Exploring TM Image Texture and its Relationship with Biomass
Transcrição
Exploring TM Image Texture and its Relationship with Biomass
ACT Publication No. 05-13 Exploring T M Image Texture and its Relationship with Biomass Estimation in Rondenia, Brazilian Amazon. Dengsheng Lu and Mateus Batistella Reprinted from: Acta Amazonica. 35(2) 2005: 249-257 -- - ~ Anthropological Center for Training and Research on Global Environmental Change Indiana University, Student Building 331, 701 E. Kirkwood Ave., 47405-71 00, U.S.A. Phone: (81 2) 855-6181. Fax: (812) 855-3000, Email: [email protected], internet: www.indiana.edu/-act ACTA AMAZONICA L VOLLIME 35 N02 * ABRII. :JUNIIO 2005 AMAZONICA I MANAUS WONAS BRASIL ACTA AMAZONICA I VOLUME 35 N." 2 * ABRII. / JUNHO 2 0 5 Esta publi@o tem seus artigos indexados em: American lnstirute of Biological Science; Biological Abstracts; Biosciences Information Service; BlREME - Centro I ~ t i n oAmerican0 de Informadon en Ciencias; CAB Abst;, Cambridge Scientific Abstracts Service; Chemical Abstracts; Current Awareness; SoilCD; TreeCD; University Microfilms International; Zoological Records, SciELO, Peri6dicos CAPES. EXPEDIENTE PRESIDENTEDA wiTB~lc.4 Luis Inicio Lula da Silva COORDENA@ODE EICIENSAO Jackson Fernando Rego MINISTRODA CI~NCIA E TECNOLOGIA Eduardo Campos k Design Editorial DIRETOR DO INPA Jost Antonio Alves Gomes EDITOR-CHEFE George Henrique Rebelo Instituto Nucionalde Posqutsas datlmaz$nia EDITORES-ADJUNTOS Antonio Carlos Webber LJniuersfdndeFederddoAmawnus Efrem Jorge Gondim Fefieira InstituteNucionaldePEsquisus&a Amazdniu COMISSAO EDITORIAL Alceu Ranzi UniuersidaJeFedcrai do Acre Augusto Loureh H e m i ues lnstituto.\ucirmaldePesqutiac 2Amadnia Edinaldo Nelson dos Santos S h Instituto NudonuldePesqui.wsdaAmuzi3nia Edineia Mascarenhas Dias PROJETO G&ICO E EDITORA~O ww.attema.com.br P R O D U EDITORIAL ~ Geo e Tokuwo Nakamura Ttto ernandes Shirley Ribeiro Cavalcante ClAudio Lima Elione Angelim Benj6 Odintia Garcia Bezerra Y ESTAG~IOS onathas Moraes Brandgo osut Lima da Silva ASSrnAM YMa Video de S. Penedo Wda de Oliveira Santos SUGESTOES, CR~TICASE D ~ D A S acta@in a.gov.br shirleyc&npa.gov.br TIRAGEM 1.000 exemplares F010 DACAPA Antonio Manzi I ~ t i t u tNacionuldePqutsasda o Amartmiu Ernest0 Renan Freitas Pinto UniwrsidadeFedarJdoAmawnar Hiroshi Noda brredefluxodoLBA,cvm SOm ds dtura, instaladn la Reservu Biol6gica &Jam IBAMA), emJi-&rat@ RO. ImtifwoNudonulde Pesquisus da Amaz6niu llse Walker Insti'IutoNucional de Pvsquisa~ du Amadnia Luiz Alberto dos Santos Monjel15 LiniuersidadeFoder~ld o ~ m Marlene Freitas da Silva UniuwsidadedoEs~dodoAmawnus Peter Mann de Toledo MuseuPu-e Emilio CoeIdi Reinaldo Imbrozio Barbosa lmtitwo NmMonaldePpsqufsasdaAmaz6nia Sandra do Nascimento Noda IlnlvasidadePederaldo Amarona BtZ0NICi-I Av.An& Araujo, 2936 A l e h CEP 69011-970 Manaus Amazonas Bmil c;lbuF'asxal478 fhe+55612)6423@8 *FQc +55(92)60.3223 E-mail: [email protected] Site: http://acta.inpa.gov.br/ ACTA AMAZONICA I 1 VOL 35(2)2005 I ii - 300 SUMMARY BERGO. Ceko ; MENDONCA, H 6 l i ~: SII* M m o s -Effect of time and kquency ofcutting in the essenti~loil pnxluction Ill of long peppcr (Piper hispidineruun~C. 1)C.) ..................... IIORI1E,Atlriana; GOMB, Isr;~el;MIRANDA, Scbastiio; SII.V.4, Maria;. Contribution to the hydrochemestry ofdrainages in the municipality of mwaus -Am 119 . .. LIMA, Jose; SAN'rOS, Joaquim dos: HIGIJCHI, Niro. - Lumher industries in thc stateofAmazonas:situation in 2000. 125 AMARN, Bmndito; -Fisheries gmd Fichiie&)nat the indigenous resemh~haninkaKaxinaxxi, river Breu, B d Y e n r . .. ........... ....... ... 