38 9 20139 GeomaticsandInformationScienceofWuhanUniversity Vol.38o.9 Sept.2013 1671-8860(2013)09-1068-05 A LandsatETM+EO-1ALI 1 2 12 (1 2350108) (2 2350108) 利用线性光谱分解模型 对同日过境的 LandsatETM+ 和 EO-1ALI 影像的不透水面信息反演结果进行对比研究 从提取精度 盖度精度两方面对两种传感器影像的不透水面反演能力进行对比结果表明 ALI 反演不透水面的能力优于 ETM+ 其提取总精度和 Kappa 系数均高于 ETM+ 高 ; 其均方根误差和系统 误差的绝对值都小于 ETM+ 两者产生差异的原因在于 ALI 的光谱分辨率和辐射分辨率均高于 ETM+ 线性光谱混合分析 ; 不透水面 ; 光谱分辨率 ;ETM+;ALI TP751 LandsatETM+ EO-1ALI ETM+ [1] ALI Goward Landsat-7ETM+ IKOOS IKOOS 1.1 ETM+ [2] ;Soudani IKO- OS SPOT ETM+ Land- ETM+ SPOT IKO- sat-7 Landsat-7 1 min [3] OS ;Chander AWiFS/ 1 EO-1 Landsat EO-1 ALI LISS- ITM/ETM+ ETM+ (1) ETM+ ALI [4] ; 10 ETM+ ASTER ETM+ ASTER ETM+ [5] 2013-04-20 (40371107); (2011Z01269)
38 9 1069 2003-03-26 Landsat ETM+ EO-1 ALI 2003-06-22 IKOOS τ= cos[(90 -θs)π/180 ] (2) 1 ETM+ALI Tab.1 SpectralCharacteristicsoftheETM+ andalisensor Blue Green Red IR SWIR TIR ETM+ ALI / μ m /m / μ m /m Band1p (0.433~0.453) 30 Band1 Band1 (0.450~0.515) 30 (0.450~0.515) 30 Band2 Band2 (0.525~0.605) 30 (0.525~0.605) 30 Band3 Band3 (0.630~0.690) 30 (0.630~0.690) 30 Band4 (0.775~0.900) 30 Band4 (0.775~0.805) 30 Band4p (0.845~0.890) 30 Band5p (1.200~1.300) 30 Band5 (1.550~1.750) 30 Band5 (1.550~1.750) 30 Band7 (2.090~2.350) 30 Band7 (2.080~2.350) 30 Band6 (10.40~12.50) 60 / / Pan 0.520~0.900 15 0.480~0.690 10 1.2 ALI ETM+ RMS= ALI (RMSE) 0.5 RMS 0.02 [10] D ( [6] ) ( ) (ICM) (IACM) [7] 4 1ETM+ ALI ALI IACM 2.2 ρλ = π ((Q λgain λ +bias λ )- (h λgain λ +bias λ ))d 2 ESU λcosθsτ (1) [8] τ [9] ChavezCOST 2 ETM+ALI ALI 1 ALI 9 (ALI );2 ALI ETM+ ALI 1p 5p ALIETM+ ( ALI ) 2.1 [10] R b = fir ib +e b fi =1fi 0 (3) R b b ; ;fi i ; R ib i b ;e b (e b ) ρ λ ;λ ; Q λ λ D ;h λ λ R imp i =flowr lowi +fhighr highi +e i (5) D ;gain λ bias λ R imp i i ;flow ;ESU λ ;d fhigh ; - ;θs R lowi R highi M b=1 1 e 2 2 ( b / M ) (4) RMS e b ;M RMS Wu [10]
1070 20139 1 ETM +ISA =0.8911ISA +0.0005 Fig.1 ResultantEndmenberImagesBasedonLSMA (R 2 =0.9998) (8) ALI ISA =0.9012ISA +0.0005 i ;e i (R 2 =0.9998) (9) 2.3 ALIISA =0.9312ISA +0.0004 3 (R 2 =0.9999) (10) (2) IKOOS 3 2 Fig.2 ExtractedImperviousSurfaceImages 2.4 3 3 ETM+ ALI ALI IKOOS 3.2 2 ALI ETM + ETM+ ALI ALI ALI 3 3 ALI 135 IKOOS RMSE SE (RMSE) (SE) RMSE = SE = (^Vi -V i ) 2 槡 (^Vi -V i ) (6) (7) ^Vi i ;V i i ;=135 3 RMSE SE 3 135 3 (ISA) (ISA ) 3 IKOOS 3 3.1 3 ALIALI Kappa ETM+ ETM+ALI ALI (83.56%) ETM + (80%) ETM+ 3.56 ; ALI(88.89%) 9 ETM+ ETM+ 9 ETM+ (2 A ) (2 B ) (2 C ) ALI ETM+ 3 R 2 0.99
38 9 1071 ALIR 2 (0.9999) (0.9312) 1 (0.0004) ;ALI R 2 ETM+ (0.9012) ETM+ 0.8911 1 ALI 9 ALI RMSE SE ETM+ 0.589 0.174 0.137-0.072 ALI 0.596 0.176 0.129-0.065 ALI 0.615 0.204 0.100-0.045 IKOOS( ) 0.661 0.275 / / ETM+ [12] Chander Thome [13-14] ALI 4 1)ALI ETM+ 9 ALI 2 ETM+ Tab.2 AccuracyAssessmentBasedonImpervious SurfacePercentage Kappa ETM+ ;ALI ; RMSE SE ETM +; 2)ALI 3.3 ETM+ALI 1) ETM+ ALI 3)ALI (1) [1] YangLHuangCHomerC Getal.AnApproach formappinglarge-scaleimpervioussurfacessyn- [411] 1 ETM+ ergisticuseoflandsat-7etm+ and HighSpatial (1 ETM+ ResolutionImagery[J].CanadianJournalofRemote A ) Sensing200329230-240 [2] GowardSDavis PFleming Detal.Empirical ComparisonofLandsat7andIKOOS Multispec- ;ALI A tral MeasurementsforSelected Earth Observation ;ALI ETM+ System (EOS)ValidationSites[J].RemoteSensing ETM+ ofenvironment20038880-99 [3] SoudaniKFrancoisCMaireGetal.Compara- ETM+ tiveanalysisofikoosspotandetm+ Data 5p ALI LeafAreaIndexEstimationinTemperatureConifer- ALI ousanddeciduousforeststands[j].remotesens- ALI ingofenvironment2006102(1)161-175 ALI [4] ChanderGCoan MScaramuzzaP.Evaluationand 2) ETM+ ComparisonoftheIRS-P6andtheLandsatSensors [J].IEEE TransactionsonGeoscienceandRemote 8bit Sensing200846(1)209-221 0~255;ALI16bit [5].ASTER LandsatETM+ 0~65535ETM+256 [J]. 