An evaluation of pH and NO3 sensor data from SOCCOM floats and
Transcription
An evaluation of pH and NO3 sensor data from SOCCOM floats and
AnevaluationofpHandNO3sensordatafromSOCCOMfloatsandtheirutilizationto developoceaninorganiccarbonproducts AsummaryofdiscussionsandrecommendationsoftheCarbonWorkingGroup(CWG)of theSouthernOceanCarbonandClimateObservationsandModelingproject(SOCCOM) March2016 AnevaluationofpHandNO3sensordatafromSOCCOMfloatsandtheirutilizationto developoceaninorganiccarbonproducts R.Wanninkhof1,K.Johnson2,N.Williams3,J.Sarmiento4,S.Riser5,E.Briggs6,S.Bushinsky4, B.Carter5,7,A.Dickson6,R.Feely7,A.Gray4,L.Juranek3,R.Key4,L.Talley6,J.Russel8,andA. Verdy6 1.AtlanticOceanographicandMeteorologicalLaboratoryofNOAA,MiamiFL 2.MontereyBayAquariumResearchInstitute,MossLandingCA 3.OregonStateUniversity,CorvallisOR 4.PrincetonUniversity,PrincetonNJ 5.UniversityofWashington,Seattle,WA 6.ScrippsInstitutionofOceanography,UCSD,LaJollaCA 7.PacificMarineEnvironmentalLaboratoryofNOAA,SeattleWA 8.UniversityofArizona,TucsonAZ Tableofcontent: Page Recommendationsandfindings 3 1.Introduction 4 2.Biogeochemicalsensorsonfloats 5 Typesoffloats 6 DetailsofsensorsonSOCCOMfloats 7 Frequencyofprofilingmeasurements 10 3.Performanceofsensors 12 4.Sensordeployments,calibration,validationandchecks 12 5.Recommendedconstantsforinorganiccarbonsystemparameters 13 6.pHcomparisonsbetweenbottleandfloatdata 16 7.UtilizingmultiplelinearregressionstoadjustpHandNO3sensorsonfloats18 8.Datamanagementandqualitycontrol/adjustments 20 9.Productdevelopment 21 10.Outlook 23 AppendixA.Chargeforthecarbonworkinggroup(CWG)ofSOCCOM 25 References 28 2 AnevaluationofpHandNO3sensordatafromSOCCOMfloatsandtheirutilizationto developoceaninorganiccarbonproducts Recommendationsandfindings: TheCarbonWorkingGroup(CWG)oftheSouthernOceanCarbonandClimate ObservationsandModeling(SOCCOM)projectistaskedwithprovidingrecommendations foracquisitionandqualitycontrolofdatafrombiogeochemicalsensorsonSOCCOMfloats: CWGFindings Ø TheNO3andpHsensorsonfloatsareindevelopmentstagebutrapidlyreachingthe pointofprovidingqualitydatatostudykeybiogeochemicalprocessesinthe SouthernOcean. Ø Thereisanimprovedunderstandingandremediesofissuescausingdriftinthe biogeochemicalsensors. Ø ThetransitionfromanadhocadjustmentprocedureofpHandNO3toasystematic approachhandledbydatacentersiswelllaidoutandprogressing. Ø TheaccuraciesoffloatO2,NO3andpHdataafterappropriateadjustmentsare2 µmolkg-1,0.5µmolkg-1,and0.01,respectively. Ø NO3andpHdatafromtheprofilingfloatsinSOCCOMcanbeadjustedasneeded usingMultiLinearRegression(MLR)algorithmsbasedonqualitybottledatafrom 1000-2100dbarintheregion(seeeqns.3and4)towithinthelevelofcurrent instrumentaluncertainties. CWGRecommendations Ø Allsamplesbroughtbacktoshorefromfloatdeploymentvalidationcastsshouldbe analyzedfortotaldissolvedinorganiccarbon,DIC,inadditiontopHandtotal alkalinity,TAlk. Ø Continuedacquisitionandassemblyofhighqualityinorganiccarbondatafrom CTD/bottlecastsiscriticalforqualitycontroloffloatsensors,modelvalidationand algorithmdevelopment/improvement. Ø Aconsistentsetofdissociationconstantsandachemicalmodelshouldbeusedfor calculatinginorganiccarbonsystemparametersasdescribedinTable4. Ø ItisimperativethatthetemperatureandpressuredependenceofpHsensors,and theircorrespondencetotheappropriatepHscalesisfurtherstudiedindetail. Ø Continuedstudyandimprovementofadjustmentapproachesandincorporationof theseeffortsintoaroutineapproachfollowingArgoguidelinesisneeded. Ø Continuedstudyandimprovementindeploymentprocedures,physical configurationofsensors,sensordesignandqualitycontrolisneeded. Ø AccommodationisneededforupdatedprotocolsinArgotoprocessthe biogeochemicalandbiologicaldatausingseparatefilesanddataacquisitioncenters priortomergingofphysicalandbiogeochemicaldata. 3 1.Introduction i.Purposeofdocument TheSouthernOceanCarbonandClimateObservationsandModeling(SOCCOM)projectis revolutionizingoureffortstoinvestigatetheSouthernOcean.Thenewapproaches,in particulartheutilizationofbiogeochemicalsensorsonfloats,requireacarefuluncertainty analysisandclearprocedurestocreateusefulproducts.Thecarbonsystemworkinggroup (CWG)ofSOCCOMischargedtoaddresstheseissueswithregardstothemeasuredand derivedinorganiccarbonsystemparameters:totalalkalinity(TAlk),totaldissolved inorganiccarbon,(DIC),partialpressureofCO2(pCO2),andpH.Theonlyinorganiccarbon systemparameteracquiredfromtheSOCCOMfloatsispH,theothersarecalculated throughdifferentmeansasdescribedbelow.Thisdocumentisanauthoritativeoverviewof procedurestoadjustsensorsforoffsetsanddrift,andtouseofthefloatdatatocreate products,suchasregionalfieldsofinorganiccarbonparameters. ii.TheCWGandSOCCOMobjectives ThecarbonworkinggroupactivitiesarepartoftheoverallobjectivesofSOCCOM.SOCCOM isfocusedonunlockingthemysteriesoftheSouthernOceananditsinfluenceonclimate. SOCCOMisawell-structuredprogramwiththreemajorthemesofobservation,modeling andbroaderimpacts.TheCWGaddressestheobservationthemewithfocusoncreating fieldsofcarbonsystemparameters,suchascarbonatemineralsaturationstates.Afull descriptionofthechargeoftheCWGisprovidedinAppendixA.Therecommendationsand implementationwillalsobenefitthecarboncyclemodelsoftheSouthernOceanthrough improvedconstraints.TheCWGwillhavesignificantbroaderimpactsastheprocedures andapproachesdescribedcanbeutilizedtoimprovethequalityandproductsresulting fromfloatsequippedwithbiogeochemicalsensorsocean-wide. iii.Outlineofdocument Thisdocumentprovidesrecommendationsandguidelinestoadjustandcheckdatafrom biogeochemicalsensorsonprofilingfloats,inparticularpHandNO3sensorsbasedon empiricalapproachesandmodels.Theadjusteddataalongwithmodelsofdifferentlevels ofsophisticationareusedtodeterminethelargerscalefieldsofthebiogeochemical parametersofinterest. Thefirstsectionofthedocumentdescribesthesensorsonfloatsandtheirpre-cruise calibrationandtypicalbehaviorwhenfirstdeployedwithfocusontheNO3andpHsensors. ThisisfollowedbyadescriptionofthepHmeasurementandthederivationofcarbon systemparametersandtheiruncertainties.AdjustmentproceduresoftheNO3andpH sensorsaredescribedbasedonmultiplelinearregression(MLR)algorithmsderivedfrom thevalidationcruisesandotherhighqualitybottledata.Thefinalsectiontouchesuponthe carbonproductsthatwillbeproducedaspartoftheSOCCOMprojectusingdifferent empiricalapproaches. 