@PhDThesis{Pereira:2017:EsCaAP,
author = "Pereira, Francisca Rocha de Souza",
title = "Sensoriamento remoto LiDAR e {\'o}ptico aplicados {\`a}
estimativa de biomassa a{\'e}rea de manguezais: estudo de caso na
APA de Guapimirim, RJ",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2017",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2016-11-18",
keywords = "LiDAR, imagem {\'o}ptica de alta resolu{\c{c}}{\~a}o,
an{\'a}lise textural, estimativa de biomassa a{\'e}rea,
manguezal, high resolution optical image, textural analysis,
biomass estimation, mangrove.",
abstract = "Os manguezais s{\~a}o ecossistemas costeiros que ocorrem na
interface entre a terra e o mar tipicamente em regi{\~o}es
tropicais, apresentando esp{\'e}cies adaptadas {\`a} salinidade
e inunda{\c{c}}{\~o}es pelas mar{\'e}s. Os manguezais realizam
fun{\c{c}}{\~o}es ecol{\'o}gicas essenciais para a
manuten{\c{c}}{\~a}o da vida terrestre e marinha e para o
sustento de comunidades costeiras. S{\~a}o importantes
transformadores de nutrientes em mat{\'e}ria org{\^a}nica e
geradores de bens e servi{\c{c}}os como a
estabiliza{\c{c}}{\~a}o e prote{\c{c}}{\~a}o da linha de
costa, controle da polui{\c{c}}{\~a}o, sequestro de carbono
atmosf{\'e}rico e regula{\c{c}}{\~a}o do clima. Seu atual
desflorestamento {\'e} preocupante tanto ambientalmente como
socioeconomicamente e sua restaura{\c{c}}{\~a}o e
conserva{\c{c}}{\~a}o s{\~a}o importantes n{\~a}o s{\'o} para
a regula{\c{c}}{\~a}o dos fluxos de carbono e controle das
mudan{\c{c}}as clim{\'a}ticas, mas tamb{\'e}m para a
manuten{\c{c}}{\~a}o de seus valiosos servi{\c{c}}os prestados
{\`a} zona costeira. No presente trabalho uma {\'a}rea
relativamente extensa (\$\sim\$58,2 km\$^{2}\$) de manguezal
inserida na APA de Guapimirim na Ba{\'{\i}}a de Guanabara, RJ
foi estudada. O objetivo geral do estudo {\'e} averiguar o
potencial uso de dados LiDAR aerotransportado de retorno discreto
para estimar a biomassa acima do solo (AGB) do manguezal com
distintos graus de altera{\c{c}}{\~a}o, e comparativamente,
investigar o potencial uso de {\'{\i}}ndices texturais derivados
de imagem {\'o}ptica de alta resolu{\c{c}}{\~a}o WordView-2
para estimar a AGB e distinguir tipos de cobertura do manguezal.
Foram extra{\'{\i}}das 26 m{\'e}tricas descritivas da altura
normalizada da nuvem de pontos LiDAR e os {\'{\i}}ndices
texturais \emph{Fourier-based textural ordination} (FOTO) e
Grey-\emph{Level Co-occurrence Matrix} (GLCM) da imagem
{\'o}ptica pancrom{\'a}tica. Foram testados os m{\'e}todos de
an{\'a}lise de regress{\~a}o Random Forest, AutoPLS e PLS para
estimativa da AGB. Foi demonstrado que o uso de dados LiDAR para
estimativa de AGB de manguezal com distintos graus de
altera{\c{c}}{\~a}o foi efetivo e superior aos resultados
obtidos com uso dos {\'{\i}}ndices texturais extra{\'{\i}}dos
da imagem {\'o}ptica. O modelo preditivo mais preciso da AGB
utilizando dados LiDAR (M2a) apresentou R\$^{2}\$(CAL)=0,89,
R\$^{2}\$(LOO)=0,80, RMSE(CAL)=11,20 t/ha, RMSE (LOO)= 14,80
t/ha e SER\% = 8,90. As vari{\'a}veis preditoras que mais
contribu{\'{\i}}ram na modelagem foram avg, min, max, d02, d03,
d04, d05 e d08 demonstrando que informa{\c{c}}{\~o}es de
densidade de pontos relativos aos estratos estruturais da floresta
s{\~a}o importantes vari{\'a}veis para a estimativa de AGB de
bosques de mangue com distintos graus de altera{\c{c}}{\~a}o,
bem como para detec{\c{c}}{\~a}o de {\'a}reas mais alteradas ou
mais preservadas. O padr{\~a}o de variabilidade textural
associado {\`a}s caracter{\'{\i}}sticas dos doss{\'e}is
florestais com distintos graus de altera{\c{c}}{\~a}o mensuradas
pelos {\'{\i}}ndices FOTO e GLCM n{\~a}o apresentou forte
rela{\c{c}}{\~a}o com os valores de AGB. Por{\'e}m, a
classifica{\c{c}}{\~a}o Random Forest baseada nos
{\'{\i}}ndices texturais apresentou bons resultados na
discrimina{\c{c}}{\~a}o de tipos de cobertura como {\'a}reas de
n{\~a}o mangue, mangue alterado e mangue mais preservado. A
presente tese demonstra a efic{\'a}cia do uso de t{\'e}cnicas de
sensoriamento remoto, em especial de dados LiDAR de retorno
discreto para estimar e mapear a AGB com boa acur{\'a}cia e para
discriminar tipos de cobertura no manguezal. Os resultados aqui
apresentados podem contribuir com as an{\'a}lises e
caracteriza{\c{c}}{\~a}o estrutural do manguezal,
quantifica{\c{c}}{\~a}o e qualifica{\c{c}}{\~a}o da AGB e
