InadditiontoenumeratingpastsuccessesofGNET,theworkshopparticipantssoughttoexaminetheyetunexploitedvaluethatthenetworkmayprovideforachievingnewscienceobjectives,suchastroposphericandionosphericmapping,gainingnewinsightsconcerningsurfacemassbalance,icedynamicsandoceantidalmapping,amongothers.Thisworkshopreportisdesignedtocapturesomeofthesenewexplorations,butwillhardlyserveasacomprehensivesurveyofallofthedetailsorallofthenewscientificdiscoveryanewdecadeofGNEToperationcouldenable,howeverthenetworkmayevolve.Thereisanemphasisplaceduponthecomponentsoficesheetsurfacemassbalanceinthisreport.Surfacemassbalanceisessentiallythenetinputmasscomponent.Thisisdrivenprimarilybytheatmosphereandnear-surfaceicesheettemperaturestructure.Theemphasisismotivatedbythreerecentsciencebreakthroughs:(i)sinceabout2006thenegativemassbalanceoftheGreenlandicesheetisdominatedbymeltprocesses(andsinceabout1997dominatedbysurfacemassbalanceingeneral);(ii)recentfindingsdemonstratethatGNEThavesensitivitiestotheloadingcomponentsofthevariouselementsofthesurfacemassbalance;(iii)the
5
zenith-delaysmeasuredinthecarrierphaseoftheelectromagneticpulsesreceivedatthestationsarecapableofsignificantlyimprovinghindcastmodelsofprecipitation,afundamentalcomponentofthesurfacemassbalance.Theprimaryrecommendationsofthisreportare:
1. ContinuetosupportthecontinuousandautonomousoperationofthecurrentconfigurationofGNET.Thespatialdistributionandlongtimeseriesofobservationsfromthecurrentnetworkhasenabledawealthofscientificdiscovery,andextendingthesetimeseriesintothefuturewillenablenewscience.
2. Maximizetheutilityofthecurrentdatabypromotingtheexistingdatadistributionmodel,whichencourageslow-latencyaccesstodata.Theseeffortsshouldbeaugmentedwithreadilyavailabledoinumberstoenableproperdatacitation.
3. EncouragingnewusesofGNETdata,suchastroposphericzenithdelayanalyses
toimproveatmosphericmodelsinGreenlandanduseofthesedataformodelingtheionosphere.Thisisoneexampleofmanyofnewsciencethatcouldbesupportedbytheexistingdata.
4. Ifpossible,densifythecurrentnetworktobetterresolvethoseareasof
maximumgradientinGIA,and/orregionsofrapidglacierchange.
5. Givenfiniteresources,weconsiderascoringschemeforevaluatingtherelativeimportanceofexistingstations.Consideredaretheroleofexistingstationsforfosteringanimprovedunderstandingof(i)GIA,(ii)surfacemassbalanceminusdischarge(SMB-D)and(iii)resolvingthelargerdiscords(>4.5mm/yr)revealedincomparisonofGIAmodelsbasedonrelativesea-leveldataversusthosemorereliantonGNETupliftratedata.TheultimatedecisionregardingtherelativeimportanceofthesecompetingscienceobjectivesshouldbeguidedbytheacceptanceofviablescienceprojectsselectedbytheNationalScienceFoundation.
6
1 CurrentstateofthenetworkTheGreenlandGeodeticNetwork(GNET;Figure1)currentlyconsistsof58geodetic-gradeGlobalPositioningSystem(GPS)receivers,locatedonfixedbedrocksitesaroundthecoastofGreenland[Bevisetal.,2012].ThecapabilityoftheextendedGlobalNavigationSatelliteSystem(GNSS)hasnowsurpassedtheprecisepositioningperformanceoftheolderGPSduetoincorporationofmoresatellitesandadvancesinon-boardtechnology.Thenetworkhasthefullcapabilityformmlevelgeodesy.GNEThasgrownovertheyears,andisoperatedincooperationbetweenseveralgroups.Currentlythereare36stationsoperatedbytheOhioStateUniversity(OSU),1operatedbyNASA’sJetPropulsionLab(JPL),1operatedbyUNAVCO,15byDenmark(DTU),and5bytheUniversityofLuxembourg.Itisimportanttonotethatthoughthereareavarietyof‘owners’ofvariousGNETstations,thereiswidespreadcooperationbetweenthenetworkpartners,intermsofsharingoflogistics,datamanagement,andcollaboratingonbestpracticesforstationinstallation.MaintenanceissharedandcoordinatedbyOSU,UNAVCO,andDTU.Eachstationconsistsofageodetic-gradeGPSantennamountedonapermanently-anchoredmonumentonbedrock,aGPSreceiver,anextensivepowerandpower-managementsystem(solarpanels,windgenerators,batteries),andatelecommunicationssystemfortelemeteringdata-thiscanbewiredorwireless,andisfrequentlyprovidedbytheiridiumnetwork.Thesestationshavethefullcapabilitytomonitorveryprecisecrustalmovements,asforexample,recentlydescribedbyHerringetal.[2016].Eachstationinthenetworkfacesitsownspecificchallengestolongevity.Simplewearandtearonequipmentcancausefailures,butextremeweatherconditionsatsomesitescanalsoposeproblems,withhighwinds,icingandheavysnowloadsdamagingstations.Satellitecommunicationfordataretrievalcanfailformultiplereasons.Somestationsarevisitedbywildlife;polarbearsareanobviouspotentialproblem(theyarenaturallycuriousandverystrong),butequallydamagingtodatareturncanbethesharpteethandbeaksofArcticfoxesandravens.Inspiteofthesepotentialproblems,regularmaintenanceofthestationshas,overtheyears,produceda92%rateofdatareturnforthenetworkandgapsinthedatastreamshavesteadilyreducedovertheyears.Currently,inanygivenyear,manystationsmustbevisitedforroutinemaintenance,upgrades,andrepairs.Thecostofthesemaintenancevisitsislargelydrivenbythehelicoptertimerequiredtoaccess
GNET 2007-2009 (-2014)
• 2007– 23 new stations– 3 field team– Ad-hoc charter helicopters
• 2008– 11 new stations– 1 field team– Long-term fixed charter
helicopter
• 2009– 12 new stations– 1 field team– Long-term fixed charter
helicopter
• 2010-2014– 3 new stations
Figure1.Mapof58currentGNETsitesandinstallationdate.
7
mostGNETsites.Continuedmaintenance,upgrades,andadditionstothenetworkwillrequiretheinvestmentofresourcesbynationalfundingagencies.Althoughexpensive,GNETdatasupportsasubstantialbodyofscienceinGreenland,acrossmanydisciplines,asoutlinedinSection2.
2 CurrentandfuturescienceusingGNETdata
2.1 SolidEarthandIsostaticAdjustmentTheelevationofEarth’scrustiscontinuallychangingowingtothecontinuummechanicalresponseofthesolidEarthtotheredistributionofmassontheEarth’ssurface.Oneoftheprimarycausesofsuchchangesisthegrowthandmeltofthegreaticesheets,suchastheGreenlandicesheet.Thechangesincrustalelevationduetotheinterrelatedchangesingroundedicemassandoceanloadingisbroadlytermedglacialisostaticadjustment(GIA)whenthesechangeshavelongtimescales,generallyoforder100-10,000years.However,veryimportantcrustalresponsesalsooccuronelastictimescales,essentiallythesametimescalesasseismicwavepropagationandabasicobservationofferedbyGNETisthatpresent-dayicemassvariabilitytendstodominatemanyoftherecordedverticalsignals[Bevisetal.,2012].OurgeophysicalandobservationalunderstandingofGIAprocesseswerenurturedbystudyoftheemergenceofthecoastlinealongtheBalticSeainSwedenandFinlandmorethan100yearsago[e.g.,Ekman,1991;Peltier,1998].Landtodayrisesatratesof2-15mm/yearnearthecentersoftheformergreaticesheetsoftheLatePleistocene,causinglocalsea-levelstorecede,producingaclearobservationalrecordofsea-leveldrop.Thisphenomenonwascalledpost-glacialrebound(PGR),yetasitbecameunderstoodthattheprocessinvolvedtheentiresphericalearthduetothe125metersofsea-levelloadchange,andassociatedgeoidchanges,theprocessisnowtermedglacialisostaticadjustment(GIA).TheoriginaldesignofGNETwasprimarilyfocusedonthemeasurementofverticallandmotionsassociatedwithdirectpasticeloadandviscoelasticresponses. ForsolidEarthapplications,werecommendcontinuedoperationofexistingGNETsitesforaslongaspossible.Thisallowsustoseparateelasticandviscoelasticcrust-mantleresponses.Theselengthenedtimeserieswillaidinaccomplishingthisgoal,especiallyastheyoverlapwithaltimetrymissions(CryoSat-2andICESat-2(launchingin2018)),WorldViewimagingsatellites,aswellasthethreeSentinelsatellitesofESA’sCopernicusprogram.Inaddition,theNASA-ISROjointNISARmission,launchingin2020thatwillenablespace-baseddischargeobservationsofGreenland’soutletglacierstobetterresolvemassbalancetimeseries.ThisperiodoftemporallycoincidentobservationswillgreatlyimprovetheelasticcorrectionforsolvingforGIA.ItisimportanttoemphasizethatGIAisatime-invarianttrend,andmoreover,thataGIAcorrectionisusedtocorrectsatelliteobservationsthatmonitorthestateoftheicesheet.GreatdiscrepanciesremaininourcurrentpredictionsofviscoelasticGIA[cf.,Lecavalieretal.,2014;Khanetal.,2016](seeError!Referencesourcenotfound.).
8
Figure2.Relativesea-level-basedreconstruction(a;Lecavalieretal.,2014)andpredictionofpresent-dayupliftrate.GIAmodelwithGPS-basedupliftdata(b;Khanetal.,2016)andrevisedicehistoryreconstruction.Thediscrepancybetweenthetwopredictionsislargeenoughtobeanimpedimenttoproperlydiscriminatingbetweenalternativeicehistoryreconstructions.ReconciliationwillallowGIAmodelsthatproperlypredicttheGPStrendstobeusedinpaleoclimatemodels.
Inaddition,thecombinationofseismicimagingdataalongwithGPS-basedcrustalupliftdataenablesimprovedunderstandingofEarthstructureproperties.TheNSF-fundedGreenLandIceSheetmonitoringNetwork(GLISN)project(e.g.,Murrayetal.,2015)willprovideimprovedseismicdata.CombiningresultsofGLISNandGNETwillhelpbetterdefinebasalice-rockinteractions,lithosphericthickness,asthenosphericstructure,lateralheterogeneitiesintheEarth’scrustandprovideanimprovedbasisfortheheatfluxboundaryconditionusedinicesheetmodels[Rogozhinaetal.,2012].EachoftheseimagingfeaturesallowsbetterconstraintsforforwardmodelingtheGIAresponsetopastloadingfromicechanges.
Figure3:MapofcurrentstationswithGIAestimates.Currentstations(blackdots)andsuggestednewstations(yellowdots)areplacedtoconstrainGIA.
9
AllGNETstations,shownasblackpointsinFigure3,arecurrentlyprovidingdatarelevanttoconstrainingGIA.Additionalstations,shownasyellowpointsinFigure3,arelocatedinplacesthatwouldenhanceourunderstandingofGIAinareashavingnotablespatialheterogeneityinsolidearthstructure.SeismicconstraintsfromprojectslikeGLISNbetterconstrainthestructureofthesolidearthparametersusedincrustalmotionmodelsofbothlongandshorttimescales.CombinationsofseismicandGPSgeodesycanbeimportantforconstrainingvariousimprovementsinEarthstructureandconstitutiveproperties[e.g.,Bosetal.,2015].
2.2 IceMassBalanceTherearetwoessentialphysicalcomponentsthatcontrolannualandinter-annualicemassbalance.Theseareinput,estimatedfromsurfacemassbalancemodels,andoutput,thatisgenerallycomputedfromicedrainagevelocitydata[Shepherdetal.,2012].Oneapproachtodeterminingtheongoingmassbalanceofanicesheetistoindependentlyestimateeachofthese,i.e.,changesinthe1)surfacemassbalanceand2)theicedischarge.Asecondapproachistoestimatetheinter-seasonalheightchangesovertheicesheet(estimatingthevolumechange)andfindanappropriatedensityscalingtodeterminemasschange.Athirdapproachistomeasurethegravitychangesfromspace.Thislatterapproachdeterminesmasschangesdirectly.Eachmethodhasadvantagesandpitfalls.Inwhatfollows,weaddresshowGNEThassupportedthegoalsofthesethreemethods,butdonotdiscussthemoreadvancedspeculationthatGNETmightofferawaytoadvancemassbalanceonitsown[e.g.Khanetal.,2010;Yangetal.,2013].