4 2.Biogeochemicalsensorsonfloats TheArgoprofilingfloatprogramstartedintheearly1990'swithanambitiousgoalof deployingaglobalarrayoffloatsequippedwithtemperature(T),salinity(S),andpressure (P)sensors.Itsprimaryobjectivewastodeterminechangesinoceanheatcontentinthe upper1500m.TheoverallvisionofArgoistogreatlyexpandourknowledgeofchangesin theoceanwith"greatlyimprovedcollectionofobservationscapabilitiesinsidetheocean throughincreasedsamplingofoldandnewquantitiesandincreasedcoverageintermsof timeandarea"1.Thecurrentprogramhaswell-definedobjectivesalongwithgood dissemination,qualitycontrol,anddatamanagementprotocols,anddatafromtheArgo programisnowusedinoverscientific200publicationsperyear.ThegreatsuccessofArgo, theurgentneedtostudytrendsinothervariablesinachangingocean,andrapidadvances insensortechnologyhaveinspiredresearcherstoincorporatenovelsensorsontothe profilingfloatplatforms.Oxygensensorswerethefirsttobeusedextensively(Körtzinger etal.2004)andarenowdeployedonabout5%oftheglobalfloatarray(seeFigure1for floatpositions).Thesensorsarewellbehavedandwithproperprecautionsyielddata believedtobeaccuratetowithin1%(≈2µmolkg-1).Oxygendataprocessingprotocolsare closetoroutinebutArgobiogeochemicalsensordatastillfollowadifferentqualitycontrol paththancoreArgodata. ThebiogeochemicalsensorsofprimaryinteresttothisdiscussionareNO3andpH.Wealso considerpropertiesthatareestimatedorcalculatedfromtheseandotherArgo measurements(e.g.aragonitesaturationΩA,TAlk,andDIC).TheO2sensorsarealso discussedasO2dataareusedextensivelytoestimatethecarbonsystemparameters.The O2sensorsserveasanexampleofacomparativelymatureefforttodeploynovel biogeochemicalsensorsontheArgoplatform. 1http://www.argo.ucsd.edu 5 Figure1.LocationsofallfloatsreportingthroughtheWorldMeteorologicalOrganization’s GlobalTelecommunicationsSystem(GMS)forJanuary20162(top)andfloatsequipped biogeochemicalsensorsasofSeptember20153(bottom) Typesoffloats FivetypesoffloatsarecommonlydeployedintheArgoprogram(APEX,Navis,SOLOII, Provor,andNOVA).SOCCOMexclusivelyusesAPEXandSea-BirdNAVISfloatsattimeof writing.TheNAVISfloatswithbiogeochemicalsensorsareastandardproductofSeabird Electronicsandcanbepurchasedwithuserspecifiedconfigurations.Theyareinthelate stagesofdevelopmentandnotyetconsideredfullymature(asoftheendof2015).The TeledyneWebbResearch(TWR)APEXfloatsaredeployedbytheUniversityofWashington (UW)andaretheprimaryplatformfordeployingthebiogeochemicalsensorsdescribed below.However,theAPEXfloatwiththeBGCsensorsusedinSOCCOMisnotastandard productofTWRandisnotcommerciallyavailable.SOLOIIfloatsareextensivelydeployed byWHOIandSIOforthecoreArgoarray,buttheyhavenotyetbeenadaptedtocarrya varietyofBGCsensors.Theplatformshavenoappreciableeffectonthedatafromthe biogeochemicalsensors,otherthanthatsomefloatshavepH,NitrateandO2sensorsina pumpedwatercircuit4.Thismayimpactsensoroperation.Oxygensensorsinthepumped streamprecludeO2aircalibrationsandanecdotalevidencesuggestsmorefoulingissues withthenitratesensorswhentheyareinapumpedloop. 2http://www.argo.ucsd.edu/statusbig.gif 3http://argo.jcommops.org/maps.html 4Theplacementinapumpedwatercircuitistominimizefouling.However,someresults suggestthat,tothecontrary,thisisnotoptimalforthepHandNO3sensors.Theyarebeing repositionedtohavedirectexposuretoseawater. 6 DetailsofsensorsonSOCCOMfloats5 Temperature(T),Pressure(P),andconductivity/"salinity"(C/S)sensorsarethecore sensorsontheArgoprofilingfloatsandhavehighaccuracy.AlloftheArgoandSOCCOM floatscarrySea-Bird(SBE)pressure,temperatureandconductivitysensors.Conductivity sensorssometimesexperienceslowdriftratesthatarecorrectedforinthedelayedquality control(QC)processusingsalinityclimatologies6. ThetemperaturesintheArgoprofilesareaccurateto±0.002°C.Pressuresarenominally accurateto±2.4dbar,buttheymayalsodrift.Pressurevaluesareadjusted,ifneeded,in nearreal-timebasedontheknownpressurewhenthesensorisatthesurface.Forsalinity theaccuracyis0.005basedonpropagationofuncertaintiesinconductivityand temperature.ThefactoryspecificationsoftheSBEsensorsareprovidedinTable1. --------------------------------------------------------------------------------------------------------------------- Table1.Accuracyanddriftofthetemperature,Pressure,andconductivitysensorsas providedbythemanufacturer(Sea-BirdSBE)a. Sensor CalibrationStandard Initialaccuracy Typicaldrift T(˚C) ITS-90 0.002 0.0002˚Cy-1 C(S/m) IAPSO/OSIL 0.0005 0.0002(S/m)y-1 P(dbar) Deadweight&Pref 2 0.8dbary-1 a:basedonapresentationbySRisertotheCWGon9/15/2015. --------------------------------------------------------------------------------------------------------------------- ThreetypesofO2sensorsarecurrentlydeployedonSOCCOMfloatsthatallworkonthe sameprinciplereferredtoasoptodes.TheseopticalO2sensorsconsistofasemi-permeable membrane,sensingelement,light-emittingdiode(LED)andphotodetector.Thesensing elementcontainsaluminescentdyethatisimmobilizedinagelmatrix.Whenexposedto bluelightfromtheLED,thedyeeitherluminescesorisquenchedbyinteractionwithO2. Somesensorsalsoemitaredlightasareferencetoenhancestability.Thisredlightis simplyreflectedbackbythedye.Theintensityorlifetimeofthereturnedluminescenceis measuredbyaphoto-detector.Becausetheintensityofresponsedriftsrelativelyquickly SBE63sandAanderaaoptodesonlyusethedecaylifetimetocalculatethedissolved oxygenconcentration.ThequantitymeasurediscloselyrelatedtothepartialpressureofO2 andthesensorcanmeasuretheO2inairaswell,thereforofferingauniquewaytocalibrate thesensorwhenthefloatsurfaces(Johnsonetal.,2015). 5Fromhttp://sio-argo.ucsd.edu/RG_Climatology.html)and http://www.argo.ucsd.edu/Data_FAQ.html 6http://www.jcommops.org/Apps/WebObjects/Argo.woa/wo/ OR63C76oBdZD8I3GtmaI1g/1.0.0.41.1.1 7 ForthefloatsdeployedduringSOCCOMtheSBE63andAanderaaOptode38307and4330 modelshavebeenused.AsshowninTable2belowtheyhavesimilarcharacteristics.The importantdifferenceisthattheSBE63isplumbedintothewaterloopthatalsohousesthe conductivitysensorandthereforcannotmakeoxygenmeasurementsinairwhenit surfaces. Theseairmeasurementsofferanaccuratecalibrationvalueifatmosphericpressureand watervaporpressureareknownattimeofsurfacing.Airimmediatelyabovetheoceanis generallyat100%relativehumidity,sothatthewatervaporpressurecanbecalculated fromairtemperature,salinity,andatmosphericpressure.Theseopportunisticcalibrations inairwhenthefloatsurfacesofferameanstoadjustforsmalldrifts(Bushinskyetal., 2016).Theaircalibrationshaveimprovedaccuraciesto2µmolkg-1comparedtothe factoryspecificationsinTable2.Theaccuracyislimitedprimarilybyknowledgeofthe barometricpressureatthelocationthefloatreachesthesurface. --------------------------------------------------------------------------------------------------------------------- Table2.SummaryoffloatO2sensorperformancebasedprimarilyonmanufacturers' specificationsa. a:FromapresentationofS.RisertotheCWG(9/15/2015) --------------------------------------------------------------------------------------------------------------------- TheNO3sensorsonAPEXfloatsareInSituUltravioletSpectroscopy(ISUS)sensorsbuilt andcalibratedatMBARI.TheNAVISfloatshaveSUNAsensorsbuiltandcalibratedby Satlantic.TheSUNAV2(SubmersibleUltravioletNitrateAnalyzer)isachemical-freeUV nitratesensorbasedontheISUSnitratemeasurementtechnologydevelopedatMBARI. ISUSandSUNAhavethesamemainopticalcomponents,butSUNAhastheopticalpath configureddifferently.ThedescriptionbelowisadoptedfromJohnsonetal.