@PhDThesis{SánchezIpia:2021:DeDeUs,
author = "S{\'a}nchez Ipia, Alber Hamersson",
title = "Detection of deforestation using remote sensing time series
analysis",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2021",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2020-06-26",
keywords = "Floresta Amaz{\^o}nica, desmatamento, aprendizagem de
m{\'a}quina, sensoriamento remoto, Amazon forest, deforestation,
machine learning, remote sensing.",
abstract = "The Amazon rainforest plays an important role in the global carbon
and water cycles, having direct influence on Earths atmosphere and
it suffers the consequences of the current climate crisis.
Deforestation monitoring systems are a source of information on
the forest condition for the scientific community, policy makers,
and the general public. In this thesis, we identified three areas
on which such systems could be improved: data processing,
information extraction, and information distribution. Processing
data of Earth observation satellites is subject to atmospheric
noise. In particular, clouds obstruct the surveying of the Amazon
rainforest. They introduce discontinuities on the the spatial and
temporal patterns, which reduce the ability of analyst to extract
information about features on the surface and reducing the
reliability of the information obtained. Any information on Earths
surface in our particular case, information on Land Use and Land
Cover change increases its value through sharing, validation, and
reuse in broader communities. Regarding data processing, we tested
several cloud detection algorithms on Sentinel-2 imagery and we
found that Fmask4 provides the best performance under frequent
cloud coverage. With this knowledge, we proceed to extract
deforestation information using time series of the Landsat 8 and
Sentinel-2 satellites, applying machine learning techniques of
Deep Learning and Random Forest, respectively. We obtained the
best results by using time series of Sentinel-2 images processed
with Random Forest. Finally, we demonstrated the best way to
provide scientists with access to massive amounts or Earth
observation data and processing tools is through collaborative
analysis environments offered through Internet, such as Jupyter
notebooks. RESUMO: carbono e {\'a}gua, tendo influ{\^e}ncia
direta na atmosfera terrestre e sofrendo as consequ{\^e}ncias da
atual crise clim{\'a}tica. Da{\'{\i}} a import{\^a}ncia dos
sistemas de monitoramento de desmatamento como fonte de
informa{\c{c}}{\~a}o sobre a condi{\c{c}}{\~a}o da floresta
para comunidade cient{\'{\i}}fica, formadores de
pol{\'{\i}}ticas e o p{\'u}blico em geral. N{\'o}s
identificamos tr{\^e}s {\'a}reas nas quais esses sistemas
poderiam ser aprimorados: processamento de dados,
extra{\c{c}}{\~a}o e distribui{\c{c}}{\~a}o de
informa{\c{c}}{\~o}es. O processamento de dados dos
sat{\'e}lites de observa{\c{c}}{\~a}o da Terra est{\'a}
sujeito ao ru{\'{\i}}do atmosf{\'e}rico; as nuvens, em
particular, dificultam o mapeamento da floresta Amaz{\^o}nica. As
nuvens introduzem descontinuidades nos padr{\~o}es espaciais e
temporais, o que reduz a capacidade dos analistas de extrair
informa{\c{c}}{\~o}es sobre os elementos da superf{\'{\i}}cie,
e tamb{\'e}m reduz a confiabilidade das informa{\c{c}}{\~o}es
obtidas. Qualquer informa{\c{c}}{\~a}o sobre superf{\'{\i}}cie
da Terra, em nosso caso particular, informa{\c{c}}{\~o}es sobre
mudan{\c{c}}a no uso e cobertura, incrementa seu valor por meio
do compartilhamento, valida{\c{c}}{\~a}o e
reutiliza{\c{c}}{\~a}o em comunidades mais amplas. Em
rela{\c{c}}{\~a}o ao processamento dos dados, testamos
v{\'a}rios algoritmos de detec{\c{c}}{\~a}o de nuvens e
descobrimos que o Fmask4 oferece o melhor desempenho em imagens de
sat{\'e}lite com frequente cobertura de nuvens. Com esse
conhecimento, procedemos {\`a} extra{\c{c}}{\~a}o de
informa{\c{c}}{\~o}es sobre desmatamento usando s{\'e}ries
temporais dos sat{\'e}lites Landsat 8 e Sentinel-2, aplicando as
t{\'e}cnicas de aprendizado de m{\'a}quina Deep Learning e
Random Forest. Obtivemos os melhores resultados usando s{\'e}ries
temporais de imagens Sentinel-2 processadas com Random Forest.
Finalmente, demonstramos que a melhor maneira de fornecer aos
cientistas acesso a grandes quantidades de dados de
observa{\c{c}}{\~a}o da Terra {\'e} com ferramentas de
processamento e atrav{\'e}s de ambientes de an{\'a}lise
colaborativa oferecidos pela Internet, como os notebooks
Jupyter.",
committee = "Escada, Maria Isabel Sobral (presidente) and Camara Neto, Gilberto
(orientador) and Andrade Neto, Pedro Ribeiro de (orientador) and
Carneiro, Tiago Garcia de Senna and Coutinho, Alexandre Camargo",
englishtitle = "Detec{\c{c}}{\~a}o de desmatamento usando an{\'a}lise de series
de tempo de sensoriamento remoto na Amaz{\^o}nia brasileira",
language = "en",
pages = "83",
ibi = "8JMKD3MGP3W34R/42PGNM8",
url = "http://urlib.net/ibi/8JMKD3MGP3W34R/42PGNM8",
targetfile = "publicacao.pdf",
urlaccessdate = "28 abr. 2024"
}