@PhDThesis{Porfirio:2018:CoMoAe,
author = "Porfirio, Anthony Carlos Silva",
title = "Uma contribui{\c{c}}{\~a}o {\`a} modelagem de aerossol e
componentes da radia{\c{c}}{\~a}o solar no modelo GL",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2018",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2017-11-29",
keywords = "radia{\c{c}}{\~a}o solar, aeross{\'o}is atmosf{\'e}ricos,
componentes da radia{\c{c}}{\~a}o solar, sat{\'e}lites
meteorol{\'o}gicos, solar radiation, atmospheric aerosols, solar
radiation components, meteorological satellites.",
abstract = "O modelo f{\'{\i}}sico GL {\'e} utilizado no CPTEC/INPE para o
monitoramento da radia{\c{c}}{\~a}o solar global {\`a}
superf{\'{\i}}cie no Brasil e na Am{\'e}rica do Sul com base em
imagens do canal vis{\'{\i}}vel (VIS) de sat{\'e}lites
geoestacion{\'a}rios. As informa{\c{c}}{\~o}es geradas pelo
modelo t{\^e}m se mostrado {\'u}teis para uma variedade de
aplica{\c{c}}{\~o}es, como estudos clim{\'a}ticos,
valida{\c{c}}{\~a}o de modelos atmosf{\'e}ricos regionais,
modelagem da evapotranspira{\c{c}}{\~a}o e levantamento do
potencial solar. Devido {\`a} sua relev{\^a}ncia,
esfor{\c{c}}os cient{\'{\i}}ficos com o intuito de investigar a
qualidade das estimativas e identificar limita{\c{c}}{\~o}es
s{\~a}o fundamentais para propor aperfei{\c{c}}oamentos e
assegurar bons n{\'{\i}}veis de acur{\'a}cia. Nesse contexto, o
presente estudo teve como objetivo realizar uma
contribui{\c{c}}{\~a}o ao modelo GL atrav{\'e}s da an{\'a}lise
do desempenho de suas estimativas, bem como com o aprimoramento de
suas hip{\'o}teses f{\'{\i}}sicas e com a inclus{\~a}o de
novas parametriza{\c{c}}{\~o}es. Primeiramente, avaliou-se o
desempenho do modelo GL aplicado {\`a}s imagens GOES
(GL\$_{G1.2}\$) e MSG (GL\$_{M1.2}\$ e GL\$_{M1.4}\$) sobre
o territ{\'o}rio brasileiro. As estimativas GL, na base mensal,
apresentaram uma boa concord{\^a}ncia com a verdade terrestre,
sobretudo nas regi{\~o}es nordeste, sudeste e sul, com erros
m{\'e}dios entre \$\pm\$ 10 W m\$^{-2}\$. Maiores
discrep{\^a}ncias nas estimativas GL foram percebidas nas
regi{\~o}es norte e centro-oeste, principalmente na {\'e}poca de
maior concentra{\c{c}}{\~a}o de aeross{\'o}is atmosf{\'e}ricos
(agosto-outubro), com desvios m{\'e}dios mensais de 36 e 18 W
m\$^{-2}\$, respectivamente. As compara{\c{c}}{\~o}es entre
GL\$_{G}\$ e GL\$_{M}\$ mostraram que para a regi{\~a}o
nordeste as estimativas s{\~a}o bastante semelhantes, o que pode
ser creditado ao fato que essa regi{\~a}o {\'e} observada pelos
sat{\'e}lites com {\^a}ngulo de visada similar. An{\'a}lise
complementar para esta{\c{c}}{\~o}es radiom{\'e}tricas
localizadas na Espanha, mostraram um excelente desempenho do
GL\$_{M1.4}\$, com um MBE de apenas -0,2 W m\$^{-2}\$ e RMSE
de 13,8 W m\$^{-2}\$. O modelo GL negligencia a presen{\c{c}}a
de aeross{\'o}is. Assim, em seguida, investigou-se o impacto dos
aeross{\'o}is atmosf{\'e}ricos nas estimativas do modelo GL.
Ciclos di{\'a}rios medidos e estimados foram comparados para dias
de c{\'e}u claro na {\'e}poca do pico de queimadas na
regi{\~a}o sul da Amaz{\^o}nia. Tal an{\'a}lise evidenciou a
tend{\^e}ncia do modelo em superestimar a medida para dias com
concentra{\c{c}}{\~a}o de aeross{\'o}is relevantes.
