@PhDThesis{Graça:2004:MoCaÁr,
author = "Gra{\c{c}}a, Paulo Maur{\'{\i}}cio Lima de Alencastro",
title = "Monitoramento e caracteriza{\c{c}}{\~a}o de {\'a}reas
submetidas {\`a} explora{\c{c}}{\~a}o florestal na
Amaz{\^o}nia por t{\'e}cnicas de detec{\c{c}}{\~a}o de
mudan{\c{c}}as",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2004",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2004-12-15",
keywords = "observa{\c{c}}{\~a}o da terra, monitoramento ambiental,
explora{\c{c}}{\~a}o forestal, detec{\c{c}}{\~a}o de
mudan{\c{c}}as, Amaz{\^o}nia, floresta tropical, sensoriamento
remoto, environmental monitoring, forest management, change
detection, Amazon region (South America), rain forest, remote
sensing.",
abstract = "A explora{\c{c}}{\~a}o florestal seletiva tem afetado extensas
{\'a}reas de florestas prim{\'a}rias na Amaz{\^o}nia. Esta
atividade quando praticada de forma predat{\'o}ria causa impactos
severos ao ambiente, tais como, eros{\~a}o e
compacta{\c{c}}{\~a}o do solo, aumento das emiss{\~o}es de
carbono, aumento da susceptibilidade aos inc{\^e}ndios florestais
e aumento do risco de extin{\c{c}}{\~a}o de esp{\'e}cies
locais. O monitoramento da atividade madeireira {\'e} fundamental
para estabelecer planos de a{\c{c}}{\~a}o e estrat{\'e}gias que
contribuam para o uso sustent{\'a}vel dos recursos florestais. No
entanto, a detec{\c{c}}{\~a}o e o monitoramento desta atividade
por imagens de sat{\'e}lite s{\~a}o dif{\'{\i}}ceis, em
consequ{\^e}ncia das altera{\c{c}}{\~o}es pouco
percept{\'{\i}}veis causadas pela redu{\c{c}}{\~a}o parcial da
cobertura florestal no processo de extra{\c{c}}{\~a}o seletiva
das {\'a}rvores. Assim, o principal objetivo deste trabalho foi
desenvolver procedimentos metodol{\'o}gicos a partir de
t{\'e}cnicas de detec{\c{c}}{\~a}o de mudan{\c{c}}as para
mapear, monitorar e quantificar as {\'a}reas submetidas {\`a}
explora{\c{c}}{\~a}o madeireira em florestas de
transi{\c{c}}{\~a}o na regi{\~a}o de Cl{\'a}udia e
circunvizinhan{\c{c}}as, no norte do estado de Mato Grosso,
utilizando dados multitemporais dos sensores {\'o}pticos ETM e TM
Landsat. Para isto, utilizaram-se as t{\'e}cnicas de
subtra{\c{c}}{\~a}o de imagens NDVI, rota{\c{c}}{\~a}o
controlada por eixo de n{\~a}o mudan{\c{c}}a (RCEN) e a
an{\'a}lise por vetor de mudan{\c{c}}a (AVM). Adicionalmente,
foi realizado um levantamento florestal com a finalidade de
caracterizar as mudan{\c{c}}as na estrutura fision{\^o}mica de
florestas submetidas {\`a} explora{\c{c}}{\~a}o seletiva de
madeira sob diferentes n{\'{\i}}veis de intensidade. A
t{\'e}cnica AVM (com o suporte do classificador por crescimento
de regi{\~o}es), seguida pela t{\'e}cnica RCEN (com suporte de
filtragem tem{\'a}tica) obtiveram um alto desempenho na
classifica{\c{c}}{\~a}o tem{\'a}tica de {\'a}reas afetadas
pela explora{\c{c}}{\~a}o madeireira. No entanto, a t{\'e}cnica
AVM foi mais sujeita {\`a} interven{\c{c}}{\~a}o humana durante
o processo de classifica{\c{c}}{\~a}o. Posteriormente, a
t{\'e}cnica AVM foi adaptada para mapear {\'a}reas florestais
afetadas por inc{\^e}ndios. Os mapas de explora{\c{c}}{\~a}o
florestal obtidos pela t{\'e}cnica AVM foram armazenados em um
SIG com objetivo de monitorar as mudan{\c{c}}as de cobertura e
uso da terra em {\'a}reas sob explora{\c{c}}{\~a}o madeireira e
florestas afetadas pelo fogo, respectivamente, no per{\'{\i}}odo
de 2001 a 2003 e de 1999 a 2003. Observou-se, com base em
informa{\c{c}}{\~o}es auxiliares obtidas de um banco de dados
hist{\'o}rico para o per{\'{\i}}odo de 1989 a 2001, que as
florestas intervindas pela atividade madeireira representaram mais
do que a metade (51%) da {\'a}rea total de floresta existente no
per{\'{\i}}odo de 1989 a 2003, com taxas anuais de
explora{\c{c}}{\~a}o acima daquelas encontradas para o
desmatamento no per{\'{\i}}odo de 2001 a 2003. No entanto,