(2013).Nitrate absorbslightintheultravioletwithpeakabsorptionmaximumnear200nmwitha 7The3830sensorswereusedinthe12"pre-SOCCOM"floatsdeployedduringP16Sin 2014.TheyarenolongerusedinSOCCOMfloats 8 moderatelystrongmolarabsorptivity.ThedeepUVabsorptionspectrumofseawateris dominatedbythecombinedsignalfromnitrateandbromide,withamuchsmaller contributionfromdissolvedorganicmatterthatisseparatedbymeasurementsatmultiple wavelengths.TheISUSmakesasingleUVspectralscanwiththelampon(lightscan)and lampoff(darkscan).Thelightanddarkscansareused,alongwithreferencespectral intensitiesfromsimilarscansofdeionizedwatermadeinthelaboratory,tocomputethe absorbancespectrumfrom200to400nm.Theabsorbancespectrumisusedwiththe concurrenttemperatureandsalinitytocomputenitrateusingthetemperature compensated,salinitysubtractedalgorithm(SWwithT/Scorr.).Theobservedsalinityand temperatureareusedtopredicttheUVspectrumduetobromide.Thecorrectedspectrum thencontainsonlycomponentsduetonitrateandanapproximatelylinearbaselinedueto organicmatterandinstrumentaldrift.Allspectraldataistelemeteredtoshoresuchthat nitrateconcentrationscanberecomputedasnecessary.Theaccuracyofthesensorisabout 2μM,andthiscanbeimprovedthroughinsituadjustments.Errorsinnitrateconcentration areconstantoffsetsovertheentireverticalprofilesuchthattheerroratthereference depthcanbeaddedbacktotheentireprofileandamuchhigher-qualitysetofnitratedata canbeobtained.Theadjusteddatahaveaccuraciesontheorderof0.5μM,relativetothe referencevalue. --------------------------------------------------------------------------------------------------------------------- Table3.AccuracyoftheSUNANitratesensorinseawatera,b withT/S withoutT/Scorrection b corr. Detectionlimit 0.5μM 2.0μM Accuracy(greaterof) ±2μM or±10%ofreading Precision(shortterm) 0.3μM 1.0μM Drift(perhourlamptime)c 0.3μM 2.4μM a:Fromhttp://satlantic.com/sites/default/files/documents/2015_datasheet_SUNAV2.pdf b:Thesevaluesarewithappropriatecorrectionsforsalinityandtemperature c:Duringatypicalprofilethelampisactivated70secondsoratotalof5.5hoursfora5- yeardeployment --------------------------------------------------------------------------------------------------------------------- The(SBE/Satlantic)pHsensorisanewbiogeochemicalsensoranditisthefirstinorganic carbonsystemparameterroutinelydeployedonprofilingfloats.TodatethepHsensorson theSOCCOMAPEXfloatshaveallbeenbuiltatMBARIaspartofatechnologytransfer agreement.ToensureasmoothtransitionfromsensorsconstructedatMBARIto commercialversions,theMBARIsensorswillbegraduallysubstitutedwithcommercially builtsensorsofsimilardesign,beginningwiththedeploymentofSBEpHsensorsonNavis floats. MeasurementofpHusingIonSensitiveFieldEffectTransistor(ISFET)technologyis summarizedinMartzetal.(2010)andbrieflydescribedhere.TheISFETisametaloxide semiconductorfieldeffecttransistor(MOSFET).Theconductionchanneliscoveredbya 9 thininsulatinglayerofamphotericmaterial.ThepHofthesolutionatthe insulator/solutioninterfacecontrolsthesite-bindingprotonation/deprotonationstateof theinsulatormaterialand,hence,thesurfacechargeattheinterface.Theinterfacialcharge determinesthestrengthoftheelectricfieldintheconductionchanneloftheFET,located betweenthesourceanddrain.TheISFETsusedinSOCCOMareoperatedbyapplyinga constantdrainsourcecurrent.Aconventionalreferenceelectrodeisusedinconjunction withtheISFET.Itisasolid-statechlorideion-selectiveelectrode(Cl-ISE)withverylittle pressurehysteresis.TheapproximatesalinitysensitivityfortheFET|CI-ISEis0.013pH salinity–1(Martzetal.,2010). TheISFETsensordeployedonprofilingfloatsiscalledtheDeep-SeaDuraFETpHsensor. ThepHisreportedonthetotalprotonscale(pHT).Itcanoperateatpressuresto2000m depthintheocean.ThereforethesystemmustbecalibratedforpHmeasurementsthrough largetemperatureandpressuregradients.TheintegratedDeep-SeaDuraFETincludesa pressuretolerant,solid-stateAgClreferenceelectrodeCI-ISEdescribedabove. Theoutputofthesensorisavoltage,VRSthatisrelatedtoH+ionactivity,temperatureand pressureby: VRS=k0+k2(T-273.15)+f(P,T)-RT/Fln(aH+aCl-)T,P =k0+k2(T-273.15)+f(P,T)-RT/F[ln(mH,fmCl)+ln(γHγCl)T,1+V$ HClP/RT] (1) (2) k0isthereferencepotential.k2isthetemperaturedependenceofVRS.f(P)isthepressure coefficientofthesensor.Unlikeaconventionalelectrode,thek0andk2valuesoftheISFET containtermsassociatedwithFETdesign,semiconductorprocessingandpower applicationvalues.Thereforethek0andk2valuesmustbedeterminedforeachindividual sensor. Giventheuncertaintyofcoefficientsineqn.(2)overtherangeofTandPintheoceanthe uncertaintycantranslatetoa0.01pHerrorduetopressureat2000dbarpressureand 0.015duetotemperature.Asaresult,theestimatedpHTvaluescanhavebiasesonthe orderof0.02atpressuresnear2000dbar,basedonpropagationofpossibleerrorsdueto uncertaintiesinTandPdependence. InitialdeploymentsoffloatswithpHshowedameanrateofdrift,diagnosedasthechange betweenpHobservedbelow1000mdepthandtheexpectedpHatthisdepth,of-0.036pH yr-1.However,withappropriatesensorconditioning,inparticularequilibrationoftheAgCl referenceelectrode(CI-ICE)withbromideinseawater,thisdriftcanbedecreasedby5-fold. (K.Johnson,pers.com.). Frequencyofprofilingmeasurements ThepHandNO3sensorsareloggedontheupcastofthefloatwithadefault10-daycycle Approximately65measurementsaretakenontheupwardprofilewiththeAPEXfloats. ThisrateisdeterminedbypowerrequirementsoftheNO3sensorandlimitationsofthe electronics.Themeasurementspacingisvariabletocapturerelevantfeatureswith 10 nominally100-mintervalsfrom1500to1000m;50mfrom1000to400m;20mfrom400 to360m;10mfrom360to100m;and5-mintervalsfrom100mto5m.TheassociatedT, S,andO2dataareprovidedatthesamedepthsinthedatafiles.TheNAVISfloats,withmore modernelectronics,arecapableofmuchhigherverticalresolution.TheyprovidepH, oxygenandbio-opticalmeasurementsat2-meterspacingintheupper1000m.Nitrate remainsatthesameresolutionobtainedwithAPEXfloats.ExamplesoftwoprofilesofO2 andpHfromSOCCOMAPEXfloat9254areprovidedinFigure2. Figure2.Profilesofoxygen(blue)andpHT(red)fromfloat9254at39˚S,155˚W,for 11/21/2015(opensymbols)and12/01/2015(closedsymbols),illustratingthedepth resolution,andtherapidbutcoherentchangesinbiogeochemicalparametersintheupper 11 watercolumnduringtheAustralSpring/Summer(dataobtainedon12/05/2015from http://soccom.princeton.edu/soccomviz.php). 3.Performanceofsensors Basedoninformationonthesensorsprovidedaboveandthemeasurementsinthefieldto datewecanprovideanassessmentoftheperformanceofthesensors.Thisisnotexact,as onlylimitedinsituvalidationispossible.Moreover,sensorsandtheshoresidepreparations andcalibrationroutinesarestillbeingimproved.Table4providesanassessmentof currentestimatesofperformanceofthesensorsbasedonmanufacturersspecificationsand calibrationsinlaboratoryandothervenues. --------------------------------------------------------------------------------------------------------------------- Table4.Sensorresponseandcurrentestimatesofaccuracyofsensorsduringdeployment Parameter Accuracy Stability(drift) P 2dbar 0.8yr-1 T 0.002˚C 0.0002yr-1 S 0.005 0.01yr-1 O2 2µmol/kg 0.6µmolyr-1 -1(a) NO3 0.5µmol 0.3μMyr pH .01 -0.036yr-1(b) (a):Basedona1-hourtotallampusageoverayear(≈52profiles)(seeTable3). (b):FromMartzetal.