2.2.1 Surfacemassbalance
2.2.1.1 Background
Thesurfacemassbalance(SMB)referstothefactorsthatdeterminehowtheatmospherecandeliverrainandsnowtoreplenishthewatersupplyfortheGrIS,whileotherfactorsareablationbymeltingandrunoffwhilealsoaccountingforrefreezing.Greenlandalsohasnon-icesheetsurfaceswhichalsoparticipateinthemassbalancethataremeasuredbybothGRACE(directly)andbyGPScrustaldisplacements(indirectly).Figure4showsthemodelrepresentationofSMBforGreenlandthatcanbereconstructedfromthecurrentlymostsophisticatedSMBmodelrunningfrom1958-2015byvandenBroekeetal.[2016].IcemassbalanceisessentiallyestimatedasSMB-D,whereDistheicedischargeduetooutletglaciersexitingtothesurroundingocean.Thisisalsocalledtheinput-outputmethod(IOM)[Shepherdetal.,2012;Hurkmansetal.,2014].Technically,thisisnottheentiremassbalancesinceoneothercomponent,groundinglinemigration(GLM)alsomaycomeintoplay[Rosenauetal.,2013].ThemotivationtostudythetwocomponentsoftheIOM(SMBandD)isprovidedbytheimportantrelationshipoftheicesheetmassbalancetoongoingandfuturesea-levelrise[Hannaetal.,2013;vandenBroekeetal.,2016].SMBuncertaintiesaredifficulttoquantify,buttheyareoftenreportedat15-20%Tedescoetal.[2017],
10
thoughlocallyandtemporallytheymaybemuchlarger[Schlegeletal.,2016;Alexanderetal.,2016].ChangesinSMBarenowaccountingforroughly2/3oftotalmassbalance[Enderlinetal.,2014;vandenBroekeetal.,2016],reducingtheuncertaintiesassociatedwithSMBestimatesisofconsiderablepracticalimportance.Inthissection,weaddressthevariouswaysinwhichGNETmighthelptoimproveSMBestimatesand/orgaugetheiruncertainty.Thesub-componentsofSMB,andtheimportanceofthesubtlephysicsthatcontroltheirbehaviorcannotbeunderemphasized.Forexample,ithasrecentlybeenarguedthatthenegativemassbalanceofGreenlandowestotheenergybalancehavingreachedatippingpoint,whereinthefirnlayerbegantoloseitscapabilitytorefreezenearsurfacemeltwater[Nöeletal.,2017].
Figure 4. Cumulative surface mass balance (SMB) for the full land area of Greenland (orange line),cumulativeicedischargeD(blueline)andresultingcumulativemassbalanceMBGreenland(redline).GRACEtimeseriesincluded(greyline)hasbeenoffsetby1000Gtforclarity.NotethatanincreaseinslopeforD(t),anddecreaseinslopeforSMB(t)eachacttopushthemassbalanceofGreenlandintoanegativestate[vandenBroekeetal.,2016].
ItisworthwhiletodisaggregatetheconstituentsofSMBandalsotheirspatialandtemporalvariability.Long-termmassstabilityofglaciercomplexesisuniversallydeterminedbyknowingthestabilityoftheannualaccumulationandablationzones.
11
Figure5.Modelbasedannualvaluesfor1958-2012.25ofSMBconstituentsforGrIS:totalprecipitation(Ptot),melt(ME),runoff(RU),refreezing(RF)andrainfall(RA).Dashedlinesindicate1991–2015trends[vandenBroekeetal.,2016].
2.2.1.2 Theconnectionofincreasedmeltondischarge
AccumulationandablationzonesofanicesheetaredefinedastheareaswherethesignofannualSMBdiffer(positiveandnegative),respectively,withthetwozonesbeingseparatedbytheequilibriumlineatwhichSMB=0.InFigure5weshowthe5mostimportantconstituentsasreconstructedina regionalatmosphericclimate(RACMO2.3)SMBmodelfrom1958throughwinterof2012[vandenBroekeetal.,2016].Notethatmelting(ME)hasthestrongestestimatedslopeduring1991-2012,andalsohasthelargestexcursionsfromthenorm.Meltisalsoacriticalcomponentofchangesindischarge,D,aswatermovementtothebaseoftheicesheetcancauselubricationandglacierspeedup[alsoseeFigure4,andthechangesinslopeafter2005].Somewhatparadoxically,overtime,extensivemeltwateratthebasemayleadtosubsequentdecreaseinglacierspeed[vandeWaletal.,2008;Soleetal.,2011].Theseprocessesarenoteasilymodeled[e.g.,Dasetal.,2008;PimentelandFlowers,2011]andcanseeevolutionarycomplexitythatarepoorlypredictedinmodels[e.g.,Tedstoneetal.,2015].Observationsofice-flowenhancementbysuchlubricationindicatesabroad-scaleinfluenceasthemeltwaterspreadsoverthebed.Thepatternofspeedupisalsoquitecomplex,reflectingthecontrolbybottomtopographyatsub-kmscales,sedimententrainment,subglaciallakefillingandflushing,andconduitreorganizations[e.g.,HewittandFowler;2008;Joughinetal.,2013].Emphasisshouldbeplacedondatasetsofhighqualityandresolutioninspaceandtime.Duringthepastdecade,Arcticsummershavebeentrendingtowardearliermeltingonsetandextrememaximumheatandduration[CrawfordandSerreze,2015].ThisfactwasevidencedmostprofoundlyonGrISbythesummer2012summermelt.AninteractivewebsiteattheNationalSnowandIceDataCenter(NSIDC)currentlytracks
12
theareaextentofmelting.Figure6showsasummaryofthemeltextentof2012(orangecurve)incomparisontothedailymedianvaluesandtheirtypicaldeviancefromthemedian.Theplotclearlyshowssummer2012tobeextraordinaryinarealextent.Wenotethattheareaisfairlyeasytomeasurefromspaceusingemissivityorscatterometry.However,thevolumeofmelt,itssubglacialroutingandtime-scaleforflushingfromthesystemispoorlyknown,atbest.ModelsthatderiveIOM,assume,generally,thatwaterwillrunthroughthesysteminstantaneously,sinceroutingmodelsarenotdevelopedforoperationalSMBmodels.Thismeansthatwhileannualerrorsmaystaybelow20-25%,duringanygivenweek,orevenmonth,themelt(ME),run-off(RU)andre-freeze(RF)maybeinerrorbymorethan100%.Asweseekfuturesea-levelriseestimates,thismaybecomeanincreasingproblem,andthereisadesiretogetabettercontrolonthespace-timeevolutionoftheseroutingsystems.
Figure6.MeltextentmeasuredfromspaceduringSpringtoFallmonthsforGrIS(1980-2010)median,withsingleyear2012showninorangeand2017inblue(fromNSDICinteractivewebsite).NotethatinJuly2012themeltextentquadruplesoverthe30-yearmedian.
Arcticsummershavebegunearlier,lastedlongerandreachedmoreextremesurfacetemperaturesduringthepasttwodecadesCrawfordandSerreze[2015].Asaconsequence,wearelikelytoseetheroleofME,anditsfeedbackonD,andthepartitioningofRUandRFcomeunderincreasedscrutinyforthepurposesofmakingcorrectmodel-basedprojectionsoffuturesea-levelrise.TypicalevaluationsofthequalityofMEinSMBmodelsarequantifiedthroughinter-comparisonofspace-based
13
melt-daysobservedinsatelliteimagestothosecapturedinthereanalysismodel.Suchinter-comparisonisshowninFigure7.Hereweseethatmelt-daysmaybediscrepantby±15days(thesearedaysinwhichmodelandobservationonanygivenpixeldisagree).Whilethemodelsinter-compareinareasonableway,theuncertaintyinroutingpath,time-lagandultimatepercentofexpulsiontotheseaisnottested.
Figure7.SummaryofcomparisonoftwoSMBmodelsandsatellite-detectedannualmeanofthetotalnumberofmeltdays(toppanel;basedonspacebornepassivemicrowavedata).BottompanelshowsthedifferencebetweenmodelsandtheT19HmeltandExtXPGRalgorithmsusedforprocessingspace-baseddatasets[Fettweisetal.,2011].ThemeannumberofGrISpixelswhenRCMandthealgorithmsdetectmelt(RCM=SAT),whenRCMdetectsmeltbuttheretrievingalgorithmsdonot(RCM>SAT)andwhenRCMdoesnotdetectmeltwhilethealgorithmsdo,(RCM<SAT)isalsolistedasapercentageofthenumberofGrISpixels°—summerdays.[ReproducedfromFettweisetal.,2011].
2.2.1.3 SMBAnalysiswithGNETData
Regardlessofthemethodsofconstruction,SMBconstituentswillcontributetoanytimeseriesrepresentationonmass-bearinggridsformedbysummingestimatesofSMB.TemporalchangestototalmassbalanceareusuallydominatedbySMB.GNETcanbeusedtotestbothtypes(intra-seasonalandsecular)ofmasschange.Insomeareas,wecanactuallytestSMBestimatesbycomparingelasticdisplacementtimeseriescomputedfromthemassgridstotheactualdisplacementsmeasuredbyGNETasinarecentpaperbyLiuetal.[2017].Forexample,amajorsnowfallevent,orrapidsequenceofevents,shouldproduceapredictablesignalintheelastictimeseriesatnearbyGPSstations.However,forthelengthscalesknownforsuchevents,thesignalislikelynotdetectedabovethenoiselevelatdistantstations.Giventwosuchmajorsnow
14
falleventsamonthapartinthesamearea,onecanroughlypredictthecorrespondingsignalsinthede-trendedelasticdeflectionscomputedfromdailyorweeklySMBgridstobesimilarlyreflectedinthede-trendedGPStimeseries.Intheory,anysustainedchangeinSMBshouldhaveacorrespondingsignaturerecordedintheGNETtimeseriesfromwhichaloadingmassmaybeinvertedfor.Inpractice,allgeophysicalsourcesofloadingofthesolidEarthmustbecarefullyaccountedfor,suchastheatmosphere,oceanandlargescaleglobalhydrology[vanDametal.,1997;Bevisetal.,2012;AdhikariandIvins.,2016;Liuetal.,2017]
2.2.1.4 UseofGNETDataforFundamentalImprovementinSMBModeling
OneobviouswaytoimproveSMBestimatesistoincreasethepredictionskillofhybridnumericalweatherandsnowmodels.WenotethatoneofthelargeuncertaintiesinFigure5isthetotalprecipitation.Coupledweather-snowmodelsarebuiltaroundhigh-resolution,regionalnumericalweatherprediction(NWP)modelsthatareembeddedwithinaglobalmodel.Typically,thisisanECMWFmodel,suchasitslatestglobalreanalysis,ERA-InterimorERA-5.
Figure8.Illustrationofanensembleanalysiswith(green)andwithout(yellow)minimizationonthestatisticalrangeofanyindividualatmosphericmodelvariable.Theexampleanalysisexampleillustratestheimportanceforthe“forecastcycle”.TheuseoftheGNSSphasedelayinformationinthemodelthatisdataassimilative(“4D-Var”)hasthepotentialtogreatlytightenhindcastmodelsthatarecriticalforSMBmodelslikeRACMOandMAR.The50+GNETstationsolutionsfortroposphereparameterscansupplydatarequiredforvariationalminimizationinfourdimensions.Itisimportanttonotethatrelativelygreaterimprovementoccursinmodelestimates(4DVartrajectories)asthedistancetoobservationsisreduced.[Baueretal.,2015].
15
Coupledweather-snowmodelslikeRACMO2andMARarequitesophisticatedintheircaptureoftheappropriatephysics,aweaknessisthattheylackdirectassimilationofthemeteorologicalobservationsacquiredwithintheirmodeldomains.GlobalmodelssuchasthoseoftheECMWFdoassimilateallavailablemeteorologicaldata,includingGPSdelaydata,usingvariationaldataassimilationmethodsknownas4DVarandensemble4DVar(En4DVar)Baueretal.[2015].AgeneralillustrationisshowninFigure8.WeadvocatethatconsiderableimprovementinSMBcanberealizedbyupgradingcoupledweather-snowmodelcodesthroughimplementationof4DVarmodulesthatdirectlyassimilategroundincludingGPSdelay,upperairandsatelliteobservations.BelowwegiveasynopsisofthegainsshouldberealizedusingGNETdata:
• improvedpredictionskillintheNWMcomponent• morereliableboundaryconditionsforthemulti-layeredsnowmodel• improvedmodelsofthestateofthesnow/firn/icelayersbeneaththesurface.