133 COSTA, Maycira; - Use of radar imagery for estimating net primary productivity of aquatic vegetation in t h e Amazon floodplain. 145 ~ SIIVA IXAS. Maria; COI-IEN,Julia; GANDU,Aclilson; - Clouds. rain and biosphere interactions in Anxwon. 215 - S01!7A FILHO, Jose; RIBEIRO, Aristides; COSrA, Marcos; COHEN, Julia; - Control mechanisms of the seasonal variation of trnnspiration in a northeast amazonian tropical rainforest . 223 .. -..... . . . .-. WILI.lAMS, Earlc;DI\LL' AN'VONIA, Naor: DALC AN'TONIA, Vitoria; ALMElD'A,Jorge;SUAIKL, Francisco; LIBBMANN, Bnnt; MALHADO, A& -.She drought of the century in the Amazon Basin. an analvsis of dle reeional variation of r&nf;lll in South America in 1926. 231 ., BATISELLA, Mateus; MORN, Emilio F.;- Hunlan dimensions ofland use and lnntl cover in the Amazon: acontribution .-~ fmm 1.DA. 239 ..-..-. -- ... - ...... - ... .- ....... .- .... ~~ EsP~wI'O-SANTO, F e n w ~ d I.uiz; MACHADO, Evandro; -Analysis of the floristic and phytosocioiogic composition of 'hpaj0s n;~tionalforest with geognphic support ofsatellite images. 155 LU, Oengsheng; BPII'IS'I'EI.IA, Mateus;- lixploring TM image texture and its relationships with biomass estimation in RomlBnia, Brazilian Atnazon. 249 ~ ALBIIECH?: Rachel; SLLVA D M , Mu*. - Microphysicalevidence of the transition between predomin;int convective/ stratifor~nrainfall associated with the intraseasonal oscillation in the Southwest Amazon. 175 .... . . . . ~ ARTAXO, Paul<>;GATTI. Luciana V.; LEAL, h a M. C6rdova: LONGO, Karla M.; PRCITAS, Saulo R.; LARA. Luciene I..; PAULIQLWS, Thcotonio M.; ~ ~ 0 ~ 6 ~ 1 0 ,S.;~1<1220, line LucianaV: Atmospheric Chetnisuyin Anlmnia: d ~ forcst e and the biomass burning emissions controlling the composition of the Anuazonian :itrnosphere. 185 - .......... -- . - KKUSCI IE. Alex: DAIIGSTGR. Maria: VICTORIA. Re~mldo: ~ NOVO, Evlyn: I:ERIU?lRA, Laerte; BARBOSA, Cliudio; CARVALHO, Claudio; SANO, Edson I?.; SHIMABUKUIW. Yosio; HUETI:, AITredo; I'O'I'I'ER, Christopher; KOBEI<l'S, Dar A.; HESS. iaura I..; MtilACK,JohnJ.;\'OSI IIOKA, IJiiki: KT.OOS'T1:R. Steven; KUMAK. Vipin; MYNliNl, Ranga; KA'SANA. Piyachat: IIIDN, Kamel; MIURA, Ibmoaki; Aclvanccd rcmotc sensing tcchiniqucs for glol>alchanges and Amazon ecosystem functioning studies. 259 ........................... -. . . COKIEIA, Fmncis; A I Y A Regina; ~ MANU, Ant6nio: GIELOW; Ralf; KUBO'SA, Pauio; -Calibration of the simplificul simplc biosphere rnodel (SSiB) for Amazonian pasture and forest sitcs using LBA data. 273 .............. ~ INOUE. Llis;AFONSO, Luis B.;IWAMA, George K.; MORAES, G i i k n o . - IiUects of clove oil o n the stress response o f - E~ectcdflandu& changes in the bi~geoch~misu~offluvih matrirlxi (!jrycorr ce~halus)stibiected to transport. 289 ~ ~ t e noftIleJi-Paran5 ls tivet basin. R o n d 6 ~ a . 1%' GIL-SANTANA, tlelcio R.; COLETTO-SILVA, Alexandre;. BsrerLgeri~Jrufasli, n. pen., 11. sp. of Keduviinae from MORAES, Bergson; COSTA, Jose: COSrA, Antonio; COSS'A, Reserva Ducke, Anlazonas State, Brazil (HemipteraMarcos; - Spatial and tempomlvariation ofprecipis~tionin Heteroptern, Reduviidae). 297 207 the State olPai-L. ~ ~~ / ACTA 8 AMAZONICA I Exploring TM Image Texture and its Relationships with Biomass Estimation in Ronddnia, Brazilian Amazon. Dengsheng LU1; Mateus BATISTELLA2 ABS'IRACT Many twrmre measures have been dewloped and used for improving landaverdassification accuracy but d y has research examined the role of textures in impmving the performance of aboveground biomass estimations. The relation5hip between texture and biomass is poorly understood. 