(2010),newconditioningapproacheshavedecreasedthisdrifttoless than0.007yr-1 --------------------------------------------------------------------------------------------------------------------- Thereareslightdifferencesinsetupsandtypesoftheinstrumentsdeployedbutoverall theyshouldhavesimilarperformance.Animportantdifferenceisifthesensorsare plumbedintoawatercirculationsystemorexposeddirectlytotheoceanwater.The circulationsystemprecludestheaircalibrationoftheO2sensor,butshouldimprovethe responsetimeofthesensor.Thecirculationloopshaveananti-foulingagentbuttheNitrate (opticalsensor)appearstohaveotherfoulingissuesintheloop.Currentlyallthe biogeochemicalsensorsontheAPEXfloatsarebeingtestedwithsensorsdirectlyexposed toseawater. 4.Sensordeployments,calibration,validationandchecks Laboratorycalibrationofsensorshasbeenchallengingduetodifficultiesrecreatingthe environmentatseawhenthesensorsaredeployed.O2sensorsareroutinelycheckedinthe laboratory.FortheAPEXfloatstheO2sensorsarecomparedside-by-sidewithacalibrated laboratorysensorofthesamemakeandmodel.pHsensorsarecheckedpriorto deploymentinartificialseawaterbutduetodifferencesinchemicalconstituentsbetween artificialseawaterandrealseawater,appreciableadjustmentsarenecessary.NO3sensors arecheckedinthelaboratoryaswellbutwhileoverallresponseandgeneralbehaviorof thesensorcanbechecked,mostdatarequireanadjustmentoncedeployed. 12 ThevalidationCTD/bottlecaststhatarepartoftheusualSOCCOMprotocolatthetimeof floatdeploymentareanimportantmeanstocheckthesensorswhentheyaredeployedand todetermineappropriateadjustments.Thefirstprofilefromthefloatoftendoesnot correspondwellwiththebottledatafromthecastbecausethesensorssometimes experiencedriftduringtheirinitialprofile,notablytheNO3sensor.TheO2sensors generallydonotundergoacheckagainstthecalibrationcast,inpartbecausethesensors showgoodagreementbetweenlabbasedcalibrationvaluesandthedeploymentvalues. However,forthesakeofconsistency,similarchecksandadjustmentsshouldbemadefor O2,pHandNO3sensors.Thedeploymentcastisagoodqualitativecheckfordriftand offsetsbutcurrentlyadjustmentstopHandNO3sensorvaluesareperformedbasedonthe MLRsdevelopedfromdeploymentcastsandotherhighqualitybottledataintheSouthern Ocean(seesection7). ThepHandNO3sensorsroutinelyneeda"soakingtime"ofafewprofilesduringwhich thereisappreciabledrift.Thedriftisattributedtoseveralfactors.ForNO3itappearsthat theopticalwindowcouldbedirtiedbyorganicsduringstorageandtransport.ForpHthe AgClinthereferenceelectrodereactswithBrinseawateruntilequilibriumbetweenAgCl andAgBrintheelectrodeisobtained.Cleaningproceduresofopticalwindowsandlonger soakingtimesofpHsensorsinrealseawaterinthelaboratoryarecurrentlybeing investigatedasaviablemeanstodecreaseinitialdrift. OncetheNO3sensorstabilizesfromlargedriftthereisasmallerresidualdrift(seeTable 3).ThustheNO3sensoradjustmentofteninvolvesanadjustmentforoffsetsoonafter deploymentfollowed,atalatertime,bycorrectionforalongertimedrift.Theadjustments aredoneinanopportunisticfashionbasedonknowledgeofclimatologiesand stoichiometriesbetweenparameters.Theprimaryadjustmentsaredoneatdepthbasedon theWorldOceanAtlasWOA,otherclimatology,orasrecommendedforSOCCOMbyaMLR algorithm(section7).ForNO3sensorsdeployedinsubtropicalregions(e.g.,nearBATS) wheresummertimenitratevaluesinthesurfacearezero,asecondcalibrationcheckis basedonthezeroconcentrationinthemixedlayer.Driftovertimeisdeterminedand correctedbasedoncomparisonoffloatandMLRalgorithmatdepth(1000-2100m)where nitrateislinearlyregressedagainsttemperature,salinity,pressureandoxygen:NO3= a+bT+cS+dP+eO2,wherethecoefficientsaredeterminedfromcruisedatainthearea. 5.Recommendedconstantsforinorganiccarbonsystemparameters AdjustmentofpHsensorsissignificantlymorechallengingthanforothersensorsand requiresinformationonthedissociationoftheinorganiccarbonspeciesinseawater.This CO2systeminformationisalsonecessarytocreateinorganiccarbonproductsfromthepH floatdata.pHsensorsonthefloatsmeasurethepHatinsitutemperatureandpressure whichisdifferentfromthecalibrationandvalidationsampleswhichareanalyzedat constanttemperatureand1atm.ThesensorsmeasurethefreeH+ionsanditismost appropriatetoexpressthepHonthefreescale(K.Johnson,pers.com.).ThepHsamples fromCTD/Bottlecastsaremeasuredatconstanttemperature(usually20or25˚C)onthe totalscaleinaship'slaboratory.OnselectoccasionspHisnotmeasuredonboardbut ratherdeterminedalongwithTAlkonbottlesamplesshippedtotheshoresidelaboratory 13 ofA.DicksonofSIO.Tocomparethebottledatawiththesensordataandapply adjustmentstothesensordata,carbonatedissociationconstantsandachemicalmodel needtobeapplied.Detailsofthebasicinorganiccarbonchemistryandrecommended analysisprocedurescanbefoundinDicksonetal.(2007). Herewefocusonusingdiscretebottlesamplesatfixedtemperatureand1atmpressureto adjustinsitupHsensordata.Excel®andMATLAB®routinesarereadilyavailableto performtheconversions.Thetemperatureandspeciesconversionsrequireachemical modelandtwoinorganiccarbonsystemspecies(DIC,TAlk,pH,orpCO2)(e.g.Pierrotetal. 2006).TherecommendedconstantsforSOCCOMeffortsareprovideinTable5.Theyare consideredthemostconsistentbasedoncurrentknowledge,butthemainpurposeofthe recommendationisforconsistency.Thatis,allSOCCOMproductsshouldbebasedonthe sameconstantssuchthatdisagreementsinresultsorinterpretationsarenotcausedby productsbeingdevelopedusingdifferentconstants. --------------------------------------------------------------------------------------------------------------------- Table5.Recommendedconstantsforcalculatinginorganiccarbonsystemparametersin SOCCOM,includingtemperatureandpressurenormalizations. Carbonatedissociationconstants(1) pHT(2) Totalscale(molkg-1seawater) pK0(3): SolubilityWeiss&Price(1980) pK1andpK2(4)Luekeretal.(2000)usingpHT Associatedconstants(5) pK(B) DissociationconstantboricacidDickson,(1990) pK(HF) DissociationconstantforhydrofluoricacidofPerez&Fraga(1987) (6) pK2P,pK3P 2ndand3rddissociationconstantsphosphoricacidDicksonetal.,(2007) Pressure(7) Pressurecorrectionsforvariousacid-basedissociationconstantsas implementedinCO2SYS TB TotalboroninseawaterparameterizedwithsalinityLeeetal.,(2010) (8) KHSO4 Dickson(1990) ΩAr,ΩCa(9) AragoniteandCalcitesaturationstatesasprovidedinCO2SYS. (1)TheCO2SYSprogramhasbeenadjustedbyD.Pierrottoaccommodatetheseconstants. (CO2SYSv2.2).ThefootnotesnotesbelowareadaptedfromA.Dickson,pers.com. (2)Spectrophotometricmeasurementusingpurifiedm-cresolpurpleat25˚Casreference (Liuetal.,2011).Thesehavebeenverifiedinthelaboratory(at25˚CandS=35)ofA. Dickson,SIOandtheyagreewithLiuetal.to0.002inpH.However,theresultingpHis notnecessarilyinperfectagreementwiththewaypHisdefinedinthedefinitionofthe variousacid-dissociationconstants.Preliminaryworksuggestsadiscrepancyofabout 0.006inpH.IfthisdiscrepancyisaccountedforthenpH,TAlk,andDICareinternally consistentwithinlikelyuncertainties. (3)pK0basedonWeiss(1974),asimplementedintheappendixofWeiss&Price(1980) 14 (4)pK1andpK2basedondataofMehrbachetal.