Towardthesegoalsthefollowingactionsarerequired:
• RoutinegenerationandassimilationofGPSdelayparametersfromGNET• DemonstratetheimpactofassimilatingGPSdelaydataintohighresolution,
regionalNWMs• GPSdelayparametersshouldberoutinelyestimated,innearly-real-timeand
usedtodemonstratethatthedelaydataimprovesthewatervaporfieldsandpredictionsforprecipitation
• Addressissuesthatarepertinenttoweather-snowassimilation.Asmanyoftheseissuesmaybeobscuretothoseunfamiliartoatmosphericprofilinganditsconsequencesforoperationalmodels,weofferthefollowingsectiononthetechnique.
2.2.1.5 GNETZenithTotalDelay:UseforRegionalAtmosphericModels
ThegoalistouseGNETGPSobservationstogenerateandassimilatezenithtotaldelay(ZTD;BennittandJupp,2012)togreatlyimprovethepredictionofsurfacemassbalance(SMB)overtheicesheetproducedbyregionalatmosphericmodels,likeMAR,RACMO,orWRFthatshowlargedifferences(Figure7).ZTDobservationsimplicitlyyieldtheverticallyintegratedwatervaporamount(precipitablewater)nottheverticalprofileofwatervaporlikeradiosondes.However,watervaporisconcentratedinthelowerpartoftheatmosphere–moisturecontent(specifichumidity)typicallydecreasesexponentiallywithheight.ZTDisavailableat~10minuteintervalsifdesiredincontrasttothetwicedailyradiosondeascents,sopotentiallydetailedtemporalbehaviorofmoistairintrusionsintoGreenlandcanbecapturedfromZTDusage.ZTDvaluesaffectedbysensoricingandsnowaccumulationwillhavetobescreenedoutbyarobustqualitycontrolprocedure.TheaccuracyofZTDvaluesinnorthernGreenland
16
duringwintermaybeanissuewithannualprecipitablewatervaluesbeing4-5mmthere(RobaskyandBromwich,1994),withmuchsmallervaluesinwinter,comparabletotheuncertaintyattachedtovaluesinferredtoZTD.AtmosphericmotionfieldsofcomparableaccuracytothewatervaporinformationprovidedbytheGPSZTDwillbeneeded.ThisisgoingtobeachallengeforGreenland,largelysurroundedbydatasparseocean.ThedrydelayderivablefromZTDcorrespondstosurfacepressuresotheseobservationsimprovetheatmosphericcirculationdepiction.OneapproachthatwillhelpistousetheGPSobservationsinconjunctionwiththeautomaticweatherstations(GC-Net)ontheicesheet.TheGNETandGC-Netshouldbeviewedasanintegratednetwork.Additionalcomplementarydeploymentsmaybeneeded.Also,allavailablenon-AWSsurfaceobservationswillbeneeded.ZTDvaluesmaybeasignificantchallengetoassimilateaccuratelyasthesitesarelocatedinthecomplexcoastalenvironmentforgeodeticapplicationsratherthanmeteorologywheresamplingofbroadscaleconditionsisdesired.Mahfoufetal.[2015]hasstudiedthechallengesofZTDassimilation.Thesearepartlyresolutiondependent.Modelgridspacingofafewkmmayberequiredtoresolvethenecessaryterraindetailsandmaynotbeadequateforallsites.Consequently,adetailedevaluationoftheGPSsitesisneededtodeterminethosebestsuitedtocharacterizingatmosphericbehavior.TheMahfoufetal.[2015]studyofEuropeisforacasewhereinsummerthunderstormsshowthatGPSZTDassimilationimprovespredictionoflargeprecipitationevents–morethan0.5cm(waterequivalent)perday.Greenlandisaverydifferentenvironment.TheeventsofconcernforGreenlandaresynopticscalecyclones,frontsandhurricaneremnantsthatresultinlargemoistureflowsintotheisland.Thegreatestimpactseemslikelytobefromsitesonthewest,southandeastcoastsofGreenlandprobablyinthewarmerpartoftheyear.Multi-layeredairflowsassociatedwithmoistureintrusionsasaresultoflow-levelterrainblockingwillcomplicateeffectiveuseofZTDvalues.ThequestionofanassimilationapproachusedtoincorporateZTDobservationsintoregionalatmosphericmodelsisbeingdebated,themuchmorecomplex4DVARversussimplerapproacheslike3DVARand/orensembleapproaches.Perhapsthemostcompellingargumentfor4DVARcomesfromMahfoufetal.[2015]where4DVAR,likethatusedbytheglobalARPEGEmodelwitha6-hourassimilationwindow,wouldinclude10timesmoreZTDobservations(orequivalent)thantheir3DVARapproach.However,thereareotherstrategiesforsimplerdataassimilationapproachestoincorporatemoreZTDobservations.Ofequal,orgreater,importanceisthecarewithwhichZTDvaluesneedtobetreatedtoachieveaccurateassimilation.OurrecommendationisthatdataforZTD(andGC-Netvalues)fromGreenlandbeassimilatedfrom2007onwardintotheArcticSystemReanalysisversion3(ASRv3)for1979-2020at15kmwith71verticallevels.Ithasapan-Arcticdomainandextendsfrom~45°NtotheNorthPole.Theprimaryfocusistoexamineextremeweatherand
17
climateconditionsinthePanArctic.AsecondarygoalistoachievebettersurfacemassbalanceestimatesfortheGreenlandicesheetforthelast14ofthe42yearstobecoveredbyASRv3.ProspectsforusingtheHIRLAMregionalmodel(Christensenetal.,2007)andits4DVARassimilationcapabilitytoperformaveryhighresolution(gridspacingofafewkm)ofGreenlandfrom2000onwardisquitepromising.AlldatawithinthedomainwillbeassimilatedincludingGNETZTDalsowiththegoalofbettersurfacemassbalanceestimatesfortheicesheet.Tosummarizeweemphasizetwokeyrecommendationsinthiswhitepaper:
1. ZTDobservationsfromthecomprehensiveGNETarrayaroundGreenlandshowgreatpromiseforimprovingsurfacemassbalance(andfirndensity)estimatesfromregionalatmosphericmodels.Thesedatahavehardlybeenexaminedbytheatmosphericsciencecommunityandtheirspatialandtemporalbehaviorarenotwellcharacterized;andthisknowledgeisneededtomakeeffectiveuseoftheseobservationsviadataassimilationintoatmosphericmodels.TwoormoregraduatestudentsshouldbesupportedtoevaluateGNETZTDdata,theirassimilationintoregionalatmosphericmodels,andcomparisonswithsimulationsproducedwithMARandRACMO2andreanalysisresultsproducedbyASRversion2.
2. SupporttheuseofGNETZTDinregionalreanalysesencompassingGreenland,ASRv3andtheHIRLAM-basedeffort.
2.2.1.6 ChangesinGrISdrivenbybothglacierdischargeandSMB
Inpractice,crustalmotionsrespondtothefullsetofloadchangesthatinevitablyinvolvebothicedischargeandSMB.WhiletheSMBchangeshavegreateramplitudesandclearcoherenceandfidelityintheirseasonality,changesinDoccursimultaneously,althoughthesechangestendtobeoflowerfrequencyintheirtemporaldomains.Observationsoficedischargevariationsaregovernedbychangesinglacierspeedandgeometry[Ahlstrømetal.,2013;Moonetal.2014;Joughinetal.,2013].However,itisaninescapablefactthattheSMBMEandDconstituentsareintimatelycoupled[e.g.,Palmeretal.,2011;PimentelandFlowers,2011;Schlegeletal.,2013].TherearerelativelynewandpromisingdirectionsthatGNETmightcontributetodisentanglingtheinterrelationshipsofcoupledicedischargesystems.IftheGNETtimeseriescanbeappropriatelyfiltered,theyshouldcontainsequentialloadsignalresponsesthataredrivenbythecouplingofMEandD.Approximately60%oftheGPSreceiversinGNETarecapableofmakingcontributionstoourunderstandingofthedischargeofGrISglaciersandicestreams.Khanetal.[2010]andLiuetal.[2017]recentlydiscussedcasesinwhichGPSdatahavebeenusedtoplaceconstraintsonglacierdynamics.Additionalresearchhasinvestigatedglacier
18
flowbyusingGPStrackersonmovingglacierice.HeredatafromGNETstationsplayacriticalroleasreferencebasestationsAndersenetal.[2010].ThebestbedrockGNETreceiversforthisobservationareeitherveryclosetotheglacierterminusbeingstudied,orexistaspairsofreceivers,whereinoneisclosetotheterminusoftheglacier,andoneisatadistancefarenoughawaytobeunaffectedbyglacierchanges.AsimilargeodeticstrategywasusedbyDietrichetal.[2007]forJakobshavnIsbræ.
InFigure9,stationSRMPhasaverytightradiusofsensitivitywithrespectto4adjacentoutletglaciers,whileUPVK,havingacoastalbedrocklocation,ismoresensitivetobroad-scaleSMBfeaturessuchasPtot(seeFigure5).AnewtreatmentofGNETdatasensitivitybyAdhikarietal.[2017]employedthe3-DcrustalmotionstostudycoupledSMB-DonseasonaltimescalesforeachGNETstation.OneexamplefortheRinkGlacierinwestcentralGreenlandusedtheGPS-determined3-Dbedrockmotiontounambiguouslyisolatemasstransportinasolitarywavethatoccursonlyduringtheintensemeltseasonsof2010and2012(seeFigures5and6).Someofthebasicconceptsofemploying3-DmotionsaredescribedbyWahretal.[2013].Inthelatterpapertwofeatureswerehighlighted:horizontalmotionsdropoffmorequicklythandoverticalmotionsasonemovesawayfromthewater-iceloadcenter.Additionally,horizontalmotiondatacontaindifferentdiagnosticinformationaboutthelocationoftheloadsource(seeFigure9).Forstudyingglacier-specificSMB-Dcouplingthelatterfeatureiscritical.AsubtlefeaturetobeextractedfromtheWahretal.[2013]example(Figure9)isthefactthatthereisaninherentdirectionalanisotropytothehorizontaldisplacements,andthesecanbeusedtolocate,undercertainconditions,thephysicallocationandsizeofthechangingwaterload.Thisisnotanewdiscovery,asSauberetal.[2000]haspointedthisoutinregardstochangesinglaciermassinAlaska,andHekietal.[2004]hasusedthedensenetworkinJapantomeasuremountainsnowloadchanges,withtheadvantageofthelaterstudybeingboththeexistenceofdenseGPSnetworkandknowledgeoftheapproximatelocationandtimingoftheload.
Figure9.AnalysisofanormalizedSMBcontributiontoverticalcrustaldisplacementforSRMP(thickline)andUPVK(thinline)stations.Theinsetshowsazoom-indistancerangeof0–100km.ThisstudyconcludedthatmodeledSMBloadingbasedonRACMO2.3actuallyincreasedtheannualvarianceoftheGPSresiduals,whichmadeitdifficulttoestimatecontributionsfromtheannualvariabilityofglacialdynamicstothetotalicemassbalance.ThestudyrevealsthepotentialuseofGNETonevaluationofSMBmodels.(FromLiuetal.,2017).