'l'his paper used Landsat Thematic Mapper (TM) data to explore relationships between'l'M image textures and atmeground biomass in Ronddnia, Brazilian Amawn. Eight grey level co-occurrence matrix (GLCM)based texture measures (i.e., mean, variance, homogeneity, contrast, dissimilarity, entropy, second moment, and correlation), associated with seven different window s i m (5~5,7~7,3x9,llxll, 15x15,19~19, and 25x25), and five TM bands (TM 2,3,4,5, and 7)wereandyed. Pearson's correlationmefficientwas used to analyze texture and hiomas relationships.This research indicates that most t m u p s are weakly correlated with successional vegetation biomass, but some textures are s ~ a n t ldy t e d with matun:forest biomass. In conmt, TMspfftral signaturesaresigru6candycorrelatedwith successional vegetation biomass, but weakly correlated with mature forest biomass. Our findings imply that textures may be critical in impmving mature forest biomass estimation,hut relativelylrss important hsuccessionalvegetation biomass estimation. KEYWORDS texture, aboveground biomass, TM image, correlation, Amazon Explorando texturas de imagens TM e sum relaghes com estimativas de Biomassa em Rondbnia. RmMo Muitas medidasde lextura t h sido desenvoluidase utilizudaspara melhorara acurdcia de c l m s i p ~de s cobertura das t m , mmrammmte t&m-seaualiadoa importdncia dessas medidas em estimatiuas de biomava. Este t r a b a l h utillzou dadmLandsat TMpara explorur as relaqcies entre texturas de imugms ZU e bionwssa em Rondhnia, M n i a Forum analisadas oito medidas de textura buseudus em m a t e e s de cwcorr&cia de tons de cinur (i.e., ddia, uaridncia, ' ,mfmste, , -d entrt@, segundomomfoemmh@o), ~ m m s e t e d t f ~ ~ 7 x 7 , 9 x 9 , 1 1 x 1 1 , 1 . 5 x 1 5 , 1 9 x 1 9 e 2 5 x 2 5e ~) b a n d u s i W ( l M 2 , 3 , 4 , 5 e7).i n d i c e s d e c o m ~ d e p e a r s o n foram utilizndospara analisar a s rel-s entre textura e biomassa. Estapesquisa indica p e a maioria das medidasde tertura sciopuco m I m a o o & c o m biomussa de vegeta@io senr&a, mas dgumas medidasde tertura t5m comh@o significativa com a biomassa deformcrgciesflorestais mudurus. Ao contrdrio, assinaturus espectrais de handas m s d o ~ ~ u i t i u a m e n t e m r r e ~ m m a h ~ d e v e g e ~mdmlf iiu a u n,n e n t e a n r e l a c i ~ m a b i o m s a deflomtas maduras. 0sresultados i n d i m que medidas de t a t u r a srio impntantes em esfimativasde b w m a w deflorestu mudura, mas relatiwmente m a o s imporantespara estimatiuasde biomaw de vegekqio senmdriria s ~ 1 EXPLORING TM IMAGE TEXTURE AND ITS REMTIONSHIPS WITH BIOMASS ESTIMATIONIN ROND~NIA,BRAZILIAN AMAZON. AMAZONICA Lp --A effectively extract biomass information? During image processing, what size of moving window is appropriate for a specillc texture measure?This paper examines image texture and vegetation aboveground biomass relationships using Landsat Thematic Mapper (TM) data in RondBnia, Brazilian Amazon, as an effort to find suitable textures for biomass estimation. INTRODUCTION Aboveground biomass governs the potential carbon emission that could be released to the atmosphere due to deforestation. Accurate biomass estimation is necessary to understand impacts of land-useAand-cover (LULC) change on globalwanningandenvimnmental degradation. In recent years, biomass estimation has attracted scientific interest because regional changes in biomass havebeen aysociated with climate DS and ecosystem changes. 