(1973),convertedtototalhydrogenion concentrationscale(Luekeretal.,2000). (5)Toconverttotalalkalinitytocarbonalkalinityneededincalculations,andtoestimate thehydrogenionactivityfromnon-carbonatespeciesinseawater.Valuesarevalidfor S≈20-38,T≈-2to40˚C) (6)pK2P,pK3PforH3PO4dissociationarebasedonMillero(1995)correctedtopHTscale (seeDicksonetal.,2007) (7)Pressurecorrectionsforacid-basedissociationconstantsasimplementedinCO2SYS, ThispressuredependenceisbasedonoriginalworkbyCulbersonetal.(1968)onthe equivalentofthefreepHscalewithsubsequentapproximationsofthe∆VofHCO3- basedonthatofHSO4-(K.Johnson,pers.com.). (8)WatersandMillero(2013)andKhooetal.(1977)showvaluesofKHSO4thataretwiceas highastherecommendedKHSO4ofDickson(1990).Theeffectsofpressureand temperaturearedifferentforthevariousKHSO4valuesthatwillleadtodifferencesin pHof≈0.01forthedifferentKHSO4valuesonthefreescale (9)Note,somecurrentbiogeochemicalmodels(e.g.TOPAZ)donotreproducecalculated valuesfromCO2SYSusingTAlkandDIC.ThisappearstobeanissuehowCa2+is determined.InCO2SYS,Ca2+=293.84*S. --------------------------------------------------------------------------------------------------------------------- ForchecksofotherCO2systemcodesthefollowingtwoexamplesareprovided. TestcasesusingCO2SYSV2.2andconstantsinTable5: Sample400217Station40,P16S(2014) Inputvalues P=1000.3dbar S=34.319 T(insitu)=5.720˚C TAlk=2286.4µmolkg-1 DIC=2169.1µmolkg-1 pHT(20,0)(a)=7.7366 PO4=1.82µmolkg-1 SiO3=17.8µmolkg-1 outputusingTAlkandpHT(20,0) pHT(5.72,1000)=7.9104 pHF(5.72,1000))(b)=7.9593 DIC=2167.8µmolkg-1 fCO2(5.72,1000)=493.4µatm ΩAr(5.72,1000)=1.14 ΩCa(5.72,1000)=1.79 Sample400210Station40,P16S(2014) P=1999.5dbar S=34.581 T(insitu)=2.626˚C TAlk=2350.6µmolkg-1 DIC=2264.3µmolkg-1 15 pHT(20,0)=7.632 PO4=2.37µmolkg-1 SiO3=82.4µmolkg-1 outputusingTAlkandpHT(20,0) pHT(2.62,2000)=7.8062 pHF(2.62,2000)=7.8459 DIC=2265.9µmolkg-1 fCO2(2.63,2000)=576.5µatm ΩAr(2.63,2000)=0.74 ΩCa(2.63,2000)=1.15 (a):pH (20,0)isthepHonthetotalscaleattemperatureof20˚Candpressureof1atm. T (b):pH (5.72,1000)isthepHonfreescaleattemperatureof5.72˚Candpressureof1000 F dbar(98.7atm).pHFiscalculatedfromDICandTAlkwhereDICinturnisdetermined fromTAlkandpHT(20,0). 6.pHcomparisonsbetweenbottleandfloatdata FortheSOCCOMeffort,useofthetotalpH(pHT)scaleisrecommended.Theotherscales commonlyusedinseawateraretheseawaterscale(pHsw)andfreescale(pHf)8.A comprehensivedescriptionofthemetrologyofpHcanbefoundinDicksonetal.(2016). TheDeep-SeaDuraFETpHsensormeasuresthefreehydratedhydrogenionsandpHf wouldbethebestscaleforthissensorinparticularbecausethepressuredependenceof bisulfateioninseawaterisnotwellknowncausinganuncertaintyinthecalibrationofthe pHdataatdepthcomparedtothesurfacereference.However,theDuraFETsensorsare calibratedandvalidatedonthepHTscaleasthisisthescaleofchoicefor spectrophotometricbottlepHmeasurementsinseawater. Table6providesanoverviewofthecalculatedpHat25and0˚Candatboththesurface and2000dbarusingdifferentpHscales.ThistableshowsthatthenumericalvaluesofpHT andpHswarequitesimilar,duetothefactthathydrofluoricacidhasasmallcontributionto hydrogenionsinseawater,butthedifferencewiththefreescaleisappreciable.Of particularnoteisthatthetemperatureandpressuredependenceofthepHfandpHTscales differwhichisacauseofambiguityintheDuraFETpHvaluesreported.Withtheadvances inpHsensortechnologyacarefulevaluationofthetemperatureandpressuredependence ofpHinseawaterisurgentlyneeded. --------------------------------------------------------------------------------------------------------------------- Table6.ComparisonofpHscalesfordifferenttemperaturesandpressuresa S T(˚C) P(dbar) pHsw pHT pHf 35 25 0 7.514 7.523 7.635 8 pHf=-log{H+},pHT=-log({H+}+{HSO4-}),andpHsw=-log({H+}+{HSO4-}+{HF0},where {H+}indicatestheactivityoffreehydratedhydrogenions. 16 35 0 0 7.876 7.883 7.824 35 25 2000 7.443 7.452 7.552 35 0 2000 7.793 7.799 7.839 a:calculatedusingtheCO2SYSV2.2programwhereTAlk=2300µmolkg-1;DIC=2230µmol kg-1;PO4=2µmolkg-1;andSi=50µmolkg-1 --------------------------------------------------------------------------------------------------------------------- AnexampleofthemagnitudeofcorrectionsofpHforadepthprofileofacalibration station,Station40onP16S,2014whereaSOCCOMfloatswasdeployed,isshowninFigure 4.TheredlineisthepHTmeasuredfromCTD/Bottlesamples(20˚C,1atm)andthegreen lineisthecalculatedpHTatinsitutemperatureandpressurewhichwouldbesimilartoa pHprofileobtainedfromaSOCCOMpHsensor.Atapressureof1600dbarandinsitu temperatureof3.15˚Cthedifferencebetweenthevalidationdataat20˚C,1atmandinsitu pHdatais0.188ofwhich0.242isassociatedwiththetemperaturedifferenceand-0.065is thepressureeffectonpH.Incomparison,thedifferencebetweenmeasuredpHT(20,0)and calculatedpHT(20,0)fromTAlkandDICfortheshipboardmeasurementis -0.005. Figure4.ExampleofpressureandtemperatureeffectonpHbasedonaCTD/bottlecaston cruiseP16.TheredlineistheinterpolationofthepHTmeasuredbyspectrophotometry(open redcircles).TheblueopensquaresarethepHTvaluescalculatedfromTAlkandDIC.The greenlineisthecalculatedpHTatinsitutemperatureandpressuremimickingafloatprofile. 17 Thedashedlinewithcrossesistheeffectofpressurerelativetothemeasureddata,andthe dashedlinewithplussymbolsistheeffectoftemperaturerelativeto20˚C. Bothtemperatureandpressureareappreciablecorrectionsthataresensitivetothe dissociationconstantsused.TocalculatetheeffectoftemperatureonpHrequiresDICor TAlkasasecondinorganiccarbonparameterbuttheaccuracyofthemeasurementofthe inorganiccarbonparameterhaslittleeffectontheestimate.Forinstance,a1%uncertainty intheICparameter(DICorTAlk)willtranslatetoanegligibleuncertaintyof0.0005inthe pHforaconversionfrom20˚Cto0˚C.Thusforvalidationsamplestakentoshoreforthe purposeofcomparingpH,theTAlkmeasurementneednotbeofhighaccuracy.The temperatureeffectonpHisabout0.2%˚C-1onlogscale.Thepressurecorrectionisabout0.04/1000dbar.Onlyonesetofconstantsisavailableforthiscorrection,andits uncertaintyisunknown.Sincethereareuncertaintiesintheconstantsandconversions frominsitupHfromthefloattospectrophotometricpHmeasurementsitisprudentto attainthehighestqualityDICandTAlkmeasurementsasthesemeasurementscanprovide anindependentestimateofpH. 