19
GNETstations,beingoriginallyplacedtomeasureGIA,arenotnecessarilyclosetotherapidlychangingoutletglaciers.Forthepurposesofusing3-Dcrustalmotions,thestationsclosesttooutletglaciersarethemostpromisingforprovidingnewconstraints.Adhikarietal.[2017]usedallthestationsthathadprocessed3-Dtime-seriesavailableandderivedasensitivitygradientmap(seeFigure10)forthesestations.Themainprinciplefollowsfromthatusedingeophysicalmodeldataassimilationforadjointsystems[e.g.,Trompetal.,2005;Larouretal.,2016].Eachstationmaybeinfluencedbyon-landmasschangesinitsimmediatelocalarea,includingatmosphericloadingandloadingfromice-snow-watermasssystems.Thelatterchanges,whentheyarelargeinamplitude,dominantlyoccurwithintheoutletglaciersystems,yetthefullSMBloading,includingthatoccurringonbedrockmustbealsocomputed.ItispossiblethatstationsmayhaveallthreeoftheircrustaldisplacementssensitivetomasschangesassociatedwithME-Dcoupledevents,ortolocalRUorPEinduced-changes.Figure10shows33stationswithmappedsensitivitytomassloadinginthevicinityofeachoftheGNETstationreceiversthathave3-componentdisplacementdataavailable.Hereathree-colorwheelisused.Cooltohotcolorsrepresentprogressivelyhighersensitivityzonesforthestations,suchthatatleast1%ofmaximumdisplacementinducedbyaunitloadappliedoveraunitareainthesezoneswouldberecordedinone(blue),two(green)andthree(red)componentsofthedisplacementvector,respectively.Thesearereferredtoaszonesofinfluence(ZOI)byAdhikarietal.[2017],anditisimportanttounderstandthatthesefullyaccountforloadvariationsoccurringata2-kmscale,sothesecanshowwhereapotentiallyMElubricatedicestreammightquicklylooseorgainmass.ForthemarineterminatingRinkGlacier,the3-Dmotionscapturealargewavetransitingdown-glacierduringsummer-to-wintermonthsduringtheintensemeltyearof2012(seeFigures5,6).Figure11showsthetrajectoryofthehorizontaldisplacementvectoroftheRINKstationduringthedown-
Figure10.Apredictionof3-Dcrustalmotionsfromasimplesinglediskwaterloadremoval.(a)showsthevertical(positiveupward)andhorizontal(positiveawayfromthedisccenter).Crustaldisplacementsarecausedbyremovingauniformdiscloadofradius20kmhavinga1-mwaterthickness.Predictionsaregivenasafunctionofthedistancetothecenterofthedisc.Theverticaldot-dashedlinemarkstheedgeofthedisc.ResultsarecomputedusingseismologicallyrealisticEarthmodel.(b)Theratiooftheverticaltohorizontaldisplacementsshownin(a).Notethedifferentpositioninframe(a)ofthepeakvalueswithrespectto0.[Wahretal.,2013].
20
glaciertransitoftheME-Dmassejectionduringsummertowinterof2010and2012.NotethatthereisnoevidenceofthewaveduringyearsofnominalME.
Verticalmotionstendtoreflectmassgainsandlossesoverlargerregions,whilehorizontaloversmallerones.ThisisreflectedattheRinkGlacierbytheexcellentcomparisonof3°GRACE-masconbasedloading,Schlegeletal.[2016](redlineinFig.12a)andthe90-daymeanoftheverticalGPStime-series.Horizontalmotionsshownoclearcorrelation(lowertwoframesinFigure12a),beingmoresensitivetothe10-50
Figure11.ZOIsfor33GNETstations.Thecentermapinthetoppanelshowsmeasuredicesurfacevelocities,locationsofall54GNETstations(redcircles),and5regionsofGreenland(blackboxesdisplayingZOIs.TheZOIsarecomputedusinghigh-resolution(~2km,seeAdhikarietal.,2016)gradientmaps.Blue,green,andredsignifytheprogressivelyhighsensitivityzonesforthecorrespondingGPSstations,suchthatatleast1%ofmaximumdisplacementinducedbyaunitloadappliedoveraunitareainthesezoneswouldberecordedinone,twoandthreecomponentsofthedisplacementvector,respectively.Foreachregion,stationIDsarelisted(fromNorthtoSouth)onlyforthosestationsthathaveZOIs.Majoroutletglaciersarealsoshown[Adhikarietal.,2017].
21
kmscaleloadingeventsandmorerapidshiftsinloadingstructurethan3months.The2-DplotofhorizontalpositionisshowninFigure12b,forallyearsofRINKdata,withamplitudestandoutatthe0.5–1.0cmlevelduringtheintensemeltyears.DuringthetwoyearsofintensemeltthetrajectoryisnorthingandeastingfromaboutJunetoOctober,andthenreversinginlateFalltoearlySpringofthefollowingyear.ThespatialdistributionoftheMARSMBloadingisshownbythebackgroundcolorsoftheframes(c)and(d)ofFigure12.
2.2.1.7 OtherpossiblesurfacemassbalancestudiesbenefitingfromGNET
InadditiontothefamiliaruseofGNETstationsasGPSreferencepointsforon-icestudiesandthemotionoftheGNETstationstoinfericedischargeandchangesiniceloadingofthecrust,thereareanumberofas-yetunexploitedtopicsthatcouldbenefitfromGNETGPSdata.
1. TousethetimeseriesofhorizontalGPSstationpositionstodeterminemassloss
overanunknownareathatwouldaffectaGPSstationatacertaindistancefromthelocusofmassloss.
Figure12.MasstransportwavesdetectedinhorizontaldisplacementsatRINK.(a)Componentsofdetrendeddailycrustaldisplacement(circles)measuredatRINKstation,with90-dayrunningmeans(blacklines).Models(redlines)derivedfromGRACE-basedsurfaceloadingreconciletheobservationsofverticaldisplacement,notabledeviationsoccurduringintensemeltyears(grayshadows)forthehorizontalcomponents.(b)Map-viewtrajectoriesofhorizontals.(Largeonlyin2010and2012.)Circlesshowmonthlystationpositions.(c)PatternofmassdeficittransitingtheRinkGlacierduring2012summer.About-7.1mofmonthlythinningovertheoptimaldomain(bluefillwithintheglaciertrunkoutlinedbywhiteline)isrequiredtoexplainthemeanmonthlydisplacement(redarrow).Alsoplottedarethemagnitudesandfulcrumpositionsofmonthlymassanomalies(circles)thatsatisfythemonthlydisplacementdata(arrows).Adown-glacierpropagationof(negative)massanomalyrepresentsthenegativephaseofthemasstransportwave.MeanmonthlySMBloadsareshowninthebackground.(d)Sameas(c),butforthefall/mid-winterseasonthatfollows.Itrequires+2.8mmonthlythickeningovertheoptimaldomain(redfillwithintheglaciertrunk)toexplainthemeanmonthlydisplacement(bluearrow).(e)Summaryof(c)and(d),revealingthesolitaryseasonalwaveoficemasstransport.
22
2. TousemultipathreflectionsofGPS/GNSSsignalstoobtainsurfaceparameters(snowdepth,oceantides).Larsonetal.[2009]providesaproof-of-conceptandSiegfriedetal.[submitted]presentareal-worldapplicationinAntarcticaofthismethodology.TheGNETarchiveprovidesameanstodothisonalargescale.
3. TouseGPSseismologytoprovideinformationaboutcalvingmechanicsresolvedathigh(~1Hz)frequencies[e.g.,Hollandetal.,2016]
4. TouseGPSonthesurfaceofmarineterminatingglacierstoobservegrounding/hingelineflexure.SuchstudieswouldbenefitfromfixedreferenceGPSstationssuchasGNET.
5. TouseburiedGPSantennatoobtainsnowwaterequivalentabovetheantennaemployingthewavepropagationprinciplesinsnow-ice,asforexampleoutlinedbyNievinskiandLarson(2014).
6. TofurthercoupleGPSobservationswithremotesensingtounderstandthenearsurfacerheologyofthecrustaroundGreenland.[e.g.,KucharandMilne,2015]
7. Examinetheinfluenceoftides(fromGPS-IR)oniceloadsandGPSresponse(e.g.,DeJaunetal.,2010).
Theabovelistofnewprojectswould,inseveralcases,requiretargeteddensificationoftheGNETnetwork.SuchadditionstoGNETwouldnecessarilybetheresultofproposal-drivenscienceinvestmentsfromNSF,NASAorotherfundingagency,andarethereforenotaprimaryrecommendationofthisreport.
2.2.2 SupportforInterpretationofAltimetryGNETcanalsohelptocalibrateorcorrectthealtimetryusedtoformSMBestimates.Althoughnotyetexploited,someICESattrackshaveverycloseapproachestoGNETstations.CrustalupliftmeasuredattheseGNETsitescouldbeusedtocorrecttheupliftofthesnowsurfacemeasuredbyICESat,allowinganassessmentofthechangeofthethicknessoftheicesheet.AsimilarstudycouldbedonewithICESat-2,followingits2018launch.Inaddition,thegroundtrackpatternofICESat-2isexpectedtobemuchdenserthanthatofICESat,expandingthenumberofGNETstationsthatcouldbeusedinaltimetryvalidation.StationsintheGNETnetworkareonexposedbedrockneartheicesheetperiphery,andrecordchangesintherocksurfaceheighttoafewmillimetersperyear.ICESat-2tracksinsomecaseswillnaturallypasswithinafewtensofmetersofaspecificstation,andinothercases,ICESat-2canbepointedtowithinafewtensofmetersofagivenGNETstation(thepointingcontrolofICESat-2being~40m).TheroughterrainandsteepslopesnearmanyoftheGNETstationswillbroadenthelaserpulsereturnofICESat-2comparedtothereturnoveraflatsurface,andshouldyieldelevationprecisiontowithinafewtensofcentimeters.Giventhelocallyroughterrain,anintermediateheightproductwouldbeneededinordertocomparethesub-centimeterprecisionfromGNETtothenearbyICESat-2measurements.Inordertobridgethisscalegap,GNETstationscouldbesurveyedusingLIDAR,stereophotography,and/oratotalstationsurvey.Withthesedata,itwillbepossibletodeterminelocationofthe
23
GPSmonument,theGPSantenna(anditsphasecenter)relativetotherocksurfaceinwhichtheantennawasinstalled.Theselocalarea,geodeticgradeDEMs,withaperturesofhundredsofmetersorevenakilometer,wouldenablethegenerationofepoch-specificDEMsusingWorldViewsatelliteimageryadjustedtothelocalantennaandrocksurfaceDEM.TheWorldViewDEMcouldhaveanapertureof10km,forexample,andallowtheGNETstation’sfootprinttobeextendedontotheicesheet,characterizingtheheightofthesurfacecrossedbyICESat-2’strackwithinadayorsooftheICESat-2measurement.
2.2.3 SupportforGRACEandtheTotalMassBalanceTimeSeriesTheotherwayinwhichGNETcanimproveourestimatesofMBistosupplementthemasschangeestimatesinferredfromGRACEandGRACE-FollowOn.Improvementsaretwo-fold:GRACEhasverylowresolution(250kmx250kmattheverybest)andcannotdealwithcomplicatedcoastlines,snowcoveredareasandsmallglaciersvs.icesheet,etc.Secondly,GRACEmasschangedeterminationisheavilyreliantonanaccurate“GIAcorrection”thatcanonlybeprovidedbyacombinedmodelingandGPS-determinedverticalmotionretrievedfromthesecularcomponentofthetimeseries.GRACEmeasuresbothiceandrockmasschangesassociatedwithGIA.TheverticalrockmassfluxisestimatedusingaGIAmodel,andthisisremovedfromthetotalmasschangesensedbyGRACEtoisolatetheicemasschange.TheproblemswiththisGIAcorrectionarethat(1)itcanbelargerthantheresultingestimateoficemasschangeinsomeregions,and(2)thepotpourriofcurrentmodelspredictdifferentratesofGIA.TheonlywayoutofthisproblemistoimprovetheobservationalconstraintsonGIA,sothattheGIAmodelsarebetterconstrainedbyawiderrangeofobservations.ThiswasoneoftheoriginalgoalsofGNET,andtherecentworkbyKhanetal.[2016]hasdemonstratedthaterrorsintheGIAcorrectiontraditionallyusedbytheGRACEanalysisgroupshaveledtomajorerrorsinmasschangeestimatesforspecificdrainagebasins.
