基于鲜叶光谱估测氮素营养的新植被指数

基于鲜叶光谱估测氮素营养的新植被指数 158 第24卷 2008年 第3期 3月 农业工程学报 TransactionsoftheCSAE V_01.24No.3 Mat.2008 Newvegetationindexforestimatingnitrogenconcentration usingfreshleafspectralreflectance Zhang]inheng,WangKe2 (1.InstituteofEco—environment&AgricultureInformation,Qinga~oUniversityofScienceandTechnology,Qingdao266042,China 2.InstituteofAgricultureRemoteSensing&InformationSystemApplication,ZhejiangUniversity,Hangzhou310029,China) Abstract:Fieldexperimentwasconductedtostudythefeasibilityofestimatingthenitrogenstatusofriceusingflesh1eaf spectralreflectance.Basedonthemechanismofnitrogentransferamongdifferentriceleaves,rededgeslopeandrededge positionofthemostupperandthe3rdexpandedleaveswereinvestigatedspecially.Andusingthetworededgeparameters(red edgeslopeandrededgeposition),anewvegetationindex(rededgecurveshoulderanglevegetationindex,abbreviationRSAVI) wascalculatedtodetectricenitrogennutritionstatus.ThecorrelationbetweennitrogenconcentrationandRSAVIatdifferentrice growthstageswasstudiedinordertoproveavailabilityofthenewindexdeterminingnitrogennutrition.Theconclusionshows thatRSAVIiscorrelatedsignificantlywiththenitrogenat1eaf1eve1andregressionanalysisindicatesthatpolynomialmodelsare creditablefordeterminingnitrogenconcentrationusingRSAVIatO.O11eve1.Thecorrelationcoe伍cients(R)anddetermination coefficient)betweenRSAVIandnitrogennutritionarealllargeandhighlysignificant,withtherangeofRfromO.867to 0.938andRfrom0.7512to0.8796.TheresultssupportthehypothesisthatRSAVIisaneffectivetoolforestimatingnitrogen concentrationinthisexperimentation. Keywords:leafspectralreflectance;vegetationindex;rice,nitrogenstatus CLCnumber:S14:S311:$435.111.31DocumentCode:AArticleID:1002—6819(2008) 3—0158O4 ZhangJinheng,WangKe.Newvegetationindex reflectance[J].TransactionsoftheCSAE.2008,24(3) OlntrOducti0n forestimatingnitrogenconcentrationusingfreshleafspectral 158,161.(inEnglishwithChineseabstract) nagriculturalresearch,thechemicalcompositionsof I cropmaterialshavebeensuccessfullyestimatedfromthe reflectancespectraofdriedgroundsamples.Approximately 42minorabsorptionfeaturesinthevisibleandnear-infrared portionsoftheelectromagneticspectrumhavebeen successfullycorrelatedwiththeconcentrationsoffoliar chemicals,suchasnitrogen,proteinandligninllJ.Many narrowbandsallowprecisequantificationofchlorophyll contentbymeasuringitspronouncedspectralcharacteristics. Asafundamentalresearchtoolfortheapplicationofcurrent andfuturehyperspectralandevenmulti—spectralsensing systems,afieldspectroradiometerwasusedtocollect reflectancedata.Somestudiesusingfieldspectrometryto Receiveddate:2007435—-16Accepteddate:2008-01—18 techResearchandDevelopment Fundationitem:ProjectsupportedbytheHi— ProgramofChina(863Program)(2007AA10Z205)andtheNationalNatural ScienceFoundationofChina(40601062,30571112) Biography:ZhangJinheng(1973一),male,Ph.D,associateprofessorat QingdaoUniversityofScienceandTechnology.Majoringinagriculturalremote sensingandinformationtechnology.Mailingaddress:53ZhengzhouRoad. QingdaoCity,CollegeofEnvironmentandSaltyEngineering,Qingdao UniversityofScienceandTechnology,Qingdao266042,China. Email:zhangjinheng@qustedu.cn measurenitrogenconcentrationofcropshavebeen conductedbyotherresearchers.Thenitrogenstatusof variouscropshasbeenrelatedtoreflectanceatspecific wavelengths,andvariousreflectanceratiosandindiceshave