Multitemporalmonitoringofwetlandwaterlevelinflorid

Multi-temporalmonitoringofwetlandwaterlevelsintheFloridaEvergladesusinginterferometricsyntheticapertureradar(InSAR)Sang-HoonHonga,b,e,⁎,1,ShimonWdowinskia,1,Sang-WanKimc,Joong-SunWondaDivisionofMarineGeologyandGeophysics,UniversityofMiami,Miami,FL33149-1098,USAbSatelliteDataApplicationDepartment,KoreaAerospaceResearchInstitute(KARI),Daejeon,305-333,RepublicofKoreacDepartmentofGeoinformationEngineering,SejongUniversity,Seoul,143-747,RepublicofKoreadDepartmentofEarthSystemSciences,YonseiUniversity,Seoul,120-749,RepublicofKoreaeDepartmentofCivilandEnvironmentalEngineering,FloridaInternationalUniversity,33174,USAabstractarticleinfoArticlehistory:Received15January2010Receivedinrevisedform19May2010Accepted22May2010Keywords:WetlandsEvergladesInterferometricsyntheticapertureradar(InSAR)Smallbaselinesubset(SBAS)Smalltemporalbaselinesubset(STBAS)AbsolutewaterlevelsInterferometricsyntheticapertureradar(InSAR)techniquescansuccessfullydetectphasevariationsrelatedtothewaterlevelchangesinwetlandsandproducespatiallydetailedhigh-resolutionmapsofwaterlevelchanges.Despitethevastdetails,theusefulnessofthewetlandInSARobservationsisratherlimited,becausehydrologistsandwaterresourcesmanagersneedinformationonabsolutewaterlevelvaluesandnotonrelativewaterlevelchanges.WepresentanInSARtechniquecalledSmallTemporalBaselineSubset(STBAS)formonitoringabsolutewaterleveltimeseriesusingradarinterferogramsacquiredsuccessivelyoverwetlands.Themethodusesstage(waterlevel)observationforcalibratingtherelativeInSARobservationsandtyingthemtothestage'sverticaldatum.WetestedtheSTBAStechniquewithtwo-yearlongRadarsat-1dataacquiredduring2006–2008overtheWaterConservationArea1(WCA1)intheEvergladeswetlands,southFlorida(USA).TheInSAR-derivedwaterleveldatawerecalibratedusing13stagestationslocatedinthestudyareatogenerate28successivehighspatialresolutionmaps(50mpixelresolution)ofabsolutewaterlevels.WeevaluatethequalityoftheSTBAStechniqueusingarootmeansquareerror(RMSE)criterionofthedifferencebetweenInSARobservationsandstagemeasurements.TheaverageRMSEis6.6cm,whichprovidesanuncertaintyestimationoftheSTBAStechniquetomonitorabsolutewaterlevels.AbouthalfoftheuncertaintiesareattributedtotheaccuracyoftheInSARtechniquetodetectrelativewaterlevels.Theotherhalfreflectsuncertaintiesderivedfromtyingtherelativelevelstothestagestations'datum.©2010ElsevierInc.Allrightsreserved.1.IntroductionWetlandsarezonewithexcesswater,wherenutrientcyclingandthesun'senergymeettoproduceaveryproductiveecosystem(Mitsch&Gosselink,2007;Rivera,2005).Theyencompassawidevarietyofaquatichabitats,homeforawidevarietyofplantandanimalspecies.Wetlandsalsohaveavaluableeconomicalimportance,astheypreventflooding,filternutrientsandpollutantsfromfreshwaterusedbyhumans,andprovideaquatichabitatsforoutdoorrecreation.Duetoseverepopulationgrowth,landproclamation,andurbanization,manywetlandsareundersevereenvironmentalstress.However,theincreasingrecognitionofwetlandimportancehasledtorestorationactivitiesinafewregions.Astheseecosystemsarewaterdependent,hydrologicalmonitoringofthewetlandsiscriticalformanagementandrestoration.Typically,hydraulicmonitoringofwetlandsisconductedbystage(waterlevel)stationsprovidinggoodtemporalresolutionatafinitenumberofobservationpoints.However,thesemeasurementshavelimitedcapabilitytodetectspatialpatterns,becausestagestationsaretypicallydistributedseveral,oreventensofkilometers,fromoneanother.Furthermore,insomeruralareasitisdifficulttogetstagerecordevenfromasinglesite,resultinginaverylimitedknowledgeofthewetlandhydrologicalconditions.Significantspatialresolutionimprovementswereachievedbyinterferometricsyntheticapertureradar(InSAR)observations,whichmeasurewaterlevelchangesoverwideareaswith5–100mpixelresolutionandseveralcentimetersverticalaccuracy(Alsdorfetal.,2000;Wdowinskietal.,2004).ThewetlandInSARtechniqueworkswherevegetationemergesabovethewatersurfaceduetothe“doublebounce”effect,inwhichtheradarpulseisbackscatteredtwicefromthewatersurfaceandvegetation(Richardsetal.,1987).InSARobservationsweresuccessfullyusedtostudywetlandhydrologyintheEverglades(Hongetal.,2010;Wdowinskietal.,2004,2008),Louisiana(Kimetal.,2008;Lu&Kwoun,2008;Luetal.,2005)andtheSianKa'aninYucatan(Gondweetal.,2010).Howeverdespitethevastspatialdetails,theusefulnessofthewetlandInSARobservationsremainedlimited,because(1)hydrol-ogistsandwaterresourcesmanagersneedinformationonabsolutewaterRemoteSensingofEnvironment114(2010)2436–2447⁎Correspondingauthor.DepartmentofCivilandEnvironmentalEngineering,FloridaInternationalUniversity,33174,USA.Tel.:[phone];fax:[phone].E-mailaddresses:[e-mail](S.-H.Hong),[e-mail](S.Wdowinski),[e-mail](S.-W.Kim),[e-mail](J.-S.Won).1Tel.:[phone];fax:[phone]-4257/$–seefrontmatter©2010ElsevierInc.Allrightsreserved.doi:10.1016/j.rse.2010.05.019ContentslistsavailableatScienceDirectRemoteSensingofEnvironmentjournalhomepage:flowisdrivenmainlybygravitationalpotentialinduce