2.3 Ionosphere,TroposphereThe ionosphere and troposphere affect GNSS signals traveling from satellites toreceivers.The ionosphereextends fromabout50-1000kmabove theEarth’ssurfaceand acts as a dispersivemedium for GNSS signals. Thus, the phase delay of a radiosignalintheionosphereisfrequency-dependentanddeterminedbythenumberoffreeelectronsintheionosphere.Dual-frequencyGNSSobservationscanbeusedtoestimateionosphericdelayandalsodeterminethetotalelectroncontent(TEC)alongthesignalpath [MisraandEnge,2011;Komjathyetal.,2016].Processesaffecting the ionosphereincludespaceweathereffectsduetosolarandgeomagneticactivity,loweratmosphericcoupling(suchasstratosphericwarming,tidalsignatures),andnaturalandman-madehazards (tsunamis, earthquakes, rocket launches).Hence, themonitoring of TEChasuses ranging from space weather monitoring [Basu et al., 2001; Coster et al., 2003;CosterandKomjathy,2008], largeandsmall-scaleweathermonitoringsuchasstudiesof global stratospheric warming signatures Goncharenko et al. [2010] to tsunami
24
trackingKomjathyetal.[2016],andthunderstormsLayetal.[2013],andevaluationofco-seismic[CalaisandMinister,1998]orco-eruptiveenergyrelease[Heki,2006].Spanning60-85° north latitude, GNETpresents a unique opportunity to observe thehigh-latitudeionospherewithpiercepointsrangingfromabout55-90°northlatitude.Ground-based GNSSmeasurements can provide a complete latitudinal profile of theArctic ionosphere. Geomagnetic processes are commonly observed with magneticstations that are co-located with GNSS stations in some places, and benefit fromsharingresourcesbetweenthetwostationtypes.Forseveralyears,GNEThasbeenusedforionosphericstudieswithafocusonstudiesof the polar cap and auroral oval processes via TEC and scintillation observations[Durgonicsetal.,2017].Routineprocessinggenerateshourly tosub-hourlyTECmapsover Greenland for near-real time application. However, insufficient data access tomost GNET stations limits this to only 10-15 stations, which impacts the spatialresolution of the results.While real-time streaming capabilities should be the long-term target for all sites, data access on a sub-hourly basis would be extremelybeneficial.Anincrease inresolutioncanbeachievedbyadditionof inlandstations inregions that move less than a few mm per day (this stability requirement can berelaxedforstationsthatonlyobservetheionosphere)andupgradingfromGPS-onlytomulti-GNSSconstellationobservationsthatincludeallavailablesatelliteconstellations.Samplingratesof1HzforTECand50-60Hzforscintillationstudiesareidealforsuchstudies. To minimize the bandwidth requirement for scintillation applications,scintillationindicescanbecalculatedlocallyandthentransmitted,instatedoftherawGNSSobservations.To fullyenable scienceduring strongscintillationconditions, therawdata shouldbe stored locally in a ring-buffer such that on-demanddownload ispossible.Thiswouldnotonlybenefitnavigationorpotentiallypowergridprotection,butenablestreamliningofthesciencesuchthatroutineanomalydetection,catalogingandanalysisispossible;ultimatelyexpandingionospherestudiesfromevent-basedtoacontinuum.The troposphere, ranging up to about 9 km high at the poles, is a non-dispersivemediumandassuchitsimpactonaradiosignalisfrequencyindependent(MisraandEnge, 2011). The effect of the troposphere on GNSS signal propagation can beseparatedintodryandwetphasedelays.TheformeriscausedbydrygasessuchasN2andO2,and the latter isdue towatervapor. InstandardGPSpositioningprocessing,these delays are corrected with models and obliquity mapping functions. Howevermaking use of observations from the global network of GNSS receivers and localmeasurements of pressure and temperature, the dry andwet terms of delay can bederived,andfromthem,parameterssuchasthezenithprecipitablewatercontentcanbeestimated[Bevisetal.,1992].Water vapor estimation should be a standard product for GNET and would benefitsatellite mission validation and calibration (e.g., improvements to the troposphericmodelsusedforICESat-2rangedelaycorrection).Thecurrentlowtemporal,buthighspatialresolutionMODISwatervaporestimatescanbemergedwithGNSSproductsto
25
improve temporal resolution.While GNET’s spatial layout makes it difficult to fullycapture the heterogeneity of the troposphere, moving to full constellation GNSStracking at high data rateswill increase the number of troposphere piercing points.The network should be densified along the coast as most of the water vapor isexpected there. However, interior Greenland stations will allow tracking of watervaporintheinterior.Recentmeltingeventsthataffectedtheicesheetinteriorcanbeexpected to increase thewatervaporconcentration in theatmosphere.Forexample,thin, lowlevel liquidcloudswererecentlyshowntooccuroverGreenlandfrequentlyand enhanced the 2012melting event [Bennartzetal.,2013]. The additional interiorGNSSstationsproposedforionospherestudiescouldalsobenefitclimateresearchfortracking the atmospheric water content and, using GNSS-reflectometry Larsonetal.[2009],recordsnowheight/snowwatercontent.Meteorologyinstrumentationshouldbe included at all stations as these observations at high temporal resolution can beusedinvalidationeffortsofnumericalweathermodels.AlloftheseapplicationswouldbeacompletelynewuseofGNETdata.
2.4 LargescalegeodesyGNETGPSstationsplayanimportantroleinrealizationofaccurateandstableglobalinternationalreferencesthatcanbeusedforobservingchangesinthegeospheresuchaschangesintheicesheets,sealevel,theEarth’stectosphereandstrainfieldacrosstheNorthAmericaplate.Asdiscussedabove,GNETalsoservesasaconsistentreferenceforindividualsurveysandcampaignstherebylinkingsuchdatatogether.Theimportanceofanaccuratereliableinternationalreferenceframe(suchastheInternationalTerrestrialReferenceFrame;ITRF)forsupportingasustainabledevelopmentofinternationalsocietyandinternationalcollaborationhavebeenacknowledgedbytheUnitedNations(Resolution69/266),andallnationsbenefitfromwell-coordinatedreferenceframedetermination.
2.4.1 GNSScoordinatetimeseriesandreferenceframes.Thecomputationofcoordinatetimeseriesisafundamentalandnon-trivialtaskthatiscrucialforobtainingthebestresultsforsubsequenttimeseriesanalysis.InthecomputationofcoordinatetimeseriesfortheGNETstationsthereareissuesrelatedtohandlingofGPSclockandorbiterrors,ionosphericdelays,andrelatedtopics.Furthermore,suchcomputationsaredependentonthestabilityofthereferenceframe.Changesincomputationstrategiesandreferenceframeinstabilitiesanduncertaintiesaffecttheresultingcoordinatessubstantially.OftenthereferenceframeneedstoberedefinedwhencoordinatetimeseriesforGNETstationsarecomputed.SuchstudieshavebecomemorepowerfulastheGNSSsatellitenetworkhasexpanded(GPS,GLONASS,and/orGALILEO)andthecorrespondingprocessingstrategieshavematured.Thesedevelopmentsareongoingamongmanyresearchgroupsaroundtheworld.Inaddition,asthenumberandlocationofreferencestationsgloballyhasevolved,theselectionofstationsusedinsecuringastablerealizationofthereferenceframehasalsoevolved.
26
Tosatisfythestringentrequirementsforreferenceframedetermination,potentialreferencestationsrequirecontinuousdataseriesfromgeodeticGNSSreceiversequippedwithchokeringantennasmountedonstablemonuments.Bydesign,GNETstationshavethatquality.TheobservablesfromallGNETstationsareconvertedtothestandardRINEXdataformat.TheGNETrawdatahasbeenarchivedastwosetsofdata:UNAVCOandLuxemburgdata.TheseareonlinethroughtheUNAVCOwebsite,andtheDTUSpaceGPSdata,whicharestoredatDTUSpaceandonlyavailableonrequest.Thedataportalsmaybeusedforpost-processingandestimationofcoordinatetimeseries,orusedasreferencestationsfortemporaryGPSstationinstallationsforawiderangeofapplicationsdescribedabove.InordertomaximizethebenefitoftheGNETnetworkforcoordinatesystemstudiesorreferenceframedetermination,stationsshouldlogGALILEOandGLONASSsignalsaswellasGPS.Furthermore,GNETstationsmayneedtobeaugmentedbyreferenceclocks,metsensors,andothergeophysicalequipmentsuchmagnetometers,radiationsensorsetc.
2.4.2 MakingITRFavailabletotheusers.Tosupportresearchanddevelopmentprojectsitisessentialthatthereferenceframeisrealizedlocallyandmadeaccessibletoend-users.Today,thereferenceframesarerealizedusingpermanentlyoperatingGNSSreferencestationsforwhichthereferencecoordinatesandGNSSdataareappliedforaccurateandconsistentpositioningofvehicles/platforms/payloadsmappingthequantityofinterest(e.g.icesheet/glaciergeometry,velocities,etc.).Currently,theSDFEandDTUareoperatingGNSSstationsinGreenlandtofulfillthistaskasthenationalauthoritytorealizenewITRFslocallyanddefinetransformationstotheofficialnationalframe.ThemainchallengestomakingITRFsavailabletousersarethatthefewon-linereferencestationsdonotprovidesufficientcoveragetodefineandmaintainasufficientlydetailedreferenceframeinallpartsofGreenlandtosupportresearchanddevelopmentprojectsandauthoritytasks,andthatdatafromallexistingreferencestationshasnotyetbeenmadeavailabletotheusersintherequiredrateandformat.ThistaskrequiresreferencecoordinatessupplementedbyhigherrateGNSSdatafromtheGNSSstations.Currently,theSDFEandDTUaresupportingthistaskasthenationalauthority.ReferencecoordinatesareavailableforallGNETstations.Onlyaminorsubsetofthestations,however,canprovidehigherratedata.GNETstationscouldpotentially,underanupgrade,providethereferencecoordinatesandhigherrateGNSSdataneededtosupportgroundandairborneoperationsinresearchanddevelopmentprojectsovertheentireisland.ByupgradingGNETstationswithhigherratereceiversandcommunicationlinksthegapscanbefilled.Furthermore,stationsmayneedtobeupgradedwithmultiGNSS(asopposedtoGPS-only)receivers.Optimally,near-real-timedatamaybeneeded.
27
2.4.3 GNSSreflectometry.AnalysesofthereflectedGNSSsignalsshownewinterestingapplicationsofGNETdataforstudyingchangesinsealevelandsnowaccumulation.Bothapplicationsmayprovidevaluablesupplementstotheexistingobservationnetworks,especiallyinGreenlandwheredeploymentofsuchinstrumentationischallenging.OnechallengeinimplementingGNSSreflectometryforsealevelwork,isthatonlyafewstationsarelocatedneartheoceanscapableofobtainingsealevelinformation.Forsnowaccumulationwork,moststationsarelocatedonbedrockontheperipheryoftheisland,andassuchareoutsideofGreenland’saccumulationarea.However,valuableestimatesofsnowheightandmassofprecipitation(snow/firndensification)isimportanttoassessintheablationzone,thusaddingtoourknowledgeofSMBprocesses.Additionally,theseapplicationsrequirereflectionsatverylowanglethatarenormallyeliminatedfromthedatastreambyanelevationmask.
ThisisanewapplicationofGNSSrecentlypresentedatconferences,butatpresentonlyafewGNETstationscanbeused.However,applicationtoGNETdatamaybedevelopedusingexistingdataacquiredbythenetworkundercertainconditions.Forbestresults,stationslocatedclosetotheoceanmaybeusedformonitoringsealevelandsupporttidalstudiesandstudiesofsealevelrise.Stationslocatedcloseto,oractuallyon,theicesheetmaybeusedformonitoringchangesintheicesheetsandsupportstudiesoficesheetmassbalanceandofsnow/firnproperties.Tosupportthiswork,morestationsinclosevicinitytotheoceanandtotheicesheetareneeded.
3 DataManagement
3.1 GeneralBackgroundForbroaderuseofGNETrawdata,itisrecommendedthatalldatabemadeavailableonline,withaminimaldelay,asfarasfeasible(afewstationslikeStationNORDarenotroutinelydownloadedforhigh-ratedata).Withjusttwoagenciesinvolved,itisprobablyeasiesttomaintaintheseparateUNAVCOandDTUsites,aslongasdataformats,RINEXversionsandsitemaintenancelogsaresimilarandwelldocumented.Itisrecommendedthatbothofthedatacentersprovidedoinumbersfortheirrespectiverawdata,forpropercitation.Havingclearformatsandsimpleaccesstoalldatawouldbenefitnumerousotherapplicationsfornon-expertGPSusers,e.g.inthemeteorologydomain,andalsomakesurethattheGNETdataisusedmuchmorewidelyforlocalGreenlandsurveying,bothforsocietyuseandmining/resourcedevelopment(thisisessentiallynothappeningtoday,mainlybecauseoflacktrainingforlocalnon-expertusers).
28
3.2 AncillaryDataAbsoluterepeatedgravitydata,observedirregularlybothbyULuxandDTUSpace,shouldsimilarlybeassigneddoi’s,andalsobepostedwithinashortintervaloftheobservations(ideally<1year)onDTUSpaceandULuxwebsite,respectively.Theurgencyforthiskindofdataislower,andonlyscienceoriented.Thefinalprocessedgravityvaluesshouldbeprovidedalongwithestimatedstandarderrors,tidalcorrections,operatorissues,hardwaretype,andverticalgravitygradientused(ifany).Itislikelynotrelevanttoincludeindividualdropdataandsimilar,butimprovedaccesstodatawouldstimulatemorescientificinvestigationsofGNETandgravitychanges,andthusbetterGIAunderstanding.
