@InProceedings{Macedo:2008:UtMoMi,
author = "Macedo, Rodrigo de Campos",
affiliation = "{Instituto Nacional de Pesquisas Espaciais (INPE)}",
title = "Utiliza{\c{c}}{\~a}o de Modelo de Mistura Espectral para
Estratifica{\c{c}}{\~a}o Et{\'a}ria em Plantios de Eucalyptus
spp",
booktitle = "Anais...",
year = "2008",
organization = "Simp{\'o}sio Regional de Geoprocessamento e Sensoriamento Remoto,
14.",
note = "Setores de Atividade: Silvicultura, Explora{\c{c}}{\~a}o
Florestal e Servi{\c{c}}os Relacionados Com Estas Atividades. and
{Informa{\c{c}}{\~o}es Adicionais: This present research has the
goal to analyse the utilization of Model of Spectral Unmixing for
separate Eucalyptus classes of age. The premise is that there is
relationship between radiometric data extracted from LANDSAT-5/TM
images and volume data estimated from Eucalyptus ssp. stands.
There were four treatments and one testimony: i) fractions
vegetation plus shadow} and {ii) fractions vegetation plus shadow
plus soil} and {iii) fractions vegetation plus shadow and band 4
(NIR)} and {iv) fractions vegetation plus shadow and NDVI} and and
the testimony: bands 3 and 4. The study area is covered by
Eucalyptus stands from VCP (Votorantim Celulose e Papel) company,
located in Cap{\~a}o Bonito town, S{\~a}o Paulo state, Brazil.
The fractions images was generated using surface reflectance from
TM bands (B3, B4 and B5). The evaluation of classifications used
three criteria: Confusion Matrix, Accuracy and Kappa Coefficient.
The best result (for Model of Spectral Unmixing) was the
utilization of fraction vegetation plus shadow, but the
classification with the testimony (bands 3 and 4) had higher
values of Accuracy and Kappa Coefficient..",
keywords = "invent{\'a}rio florestal, processamento digital de imagens,
sensoriamento remoto.",
abstract = "O presente trabalho teve como objetivo avaliar a
utiliza{\c{c}}{\~a}o do Modelo de Mistura Espectral para
discriminar classes et{\'a}rias de Eucalyptus spp. A premissa
{\'e} que h{\'a} rela{\c{c}}{\~a}o entre dados
radiom{\'e}tricos extra{\'{\i}}dos da imagem LANDSAT-5/TM e
volume estimado dos talh{\~o}es de Eucalyptus ssp. Realizaram-se
quatro tratamentos e uma testemunha: i) fra{\c{c}}{\~o}es
vegeta{\c{c}}{\~a}o + sombra; ii) fra{\c{c}}{\~o}es
vegeta{\c{c}}{\~a}o + sombra + solo; iii) fra{\c{c}}{\~o}es
vegeta{\c{c}}{\~a}o + sombra + banda 4 (NIR); iv)
fra{\c{c}}{\~o}es vegeta{\c{c}}{\~a}o + sombra + NDVI; e como
testemunha, bandas 3 e 4. A {\'a}rea de estudo {\'e} coberta por
Eucalyptus, localizada em Cap{\~a}o Bonito, S{\~a}o Paulo. As
imagens-fra{\c{c}}{\~a}o foram geradas utilizando
superf{\'{\i}}cie de reflect{\^a}ncia das bandas 3, 4 e 5. A
avalia{\c{c}}{\~a}o da classifica{\c{c}}{\~a}o utilizou
tr{\^e}s crit{\'e}rios: Matriz de Confus{\~a}o, Acur{\'a}cia e
Coeficiente Kappa. O melhor resultado foi a utiliza{\c{c}}{\~a}o
da fra{\c{c}}{\~a}o vegeta{\c{c}}{\~a}o + sombra, mas a
classifica{\c{c}}{\~a}o com a testemunha (bandas 3 e 4) obtive
os maiores valores de Acur{\'a}cia e Coeficiente Kappa.
Palavras-chaves: Modelo de Mistura, biomassa, Eucalyptus spp.
Model of Espectral Unmixing for class age stratification on
Eucalyptus spp stand ABSTRACT: This present research has the goal
to analyse the utilization of Model of Spectral Unmixing for
separate Eucalyptus classes of age. The premise is that there is
relationship between radiometric data extracted from LANDSAT-5/TM
images and volume data estimated from Eucalyptus ssp. stands.
There were four treatments and one testimony: i) fractions
vegetation plus shadow; ii) fractions vegetation plus shadow plus
soil; iii) fractions vegetation plus shadow and band 4 (NIR); iv)
fractions vegetation plus shadow and NDVI; and the testimony:
bands 3 and 4. The study area is covered by Eucalyptus stands from
VCP (Votorantim Celulose e Papel) company, located in Cap{\~a}o
Bonito town, S{\~a}o Paulo state, Brazil. The fractions images
was generated using surface reflectance from TM bands (B3, B4 and
B5). The evaluation of classifications used three criteria:
Confusion Matrix, Accuracy and Kappa Coefficient. The best result
(for Model of Spectral Unmixing) was the utilization of fraction
vegetation plus shadow, but the classification with the testimony
(bands 3 and 4) had higher values of Accuracy and Kappa
Coefficient.",
conference-location = "Aracaju/SE",
conference-year = "2008",
label = "lattes: 2659602494903108 1 Macedo:2008:UTMOMI",
language = "pt",
targetfile = "Macedo_Geonordeste_2008.pdf",
urlaccessdate = "09 maio 2024"
}