@MastersThesis{FloresJśnior:2019:PaAlEm,
author = "Flores J{\'u}nior, Rog{\'e}rio",
title = "Parametriza{\c{c}}{\~a}o de algoritmos emp{\'{\i}}ricos e
algoritmo quasi-anal{\'{\i}}tico QAA para estimativa de
clorofila-a em lagos da v{\'a}rzea do rio Amazonas",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2019",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2019-03-14",
keywords = "Qualidade da {\'a}gua, sensoriamento remoto da {\'a}gua,
fitoplancton, algoritmos bio-{\'o}pticos, lagos amaz{\^o}nicos,
water quality, water remote sensing, phytoplankton, bio-optical
models, Amazon lakes.",
abstract = "O monitoramento sistem{\'a}tico, essencial para a
manuten{\c{c}}{\~a}o dos servi{\c{c}}os ecossist{\^e}micos,
visa auxiliar na gest{\~a}o, manuten{\c{c}}{\~a}o e
recupera{\c{c}}{\~a}o de recursos h{\'{\i}}dricos diante das
a{\c{c}}{\~o}es antr{\'o}pica. Suplementando as metodologias
tradicionais, o sensoriamento remoto (SR) de ambientes
aqu{\'a}ticos relaciona as propriedades {\'o}pticas e
biogequ{\'{\i}}micas do meio, possibilitando uma vis{\~a}o
sin{\'o}ptica de seus padr{\~o}es de distribui{\c{c}}{\~a}o no
tempo e espa{\c{c}}o. A determina{\c{c}}{\~a}o do estado
tr{\'o}fico e da produtividade prim{\'a}ria do ambiente,
estimadas por meio da concentra{\c{c}}{\~a}o de clorofila-a
(Chl-a), podem ser utilizados como {\'{\i}}ndice da qualidade do
recurso h{\'{\i}}drico. A quantifica{\c{c}}{\~a}o de Chl-a por
SR embasa-se na aplica{\c{c}}{\~a}o emp{\'{\i}}rica e/ou
semi-anal{\'{\i}}tica de algoritmos, geralmente desenvolvidos
para {\'a}guas oce{\^a}nicas e costeiras, necessitando assim
serem calibrados para sua utiliza{\c{c}}{\~a}o em {\'a}guas
continentais complexas como {\`a}s da plan{\'{\i}}cie de
inunda{\c{c}}{\~a}o amaz{\^o}nica. Assim, este trabalho
objetivou: i) A aplica{\c{c}}{\~a}o e valida{\c{c}}{\~a}o de
algoritmos emp{\'{\i}}ricos em dados in situ e imagens orbitais
simulados para os sensores OLI e MSI, por meio de
simula{\c{c}}{\~a}o de Monte Carlo (MC); ii)
Aplica{\c{c}}{\~a}o do algoritmo quasi anal{\'{\i}}tico QAA da
literatura, calibra{\c{c}}{\~a}o e Parametriza{\c{c}}{\~a}o de
uma vers{\~a}o para {\`a}s {\'a}guas complexas da
plan{\'{\i}}cie amaz{\^o}nica (QAALCG) e aplica{\c{c}}{\~a}o
de {\'{\I}}ndices para a estimativa de Chl-a com os produtos
gerados, utilizando dados in situ (Rrs(\λ)) simulados para o
sensor OLCI. Para isto, foram obtidos dados {\'o}pticos e
limnol{\'o}gicos em lagos da plan{\'{\i}}cie de
inunda{\c{c}}{\~a}o do baixo amazonas para quatro campanhas de
campo entre 2015 e 2017, formando um conjunto de 94 pontos
amostrais. Foram utilizadas as m{\'e}tricas estat{\'{\i}}sticas
R2, MAPE, NRMSE e bias, para a avalia{\c{c}}{\~a}o do desempenho
dos algoritmos. A calibra{\c{c}}{\~a}o dos algoritmos
emp{\'{\i}}ricos esbarra na heterogeneidade dos dados das
campanhas, coletadas em diferentes fases da hidr{\'o}grafa,
apresentando resultados insatisfat{\'o}rios no conjunto com todos
os dados. A calibra{\c{c}}{\~a}o e valida{\c{c}}{\~a}o (MC)
apenas para a campanha de 2017 foi satisfat{\'o}ria e
possibilitou a aplica{\c{c}}{\~a}o do algoritmo {\`a}s imagens
dos sensores OLI e MSI de mesma data. Os algoritmos
emp{\'{\i}}ricos aplicados {\`a}s imagens de sat{\'e}lite
n{\~a}o foram satisfat{\'o}rios para quantifica{\c{c}}{\~a}o
da Chl-a, o que pode ser atribu{\'{\i}}do a alta din{\^a}mica
observada entre a data de aquisi{\c{c}}{\~a}o das imagens e a
aquisi{\c{c}}{\~a}o de medidas in situ, mesmo considerando-se
uma defasagem de apenas 2 dias. A calibra{\c{c}}{\~a}o e
parametriza{\c{c}}{\~a}o das rela{\c{c}}{\~o}es
emp{\'{\i}}ricas do QAALGC foram essenciais para o bom
desempenho do algoritmo. Os coeficientes de absor{\c{c}}{\~a}o
derivados do QAALGC, obtiveram resultados satisfat{\'o}rios para
o conjunto de dados testado, por{\'e}m com tend{\^e}ncia a
subestimar os valores como os demais algoritmos avaliados. A alta
concentra{\c{c}}{\~a}o de part{\'{\i}}culas inorg{\^a}nicas
na campanha independente, utilizada para validar o QAALCG, limitou
a obten{\c{c}}{\~a}o de resultados satisfat{\'o}rios para a
estimativa de Chl-a utilizando os coeficientes de
absor{\c{c}}{\~a}o modelados pelo algoritmo. Os resultados