201131 (7)1902-1907 ETM+ AS- [6] IrishR.Landsat7ScienceDataUsers Handbook. TER 12bit [OL].htp//ltpwww.gsfc.nasa.gov/ias/hand- 8bit book/handbook toc.html2009
1072 20139 [7]. LandsatTM/ETM+ [J]. 2007 32(1)62-66 [8] ChanderGMarkham BHelderD.Summaryof Current Radiometric Calibration Coeficients for Landsat MSSTMETM+andEO-1 ALISen- sors[j].remotesensingofenvironment2009 113893-903 [9] ChavezPS.Image-basedAtmosphericCorrections- revisitedand Revised[J].Photogrammetric Engi- 1036 neeringandremotesensing199662 (9)1025- [10] WuCMurrayA T.EstimatingImperviousSurface DistributionbySpectralMixtureAnalysis[J].Re- motesensingofenvironment200384493-505 [11] WengQHuXLuD.ExtractingImperviousSur- ternationaljournalof Remote Sensing200829 (11)3209-3232 [12].ASTER LandsatETM + [J]. 201136(8)936-940 facesfrom MediumSpatialResolution Multispectral andhyperspectralimagerya Comparison[J].In- [13]ChanderGMeyerDJHelderDL.CrossCalibra- tionofthelandsat-7etm+ andeo-1alisensor [J].IEEE TransactionsonGeoscienceandRemote Sensing200442(12)2821-2831 [14]Thome K JBiggar S F Wisniewski W.Cross ComparisonofEO-1Sensorsand OtherEarth Re- sourcessensorsto Landsat-7 ETM + Using Rail- scienceandremotesensing200341(6)1180-1 188 E-mailtang-chu-jun@163.com ALSMA-basedComparisonofthePerformancesinRetrievingImpervious SurfaceBetweenLandsatETM+ andeo-1ali TAG Fei 12 XU Hanqiu 12 (1 ColegeofEnvironmentandResourcesFuzhouUniversity2XueyuanRoadFuzhou350108China) (2 InstituteofRemoteSensingInformationEngineeringFuzhouUniversity2XueyuanRoadFuzhou350108China) roadvaleyplaya[j].ieee Transactionson Geo- AbstractTheextractionofimpervioussurfacesfromsateliteimageryhasbeenahottopicin theremotesensingfieldoverthepastdecade.everthelesswhethertheimpervioussurface informationextractedfromdiferentsensorimagesiscomparableisstilunknown.thispa- perimplementedacomplementarystudybasedonacomparisonoftheretrievedimpervious surfaceinformationfrom LandsatETM + and EO-1 ALIsensordata.Impervioussurface featureswerederivedfrom adate-coincidentimagepairofthetwosensorsbyusinglinear spectralmixtureanalysis(lsma).theaccuracyofretrievedimpervioussurfaceinformation ofthetwosensorswasassessedandcompared.theresultsshowthatthe ALIimagehas higheraccuracythanetm+assuggestedbyitshigheroveralaccuracyand Kappacoefi- cientandlowerrootmeansquareerrorandsystematicerror(inabsolutevalue).thedifer- encesinspectralresolutionandradiometricresolutionbetweenthetwosensorsarebelieved tobethemainfactorscausingthesediferenceswhenretrievingimpervioussurfaces.anin- creaseinspectralinformationin ALIsensorcanbeofhelp whendistinguishingdiferences betweenlandcovertypeswhiletheenhancementinradiometricresolutioninthealisensor canmakethesensormoresensitivitewhendetectinggroundsurfacefeatures. Key wordslinear spectral mixture analysis;impervious surface;spectral resolution; ETM+;ALI AboutthefirstauthorTAGFeiPh.Dcandidate.Hemajorsinremotesensingapplicationsinenvironmentandnaturalresources. E-mailtang-chu-jun@163.com