3.3 GenerationofTimeSeriesAcentralGNETwebpageshouldprovidethenecessarydatalinkstoalldata,bothrawandprocessedGPSdata,aswellasgravity.Itcouldalsolinktoother3rdpartyrelevantwebsites,suchasthenationalGreenlandmonitoringsitewww.promice.dkandotherGreenlanddataportals.Itwouldalsoberelevanttoincludesomeresourcematerialforlocalnon-expertusers.Inadditiontotherawdataaccessdescribedabove,processedGPSdataintheformoftimeseriesofstandardizedlatitude,longitudeandheightsate.g.30secresolutionarerecommendedforprovisionbyoneormore“official”dataprocessingcenters.IthasbeenamajorlimitingfactorforthewidespreaduseofGNETdatathatonlythe“raw”GPSdataareavailable.Processingsuchdataarebeyondthecapabilityofmanyusers,e.g.forglaciologistsandothergeoscientistsprimarilyinterestedinusingGNETfordirectGreenlandmassbalanceestimation.A“level2”GNETproductshouldberoutinelyprovidedbyoneormoredatacenters(e.g.OSU,UNevada,UNAVCOand/orDTUSpace),usingawell-describedanddocumentedmethodology.Onlybyprovidingthe“Level-2”coordinateproductwillabroaderusercommunityreallybeabletoutilizeGNET(andANET)dataforinnovativescience.TheavailabilityofL2-datawouldnourishnewscientificadvances,possiblylendinggreaterstrengthforannualmassbalance,especiallywheninterwovenwithothersatellitedata.WiththehightemporalresolutionofGNET,itwouldthenalsobepossibletoturnthisupsidedown,andimprovesatellitemonitoringoficesheetchangesbyGRACE,altimetryandiceflowvelocitymapping.Thereisaparalleltosatellitemissionshere:imaginethattheGRACEteamonlyprovidedLevel-1data(satelliterangeandrange-ratedata);thiswouldhaveseverelylimiteduseofGRACEdatatoahandfulofusers.ForGRACEthe“competition”inprovidingthebetterLevel-2databyimprovedmethodologyhasindeedhelpedtogeneratebetterLevel-2products,forthebenefitsofallusers.Tobeabletobenefitfromthesatellite/phonelinedatacommunicationtomostGNETstations,theLevel-2processingshouldideallybesetupinanearreal-timefashion,
29
withalatencyoffewdays,tomatchthepresent6-daynear-realtimeprocessingofSentinel-1iceoutletglaciervelocitydata,alreadyrunningroutinelyaroundtheentireGreenlandicesheet(e.g.,http://cryoportal.enveo.at/).Also,thereferencesystemandtidal/atmosphericcorrectionsappliedshouldbeagreeduponbetweenallprocessingcenters,tothedegreepossible.TheLevel-2GNETdataprovidersshouldbeprovidedsecurelong-termfundingforrunningtheroutineprocessingservice,justlikeUniversityofTexas,JPL,andGFZ(Germany)doesforGRACE.Givingthe10-yearexperienceinanalysingGNETdatabykeyPI’s,thisshouldnotrequirelargeresources(comparede.g.totheGNETmaintenancecosts),asstandardprocessingtoolssuchasGIPSY,GAMITandBerneseareavailableandapplicationsfornetworkanalysisproven.TheLevel-2processingcouldalsobeaugmentedwithotherparameters,suchasionosphereandatmosphericrelatedprocessing,butthismightbeabiggerchallenge(andpotentiallyfewerusers)thanthe“clean”L2coordinateproducts.Derivedhigher-levelproducts,suchasimprovedGIAmodels,wouldstillbe“pure”scienceandkeptoutoftheL2-service,althoughlinksandrecommendationsshouldbeprovidedontheGNETdataportal/website.
3.4 DataStreamandArchivingFundedbyNSFDatafromtheNationalScienceFoundation(NSF)-fundedGNETnetworkisdownloadedeitherthroughTeleGreenlandADSLmodems(3sites)orthroughIridiumsatellitemodems(38sites).GNETmakesuseoftwodifferenttypesofIridiummodemstoconnecttotheremoteGPSreceivers.Onetypeinvolvesdialing29remotesitesfromanIridiumdownloadhubbasedinBoulder,Colorado.Thisdownloadhubconsistsofonebasemodemforapproximatelyevery6remotesites.ThesecondinvolvesusingIridiumRouter-BasedUnrestrictedDigitalInternetworkingConnectivitySolutions(RUDICS)modem,whichusesmulti-protocolMobileOriginated(MO)andMobileTerminated(MT)circuitswitcheddataconnectivity.InthecaseofRUDICSmodem,dataareroutedthroughanethernettunnelmaintainedatUNAVCO.TheRUDICSmodemprovidesahigherbandwidthconnection,andthuscanpassthroughhigherGPSdatarates.Themaintradeoffisthetwo-foldincreaseincostfortheRUDICSmodem.Thedial-upandRUDICSIridiummodemsrequireaserverandautomatedscriptstoaccesstheGPSdataandreceiverstateofhealthinformation.Onceconnected,ascriptlooksforthelastsuccessfullyarchivedfile,andproceedstoretrieveallremainingwholedatafiles.Iftheconnectionisinterrupted,thedownloadscriptwillreattempttheretrievalonthenextconnection,whichoftenoccursmultipletimesperday.Enoughbandwidthoverheadismaintainedtocatchupondatathatmighthavebeenmissedordelayedduetotechnicalissueswiththesite,regardlessofthedurationofthecommunicationsoutage.
30
Rawdatainreceiver-specific,binarycompressedformatfilesdownloadedfromtheGPSreceiversthroughtheIridiumnetworkorADSLmodemsarepromptlyarchivedatUNAVCO.Additionally,therawdataareconvertedtoRINEX2.11andpostedasdailyfilesontheUNAVCOpublicftpserver,whichisaccessibleviaanonymousftp.DataqualitychecksareperformedwithTEQC[EsteyandMeertens,1999],andresultingsummaryfilesarearchivedalongwiththeobservationandnavigationfiles.OngoingdataprocessingisperformedbythePIteamatOhioStateUniversity,whichyieldspositionestimatesandotherancillaryinformation.UnliketheGeodesyAdvancingGeosciencesandEarthScope(GAGE)Facilityprocessingsystem,whichproducesdailyGNSSsolutionsfor>1700continuousstationsincludingtheNSF-fundedPlateBoundaryObservatory,TLALOCNet(Mexico),andCOCONet(panCaribbean),neitherdailypositionestimatesandtheirassociateduncertaintiesnorvelocityestimatesandtheiruncertaintiesareprovidedbyOSUtoUNAVCOforpublicationonitswebsite.IridiumserviceisprovidedatsubsidizedDepartmentofDefenseratestotheNSF.Iridiumcostsareapproximately$285/monthperdevice.Thiscostisnotcurrentlybasedondatausage,butinsteadisasetcosttoNSFthatisdividedequallyamongusers.TheDODratesarecurrentlyunderreviewandweanticipatechangesittheratestructureinthecomingmonths.
4 Bestconfigurationmovingforward
4.1 EvaluationofGNETstationsAnecessarydetailofthisreportistoestablishtherelativemeritsoftheGNETstationsastheyhaverecordedscientificdatasince2007.GiventhattheoriginalscientificgoalofGNETwastoretrievetheGIAsignalandimprovemodels,thismustbeoneofthemaincriteriaweusetowardaquantitativeevaluationofperformance.Additionally,asthedataclearlyhaveshown,apresent-daymassbalancecomponentispresentinthetimeseriesofcrustalmotionsrecordedateachofthestationsthatisusefulformakingconnectionstoSMBandD,orin-other-words,theongoingmassbalanceofcoastalGreenland[Jiangetal.,2010;Bevisetal.,2012].Inordertomakethiscomparisoninabothstraightforwardandintelligibleway,werelyheavilyonthreerecentanalysesofGIAandSMB-Dresponses.ApowerfulsourceofconstraintonGIAmodelscomesfromrelativesea-level(RSL)datasets,asthesesamplethecrustalandsea-levelresponsesatdifferenttimesduringtheGIAprocess.ArecentmodelofthistypeisthemodelbyLecavalieretal.[2014](seeFigure2a).AnalternativemodelforcomparisonisonethathasbeenheavilyinfluencesbytheGPSdatafromGNETbyKhanetal.[2016](Figure2b).WealsorelyontheZOIanalysisofAdhikarietal.[2017](Figure11).Weadoptthefollowingscoringtechniquefor40oftheGNETstationsandpresenttheseresultsonTable1.GIAisevaluatedintwoways.First,thestationhasanintrinsicrelativevalueinisolatingtheviscoelasticsignalbyitsproximitytocrustthatisbothfreefrompresent-dayicesheetchangesandnearplaceswherethepasticesheetadvancesoverland.Wetermthislattercategory“GIA-merit”.ForevaluationofastationsvalueinmeasuringSMB-D,weusetwocriteria:thelocationofthemappedZOI
31
(Figure11)andthelocationofthestationwithrespecttoannualaveragesofoutletglaciervelocity(seeFigure11,inset).Thirdly,astationmayalsobeshowntobeofvalueinhelpingtosortoutdifferencesinRSLvs.GNET-weightedGIAmodelsolutions(Figures2and3).Thelaterevaluationcategoryweterm“GIA-discrepancy”,andthescoringsystemisslightlydifferentthanthatforGIA-meritorSMB-D.Fortheformer,scoresareeither0or2,foreachstation,whilethelattertwocanscore0,1and3.Ascoreof3inthelattercasesmeansthatthestationlacksambiguityinitssensitivity,andtherefore,meritinevaluatingthesignaturesrelatedtoeitherGIAortoSMB-D.Table1.Scoresfor40oftheexistingGNETstations.A‘*’indicatesthatadditionalinformationmaybefoundinBevis
etal.[2012]orAdhikarietal.[2017].
Northwest GIA-merit SMB-D GIA-discrepancyHRGD 1 3 0KMOR 3 0 2KAGZ 3 0 2SCBY 3 3 2THU2 3 0 2DSKG 3 3 0
Northeast KMPJ 3 0 0NORD 3 1 0NRSK 3 0 0JBLG* 3 0 0LBIB 1 3 0DMHN 3 0 0GROK 1 3 0LEFN 1 3 0
EastCentral DANE 3 0 0HMBG 1 3 0MSVG 3 1 0DGJG 1 3 0SCOR 3 0 0VFDG 1 3 0KUAQ 3 3 2MIK2 3 0 2
SouthGreenland KSNB 3 1 0KELY 3 0 0KULU 3 1 0KBUG 3 1 2KAPI 3 0 2HJOR 3 0 2TIMM 3 0 2NNVN 3 1 0SENU 3 1 0NUUK* 3 0 2
WestCentralGreenland KULL 3 0 2UPVK 1 3 2
32
RINK 1 3 0QEQE 3 0 0AASI* 3 0 0ILUL* 3 1 0SRMP* 1 3 2KAGA 1 3 0
ForinterpretingthescoresshowninTable1,itisadvisedthatatotaladdedscoreisinappropriatetoemployforevaluation.Forexample,Adhikarietal.[2017]showedthemeritsofthesensitivityoftheRINKstationtoSMB-Dwithveryhighfidelityduringextrememeltyears,whereas,stationHMBGwouldhaveidenticaltotalscore,eventhoughnodemonstrationofhorizontalcrustalmotionsensitivityhasyetbeenfullycarriedout.Also,itmightbetemptingtoaddthetwoGIArelatedscores,butsomedifferencescouldbeduetolateralheterogeneityinmantleviscosity,orsomeotherpartoftheicehistory,thatiscurrentlyunderdevelopment[GlennMilne,personalcommunication,July2017].Obviously,thistableismosthelpfuloncedecisionsaremadeastothepreferredapproachtothefutureofGNET:toisolateSMB-Dononehand,orGIAontheother.TheultimatedecisionregardingtherelativeweightingofthescoresinTable1shouldbeguidedbytheacceptanceofviablescienceprojectsselectedbytheNationalScienceFoundation.