7.UtilizingmultiplelinearregressionstoadjustpHandNO3sensorsonSOCCOM floats SeveralapproachescanbeappliedtoadjustthepHandNO3datafromsensorsonthefloats butrobustmeanstodosoinsituarelimited.DuringthefirstyearsofSOCCOMitwas realizedthatapplyingmultiplelinearregressions(MLR)developedfromhighquality shipboarddataintheregionisapowerfulmeanstodiscernoffsetsinfloatsensorsat depth.ThepHandNO3sensorsappeartovaryinaconsistentmannerwithinthepressure andtemperaturerangesoverwhichtheyprofile.Thismakesitpossibletoadjusttheprofile basedonsensor-offsetsobtainedbycomparingthesensormeasurementsatdepthwith MLRscreatedfordepthsof1000-2100dbar. TheMLRsforNO3andpHarecreatedwiththesameindependent/predictorvariables:P,T, S,andO2asthesevariablesaremeasuredoneachofthefloatsthatcarrythepHandNO3 sensors.Therefore,eachfloatcontainstheinformationforchecksoftheNO3andpH sensorsonboard.TheoverallapproachtocreatetheMLRsisdescribedinJuraneketal. (2009)withconsiderationofcrosscorrelationandoverfittingissuesthatcanplagueMLR approaches. Theempiricalmultiplelinearregressionrelationshipsaredevelopedfromcruisedatain 2011(S04P)and2014(P16S)intheSouthPacificsouthof45˚Sfordepthfrom1000to 2100dbar.Thedepthrangeandassociatednarrowparameterspaceincreasesthe goodnessoffitandavoidsspuriousresultsthatcanoccurwhenindependent/predictor variablesarerelatedasoftenisthecaseforbiogeochemicalandphysicalparametersinthe ocean. TheMLRalgorithmsdeterminedoverthe1000to2100mrangeare: 18 𝑝𝐻! 𝑇, 𝑃 = 1.380 + 1.8020×10!! 𝑂! + 0.17859 𝑆 + 7.4820×10!! 𝑇 − 3.966×10!! 𝑃, 𝑟 ! = 0.98, 𝑠. 𝑑. = 0.004 (3) 𝑁𝑂! = 544 − 0.108 𝑂! − 11.4 𝜎! − 4.92 𝑆 − 2.69 𝑇 − 3.14×10!! 𝑃, 𝑟 ! = 0.85, 𝑠. 𝑑. = 0.3 (4) DetailsonthealgorithmscanbefoundinthesupportinginformationofWilliamsetal. (2016).AlgorithmsforpHTutilizingNO3insteadofO2havebeendevelopedaswellwith similaruncertaintiesbutsincetheO2sensorsonthefloatsaremoreplentifulandmore accurate,thealgorithmswithO2willbeusedexclusivelyforthepurposeoffloatpH adjustment.ForthepHalgorithmsthediscretepHmeasurementsat20or25˚Catthe surfacewerecorrectedtoinsituvaluesusingthedissociationconstantsinTable4.The outputsfromalgorithmshavebeencomparedwithindependentcruisedataintheSouth Pacific(P18S-2007),SouthernIndianOcean(S04I-2012),andAtlanticOcean(A16S-2014) anddifferencesaregenerallywithinthelimitsof0.01forpHand0.5µmolkg-1forNO3 (Figure5).Atthegeographiclimitsandinareasofdeep-waterformationthedifferences tendtoincreaseandthisshouldbetakenintoconsiderationwhenfloatdataareadjustedin theseregions. 19 Figure5.ComparisonoftheMLRoutputforpHT(T,P)withmeasuredvaluesnotusedinthe MLRdevelopmentintheSouthernOcean.A16SisameridionallineintheSouthAtlanticand P14Sisazonalline(≈62˚S)SouthofAustralia.Thecirclesdepictthesamplingdepthand thosewithcolorlesscentersindicatethattheMLRoutputiswithintheuncertaintylimitsof 0.01oftheanalysis. ForSOCCOMallpHandNO3sensordataareadjustedbasedonthealgorithmlistedabove. AdjustmentsareperformedbasedonpHandNO3floatdataacquiredatdepth.ForpHan additiveadjustmentismadetothereferencepotentialk0(seeeqn.1)toreach correspondencewiththealgorithmsvalues.ForNO3theadjustmentisadditivetoits reportedconcentrationvalue. 8.Datamanagement,qualitycontrolandadjustments SuccessfulfullimplementationoftheSOCCOMfloatprogramrequiresimprovedefficiencies andprotocolsindataacquisitionandmanagement.ContinuedinteractionswithArgodata acquisitioncenters(DAC)areamusttoassurefullincorporationofbiogeochemicalsensors andbiologicalsensorsintotheArgodatasuite.ThisisparticularlytrueiftheSOCCOM effortisconsideredaprecursorofaglobalbiogeochemicalandbiologicalsensorarray.The challengesofmakingadevelopmentaleffortlikeSOCCOM,whichrequiresflexibilityand adaptationsbutalsothemorerigidrequirementsofthemorematurearrayaresignificant. Atthesametimepiggy-backingontheArgoarrayisagreatbenefitforlessonslearnedand establishedinfrastructure. TheSOCCOMbiogeochemicalparametersmeasuredonfloatsO2,NO3andpHwillbe transitionedfrominvestigatorbasedmanagement(K.JohnsonandS.Riserlead)toa streamlinedandconsistentapproachmanagedbytheUniversityofWashington(S.Riser). ThebiogeochemicaldatawillbeprocessedinparallelwiththecoreArgodatafollowing agreementswiththeArgosteeringgroup. 20 TheArgoprocessingisasfollows9.Therearethreerelevantfilesthatarecreatedforeach float:acore-Argo,B-ArgoandM-Argoprofileandtrajectoryfiles.Thecore-Argoprofileand trajectoryfilescontainonlytheCTDdata.Theparametersincludedinacore-Argofileare pressure,temperature,salinity,andconductivity.TheB-files,whereBstandsfor biogeochemistryorbiology,willincludeallotherparametersexcepttemperature,salinity, andconductivity.Theywillincludeallintermediateparametersthatarenecessaryfor calculatingparametersofinterestsuchastheUVspectraforNO3andluminescencedecay timefortheO2optodes.ThecoreandBfilesarecreatedprimarilyforeaseofcalculating and,whennecessary,adjustingparameters.ThecoreandBfilesarethenmergedintoMfileswhichwillbeofmostinteresttoinvestigators.Theywillcontainalloceanvariables thatthefloatmeasures.Thisfilewillcontaintheconcentrationofinterestforthecore parametersandparametersfoundintheB-filebutnottheintermediateparameters.The transitiontothisnewdataprocessingscheme(calledversion3.1)willbechallengingand possiblyslowdownthereleaseofbiogeochemicaldatafromfloatswhiletheprotocolsare beingimplemented. 9.Productdevelopment ApriorityforSOCCOMistoutilizethelargeincreaseinhigh-resolutiondatainthe SouthernOceantoestimatethestateofbiogeochemicalproperties,andtheprocessesand ratesinvolved.Inparticular,themagnitudeandcausesofseasonalvariabilitywillbe establishedforthefirsttime.Theyear-roundpresenceoftheprofilingfloatsandabilityto sampleundericewillofferseasonallyandspatiallyunbiasedresults. Twoapproachescanbefollowedfordeterminingfieldsandprocessesinvolvinginorganic carbonparametersbasedonthefloatandadditionaldata.Thefirstapproachisto determineTAlkempiricallyfromfloatdataorothermeans.ThederivedTAlkand measuredpHfromthefloatcanthenbeusedwiththethermodynamicrelationshipsand constants(Table4)todetermineanyoftheothercarbonsystemparametersofinterest suchasDIC,pCO2,ΩArorΩCa.RobustrelationshipsofTAlkandsalinityandtemperature havebeenderivedforsurfacewater(Leeetal.,2006;Takahashietal.,2014)basedonthe conservativebehaviorofTAlkonregionalscales.Forsubsurface,remineralizationand mixingprecludessimplerelationshipsandmoresophisticatedalgorithmsareestablished basedontheglobaloceanwatercolumncarbondataset(GLODAP-2,Keyetal.,2004). Carteretal.