beenusedasindicatorsofnitrogendeficienciesinplants”. Spectraldeterminationsprovideanautomatic,quickand nondestructivemethodofestimatingcropnutrientstatus. Observeddifferencesinspectralcharacteristicsbetweenrice cropsofdifferingnitrogenstatussuggestthatitoughttobe feasibletoevaluatericenitrogenstatusfromleafreflectance measurements. Derivativespectraindicatetherateofchangeof reflectancewithwavelength(dR(k)/),whichistheslope ofthereflectancecurveatwavelength九.Derivativeanalysis allowsonetocorrelatetheshapeofthereflectancepa~emto nitrogenconcentrations.Derivativeanalysishasbeen appliedbyresearchersinstudyingthespectralcharacterist— icsofchlorophyllcontentandnitrogenconcentrationin crops[12-14]. Thepurposeofthisinvestigationistodevelopanew vegetationindexfromderivativeanalysisofleaf hyperspectralreflectancemeasurementsdataandthen determinethefeasibilityofestimatingtheNstatusofrice basedonthenewvegetationindex. 第3期张金恒等:基于鲜叶光谱估测氮素营养的新植被指数159 1Fieldexperimentanddatameasurements Ricefieldexperimentwasconductedinaploughedfield, 3.02hminarea,latitude31.to30.55’,longitude120.43’to 120.50’.Theannualaveragetemperaturewas15?.The lowesttemperaturewas4?.andthehighest29?.Thefield wasdividedintothreeareastowhichdifferentratesof nitrogenwereappliedasfollows,anareaof0.47hmwhich didnotreceiveanynitrogenfertilizer(N0);anareaof0.23 hm2whichreceivedhalfoftherecommendedrateof nitrogenfertilizer(N1);theremainderofthefieldreceived therecommendedrate(N2)Therecommendednitrogen applicationrateswere135kg/hm2urea, 202.5kg/hm2urea and120kg/hm2ureaatthetillefing, bootingandheading stages,respectively.Theareaswerecroppedwiththree genotypesoffice,G1(1aterice101),G2(1atericeC一67)and G3(1aterice004). Riceleafreflectancemeasurementsinthe350nmto2500 nlTlwavebands(1aminwidth)weremadeusingaportable spectroradiometerFieldSpecFR(350—2500nm).Thesensor, afiberopticof8.fieldofview,washelddirectlyabovethe leafthatwasmeasured.Alampwasusedtoilluminatethe plantsfrom10cmabovetheleaf.Themeasurementswere madeatthetillagestage,bootingstageandheadingstage. O.O12 0.010 0.008 0.006 0.004 0.002 O Wavelength/n Tenleafspectralreflectancemeasurementswereconducted. Leavesofricewerethendetached,oven—driedat70.Cto constantweightandthenanalyzedfortotalnitrogenbythe .Thenitrogenconcentrationsinleaves Dumasmethod【 wereexpressedingN?kg—dryweight. 2Resultsandanalysis Fromthedataofmisresearch.theaverageofrededge1st derivativespectralcurvesofthemostupperandthe3rd expandedleaveshadtwoobvioustransfigurationpoints,the wavelength699,704nm(1sttransfigurationpoint)and (720~1)nn’l(2ndtransfigurationpoint).Thetwo characteristics,rededgeslopeandposition,resultin shouldersbetweenthewavelength699~704nmand(720~1) nn’lofthe1stderivativespectralcurves.Andanglesexisted betweentheshouldersofthemostupperandthe3rd expandedleaves,whichisduetothetransferofnitrogen nutntionamongdifferentleavesresultedinrelevantchanges oftherededgeslopeandposition.Thusbasedontheangle oftheshoulders(699--704nlTlto720nm)betweenthemost upperandthe3rdexpandedleaves,anewvegetationindex, calledrededgecurveshoulderanglevegetation index(abbreviationRSAVI,seeFig.1)wascalculated. 67069071O730750770 Wavelength/nln Fig.1VariationsoftheRSAVI(L1denotesthemostupperexpandedleafandL3denotes the3rdexpandedleafThemeaningofa1,andRSAVIseeformulas1to5) ‘I_heformulasareasfollows: 1)RSAVI=a1+a3 2)Where,aWarctg((r1Im699.704)一r720(1))/A~,1) 3)0【3arctg((r3m”(699.704)一r720(3))/A~,3) 4)AX1I,1m699.7041.