deriving water fraction and flood maps from modis images using a decision tree approach

《deriving water fraction and flood maps from modis images using a decision tree approach》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。

1、814IEEEJOURNALOFSELECTEDTOPICSINAPPLIEDEARTHOBSERVATIONSANDREMOTESENSING,VOL.4,NO.4,DECEMBER2011DerivingWaterFractionandFloodMapsFromMODISImagesUsingaDecisionTreeApproachDonglianSun,YunyueYu,andMitchellD.GoldbergAbstract—Thisstudyinvestigateshowtoderivewaterfractionsurface[1].Allofthevisiblean

2、dinfraredinstrumentsarepas-andfloodmappingfromtheModerate-ResolutionImagingSpec-sivesensors,suchastheMulti-SpectralScanner(MSS),Landsattroradiometer(MODIS)onboardtheEarthObservingSystemThematicMapper(TM),theAdvancedVeryHighResolution(EOS)satellitesusingthelinearmixturemodelanddecision-treeRadio

3、meter(AVHRR),theSatellitePourl’Observationdelaapproach.TherecentfloodsintheMidwesternUnitedStatesinJune2008andintheNewOrleansareainAugust2005werese-Terre(SPOT)andtheAdvancedSpaceborneThermalEmis-lectedforthisstudy.MODISsurfacereflectancewiththematchedsionandReflectionRadiometer(ASTER).TheModerate

4、-Res-landcoverdataintheMidwestpriortothefloodingeventswereolutionImagingSpectroradiometer(MODIS),whichisusedinusedforthetrainingdataset,withthesplittestmodeof50%forthisstudy,alsobelongstothistypeofsensor.Microwavesen-trainingandtheremaining50%fortesting.Theprecision,orsorssuchasRADARSAT[2],anac

5、tive(radar)sensor,areexcel-accuracyrate,ofthewaterclassificationreachesover90%fromthetest.Ourresultsdemonstratethatthereflectancedifferencelenttoolsformonitoringfloodssincetheycanpenetrateclouds,(CH2–CH1)betweentheMODISchannel2(CH2)andchannel1whichusuallyoccurduringfloodperiods..Ifwithoutvegeta-(C

6、H1)isthemostusefulparametertoderivewaterfractionfromtionortrees,radarreturnsareusuallylowoverthesmoothopenthelinearmixturemodel.Rulesandthresholdvaluesfromthewatersurface.Thischaracteristicallowsfloodextentstobede-decisiontreetrainingwereappliedtorealapplicationsondifferentterminedwithgoodaccur

7、acyundermanyconditions[3]–[6].dates(June1,17,and19,2008fortheMidwesternregionoftheU.S.)andatdifferentlocations(NewOrleansin2005)toidentifyHowever,turbulence,wind-inducedwaves,orvegetationand/orstandingwaterandtocalculatewaterfraction.Th