4.2 AddingstationstotheexistingnetworkManydisciplineswouldbenefitfromtheadditionofstationstothenetwork,buttheadditionwouldbetargeted.TheGlacialIsostaticAdjustmentcommunityhaspointedoutahandfulofkeylocationswheretheadditionofonlyafewnewstationswouldhelpconstrainGIAestimates.TheDynamicMassBalance(iceflow)communitywouldalsosuggesttargeteddensificationofthenetwork,butdrivenbyspecificsciencegoalsandlikelytargetingspecifickeyoutletglaciers.Thetroposphericandionosphericcommunitieswouldalsobenefitfromadensernetworkalongthecoast.Tobenefitreflectometrystudiesofsealevel,newstationswouldneedtobeclosertotheoceanthanexistingstations,allowingmultipathreflectionsfromtheoceansurface.Anotherdesireexpressedbymultiplegroupswastohavestationsaddedontheinlandice.Boththereflectometrycommunity(forstudiesofaccumulationandfirndensification)andthetropospheric/ionosphericcommunityexpressedthisdesire.Suchstationswouldposeuniquechallenges,dependingonwheretheywereplacedontheicesurface.Inparticular,stabilitywouldbeasignificantproblem.Intheablationzone,icewouldmeltoutfromunderneathanchorsandmonuments,increasingthechallenge-thishasbeenhandledwithvaryingdegreesofsuccessbyshorter-termcampaigngroups,butmaintainingalong-termstationundersuchconditionswouldbedifficult.Intheaccumulationzonetheproblemissimpler,butnolessaproblemtobesolved-inparticulartheantennawouldneedtoberaisedperiodicallytokeepaheadoftheaccumulatingsnow,andonallpartsoftheicesheet,theiceisflowing,resultinginsignificantchangesinthepositionofastationoverthecourseofevenoneyear.Thesechallengesaresignificantbutarenotimpossibletoovercome.
33
4.3 Detailedsitecharacterizationofexistingsites,additionalmeasurementsSeveralgroupsexpressedadesireforamoredetailedsitecharacterizationofexisting(andfuture)GNETsites.Thesitecharacterizationsrangedfromdetailedtopographicalsurveyingviaterrestriallidarorstructure-from-motionphotography,toadditionalatmosphericandgeophysicalmeasurements.Gravitymeasurementsexistatsomesites,butexpandedmeasurementsatotherGNETsitesisdesirable.Finally,agreaterunderstandingofthemeteorologicalconditionsatthesiteswasdesired.Thiscouldtaketheformofanythingfromafull-scaleweatherstation(unlikely)toamicro-meteorologicalstation,someofwhichcanplugdirectlyintotheGPSloggerpoweranddatastreams.
4.4 ExpandedskycoverageanddataaccessforexistingsitesThereareseveralimprovementsthatcanbemadetothenetworkwithouttheneedforadditionalstationsormeasurementsatexistingstations,butwhichwilladdvaluetothedataformultipleusers.ManyofthegroupsexpressedthatlimitingthedatacollectiontotheGPSconstellationwasaproblem,andwantedGLONASSandGALLILEOcoverageaswell,makingthenetworktrulyGNSSandnotonlyGPS.Manyapplicationsincludingspaceweather,geodesy,andothersneedaccesstodatainnear-real-timeforittobeuseful,sorapidaccesstodataisessential.Inadditiontofastaccesstothedata,manyapplicationswouldbenefitfromhigher-ratedatathanwhatiscurrentlyavailablethroughGNET.Finally,traditionalGPSusagerequiresthatthesignalpathbetweensatelliteandgroundstationbeasinglestraightline,andanyinterferencewiththisisaproblem.Forthisreason,itiscommontolimitdatacollectiontoonlysatellitesthatareaboveacertainangleinthesky,reducingtheriskofmultipathreflections.Foruserswhoutilizethemultipathreflectionsforadded-valuedataproducts,however,theselow-anglemeasurementsareessential.Thus,removingthemaskingforlow-angledataisdesirable.
34
5 REFERENCES:
Adhikari,S.andE.R.Ivins,(2016),Climate-drivenpolarmotion:2003–2015,ScienceAdv.,doi:10.1126/sciadv.1501693.
Adhikari,S.,E.R.Ivins,andE.Larour,(2016),ISSM-SESAWv1.0:mesh-based
computationofgravitationallyconsistentsea-levelandgeodeticsignaturescausedbycryosphereandclimatedrivenmasschange,Geosci.ModelDev.,9,1087–1109,doi:10.5194/gmd-9-1087-201.
Adhikari,S.,E.R.IvinsandE.Larour(2017),Masstransportwavesamplifiedbyintense
GreenlandmeltanddetectedinsolidEarthdeformation,Geophys.Res.Letters,44,doi:10.1002/2017GL073478.
Ahlstrøm,A.P.,S.B.Andersen,M.L.Andersen,H.Machguth,F.M.Nick,I.Joughin,C.H.
Reijmer,R.vandeWal,J.P.Boncori,andJ.E.Box(2013),Seasonalvelocitiesofeightmajormarine-terminatingoutletglaciersoftheGreenlandicesheetfromcontinuousinsituGPSinstrument,EarthSyst.Sci.Data,5(2),doi:10.5194/essd-5-277-2013.
Alexander,P.M.,M.Tedesco,N.-J.Schlegel,S.B.Luthcke,X.Fettweis,andE.Larour,
(2016),GreenlandIceSheetseasonalandspatialmassvariabilityfrommodelsimulationsandGRACE(2003–2012),TheCryosphere,10,1259-1277,doi:10.5194/tc-10-1259-2016.
Andersen,M.L.,etal.,(2010),SpatialandtemporalmeltvariabilityatHelheimGlacier,
EastGreenland,anditseffectonicedynamics,J.Geophys.Res.,115,F04041,doi:10.1029/2010JF001760.
Basu,S.etal.,(2001),Ionosphericeffectsofmajormagneticstormsduringthe
InternationalSpaceWeatherPeriodofSeptemberandOctober1999:GPSobservations,VHF/UHFscintillations,andinsitudensitystructuresatmiddleandequatoriallatitudes,J.Geophys.Res.Sp.Phys.,106(A12),30,389–30,413,doi:10.1029/2001JA001116.
BauerP.,A.ThorpeandG.Brunet,(2015),Thequietrevolutionofnumericalweather
prediction,Nature,525,47-55.Bennartz,R.,M.D.Shupe,D.D.Turner,V.P.Walden,K.Steffen,C.J.Cox,M.S.Kulie,N.
B.Miller,andC.Pettersen,(2013),July2012Greenlandmeltextentenhancedbylow-levelliquidclouds,Nature,496(7443),83–86,doi:10.1038/nature12002.
Bennitt,G.V.andA.Jupp,(2012),OperationalassimilationofGPSzenithtotaldelay
observationsintotheMetOfficeNumericalWeatherPredictionModels,MonthlyWeatherReview,140,2706-2719,doi:10.1175/MWR-d-11-00156.1.
35
Bevis,M.,etal.,(2012),BedrockdisplacementsonGreenlanddrivenbyicemassvariations,climatecyclesandclimatechange,Proc.Nat.Acad.Sci.,109,11,944-11,948, doi/10.1073/pnas.1204664109.Bevis,M.,S.Businger,T.A.Herring,C.Rocken,R.A.Anthes,andR.H.Ware,(1992),GPS
Meteorology:RemotesensingofatmosphericwatervaporusingtheGlobalPositioningSystem,J.Geophys.Res.,Atmos,97,15,715-15,801.
Bevan,S.L.,A.Luckman,S.A.KahnandT.Murray,(2015),Seasonaldynamicthinningat
HelheimGlacier,EarthPlanetarySci.Lett.,415,47–53,doi:10.1016/j.epsl.2015.01.031.
Bos,M.S.,N.T.Penna,T.F.Baker,andP.J.Clarke,(2015),Oceantideloading
displacementsinwesternEurope:2.GPS-observedanelasticdispersionintheasthenosphere,J.Geophys.Res.SolidEarth,120,6540–6557,doi:10.1002/2015JB011884.
Calais,E.,andJ.B.Minster,(1998),GPS,earthquakes,theionosphere,andtheSpace
Shuttle,Phys.EarthPlanet.Inter.,105(3–4),167–181.Christensen,O.B.,M.Drews,J.H.Christensen,K.Dethloff,K.Ketelsen,I.Hebestadtand
A.Rinke,2007:TheHIRHAMRegionalClimateModelVersion5(beta).TechnicalReport06-17,DanishMeteorologicalInstitute,Copenhagen.
Coster,A.J.,J.C.Foster,P.J.Erickson,(2003),MonitoringtheIonospherewithGPS,GPS
World, Vol.l4, No.5,42-45.Coster,A.andA.Komjathy,(2008),SpaceWeatherandtheGlobalPositioningSystem,
SpaceWeather,6,doi:10.1029/2008SW000400.Crawford,A.andM.Serreze,(2015),AnewlookatthesummerArcticFrontalZone,J.
Climate,28,737-753,doi:10.1175/JCLI-D-14-00447.1Das,S.B.,Joughin,I.,Behn,M.D.,Howat,I.M.,King,M.A.,Lizarralde,D.,andBhatia,
M.P.,(2008),FracturepropagationtothebaseoftheGreenlandicesheetduringsupraglaciallakedrainage,Science,320,778–781.
DeJuan,J.,P.Elósegui,M.Nettles,T.B.Larsen,J.L.Davis,G.S.Hamilton,L.A.Stearns,
M.L.Andersen,G.Ekström,A.P.Ahlstrøm,S.A.KhanandR.Forsberg,(2010),SuddenincreaseintidalresponselinkedtocalvingandaccelerationatalargeGreenlandoutletglacier.Geophys.Res.Lett.,37,L12501,doi:10.1029/2010GL043289.
Dietrich,R.,H.-G.Maas,M.Baessler,A.Rülke,A.Richter,E.Schwalbe,andP.Westfeld,
(2007),JakobshavnIsbræ,WestGreenland:Flowvelocitiesandtidalinteractionof
36
thefrontareafrom2004fieldobservations,J.Geophys.Res.,Surface112,F03S21,doi:10.1029/2006JF000601.
Durgonics,T.,A.Komjathy,O.Verkhoglyadova,E.B.Shume,H.-H.Benzon,A.J.Mannucci,M.D.Butala,P.Høeg,andR.B.Langley,(2017),Multi-instrumentobservationsofageomagneticstormanditseffectsontheArcticionosphere:Acasestudyofthe19February2014storm,RadioSci.,52(1),146–165,doi:10.1002/2016RS006106.
Enderlin,E.M.,I.M.Howat,S.Jeong,M.-J.Noh,J.H.vanAngelenandM.R.vandenBroeke,(2014)AnimprovedmassbudgetfortheGreenlandicesheet,Geophys.Res.Lett.,41,866–872,doi:10.1002/2013GL059010.
Ekman,M.,(1991)Aconcisehistoryofpostglaciallandupliftresearch(fromitsbeginningto1950),TerraNova,3,358-365,doi:10.1111/j.1365-3121.1991.tb00163.x.
EsteyL.H.andC.M.Meertens(1999)TEQC:Themulti-purposetoolkitforGPS/GLONASSdata,GPSSolutions,3,42-49,doi:10.1007/PL00012778
Fettweis,X.,M.Tedesco,M.vandenBroeke,andJ.Ettema(2011),Meltingtrendsover
theGreenlandicesheet(1958–2009)fromspacebornemicrowavedataandregionalclimatemodels,TheCryosphere,5,359–375,doi:10.5194/tc-5-359-2011.
Goncharenko,L.P.,A.J.Coster,J.L.Chau,andC.E.Valladares, (2010),Impactof
suddenstratosphericwarmingsonequatorialionizationanomaly,J.Geophys.Res.,115,AOOG07,doi:10.1029/20lOJAO15400.
Hanna,E.,F.Navarro,F.Pattyn,C.M.Domingues,X.Fettweis,E.R.Ivins,R.J.Nicholls,C.
Ritz,B.Smith,S.Tulaczyk,P.L.Whitehouse&H.J.Zwally,(2013),Icesheetmassbalanceandclimatechange,Nature,497,51-59,doi:10.1038/nature12238,2013.
Heki,K.,(2004),inTheStateofthePlanet:FrontiersandChallengesinGeophysics,150
editedbyR.S.J.SparksandC.J.Hawkesworth,pp.177–196.Heki,K.,(2006),Explosionenergyofthe2004eruptionoftheAsamaVolcano,central
Japan,inferredfromionosphericdisturbances,Geophys.Res.Lett.,33,L14303,doi:10.1029/2006GL026249.