~r720(1)I 5)AX3=l,3m699.704)-~r720(31l Wherer1m”f699-704)一一memaximumvalueofthe1st derivativespectralreflectanceofthemostupperexpanded leavesatwavelengthrange699--704nm;r3ma)【f699- 704)—— themaximumvalueofthe1stderivativespectrareflectance ofthe3rdexpandedleavesatwavelengthrange699~704 nm;r720(1)--the1stderivativespectralreflectanceofthe mostupperexpandedleavesat720nm;r72o(3)the3rd derivativespectralreflectanceofthemostupperexpanded leavesat720nm;_—一waVelengthposition.nm;?—一 wavelengthrange,nlT1. Decreasednitrogenlevelsresultedinnitrogendeficiencies inrice,asaresultnitrogennutritionwastransferredtoupper portionfromlowerportionofriceobservably.Andthe changeofshouldersofrededge1stderivativespectracurves wassteep.Asaresulttheanglesofthe1stderivativered edgespectracurvesbetweenthemostupperandthe3rd expandedleavesincreased,thereforethevaluesofRSAVI increased.However,whenenoughnitrogenwassuppliedto riceatthewholegrowthcycle,nitrogennutritiontransfer indistinctivelyindifferentportionsofrice,andthechanges ofshouldersofrededge1stderivativespectracurveswere notsteep,asaresultthevaluesofRSAVIdecreased. Theghercorrelationwifhni~ogenconcentration. thehighercorrelationcoefficientsrRvaluesfrom0.867to 0.938)werefoundatvariousgrowthstages.Andregression modelswereestablishedtodeterminethefeasibilityof estimatingtheni~ogenstatusofricebasedonRSAVI. Polynomia1regressionmodelswereusedtorelateleaves reflectanceinRSAVItotheconcentrationofnitrogeninleaf tissueatthevailOUSkeygrowthstages. Table1Modelsforestimatingleafnitrogenconcentration Note:xdenotesRSAVI,YdenotesriceNconcentrationme~edby samplesofthisexperiment,Rdenotesthedeterminationcoefficient,andthe FvaluesofFtestindicatethatthesemodelsaresignificantatO.O1leve1. ThemodelswereevaluatedbytheR(determination coemcient),thevalueoftheFtestandtheRMS(Residua1 MeanSquare).Theresultsoftheregressionanalysisandthe FtestsweregiveninTab1e1.Itisevidentfromthetablethat theR2valuesoftheregressionmodelswereal1largeand highlysignificantfrom0.75lt00.880)at0.01leve1. A= ?(一i=1 r/——2 WhereYi(f=l,2,…,n)denotesthetotalnitrogenofrice leavesobtainedbytheDumasmethod,and多(i=1,2,…,n) denotesthevaluesofregressionanalysis,andndenotesthe numberofthesamples.ThelowerthevalueofRMSandthe higherofdeterminationcoefficientmeanthehigheraccuracy ofRSAVItoestimatenitrogenstatususingtheregression mode1s 3Conclusions Anewvegetationindexcalledrededgecurveshoulder anglevegetationindex(abbreviationRSAVI)fromthelst derivativespectra1reflectanceofrededgerwavelength rangingfrom699—704rimto720rim)wasgivenbasedonthe angleoftheshoulders(699—704nmto720nm)betweenthe mostupperandthe3rdexpandedleavesinthisinvestigation. Theobjectiveofinvestigatingthecorrelationbetween RSAVIandni~ogenconcentrationofriceistotrytofindout thefeasibilityofdeterminingnitrogennutrition.So regressionmethodwasselectedbasedonRSAVIand nitrogenconcentrationattheprobabilityofFdistributingat 0.0l1eve1withthesupportofSPSSsoftware.Theresultsof thisresearchshowedthatsignificantcorrelationexisted betweenRSAVIandnitrogenconcentrationinleaftissueat thediversegrowthstages.AndFtestofregressionmodels indicatethatthesepolynomia1modelsarecreditableat0.01 leve1.Thehighercorrelationcoefficientsanddetermination