data delivery and real me data quality flags
TRANSCRIPT
Changingdynamicsofwaterqualityandnutrientmonitoringthroughtheuseoflongtermuna6endeddeployments,real9me
datadeliveryandreal9medataqualityflags
Nichole Halsey, Corey Koch,
Bruce Rhoades Adam Dutton
Sea-Bird Scientific Doug Wilson
Caribbean Wind LLC
ChangingDynamicsofWaterQualityandNutrientMonitoring
HydroCycle-PO4PhosphateSensor
Range: PO4
3- : 2.3-1200 µgP/L Accuracy: PO4
3- : ~4.6 µgP/L (extensive grab sample in field operation) Endurance: PO4
3- : 9-18 weeks (hourly-bihourly interval) Field service: PO4
3- : optics clean, new reagent cartridges, filters
ABSTRACTTheCycle-PO4wasdesignedforestuarineresearch,buthasbeenadoptedbymanywater
qualitysensorusers.Thelatestversion,theHydroCycle-PO4,offersanimprovedplaEormtoprovideawetchemicalinstrumentforcon9nuousmonitoringinbothfreshandcoastalwaters.TheHydroCycle-PO4hasincreasedreliabilityoveritspredecessorbymaximizinginstrumentup9me.Fluidicshavebeenadvancedtoprovide>90%up9mefreeofbubblesforopera9oninhighoxygensatura9onenvironments.Filterlifehasbeenextendedtominimizecloggingandincreasethe9mebetweenserviceintervals.Theseimprovementspermitphosphatemeasurementinawidervarietyofcondi9onsandprovideaninstrumentbe6ersuitedforapplica9onssuchasagriculturalrunoffmonitoring.Achallengeinservingandvisualizingreal-9medatafromin-situsensorsisqualifyingthedata.With“QualityAssuranceofReal-TimeOceanData”asaguide,wehavedevelopedsensor-specificQualityControl(QC)flagstoalertusersofSuspect/HighInterest,Bad,orGooddata.Flagsarebasedonalgorithmson-boardthesensorthatusesensorspecificmeta-dataandsitedefinedthresholdstoqualifysensorperformanceanddataqualityforeachdatum.Examplesincludeout-of-range,lowsignal,etc.So]wareisusedtoinves9gateHighInterestdataandprovidequalitysta9s9cs.Qualifieddatacanbepresentedtothecommunityinreal9me.Ul9mately,flaglevelscanbeusedtohelpautomateclassifica9onofdatasostakeholderscanmakebe6erdecisionsandscien9stscanquicklyiden9fyqualitydataforuseinanalysisandmodels.Pilotcollec9onofautoma9callyQC’edphosphatedataishappeningnowinanumberofsites.Usersarestudyingphosphorusfrom9dal,estuarine,andterrestrialsources,requiringhighfrequencyobserva9onstounderstandecosystemimpacts.MechanismsbywhichautomatedQCareperformed,howqualitymeta-datacanbeusedinthedatamanagement,web-publishing,visualiza9on,andthesignificanceofreal-9mephosphatewillbepresented.
TheHydroCycle-PO4isawetchemicalsensorengineeredforenvironmentalmonitoringtoenablevalid,scien9ficallydefensibleresultsthatleadtobe6ermanagementrecommenda9ons.ANISTtraceableon-boardstandard,factorycalibra9oncheck,a2.3µgP/Ldetec9onlimit,andanaccuratecalibra9onmeanyoucantrustdata,lineupgrabsamplerecords,andswapsensorswithminimaloffsets.Autonomousopera9ons,low-power,fivemonthreagentstabilityenableextendeddeploymentsinremoteloca9ons.Over1000samplesbetweenservices,uptofoursamplesperhourandbothSDI-12andRS232permitcapturinghigherfrequencyphosphatedatathanbefore.TheHydroCycle-PO4hasincreasedreliabilityoveritspredecessor,theCycle-PO4,andmanywet-chemicalsensors,bymaximizingdatareliabilityandsensorup9me.Fluidicshavebeenadvancedtoprovide>90%up9mefreeofbubblesfordataqualityinhighoxygensatura9onenvironmentsandtoensureitstabilizesrapidlya]erdeployment.Filterlifehasbeenextendedtominimizecloggingandenhancedataqualityduringhighsedimentevents.
RealTimeQualityControlFlagsThissystemofautomatednutrientsensorswilldeliverhigh-frequency,quality-controllednutrientdatatoscien9sts,managers,andpolicymakerstohelpunderstandnaturalandanthropogenicinfluencedcoastalnutrientdynamicsthroughouttheregion.Qualitycontrol(QC)flagsprovideconfidenceinthedataqualityandup9mesouserscanknowtheirinstrumentisworkingproperly.QCflagscanhelpspeeduptroubleshoo9ng,helpjus9fypublisheddata,andpermituserstospendmore9meonadvanceddataanalysisratherthanreviewingdatatoseparategooddatafrombad.UsingthissystemofQCflagseachdatapointreceivesageneralqualityflag(good,suspect,orbad)inreal9me.Thisflagisacompositeofthesixindividualqualityflags.Thoseflagsinclude:out-of-rangemeasurement,op9calnoise,lowsignal,bubblesandcalibra9onandmixingspikes.Theincludedso]warecanbeusedtoinves9gatehighinterestdataandprovidequalitysta9s9cs.Dataisautoma9callycolorcodedGreen(Good),Yellow(Suspect),Red(Bad)andBlue(NotEvaluated).