Heki,K.,(2004),DenseGPSarrayasanewsensorofseasonalchangesofsurfaceloads.
inTheStateofthePlanet:FrontiersandChallengesinGeophysics(editedbySparks,R.S.J.,andHawkesworth,C.J.),177-196.
Herring,T.A.,T.I.Melbourne,M.H.Murray,M.A.Floyd,W.M.Szeliga,R.W.King,D.A.
Phillips,C.M.Puskas,M.Santillan,andL.Wang,(2016),PlateBoundary
37
Observatoryandrelatednetworks:GPSdataanalysismethodsandgeodeticproducts,Rev.Geophys.,54,759–808,doi:10.1002/2016RG000529.
Hewitt,I.J.,andA.C.Fowler,(2008),Seasonalwavesonglaciers.Hydrol.Process.,22,
4653919-3930. Holland,D.M.,D.Voytenko,K.Christianson,T.H.Dixon,M.J.Mel,B.R.Parizek,I.
Vaňková,R.T.Walker,J.I.Walter,K.NichollsandD.Holland.(2016),AnintensiveobservationofcalvingatHelheimGlacier,EastGreenland,Oceanography,29,46-61.
Hurkmans,R.T.W.L.,J.L.Bamber,C.H.Davis,I.R.Joughin,K.S.Khvorostovsky,B.S.
Smith,andN.Schoen(2014),Time-evolvingmasslossoftheGreenlandIceSheetfromsatellitealtimetry,TheCryosphere,8,1725–1740,doi:10.5194/tc-8-1725-2014.
Jiang,Y.,T.H.DixonandS.Wdowinski(2010),AcceleratingupliftintheNorthAtlantic
regionandanindicatoroficeloss.NatureGeosci.,3,404–407.Joughin,I.,S.B.Das,G.E.Flowers,M.D.Behn,R.B.Alley,M.A.King,B.E.Smith,J.L.
Bamber,M.R.vandenBroeke,andJ.H.vanAngelen.(2013),Influenceofice-sheetgeometryandsupraglaciallakesonseasonalice-flowvariability,TheCryosphere,7,1185-1192,doi:10.5194/tc-7-1185-2013.
Khan,S.A.,I.Sasgen,M.Bevis,T.vanDam,J.Bamber,J.Wahr,M.Willis,K.Kjaer,B.
Wouters,V.Helm,B.Csatho,K.Fleming,A.Bjork,A.Aschwanden,P.Knudsen,P.KulpersMunneke,(2016),Similaritiesbetweenpost-LastGlacialMaximumandpresent-daymasslossfromtheGreenlandicesheet,ScienceAdvances,2,e1600931.
Khan,S.A.,L.Liu,J.Wahr,I.Howat,I.Joughin,T.vanDam,andK.Fleming,(2010),GPS
measurementsofcrustalupliftnearJakobshavnIsbræduetoglacialicemassloss,J.Geophys.Res.,SolidEarth,115,B09405,doi:10.1029/2010JB007490.
Khan,S.,J.Wahr,M.Bevis,I.VelicognaandE.Kendrick,(2010),Spreadoficemassloss
intonorthwestGreenlandob-servedbyGRACEandGPS,Geophys.Res.Lett.,37,L06501,doi:10.1029/2010GL042460.
Komjathy,A.,Y.-M.Yang,X.Meng,O.Verkhoglyadova,A.J.Mannucci,andR.B.Langley,
(2016),Reviewandperspectives:Understandingnatural-hazards-generatedionosphericperturbationsusingGPSmeasurementsandcoupledmodeling,RadioSci.,51,doi:10.1002/2015RS005910.
Kuchar,J.andG.A.Milne,(2015),Theinfluenceofviscositystructureinthelithosphere
onpredictionsfrommodelsofglacialisostaticadjustment,J.Geodynamics,86,1–9,doi: 10.1016/j.jog.2015.01.002.
38
Larour,E.,J.Utke,A.Bovin,M.Morlighem,andG.Perez,(2016)AnapproachtocomputingdiscreteadjointsforMPI-parallelizedmodelsappliedtoIceSheetSystemModel4.11,Geosci.ModelDev.,9,doi:10.5194/gmd93907.
Larson,K.M.,E.D.Gutmann,V.U.Zavorotny,J.J.Braun,M.W.Williams,andF.G.
Nievinski,(2009),CanwemeasuresnowdepthwithGPSreceivers?,Geophys.Res.Lett.,36(17),L17502,doi:10.1029/2009GL039430.
Lay,E.H.,X.-M.Shao,andC.S.Carrano,(2013),Variationintotalelectroncontent
abovelargethunderstorms,Geophys.Res.Lett.,40,1945–1949,doi:10.1002/grl.50499.
Lecavalier,B.S.,etal.,(2014),AmodelofGreenlandicesheetdeglaciationconstrained
byobservationsofrelativesealevelandiceextent,QuaternarySci.Rev.,102,54—84,doi:10.1016/j.quascirev.2014.07.018
Liu,L.,S.A.Khan,T.vanDam,J.H.Y.Ma,andM.Bevis,(2017),Annualvariationsin
GPS-measuredverticaldisplacementsnearUpernavikIsstrøm(Greenland)andcontributionsfromsurfacemassloading,J.Geophys.Res.,SolidEarth,122,677–691,doi:10.1002/2016JB013494.
Misra,P.andP.Enge,(2011)GlobalPositioningSystem:Signals,Measurementsand
Performance,Ganga-JamunaPress,ed.2.Mahfouf,J.-F.,F.Ahmed,P.Moll,andF.N.Teferle,(2015)Assimilationofzenithtotal
delaysintheAROMEFranceconvectivescalemodel:arecentassessment.TellusA,67,26106,doi:10.3402/tellusa.v67.26106.
Moon,T.,I.Joughin,B.Smith,M.R.vandenBroeke,W.J.vandeBerg,B.Noël,andM.
Usher(2014),DistinctpatternsofseasonalGreenlandglaciervelocity,Geophys.Res.Lett.,41,7209–7216,doi:10.1002/2014GL061836.
Murray,T.,M.Nettles,N.Selmes,L.M.Cathles,J.C.Burton,T.D.James,S.Edwards,I.
Martin,T.O’Farrell,R.Aspey,I.Rutt,andT.Baugé,(2015)Reverseglaciermotionduringicebergcalvingandthecauseofglacialearthquakes,Science,349,305–308,doi:10.1126/science.aab0460.
Nievinski,F.G.andK.M.Larson,(2014),Inversemodelingofgpsmultipathforsnow
depthestimation,parti:Formulationandsimulations.IEEETrans.Geosci.RemoteSensing,52,6555–6563.
Noël,B.,W.J.vandeBerg,S.Lhermitte,B.Wouters,H.Machguth,I.Howat,M.Citterio,
G.Moholdt,J.T.M.LenaertsandM.R.vandenBroeke,(2017),AtippingpointinrefreezingacceleratesmasslossofGreenland’sglaciersandicecaps,NatureCommunications,8,(14730),doi:10.1038/ncomms14730.
39
Palmer,S.,A.Shepherd,P.Nienow,andI.Joughin,(2011),SeasonalspeedupoftheGreenlandIceSheetlinkedtoroutingofsurfacewater.EarthPlanet.Sci.Lett.,302,423-428.
Peltier,W.R.(1998),Postglacialvariationsinthelevelofthesea:Implicationsfor
climatedynamicsandsolid-Earthgeophysics,Rev.Geophys.,36,603—689,doi:10.1029/98RG02638.
Pimentel,S.andG.E.Flowers,(2011),Anumericalstudyofhydrologicallydriven
glacierdynamicsandsubglacialflooding,Proc.Roy.Soc.A,467,537–558,doi:10.1098/rspa.2010.0211.
Robasky,F.M.,andD.H.Bromwich,(1994)Greenlandprecipitationestimatesfromthe
atmosphericmoisturebudget.Geophys.Res.Let.,21,2495-2498,doi:10.1029/94GL01915.
Rogozhina,I.,J.M.Hagedoorn,Z.Martinec,K.Fleming,O.Soucek,R.Greve,andM.
Thomas,(2012),EffectsofuncertaintiesinthegeothermalheatfluxdistributionontheGreenlandIceSheet:Anassessmentofexistingheatflowmodels,J.Geophys.Res.,Surface117,F02025,doi:10.1029/2011JF002098.
Rosenau,R.,E.Schwalbe,H.-G.Maas,M.Baessler,andR.Dietrich(2013),Grounding
linemigrationandhigh-resolutioncalvingdynamicsofJakobshavnIsbræ,WestGreenland,J.Geophys.Res.EarthSurf.,118,382–395,doi:10.1029/2012JF002515.
Sauber,J.,G.Plafker,B.F.MolniaandM.A.Bryant,(2000),Crustaldeformation
associatedwithglacialfluctuationsintheeasternChugachMountains,Alaska,J.Geophys.Res.,SolidEarth,105,8055-807.
Schlegel,N-J.,E.Larour,H.Seroussi,M.Morlighem,andJ.E.Box,(2013),Decadal-scale
sensitivityofnortheastGreenlandiceflowtoerrorsinsurfacemassbalanceusingISSM,J.Geophys.Res.-EarthSurf.,118,doi:10.1002/jgrf.20062.
Schlegel,N.-J.,D.N.Wiese,E.Y.Larour,M.M.Watkins,J.E.Box,X.FettweisandM.R.van
denBroeke,(2016)ApplicationofGRACEtotheevaluationofaniceflowmodeloftheGreenlandIceSheet,TheCryosphere,10,1965-1989,doi:10.5194/tc-10-1965-2016.
Shepherd,A.,etal.,(2012),Areconciledestimateoficesheetmassbalance,Science,
338,1183-1189.Sole,A.J.,D.W.F.Mair,P.W.Nienow,I.D.Bartholomew,M.A.King,M.J.Burke,andI.
R.Joughin(2011),SeasonalspeedupofaGreenlandmarine-terminatingoutletglacierforcedbysurfacemelt–inducedchangesinsubglacialhydrology,J.Geophys.Res.,-EarthSurface,116,doi:10.1029/2010JF001948.
40
Tedesco,M.,Alexander,P.,Bell,R.,BriggsK.,DasI.,MacFerrin,M.,Hanna,E.,Koenig,L.,Overeem,I.,Rennermalm,A.,(2016),UnderstandingthefundamentalprocessescontrollingthesurfacemassbalanceoftheGreenlandicesheetandimprovingestimates.WorkshopReport,LDEO-ColumbiaUniversity.
Tedstone,A.J.,P.W.Nienow,N.Gourmelen,A.Dehecq,D.GoldbergandE.Hanna
(2015)Decadalslowdownofaland-terminatingsectoroftheGreenlandIceSheetdespitewarming,Nature,526,692-697,doi:10.1038/nature15722.
Tromp,J.,C.Tape,andQ.Liu,(2005),Seismictomography,adjointmethods,time
reversalandbanana-doughnutkernels,Geophys.J.Int.,160,195-216.VanDam,T.M.,J.Wahr,Y.Chao,andE.Leuliette,(1997),Predictionsofcrustal
deformationandofgeoidandsea-levelvariabilitycausedbyoceanicandatmosphericloading,Geophys.J.Int,99,507-517.
vandeWal,R.S.W.,W.Boot,M.R.vandenBroeke,C.J.P.P.Smeets,C.H.Reijmer,J.J.
A.Donker,andJ.Oerlemans(2008),Largeandrapidmelt-inducedvelocitychangesintheablationzoneoftheGreenlandIceSheet,Science,321,2008–2011,doi:10.1126/science.1158540.
vandenBroeke,M.R.,E.M.Enderlin,I.M.Howat,P.K.Munneke,B.P.Y.Noël,W.J.van
deBerg,E.vanMeijgaard,andB.Wouters,(2016),OntherecentcontributionoftheGreenlandicesheettosealevelchange,TheCryosphere,10,1933–1946,doi:10.5194/tc-10-1933-2016.
Wahr,J.,S.A.Khan,T.vanDam,L.Liu,J.H.vanAngelen,M.R.vandenBroeke,andC.M.
Meertens,(2013),TheuseofGPShorizontalsforloadingstudies,withapplicationstonorthernCaliforniaandsoutheastGreenland,J.Geophys.Res.SolidEarth,118,doi:10.1002/jgrb.50104.
YangQ.,S.Wdowinski,andT.H.Dixon,(2013),Annualvariationofcoastalupliftin
Greenlandasanindicatorofvariableandacceleratingicemassloss,Geochem.Geophys.Geosyst.,14,1569–1589,doi:10.1002/ggge.20089.