coefficientofestimationmodelsshowthatitshouldbe possibletomonitornitrogenstatusofriceusingRSAVI theoretically.Becauseofthelimitofexperimentconditions (suchastraffic,time,manpowerandmateria1resources),the projecthadbeencarriedoutjustduringonegrowthcycle. ThemechanismofRSAVIforestimatingnitrogennutrientis notveryclear,butatleasttheresearchsupportsthe hypothesisthatRSAVIisaneffectivetoolforestimating nitrogenconcentration. IReferences1 [1]CurranPJ.Remotesensingoffoliarchemistry[J].Remote SensEnviron,1989,30:271--278. [2]ZhangJinheng.Ricenitrogennutritiondiagnosisusing continuum-removedreflectance[J].JournalofPlantEcology, 2006,3O(1):78—82(inChinesewithEnglishabstract). 【3】ZhangJinheng,WangKe,BaileyJS,eta1.Predicting nitrogenstatusofriceusingmultispectraldataatcanopy scale[J].Pedsphere,2006,l6(1):108—117. [4]GoelPK,PrasherS0,LandryJA,eta1.Potentialofairborne hyperspectralremotesensingtodetectnitrogendeficiency andweedinfestationincorn[J].ComputersandElectronicsin Agriculture,2003,38:99—124. [5]StrachanIB,ElizabethP,JohanneB,eta1.Impactofnitrogen andenvironmentalconditionsoncornasdetectedby hyperspectralreflectance[J].RemoteSensingofenvironment, 2002,8O:213--224. [6]SerranoL,FilellaI,PefiuelasJ.Remotesensingofbiomass andyieldofwinterwheatunderdifferentnitrogensupplies[J]. CropSci,2000,40(3):723--730. [7]LukinaEV,RaunWR,StaneML,eta1.Effectofrow spacing,growthstage,andnitrogenrateonspectralirradiance inwinterwheat[J].JournalofPlantNutrition,2000,23(11: 103—122. [8]BlackmerTM,WhiteSE.Usingprecisionfarming technologiestoimprovemanagementofsoilandfertilizer nitrogen[J].AustJAgricRes,1998,49(3):555--564. [9]MaBL,MorrisonMJ,DwyerLM.Canopylightreflectance andfieldgreennesstoassessnitrogenfertilizationandyield ofmaize[J].AgronJ,1996,88(6):915--920. [10]BuschmarmC,NagelE.Invivospectroscopyandinternal opticsofleavesasbasisforremotesensingofvegetation[J]. IntJRemoteSens,1993,14(4):711--722. [11]GamonJA,PefiuelasJ,FieldCB.Anarrowwaveband spectralindexthattracksdiurnalchangesinphotosynthetic efficiency[J].RemoteSensingofEnvironment,1992,41: 35—44. [12]LuohengHan.Estimatingchlorophyll-aconcentrationusing 第3期张金恒等:基于鲜叶光谱估测氮素营养的新植被指数161 first’derivativespectraincoastalwater[J].International 5244. JournalofRemoteSensing,2005,26(23):5235-- [13]BrogeNH,MortensenJVDerivinggreencropareaindex andcanopychlorophylldensityofwinterwheatromspectral reflectancedata[~.RemoteSensingofEnvironment,2002, 81:45—57. [14]BrogeNH,LeblancE.Comparingpredictionpowerand stabilityofbroadbandandhyperspectralveg~ionindices f0restimationofgreen1eafareaindexandcanopy chlorophylldensity[~.RemoteSensingofEnvironment, 2001,76:156—172. [15]BaoShidan.SoilandAgriculturalChemistryAnalysis[M]. Beiiing:ChinaAgriculturePress,2000:264. 基于鲜叶光谱估测氮素营养的新植被指数 张金恒,王珂 (1.青岛科技大学生态环境与农业信息化研究所,青岛266042; 2.浙江大学农业遥感与信息技术应用研究所,杭州310029) 摘要:采用田问试验的方法开展利用鲜叶光谱反射率估测水稻氮素营养状况的 研究.基于氮素在水稻不同功能叶片之间运转规律 的机理,文章重点分析了第一和第三完全展开叶红边斜率和红边位置的变化,并 基于红边位置和红边斜率构建了一个新的植被指数 (命名为”红边曲线肩夹角植被指数”,简称为RSAVI)监测水稻氮素营养状况.为 了 证明 住所证明下载场所使用证明下载诊断证明下载住所证明下载爱问住所证明下载爱问 RSAVI在监测水稻氮素营养状况的可行 性,分析了不同生育时期氮素含量和RSAVI之间的相关性.结果 表 关于同志近三年现实表现材料材料类招标技术评分表图表与交易pdf视力表打印pdf用图表说话 pdf 明RSAVI和叶 片氮素含量显着相关,相关系数介于0.867~0.938 之间.并且RSAVI和氮素含量之间建立的回归模型以多项式模型效果最佳,决定系数介于0.7512,0.8796之间,模型均通过0.01 水平检验.因此研究结果表明,在本次试验中使用RSAVI估测水稻氮素营养是可行的. ;水稻;氮素状况 关键词:叶片光谱反射率;植被指数