TradiEonalMethod:GrabSampleSite~12–24measurementsperyear
CurrentMonitoringTrend:RealTimeSitewithNutrientSensors~315,000measurementperyear
Waterisafiniteresourcevitalforhumanlife.Itisakeyingredientinagriculture,manufacturing,energycrea9on,andtransporta9on.Currentlyouraqua9cecosystemsarethreatened,locallyandgloballybyavarietyofpollutantsasresultofpopula9ongrowth,industrialgrowthandlarge-scaleagriculture.Pointandnon-pointsourcesincludedomes9csewage,agriculturalrun-offandindustrialeffluents.Therateandimpactofnutrientpollu9onisgrowingandiscontribu9ngtoacceleratedeutrophica9oninlakesandimpoundedrivers.Impactstoaqua9csystemscanoccurinawiderangeof9mescalesandbasedonanumberoffactors.
NERRSMyrtleBeachDeploymentACycle-PO4sensorwasdeployedintheWithersSwash,inlatewinterof2013throughthesummer.Itwasmountedinacageanddeployedalongthebo6omatasitebyFamilyKingdom.ALOBOsystem,includingaCycle-PPhoshatesensor,wasdeployedatWithersSwash,SC,partoftheNorthIsland–WinyahBayNa9onalEstuarineResearchReserve.Figure1showsPO4readingstakenoverthefirsthalfof2013.Alldataareshownwithreddatapointshighligh9ngvaluesdeemedinvalidbyautomatedanalysiswithintheCycle-Psampleprocessingsystem.SimilaranalysisisnowincludedandoutputbyCycleinstrumentsalongwithreal9medata.
Conclusions• Anaddi9onalfeatureoftheQualityControlflagsisthattheycanbeappliedtoanyCycledata-Herewepost-processeddatafroma2013deploymentbeforewehaddevelopedQCflagsorimprovedthesensor’sfilters• Over4113samples—6monthsofcon9nuoushourlydata—73%ofthedatawasgood,10%suspect,and17%ofthedatawasbad
-Unitservicedon3/12,4/22,and6/4-SomebaddatainMarchandJuneduetorunningoutofreagentandlatentreagentprimingissues-ThemajorityofthebaddataisduetothebubbleflagonApril1thatgoesun9lMay,thenshorterstretchesofbaddatain
mid-Juneandmid-July• Consistentbadbubbleflagshalfwaythroughthedeploymentisasignthatfiltersneedtobeserviced.A]erchangingthefiltersthebaddatabecomesgoodagain
-Theeventhadseveralsuspectflagsandintermi6entbadflagson3/31that,ifavailablein2013,wouldhavealertedustochangethefiltersandnotlosedataqualityforthemonthofApril
-Theregularfrequencyoffilterclogginginthissystemisaboutevery4-5weeksduringtheproduc9veseason,nowHydroCycle-PO4'sfilterslastover3Xlonger,butweshouldhaveconsideredchangingourserviceschedule!
RESULTS
Measurementsofnutrientsprovidedformoreaccuratees9matesofphosphateloadings.Threemethodsare
comparedbasedon:1)Cycle-PO4measurement,2)LinearInterpola9onofgrabsamples,and3)Linearregressionofgrab
sampleswithdischarge.
Myrtle Beach is a popular tourist site in a 60 mile stretch of beaches that support a multi-billion dollar tourism industry in South Carolina. In this area much of the storm water flows in to tidal creeks known as swashes that travers the beaches and flow directly in to the ocean. Because of this local scientists and regulators were interested in studying how swashes collect, transform, and export the nutrients from the runoff in to the coastal ocean waters. Cycle-PO4 Linear
Interpolation Discharge
Regression Aug Load (kg) 5.8 7.8 8.4 Sept Load (kg) 12.5 13.1 10.4
0
2
4
6
8
10
12
14
Load
(kg/
mon
th)
OverallQCBubbleSpikeFlag QCCOVFlag
QCLowSignalFlag
QCOutOfRangeFlag
QCMixingSpikeFlag
QCCalSpikeFlag
samplecount 4113 4113 4107 4107 3737 4023 603
%Good 73 86 93 100 97 83 96%Suspect 10 2 7 0 2 8 2
%Bad 17 12 0 0 1 9 2
OverallQCBubbleSpikeFlag
QCCOVFlag
QCLowSignalFlag
QCOutOfRangeFlag
QCMixingSpikeFlag
QCCalSpikeFlag
samplecount 4113 4113 4107 4107 3737 4023 603%Good 73 86 93 100 97 83 96%Suspect 10 2 7 0 2 8 2%Bad 17 12 0 0 1 9 2
Whythewidedatarange?WithersSwashischaracterizedbyfree9dalexchangewiththeAtlan9cOceanthroughanopenchannelcrossingthebeach.Thedrainagebasinisalsosignificantlyaffectedbystormwaterrunoff.Figure2showsthefirstdeployment(JFM)PO4(aswellasNO2+3fromacoincidentallylocatedSUNANitrateSensor)asafunc9onofHighWater/LowWateraswellaslocalprecipita9on,showingthestronginfluencesofbothoceanfloodingandstormwaterrunoff.Withthislevelofvariability,havingaccesstoreal9medataqualityinforma9onisvitaltomaintainingcon9nuousconfidenceinthevalidityofthedata.