at{\'e} 2003, apenas 17% da atividade do desmatamento ocorreram
em florestas que j{\'a} foram exploradas para a retirada de
madeira. Este fato demonstra um papel importante da atividade
madeireira na conserva{\c{c}}{\~a}o parcial da cobertura
florestal. Por{\'e}m, a explora{\c{c}}{\~a}o predat{\'o}ria,
atrav{\'e}s da {"}garimpagem florestal{"}, associada {\`a}
ocorr{\^e}ncia de inc{\^e}ndios florestais pode resultar em um
alto n{\'{\i}}vel de degrada{\c{c}}{\~a}o florestal,
comprometendo a sustentabilidade dessa atividade na regi{\~a}o.
ABSTRACT: Selective logging has been affecting extensive areas of
primary forests in Amazonia. This activity, when practiced in a
predatory way, causes severe environmental impacts such as erosion
and soil compaction, increased carbon emissions, increased
forest-fire susceptibility and increased risk of local species
extinction. Monitoring logging activity is fundamental to
establishing action plans and strategies to promote sustainable
use of forest resources. However, detection and monitoring of
selective logging from satellite imagery is difficult because of
the subtle disturbances caused by the partial reduction of forest
canopy cover during selective tree harvesting. The main purpose of
this study was to develop methodological procedures using
change-detection techniques to map, monitor and quantify
selectively logged areas in transitional tropical forests in the
area of Cl{\'a}udia and neighborhood, in northern Mato Grosso,
using multitemporal data from the ETM and TM Landsat optical
sensors. For this, techniques were used such as subtraction of
NDVI images, rotation controlled by the axis of no change (RCANC)
and change vector analysis (CVA). Additionally, a forest inventory
was carried out for the purpose of characterizing the structural
and physiognomic changes in forests logged selectively at
different intensity levels. The best thematic classification of
selective logging areas was obtained using the CVA technique (with
the support of the region growth classifier method) followed by
application of the RCANC technique. However, the CVA technique
required more human intervention during the classification
process. Later, the CVA technique was adapted to map areas
affected by forest fires. The maps of logged areas obtained using
the CVA technique were stored in a GIS with the objective of
monitoring land-cover and land-use changes in areas subjected to
logging (in the 2001-2003 period) and in forests affected by fire
(in the 1999-2003 period). It was found, based on supplementary
information obtained from a historical databank for the 1989-2003
period, that the forests affected by logging represented more than
half (51%) of the total area classified as forest, and that with
annual rates of logging were greater than the rates of
deforestation over the 2001-2003 period. However, the logged
forests contributed only 17% to the deforestation total up to
2003. This fact demonstrates the important role of logging in the
partial conservation of forest cover. Nevertheless, predatory
logging (or forest mining) associated with the occurrence of
forest fires and can result in a high level of forest degradation,
jeopardizing the sustainability of this activity in the region.",
committee = "Santos, Jo{\~a}o Roberto dos (presidente/orientador) and Soares,
Jo{\~a}o Vianei (orientador) and Shimabukuro, Yosio Edemir and
Nelson, Bruce Walker and Couto, Hilton Thadeu Zarate de and
Soares, Vicente de Paulo",
copyholder = "SID/SCD",
englishtitle = "Monitoring and characterization of areas submitted to logging in
Amazonia by change-detection techniques",
language = "pt",
pages = "275",
ibi = "6qtX3pFwXQZ3P8SECKy/FbrQK",
url = "http://urlib.net/ibi/6qtX3pFwXQZ3P8SECKy/FbrQK",
targetfile = "publicacao.pdf",
urlaccessdate = "05 maio 2024"
}