obtidos pelo QAALCG para as {\'a}guas complexas da
plan{\'{\i}}cie de inunda{\c{c}}{\~a}o amaz{\^o}nica foram
satisfat{\'o}rios. A alta din{\^a}mica sazonal e espacial da
regi{\~a}o de estudo dificulta a modelagem dos constituintes
pressentes no meio aqu{\'a}tico, por{\'e}m os algoritmos
(emp{\'{\i}}ricos e semi-anal{\'{\i}}tico) testados mostraram
potencial para a quantifica{\c{c}}{\~a}o da Chl-a. ABSTRACT:
Systematic monitoring, essential to maintain the ecosystem
services, aim to assist the management, maintenance, and recovery
of the water resources in contrast to anthropic actions.
Supporting the traditional methodologies, the remote sensing (SR)
of aquatic environments associate optical and biogeochemical
properties of the medium, allowing a synoptic view of its
distribution patterns in time and space. The determination of the
environment trophic state and primary productivity, estimated by
the concentration of chlorophyll-a (Chl-a), can be used as quality
index to water resources. The quantification of Chl-a by SR is
based on empirical and/or semi-analytical application of
algorithms, generally developed for oceanic and coastal waters,
needing to be calibrated for use in complex inland waters such as
those on the Amazon floodplains. Thus, this study aimed to: i)
application and validation of empirical algorithms in in situ data
and satellite images simulated to OLI and MSI sensors, through
Monte Carlo simulation (MC); ii) Application of the
quasianalytical algorithm QAA from literature, calibration and
parameterization of a version for the complex waters of the Amazon
floodplain (QAALCG), and application of Indices to estimate Chl-a
with the products generated, using in situ data (Rrs(\λ))
simulated for the OLCI sensor. So, optical and limnological data
were obtained in lakes from the lower Amazon floodplains in four
field campaigns between 2015 and 2017, composing a dataset of 94
sampling points. The statistical metrics R2, MAPE, NRMSE and bias
were used to evaluate the performance of the algorithms. The
calibration of the empirical algorithms faces the heterogeneity of
the campaigns, collected in different phases of the hydrograph,
presenting unsatisfactory results when calibrated with all data.
However, Calibration and validation (MC) only for the 2017
campaign was satisfactory and allowed the application of the
algorithm to OLI and MSI sensor images of the same date. Yet, the
empirical algorithms applied to the satellite images were not
satisfactory to quantify Chl-a, which can be attributed to high
dynamics observed between the images date of acquisition and in
situ measurements, even only with two days gap. Therefore,
calibration and parameterization of the empirical relations on the
QAALCG were essential for optimizing its performance. The
absorption coefficients derived from the QAALGC obtained
satisfactory results for the dataset tested, but with a tendency
to underestimate it as well as the other algorithms evaluated. The
high concentration of inorganic particles in the independent
campaign, used to validate the QAALCG, limited the achievement of
satisfactory results of Chl-a using the absorption coefficients
modeled by the algorithm. However, the results obtained by QAALCG
for complex waters of the Amazon floodplain lakes were
satisfactory considering that the algorithm was developed for
oceanic waters. The highly seasonal and spatial dynamics of the
study region makes it difficult to model the constituents present
in the water bodies, but the algorithms (empirical and
semi-analytical) tested showed potential for quantification of
Chl-a.",
committee = "Novo, Evlyn M{\'a}rcia Le{\~a}o de Moraes (presidente) and
Barbosa, Claudio Clemente Faria (orientador) and Lobo, Felipe de
Lucia (orientador) and Watanabe, Fernanda Sayuri Yoshino and
Carvalho, Lino Augusto Sander de",
englishtitle = "Parametrization of empirical algorithms and quasi-analytical
algorithm QAA to estimate chlorophil-a in lakes of the Amazon
river floodplains",
language = "pt",
pages = "157",
ibi = "8JMKD3MGP3W34R/3SUQ3U2",
url = "http://urlib.net/ibi/8JMKD3MGP3W34R/3SUQ3U2",
targetfile = "publicacao.pdf",
urlaccessdate = "18 mar. 2024"
}