'She Large Scale BiosphereAtmosphere Experiment in AmazBnia (LBA) has driven its attention to these issues contributing to the development of study area techniques for sustainable use of the land in the Amazon environment. The results presented in this article belong to a ~ ~has experienced ~ intensive d LULC ~ change ~ in thei r m c h proiectletlb~IndianaUnivmityand E m b q Satellite past three decades. 'The deforestation rates in this State range Monitoring with the support of LBA. from 1.14 to 2.62% per year between 1991 and 2000, much The advantages of remotely sensed data over traditional higher than the overall deforestation rate in the Brazilian field inventory methods for biomass estimation have been Amazon region, which ranges from 0.37 to 0.80%per year at indicated by a number of publications (Saderet a/., 1989; Roy & the same period (INPE, 2002). Following the national stntegy Ravan, 1996;Boyd et a/., 1999; Nelson et al.,2000; Steininger, ofregional occupation and development,colonizationprojem 2000; Luetal., 2002a). However, previous researchhas shown initiated by the Brazilian government in the 1970s played a the difficulty of biomass estimation based purely on remote- major role in this process (Moran, 1981). Most colonization sensing spectral signatures because of the influence of projects in thestate weredesigned to settle landless migrants. increased canopy shadowing within large stands, the The immigrants transformed the forested landscape into a heterogeneity ofvegetation stand srmaures, and spectral data mosaic of cultivated crops, pastures, and different stages of saturation (Spanneretal., 19% Roy&Ravan, 1996;Steininger, successional forests (Batistellaetal., 2003). Over the Amazon 2000). The complexity of forest stand structure and species composition in the BSI~ mw moist tropical regions result in highly variable , standing stocky of biomass and even more /' /' variable rate of biomass accu~nulation /,' following a deforestation event. This I' presents a challengefor biomass estimation, /' achallenge that must be addressed. Texture often refers to the pattern of intensityvariations in an image. Many texture measures have been developed (Haralick et d., 1973;He &Wang, 19W U r n , 1995;Riou RondOnls & Seyler, 1997).In previous research, texture measures were mainly used for LULC classification (Franklin & Peddle, 1989; Marceau etal., 1 9 9 ;Augusteijn etal., 1995; Franklin et a/.. 2000: ndi Nvoungui et al.. 2002; Podest & Saatchi, 2002). 0fjle many texture measures, the grey-level cooccurrencematrix (GIKM) may be the most common texture used for improving LULC classification (Marceau et a/., 1990: Franklin et al., 2000; 'ndi ~ ~ o u n g et u ial., 2002). MACHADINHO D'OESTE However, rarely has research focused on extraction of biomass information using texture measures.?he relationships between textures and aboveground biomass are especially poorly understood. For example, how are textures related to biomass estimation? Which texture measure can Figure 1- Location of the study area in the State of RondBnia, Brazilian Amazon. A ,' #' 250 1 VOL. 35(2) 2005: 249 - 257 - LU & BATISTELM ~ . ACTA EXPLORING TM IMAGE TEXTURE AND 11s REtATloNsHlPs WITH BoltvU6-S ESTIMATIONIN R O N ~ N I A . BRAZILIANAMAZON. AMAZONICA region, a significant amount of the deforested area is in some stage of secondary succession, requiring that we carefully evaluate biomass and carbon dynamics represented by their heterogeneous vegetation stands. To ~laluatethe relationship between image texrures and aboveground bionlass estimation, we chose the settlement of Machadinho d'Oeste,located in northeastern RondBnia (Figure 1). This is anewercolonizationnmiect thanareas alonr BR-364 (kuiahh-PortoVelho ~i~hway).'Th;setdement of ~ac'hdinho covers about 2,000 km2.Machadinho is adjacent to thehordes with the Statesofhiwunas and Mato Crosso,which may offer potentials and consmints for future initiativesof conservation anddevelopment.