@MastersThesis{Pessôa:2016:CaEsEs,
author = "Pess{\^o}a, Ana Carolina Moreira",
title = "Caracteriza{\c{c}}{\~a}o espectral de est{\'a}gios sucessionais
no dom{\'{\i}}nio Mata Atl{\^a}ntica em diferentes
condi{\c{c}}{\~o}es de ilumina{\c{c}}{\~a}o local utilizando
imagens TM/Landsat 5",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2016",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2016-03-14",
keywords = "est{\'a}gios sucessionais, Tasseled Cap, ilumina{\c{c}}{\~a}o
local, successional stages, local illumination.",
abstract = "Florestas secund{\'a}rias possuem grande import{\^a}ncia na
absor{\c{c}}{\~a}o de CO\$_{2}\$ atmosf{\'e}rico e no
equil{\'{\i}}brio ecossist{\^e}mico perdido com a
supress{\~a}o de florestas prim{\'a}rias. A Mata Atl{\^a}ntica,
ecossistema altamente afetado pela a{\c{c}}{\~a}o
antr{\'o}pica, possui fragmentos florestais em diferentes
est{\'a}gios sucessionais, os quais s{\~a}o regulados legalmente
mediante o estabelecimento de diretrizes espec{\'{\i}}ficas. A
identifica{\c{c}}{\~a}o desses est{\'a}gios nos quais
fragmentos florestais se encontram {\'e} interesse de diversos
seguimentos da sociedade e de suma import{\^a}ncia para a
prote{\c{c}}{\~a}o de diferentes ecossistemas. T{\'e}cnicas de
sensoriamento remoto tem ganhado destaque em estudos de
identifica{\c{c}}{\~a}o e de caracteriza{\c{c}}{\~a}o de
est{\'a}gios sucessionais, por{\'e}m, o Dom{\'{\i}}nio Mata
Atl{\^a}ntica acumulou poucos estudos utilizando esta abordagem.
Neste contexto, este trabalho tem como objetivo caracterizar
espectralmente diferentes est{\'a}gios sucessionais de fragmentos
florestais no Dom{\'{\i}}nio Mata Atl{\^a}ntica, atrav{\'e}s
de uma an{\'a}lise multitemporal de imagens TM/Landsat 5, levando
ainda em considera{\c{c}}{\~a}o aspectos relacionados {\`a}
topografia. Dezoito cenas TM/Landsat 5 abrangendo um intervalo
total de 25 anos foram utilizadas para selecionar e classificar
fragmentos florestais (aqui denominados de pol{\'{\i}}gonos) em
diferentes est{\'a}gios sucessionais por
interpreta{\c{c}}{\~a}o visual de imagens. Essa
sele{\c{c}}{\~a}o levou em considera{\c{c}}{\~a}o a
condi{\c{c}}{\~a}o de ilumina{\c{c}}{\~a}o de cada
pol{\'{\i}}gono. Pol{\'{\i}}gonos em diferentes classes de
ilumina{\c{c}}{\~a}o foram analisados independentemente. A
partir da transforma{\c{c}}{\~a}o \emph{Tasseled Cap}, as
fei{\c{c}}{\~o}es \emph{Brightness, Greenness e Wetness}
compuseram um espa{\c{c}}o tridimensional no qual foi avaliada a
din{\^a}mica espectral resultante da mudan{\c{c}}a na estrutura
de dossel entre est{\'a}gios sucessionais. De acordo com a
metodologia adotada, os resultados sugerem que {\'e}
poss{\'{\i}}vel distinguir est{\'a}gios sucessionais por
interpreta{\c{c}}{\~a}o visual de imagens TM/Landsat 5 levando
em considera{\c{c}}{\~a}o a classe de ilumina{\c{c}}{\~a}o
considerada. A classe Pouco Iluminado (PI) registrou maior
n{\'u}mero de inconsist{\^e}ncias e, consequentemente, maior
susceptibilidade a erros de classifica{\c{c}}{\~a}o. A
metodologia por fatiamento em classes de ilumina{\c{c}}{\~a}o
local destacou a necessidade de diferentes condi{\c{c}}{\~o}es
de ilumina{\c{c}}{\~a}o serem tratadas independentemente. A
caracteriza{\c{c}}{\~a}o espectral se mostrou importante como
ferramenta de suporte para a interpreta{\c{c}}{\~a}o visual de
imagens. ABSTRACT: Secondary forests have a remarkable weight in
the absorption of atmospheric CO\$_{2}\$ and in the ecosystem
balance lost due to the primary forests suppression. The Atlantic
Forest, which is a highly affected ecosystem by human action,
presents forest remnants in different successional stages. These
successional stages are legally regulated through the
establishment of specific legal guidelines. Identifying forest
secondary succession stages has been one of the most important
interests of several society segments and it has great importance
to the legal protection of different ecosystems. Several studies
have been carried out using remote sensing techniques for the
identification and characterization of successional stages of
forest fragments. However, the Atlantic Forest Domain accumulated
few studies using this approach. In this context, the present
study aims to spectrally characterize different successional
stages of forest fragments in the Atlantic Forest Domain through a
multitemporal analysis of TM/Landsat 5 images, taking into account
aspects related to topography. Eighteen TM/Landsat 5 scenes
covering a total period of 25 years were used to select and
classify polygons in different successional stages through visual
image interpretation. This selection considered the illumination
condition of each polygon, so that polygons on different
illumination classes could be analyzed independently. From
Tasseled Cap transformation, the Brightness, Greenness and Wetness
features composed a three-dimensional space in which the spectral
dynamics resulting from canopy structural changes between
successional stages could be characterized. According to the
methodology adopted, the results suggest that successional stages
can be distinguished by visual interpretation of TM/Landsat 5
images as long as illumination conditions are considered. The less
illuminated class recorded more inconsistencies and, consequently,
higher susceptibility to misclassification. The methodology of
slicing the images into illumination classes highlighted the need
for different illumination conditions be treated independently.
The spectral characterization showed to be useful as a support
tool for the visual image interpretation.",
committee = "Ponzoni, Fl{\'a}vio Jorge (presidente/orientador) and Sanches,
Ieda Del'Arco (orientador) and Galv{\~a}o, L{\^e}nio Soares and
Renn{\'o}, Camilo Daleles and Ribeiro, Milton Cezar",
copyholder = "SID/SCD",
englishtitle = "Spectral characterization of successional stages in the Atlantic
Forest domain under different local illumination conditions using
TM/Landsat 5 images",
language = "pt",
pages = "152",
ibi = "8JMKD3MGP3W34P/3L8MD8P",
url = "http://urlib.net/ibi/8JMKD3MGP3W34P/3L8MD8P",
targetfile = "publicacao.pdf",
urlaccessdate = "27 abr. 2024"
}