estoques de carbono, bem como, contribuir com o monitoramento,
formula{\c{c}}{\~a}o de pol{\'{\i}}ticas p{\'u}blicas de
conserva{\c{c}}{\~a}o e prote{\c{c}}{\~a}o deste ecossistema,
auxiliando a sua gest{\~a}o. ABSTRACT: Mangroves form important
intertidal ecosystems that link terrestrial and marine systems
typically in tropical and subtropical regions, presenting
physiological and morphological adaptations to environmental
stresses of high salinity and flooding by tides. Mangroves perform
essential ecological functions for the maintenance of terrestrial
and marine life and the livelihoods of coastal communities. They
provide valuable ecological and economical ecosystem goods and
services transforming nutrients in organic matter, contributing to
coastal erosion protection, pollution control, atmospheric carbon
sequestration and climate regulation, among many other factors.
Nevertheless, mangroves have experienced a dramatic decline in
area caused by overexploitation and conversion to other uses.
Their restoration and conservation are important not only for the
regulation of carbon fluxes and climate change control, but also
to maintain their valuable services for the coastal zone. Remote
sensing techniques offer a useful tool of estimating forest
biomass contributing with the monitoring of land use and land
cover dynamics and the effectiveness of environmental policies. In
the present work a relatively large area (\$\sim\$58.2
km\$^{2}\$) of mangroves inserted in the Environmental
Protection Area of Guapimirim, Guanabara Bay, RJ was studied. The
main goal of this study is to investigate the potential use of
discrete return LiDAR data to estimate the aboveground biomass
(AGB) of a mangrove forest with different degrees of disturbance,
and comparatively investigate the potential use of textural
indices derived from a high resolution WorldView-2 image to
estimate AGB and to distinguish types of mangrove coverage.
Twenty-six descriptive LiDAR metrics were extracted from the
normalized height of the LiDAR point cloud data together with the
Fourier-based textural ordination (FOTO) and Grey-Level
Co-occurrence Matrix (GLCM) textural indices from the panchromatic
optical image. Random Forest, AutoPLS and PLS regression methods
were tested to estimate AGB. The results obtained using LiDAR data
for estimating AGB were effective and superior to the results
obtained using the textural indices. The most accurate predictive
model of AGB using LiDAR data (M2a) presented R\$^{2}\$(CAL) =
0.89, R\$^{2}\$(LOO) = 0.80, RMSE(CAL) = 11.20 t/ha, RMSE(LOO) =
14.80 t/ha and RSE\% = 8.90\%. The most important predictor
variables for the M2a model were avg, min, max, d02, d03, d04, d05
and d08 demonstrating that point density relative to the forest
structural strata are important variables for the AGB estimation
in mangrove forests with different degrees of disturbance as well
as for detecting more altered or preserved areas. The textural
variability pattern associated with the canopy characteristics
with different degrees of disturbance measured by FOTO and GLCM
indices showed weak relationships with AGB values. However, the
Random Forest classification based on the textural indices showed
good results on the discrimination of different types of coverage
such as non-mangrove, altered and preserved mangroves. This thesis
demonstrates the effectiveness use of remote sensing techniques,
particularly discrete return LiDAR data to accurately estimate and
map the AGB and to discriminate types of mangrove coverage. The
results presented here can contribute to the analysis and
structural characterization of mangroves, its AGB and carbon
stocks quantification and qualification, also contributing with
the monitoring and formulation of public policies for the
conservation and protection of this ecosystem.",
committee = "Kampel, Silvana Amaral (presidente) and Kampel, Milton
(orientador) and Arag{\~a}o, Luiz Eduardo Oliveira e Cruz de and
Valeriano, Dalton de Morisson and Bentz, Cristina Maria and Longo,
Marcos",
copyholder = "SID/SCD",
englishtitle = "LiDAR and optic remote sensing applied to mangrove aboveground
biomass estimates: study case APA de Guapimirim, RJ.",
language = "pt",
pages = "211",
ibi = "8JMKD3MGP3W34P/3MM3DAP",
url = "http://urlib.net/ibi/8JMKD3MGP3W34P/3MM3DAP",
targetfile = "publicacao.pdf",
urlaccessdate = "27 abr. 2024"
}