~lheterrainundulat& ranginghm 100to 450m above sc-d level. The climate is classifiedaqequatorialhot and humid, with tropicaltlansition.Thewelldefined+season 1astsfmmJunetoAugustthe anndanrxgcpmipitation is 2,016 mm,and the annuai a\.eragc trmpratw& is 2 5 . i (.(Kond6nia. 199898).Scvetal soil noes. includinealfisols. oxisols. ulti.wls. and &d soils have &n identitimi (&nola & k, 1999) Field vegetation Inventory and biomass estimation Fieldwork was conducted during the dry season of 1999. Prrliminary image clasi6cation and band co~npositeprintoub indicated candidate areastobe surveyed, and a flight overthe amas provided visual insights a h u t the size, condition, and accessibilityof each site. The surveys were conducted in m a s with relatively homogeneous ecological conditions (e.g., topography, distance from water, and lancl use) and uniform physiognomic characteristics. ffier defining the area to be surveyed (plot sample), three subplots (1 m', 9 m2,and 100 m') were randomly selected to accurately represent the variabilitywithin the plot sample. The center of each subplot was randomly selected. 'lbtal tree height, stem height (the height ofthe first main branch), and diameter at breast height (DBH) were measured for all trees in the 100m'ma. Height and DBH were measured for aU saplings in the 9 m2area. (;round cover estimation and counting of individuals were conducted for seedlings and herbaceous vegetation in the 1 m2m a . Here, seedlingswere defined as youngupes or shrubs with a stem diameter smaller than 2 cm, saplings as young trees with a stem DBH greater than 2 cm and smaller than 10 cm, and trees as woody plants with a DBH greater than or equal to 10 cm. Adetailed description of field data collection can be found in Batistella (2M)l)and Lu e t d . (2004). A total of43 sample plots were inventoried, including 29 plots for d i i r e n t stages of successional vegetation and 14 plots for mature forests. Table 1 summarizes sample plot distributions as well as the statistical ch~racteristicsof abovegmund biomass. The vegetation ages of successional forests range h m 2 to 13 years. 4d3rabnst.w huilt to integrate all vcgctation data collect~rl dunt~etieldwork.Eauati(~nI 1 ) \rasuslrl tndculate inditidual tree Gkass &el al., 1%) and ~ q n t i o (2) n for individual saplingbiomac8(Horuaketd., 19%). YT = 0.0326"' (DT)' '"Hand (1) YS = exp[-3.068 0.957 In (DS' "'H)], (2) where D1' and DS are the tree and sapling DBH in centimeters, respectivrly; ki is the total tree or sapling height in meten; and YT and YS are the individual tree and sapling biomassin kilograms, respaively,Abmqmund biomass (AGB: w m 2 )was then calculated thmugh Eqwation (3). + where m is the tor4 tree number in a selected plot (three 100m2subplotsin aplot) ,and n is the total saplingnumber in aselected plot (three9 m2subplntsin aplot). PAand SPAmthe plot areas(in squarr-meters) for tree and sapling measurements. - Table 1 Distribution of sample plots and sutistical characteristics. -~~ Successional for& Sample plot distributions Biomass (kglrn2) Plots (N) 2-4 5 5 4-6 4 6-8 5 8-10 1 10-12 4 12-14 4 14-16 16.18 1 TO p l~o t ~ 29 Mean 8.350 Minimum Statistical characteristics Maximum Std. Dev. 4.455 - --~ 251 1 VOL. 35(2)2005: 249 - 257 LU & BATISTELLA -. - ~p~ ~ - Mature forests Biomass (kglrn2) 10.14 14-18 18-22 22-26 26-30 30.24 34-38 >38 2 -14 24.76r ~ - 11.805 -