(2016)developedanalgorithm(andassociatedMATLABcode)toestimate alkalinityglobally,referredtoas“locallyinterpolatedalkalinityregression.”Specific algorithmswillalsobedevelopedfortheSouthernOceanaspartofSOCCOM(Williams, pers.com.) 9www.argo.ucsd.edu/Data_FAQ.html 21 Figure6.Diagramofpathwaysfordataflowandqualitychecks.Thebluelinesindicatethe datastreamsusedforinterpretation,thegreenlinesarethepathwaysusedtoadjustthefloat dataandtheredlinesarethepathwaysofqualitycontrolleddataflow.Thedoublearrows indicatethatthedatarepositoriesbothingestandcontributetothequalitycontrol.The proceduresinthedashedrectanglearethefocusofthisreport.Notethatmostdepositories havemultiplefunctions. Asecondapproachistoestablishempiricalrelationshipswiththeparameterofinterest andpredictorvariablesthatcanbemeasuredonfloats,ordeterminedfrombottledata fromcruisesandappliedtofloatdata.ThisapproachissimilartotheMLRalgorithmsused toadjustthepHandNO3dataatdepthforfloats.Thismethoddoesnotnecessarilydepend onpHdataorTAlkestimates.However,itdoesassumethattheMLRwithpredictor variablesthatarenotinorganiccarbonsystemparameterscanfaithfullypredictthe inorganiccarbonsystemparametersofinterestonscalesthatarerelevanttothecarbon cycle,fromsub-mesocaletomesoscaleandbeyond;andtimescales,fromseasonalthrough interannualtodecadalandbeyond.Anyprocessthatchangesthecarbonsystem parameterswithrespecttothosepropertiesusedintheMLRwilldegradetheestimatesof 22 theevolvingcarbonsystemparameters.Forexample,inthesurfacemixedlayerfunctional dependenciescanchangerapidlyduetogastransferandbiologicalproductivityandoften cannotbewellrepresentedbyMLRs.OndecadalscalesanthropogenicCO2increaseswill invalidateMLRspredictingDIC,pCO2orpH.Whilebothapproachesneedtobeinvestigated infull,itisprobablethatparametersthatcanbebetterpredictedinthesurfacelayersuch asTAlkalongwithmeasuredfloatpHwillhavesomeadvantagesoverlarge-scaleMLR approachesnotusinginorganiccarbonsystemparametersinthesurfacemixedlayer. 10.Outlook TheSOCCOMeffortisinitsinfancyandmakingthenecessarypreparationstofullyutilize theobservationsandmodelswhenfullyinplace.AsofFebruary2016,37,or18%ofthe planned200total,floatsaredeployed;thisnumberwillbecloseto50,or25%,bytheend ofMay2016.Implementationofrobustdatamanagementandbiogeochemicaldata adjustmentschemesisunderwaybutwillrequirecontinuedcloseinteractionwithArgo andsignificantresourcesfromSOCCOM.Asdescribedabove,creativeandrobust approachesusingmultiplelinearregressionsfromCTD/bottledataareusedtocorrectNO3 andpHdatatowithin0.5µMand0.01,respectively,whichiswithincurrentinstrument specifications.ContinuedchecksandimprovementsofthealgorithmsutilizingCTD/bottle dataandbottlesynthesisproductsthatwillbedeliveredduringtheSOCCOMprojectare advisable.Further,theprogrammustrigorouslyassesssensorperformanceandfailuresto ensurethatsensorreliability,accuracyandprecisionarecontinuouslyimproved.Thiswill minimizethedependenceonadjustmentproceduresandresultindataofevergreater utility. Biologicaldatafromtheprofilingfloats,"bio-optics",arenotaddressedinthisreportbut willrequireasimilarscrutinyonsensorperformanceandapproachestoadjustsensor values.Asmanyofthebio-opticalparametersarenotstatevariablesandsometimesare operationallydefinedthiswillbeamuchgreaterchallenge.Uncertaintyestimateswillalso begreaterandapplicabilitytosomeofthelarge-scalequestionsraisedinSOCCOMwillbe limited.However,theiruseinstudyofseasonalityofthebiologicalcycleandspatial heterogeneity(bothinthehorizontalandvertical)ofbiologicalparameters,andtheir influenceontheinorganiccarboncyclewillbeinvaluable.Moreover,thebio-optical parameterslinkcloselytoseveralofthespacebornemeasurementsofferingthepossibility toexpandthe2-Dviewoftheoceansurfacefromspacetoa3-Doreven4-Dviewthat includestimeanddepthdimensions. AmajorfocusofSOCCOMaspartofunravelingthemysteriesoftheSouthernOceanisto vastlyimproveourunderstandingofbiogeochemicalprocessesandassociatedestimatesof biogeochemicalfieldsintheSouthernOceanasawhole.Themeasurementsandtoolsto tacklethislargescaleproblemarebeingdevelopedanddeployedoverthefull6yeartime scaleofthegrant,butitiscriticalthatinitialestimatesbemadeasearlyaspossiblewitha smallsubsetsofthearrayinordertoidentifyandaddressasearlyaspossibleimportant issuessuchasthoseaddressedinthisreport.Similarquestionshavetobeaddressedwith respecttomodelsthatwillimprovethroughouttheSOCCOMperiod. 23 Evenwithlimitedfloatdeploymenttodateseveralnuggetsor"firsts"canbeobtainedfrom thequality-controlleddata.Thispertainsparticularlytosomeofthefirstbiogeochemical observationsinwintertimewhenconditionsaresuchthatship-basedobservationsare challengingandabilitytosampleunderice.Undericeprocessesaspertainingto biogeochemistryinseasonalsea-iceregionscannowbedetermined.Biogeochemicalstate estimatescanbevalidatedbyfloatmeasurementsparticularlyforthewintertime. ObservingSystemSimulationExperiments(OSSEs)fornetworkdesigncanbecheckedwith thefloatscurrentlydeployedandthevalidityoftheassumptionusedintheOSSEscanbe verified. ThenextmajordevelopmentforbiogeochemicalinvestigationinSOCCOMwillbethe developmentofso-callcarbonproducts,orfieldsofcarbonparameterssuchasTAlk,DIC, ΩArandΩCausingavarietyofapproachesrangingfrominterpolationoffloatdata,to regressionanalysesandnumericalmodels.Byproducingthesefieldsthroughtimekey processessuchasnetcommunityproduction,calcification,andrespirationcanbeinferred. Moreover,thetemporalevolutionofpCO2patternsandsaturationhorizonscanbestudied andtheroleoftheSouthernOceanasamajoranthropogenicCO2sink. 24 AppendixA.Chargeforthecarbonworkinggroup(CWG)ofSOCCOM CHARGEFORTHESOCCOMINORGANICCARBONSYSTEMWORKING(CSW)GROUP [OriginalJuly11,2015;EditedDecember2,2015] RATIONALE- AmajorobjectiveofSOCCOMistoassessthechangingcarbonsinkandassociatedchanges ininorganiccarbonintheSouthernoceanthroughutilizationofnewautonomous technology.Formanyproperties,suchasaragonitesaturationordissolvedinorganic carbon,theautonomoussensorsdonotdirectlymeasuretheparametersofinterest.In particular,fortheinorganiccarbonsystem,SOCCOMcurrentlymeasuresonlypH.In additiontotheobservedpH,asecondcarbonparametermustbeobtainedinordertosolve thefullinorganiccarbonsystem.Thiswillbeaccomplishedbyusingstoichiometricand/or statisticalrelationshipsderivedfromthebottledatasetsavailable.Thesealgorithmswill thenbeappliedatthelocationofthefloatmeasurementsusingtheinsitu[float]data.Such relationshipsmayalsobeusedtogeneratepHandothertracerestimatesthatcanbeused toqualitycontrolandadjustsensordata.Theoverallapproachwillprovideadatasetwitha resolutionofthatofthefloatobservations.Timeresolutionwillbeontheorderofaweek andspatialresolutionwillbe100’sofkminthehorizontalandmetersintheverticaldown to2000m. SeveralgroupswithinSOCCOMaredevelopingandapplyingthesenovelempirical approachestodeterminetheinorganiccarbonsystemparametersofinterestfromthe autonomoussensors.AworkinggroupwithinSOCCOMisdesiredtoexchangeideas;to provideamechanismtoshareresults;andtoapplytheapproachesinnearreal-time utilizingtheautonomoussensors.Theworkinggroupaimstofacilitaterapiddissemination ofresultswithsoundinterpretation,providingclearconfidenceintervalsandspecifying appropriatecaveatsandlimitations. GOALS- Theworkinggroupwilladdressseveralspecificquestionswiththeprincipalgoaltoobtain ameasurement-basedestimateofthefullinorganiccarbonsystemincludingpH(measured andcalculated),DIC,TAlk,pCO2,CO32-,andAragoniteandCalcitesaturationindexes. Theworkinggroupwillengageinthefollowingactivities: 1.Assessqualityandissueswiththeindependent[floatbased]variablesincludingdriftand hysteresis: -Nitrate, -Oxygen -pH 25 -Temperature -Pressure Prioritytask:createaninterpolatedestimateoftheoceanicpHusingalgorithmsfromhigh qualityshipboard/laboratoryinorganiccarbonsystemmeasurementsandrelated parametersthatcanbeusedtotestthebehaviorofthepHsensor. 