
RemoteSensingofEnvironment158(2015)28–41
ContentslistsavailableatScienceDirect
RemoteSensingofEnvironment
journalhomepage:doc.guandang.net
Usingmulti-temporalLandsatimageryandlinearmixedmodelsforassessingwaterqualityparametersinRíoTerceroreservoir(Argentina)
MatiasBonanseaa, ,MaríaClaudiaRodriguezb,LucioPinottic,SusanaFerrerod
ConsejoNacionaldeInvestigacionesCientí casyTécnicas(CONICET),DepartamentodeEstudiosBásicosyAgropecuarios,FacultaddeAgronomíayVeterinaria(FAyV),UniversidadNacionaldeRíoCuarto(UNRC),RutaNacional36Km601,5800RíoCuarto,Córdoba,Argentinab
DepartamentodeEstudiosBásicosyAgropecuarios,FAyV,UNRC,RutaNacional36Km601,5800RíoCuarto,Córdoba,Argentinac
CONICET,Rivadavia1917,C1083AAJBuenosAires,Argentinad
DepartamentoMatemática,FacultaddeCienciasExactas,Físico-QuímicasyNaturales,UNRC,RutaNacional36Km601,5800RíoCuarto,Córdoba,Argentina
a
articleinfoabstract
Theapplicationofremotesensingtechnologytowaterqualitymonitoringhasspecialsigni canceforlakemanagementatregionalscales.ManystudieshaveproposedalgorithmsbetweenLandsatdataandin-situwaterqualityparametersusingclassicalregressionmodels.Thenoveltyofthispaperisthatwedevelopedalgorithmstodeterminelog-transformedchlorophyll-aconcentration(Chl-a)andSecchidisktransparency(SDT)inRíoTerceroreservoirusingLandsatTMandETM+imagery,ancillaryenvironmentalfactorsandlinearmixedmodels(LMM),obtaininganincreaseintheaccuracyoftheestimates.ThevalidationresultsshowedthatLMMwithspatialcorrelationstructurethattakeintoaccountwatersurfacetemperature(WST)andrainfallwerethemostsuitablemethodforestimatingtheseparameters.WSTderivedfromtheLandsatthermalbandwasalsovalidated.Thealgorithmswereusedtogeneratequantitativemapsprovidingspatiallyandtemporallyrichinformationonpatternsofwaterqualitythroughoutthereservoir.Waterqualityfeaturesrelatedtothehydrogeomorphologyofthereservoir,typicalseasonalityandin uxfromthecoolingsystemofalocalnuclearreactorwereidenti edinthetimeseriesmaps.
©2014ElsevierInc.Allrightsreserved.
Articlehistory:
Received2October2013
Receivedinrevisedform27October2014Accepted30October2014AvailableonlinexxxxKeywords:
RemotesensingReservoirLandsat
Waterquality
LinearmixedmodelsAlgorithms
1.Introduction
Waterisessentialforthesurvivalofalllivingthings.Partofthisresourceisstoredinlakesandreservoirs,whicharefreshwaterresourcesusedtosatisfyenvironmentalandhumanneeds.Unfortunately,waterqualityoftheseresourcesisbecomingdegraded,anddecisionmakersaredemandingnewtoolsformonitoringwaterquality.Althoughconventional eldmethodsgiveaccuratemeasurements,theyaretime-consumingandexpensive.Satelliteremotesensingisapowerfulsupportivetoolforassessingofspatialandtemporalvaria-tionsinwaterquality(Giardino,Pepe,Brivio,Ghezzi,&Zilioli,2001;Sriwongsitanon,Surakit,&Thianpopirug,2011;Triveroetal.,2013).
Anumberofsatellitesensorshavebeenusedforthestudyofsurfacewaterquality(Matthews,2011;Odermatt,Gitelson,Brando,&Schaepman,2012a;Yacobietal.,2011).AlthoughLandsatsensorswerenotdesignedforaquaticapplications(Kutser,2012;McCullough,Loftin,&Sader,2012a),severalstudieshaveproposedreliablealgorithmsbetweenLandsatdataandwaterqualityparameters,includ-ingchlorophyllandphycocyaninconcentrations(Brezonik,Menken,&
Correspondingauthor.Tel.:+543584680385.
E-mailaddresses:mbonansea@ayv.unrc.edu.ar(M.Bonansea),
crodriguez@ayv.unrc.edu.ar(M.C.Rodriguez),lpinotti@exa.unrc.edu.ar(L.Pinotti),ferrsusana@gmail.com(S.

Ferrero).Bauer,2005;Karakaya,Evrendilek,Aslan,Gungor,&Karakas,2011;Tebbs,Remedios,&Harper,2013;Torbicketal.,2008;Vincentetal.,2004;Yacobi,Gitelson,&Mayo,1995),waterclarity(Guan,Li,&Booty,2011;Hadjimitsis,Hadjimitsis,Clayton,&Clarke,2006;McCulloughetal.,2012a;Zhaoetal.,2011),totalsuspendedsediments(Bonansea&Fernandez,2013;Guang,Wey,&Huang,2006;Kulkarni,2011;Onderka&Pekárová,2008),amongothers.Thesestudiesconsid-erablyincreasedknowledgeofwaterquality,howevertheirproceduresrelysolelyonsatelliteimagerydataanddonotconsiderenvironmentalfactorsthatpotentiallyaffectwaterquality.Further,thedevelopedalgo-rithmsarecommonlybasedonempiricalrelationshipsusingclassicalsimpleormultiplelinearregressionmodelsbetweenremotelysensedre ectancevaluesandmeasurementscollectedsimultaneouslyinthe eld.Traditionalregressionmodelsassumethaterrorsmustbeinde-pendentandidenticallydistributed(iid)randomvariables.However,inenvironmentalstudiestheseassumptionsarenotfrequentlysatis ed.Linearmixedmodels(LMM),proposedbyLairdandWare(1982),arepresentedasanalternativestatisticaltool,allowingthestudyofnaturalresourceswhenclassicalregressionmodelsaredif culttouse(Zeller,Labra,Lachos,&Balakrishnan,2010).AccordingtoLarkandCullis(2004),theLMMprovidetheabilitiestohandlenon-probabilitybasedsamplingschemesandspatialcorrelationstructurebetweensamples.Inthisscenario,LMMbasedestimationsshouldprovidelessbiasedesti-matesofthepredictionvariancethanconventionalmethods(Lessels&
doc.guandang.net
0034-4257/©2014ElsevierInc.Allrights

reserved.
Mapping badland areas using LANDSAT TMETM satellite imagery and morphological...multi-temporal analyses of erosion processes at the medium scale, we ...
of Poyang Lake using multitemporal Landsat imagery...linear interpolation of the temporal sequence of ... GIS, remote sensing and hydrological models. ...
2014_RSL_Analysis of Landsat-8 OLI imagery for... images for LSWM using three NDWI models. ... certain small ponds in Region I) and mixed ...
Landsat Thematic Mapper (TM) imagery using a ...(1993) used a linear correlation between the ...and SCS models (Dymond and Shepherd 1999b, 2004...
using mathematical morphology Izzet Canarslan and ...LANDSAT 7 Ji LIANG, Xin-huan ZHANG and Jian ...A comparison of two temporal BRDF models with ...
were derived from the SRTM 90m Elevation Imagery...Landsat Thematic Mapper (TM and ETM+) and the...multi-temporal VI signatures for some typical land...
were derived from the SRTM 90m Elevation Imagery...Landsat Thematic Mapper (TM and ETM+) and the...multi-temporal VI signatures for some typical land...
热门文档