2.DevelopastreamlinedapproachforinitialQCandadjustmentoffloat-basedvariables andadocumentedapproachforuserstosuggestadjustmentsandflags. 3.Documentuncertaintyandvalidity(time,distance,depthrange)ofalgorithms.A particularfocuswillbeondistributionoferror/uncertaintyintimeandspace. Prioritytask:ProvideaninterpolatedestimateofalkalinitythatcanbeusedwiththepH sensorsdatatoproduceafirstsetofpapersonthecarbonsystemasdeterminedusingthe floats(includingestimateofuncertainty) Longer-termobjective:Examineallaspectsofthecarbonsystemmeasurementsand calculationssoastoproducethehighestqualityestimatesofDIC,pCO2andpH, 4.Createabestpracticesmanualforad-hocfloatsensorvalidationsandchecks *[5.ProvideadvicetotheSOCCOMexeconthecoordinationofmanuscriptsand presentations] *[6.Assessqualityandissueswiththevariablesfromopticalsensorsonfloatstomeasure chlorophyllandbackscatter.] *Possibleactivitiesatalaterpoint SOCCOMCarbonWorkingGroupmembers: Briggs,Ellen:surfacewaterpCO2,iceeffects Bushinsky,Seth:floatsensors(oxygen) Carter,Brendan:Alkalinityalgorithmdevelopmentandapplication Dickson,Andrew:shorebaseddiscreteinorganiccarbonsystemmeasurements Feely,Richard:P16Sinorganiccarbondata Gray,Alison:algorithmdevelopmentandapplication Johnson,Ken:floatdata Juranek,Laurie:algorithmdevelopmentandapplication Key,Bob:Bottledataassembly,includinghistoricaldata;hydrography Riser,Steve:floatdata Talley,Lynne:P16Smeasurements,CTDhydrography Williams,Nancy:algorithmdevelopmentandapplication Sarmiento,Jorge:modelapplication Russell,Joelle:modelapplication Verdy,Ariane:BiogeochemicalSouthernOceanStateEstimate(SOSE) 26 Wanninkhof:Coordinator APPROACH: -Bi-Weeklyteleconferenceswithassignmentsoftasks&reports -Protectedsiteforsharingdataalgorithmsandpreliminaryfindings -Inquiryto[outside]expertsonregimes,approaches,methods,&analyses 27 References Bushinsky,S.M.,S.R.Emerson,S.C.RiserandD.D.Swift(2016)."Accurateoxygen measurementsonmodifiedArgofloatsusinginsituaircalibrations."Limnologyand OceanographyMethodsSubmitted. Carter,B.R.,N.L.Williams,A.R.Gray,R.A.Feely(2016).LocallyInterpolatedAlkalinity RegressionforGlobalAlkalinityEstimation,LimnologyandOceanography:Methods,in press. Culberson,C.andR.M.Pytkowics(1968)."Effectofpressureoncarbonicacid,boricacid, andthepHinseawater."LimnologyandOceanography13:403-417. Dickson,A.G.,C.L.SabineandJ.R.Christian(2007).GuidetobestpracticesforoceanCO2 measurements.,PICESSpecialPublication3,191pp. Dickson,A.G.(1990)."Thermodynamicsofthedissociationofboricacidinsynthetic seawaterfrom273.15to318.15K."Deep-SeaRes.37:755-766. Dickson,A.G.,M.F..Camões,P.Spitzer,P.Fisicaro,D.Stoica,R.PawlowiczandR.Feistel (2016)."Metrologicalchallengesformeasurementsofkeyclimatologicalobservables.Part 3:SeawaterpH."Metrologia53:R26–R39. Johnson,K.S.,L.J.Coletti,H.W.Jannasch,C.M.Sakamoto,D.D.SwiftandS.C.Riser(2013). "Long-TermNitrateMeasurementsintheOceanusingtheinsituUltraviolet Spectrophotometer:SensorIntegrationintotheAPEXProfilingFloat."Journalof AtmosphericandOceanicTechnology30(8):1854-1866. Johnson,K.S.,H.W.Jannasch,L.J.Coletti,V.A.Elrod,T.R.Martz,Y.Takeshita,R.J.Carlson andJ.G.Connery(2016)"Deep-SeaDuraFET:ApressuretolerantpHsensordesignedfor globalsensornetworks."AnalyticalChemistryinpreparation Johnson,K.S.,J.N.Plant,S.Riser,C.andD.Gilbert(2015)."Airoxygencalibrationofoxygen optodesonaprofilingfloatarray."JournalofAtmosphericandOceanicTechnology32: 2160-2172. Juranek,L.W.,R.A.Feely,W.T.Peterson,S.R.Alin,B.Hales,K.Lee,C.L.SabineandJ. Peterson(2009)."Anovelmethodfordeterminationofaragonitesaturationstateonthe continentalshelfofcentralOregonusingmulti-parameterrelationshipswithhydrographic data."Geophys.Res.Let.36:doi:10.1029/2009GL040778. Key,R.M.,A.Kozyr,C.L.Sabine,K.Lee,R.Wanninkhof,J.L.Bullister,R.A.Feely,F.J. Millero,C.MordyandT.H.Peng(2004)."Aglobaloceancarbonclimatology:Resultsfrom GlobalDataAnalysisProject(GLODAP)."GlobalBiogeochemicalCycles18(4):GB4031, doi:4010.1029/2004GB002247. 28 Khoo,H.K.,Ramette,R.W.,Culberson,C.H.andBates,R.G.(1977)."Determinationof hydrogenionconcentrationinseawaterfrom5to40°C:standardpotentialsatsalinities from20to45."Anal.Chem.49:29-34. Körtzinger,A.,J.Schimanski,U.SendandD.Wallace(2004)."Theoceantakesadeep breath."Science306:1337. Lee,K.,T.-W.Kim,R.H.Byrne,F.J.Millero,R.A.FeelyandY.-M.Liu(2010)."Theuniversal ratioofborontochlorinityfortheNorthPacificandNorthAtlanticoceans."Geochimicaet CosmochimicaActa74(6):1801-1811. Lee,K.,L.T.Tong,F.J.Millero,C.L.Sabine,A.G.Dickson,C.Goyet,G.-H.Park,R. Wanninkhof,R.A.FeelyandR.M.Key(2006)."Globalrelationshipsoftotalalkalinitywith salinityandtemperatureinsurfacewatersoftheworld’soceans"Geophys.Res.Let.33: L19605,doi:19610.11029/12006GL027207,022006. Lewis,E.andD.W.R.Wallace(1998).ProgramdevelopedforCO2systemcalculations.Oak Ridge,OakRidgeNationalLaboratory.ORNL/CDIAC-105 Liu,X.,M.C.PatsavasandR.H.Byrne(2011)."PurificationandcharacterizationofmetacresolpurpleforspectrophotometricseawaterpHmeasurements."Environ.Sci.Technol. 45:4862–4868. Lueker,T.J.,A.G.DicksonandC.D.Keeling(2000)."OceanpCO2calculatedfromdissolved inorganiccarbon,alkalinity,andequationsforK1andK2;validationbasedonlaboratory measurementsofCO2ingasandseawateratequilibrium."Mar.Chem.70:105-119. Martz,T.R.,J.G.ConneryandK.S.Johnson(2010)."TestingtheHoneywellDurafetfor seawaterpHapplications."LimnologyandOceanographyMethods8:172-184. Mehrbach,C.,C.H.Culberson,J.E.HawleyandR.M.Pytkowicz(1973)."Measurementof theapparentdissociationconstantsofcarbonicacidinseawateratatmosphericpressure." LimnologyandOceanography18:897-907. Millero,F.J.(1995)."Thermodynamicsofthecarbondioxidesystemintheoceans." GeochimicaetCosmochimicaActa59:661-677. Perez,F.F.andF.Fraga(1987)."Apreciseandrapidanalyticalprocedureforalkalinity determination."Mar.chem.21:169-182. Pierrot,D.,E.LewisandD.W.R.Wallace(2006).MSExcelprogramdevelopedforCO2 systemcalculations.OakRidge,Tennessee. doi:10.3334/CDIAC/otg.CO2SYS_XLS_CDIAC105a,CarbonDioxideInformationAnalysis Center,OakRidgeNationalLaboratory,U.S.DepartmentofEnergy. 29 Takahashi,T.,S.C.Sutherland,D.W.Chipman,J.C.Goddard,C.Ho,T.Newberger,C. SweeneyandD.W.Munro(2014)."ClimatologicaldistributionsofpH,pCO2,totalCO2, alkalinity,andCaCO3saturationintheglobalsurfaceocean,andtemporalchangesat selectedlocations."Mar.Chem.164:95-125. Waters,J.F.andF.J.Millero(2013)."ThefreeprotonconcentrationscaleforseawaterpH" Mar.Chem.149:8-22. Weiss,R.F.(1974)."Carbondioxideinwaterandseawater:thesolubilityofanon-ideal gas."Mar.Chem.2:203-215. Weiss,R.F.andB.A.Price(1980)."Nitrousoxidesolubilityinwaterandseawater."Mar. Chem.8:347-359. Williams,N.L.,L.W.Juranek,K.S.Johnson,R.A.Feely,S.C.Riser,L.D.Talley,J.L.Russell6, J.L.SarmientoandR.Wanninkhof(2016)."EmpiricalAlgorithmstoEstimateWater ColumnpHintheSouthernOcean."Geophys.Res.Let.submitted. 30 31