An{\'a}lises te{\'o}ricas utilizando o c{\'o}digo de
transfer{\^e}ncia radiativa SBDART permitiram o desenvolvimento
de parametriza{\c{c}}{\~o}es para os aeross{\'o}is no modelo
GL. Adotaram-se propriedades {\'o}pticas de aeross{\'o}is
referentes a dois tipos (rural e queimadas). As primeiras
valida{\c{c}}{\~o}es do novo esquema, denominado de
GL\$_{2.0}\$, mostraram um aperfei{\c{c}}oamento na estimativa
dos ciclos di{\'a}rios e da irradia{\c{c}}{\~a}o solar em
rela{\c{c}}{\~a}o a vers{\~a}o GL\$_{1.2}\$. O impacto na
irradi{\^a}ncia solar global m{\'e}dia di{\'a}ria variou entre
-50 e -15 W m\$^{-2}\$ para profundidades {\'o}pticas de 1,0 e
0,3, para um dia com evidente presen{\c{c}}a da pluma regional de
aeross{\'o}is de queimada. Por fim, o procedimento aplicado para
a estimativa particionada das componentes da radia{\c{c}}{\~a}o
solar global, direta e difusa, a partir do GL\$_{2.0}\$ e de um
modelo de radia{\c{c}}{\~a}o direta por sat{\'e}lite se mostrou
promissor. As estimativas de irradia{\c{c}}{\~a}o global, direta
e difusa por sat{\'e}lite mostraram boa concord{\^a}ncia com as
medidas de superf{\'{\i}}cie, com valores m{\'e}dios para o MBE
de -0,2, 1,0 e -1,0 MJ m\$^{-2}\$ e para o RMSE de 1,6, 2,0 e
2,0 MJ m\$^{-2}\$, respectivamente. Em geral, os resultados
obtidos foram consistentes, e conceberam de maneira natural, duas
novas vers{\~o}es com grande potencial para
aplica{\c{c}}{\~o}es em pesquisa e para fins operacionais.
ABSTRACT: The physical model GL is used at CPTEC/INPE for
monitoring solar radiation over Brazil and South America based on
geostationary satellite VIS imagery. The information generated by
GL has shown to be useful for a variety of applications, such as
climate studies, validation of regional atmospheric models,
evapotranspiration modeling and renewable energy studies. Thus,
scientific efforts to investigate the GL's quality and to identify
limitations are essential in order to propose refinements and
ensure good levels of accuracy. The aim of this work was to
contribute to GL model structure through validations studies as
well as the improvement of its physical hypotheses and the
inclusion of new parametrizations. Firstly, the performance of the
GL model applied to GOES (version GL\$_{G1.2}\$) and MSG
(version GL\$_{M1.2}\$ and GL\$_{M1.4}\$) over the Brazilian
territory was evaluated. The global solar irradiance estimates on
a monthly basis showed good agreement with ground truth
(especially in the northeast, southeast and south regions) with
mean bias errors lower than 10 W m\$^{-2}\$. Larger
discrepancies in the GL estimates were observed in the north and
center-west regions, especially during biomass burning season
(August-October), with mean monthly deviations of 36 and 18 W
m\$^{-2}\$, respectively. Comparisons between GL\$_{G}\$ and
GL\$_{M}\$ showed quite similar estimates in the northeast
region, probably due to the fact that this region is observed with
similar angle of view from both satellites. Complementary analysis
for stations located in Spain showed excellent performance of
GL\$_{M1.4}\$ estimates, with MBE of -0,2 W m\$^{-2}\$ and
RMSE of 13,8 W m\$^{-2}\$. GL model neglects the presence of
aerosol. So, the next step was to investigate the impact of this
atmospheric constituent. Daily cycles for clear sky days during
burning biomass peak in the Amazon were considered. This analysis
made evident the tendency of the model to overestimate ground
truth for days with high aerosol loading. New algorithms were
developed for describing this influence. Simulations using the
radiative transfer code SBDART allowed to find proper
parameterizations for the new variables. Two aerosol optical
models were considered (rural and smoke). The first validations of
the new scheme (called GL\$_{2.0}\$) showed an improvement
related to previous versions GL\$_{1.2}\$, for estimation of
daily cycles as well as daily irradiation. The impact on daily
mean global solar irradiance ranged from -50 to -15 W m\$^{-2}\$
in one day with evident presence of regional aerosol plumes
showing aerosol optical depths between 1,0 and 0,3, respectively.
Last but not least, a method for estimating direct normal
irradiance and its daily integral was improved. It allows
estimation of diffuse irradiation by a simple partition of global
in direct and diffuse irradiance. The results are promising.
Validations against four Brazilian stations show mean values for
MBE of -0,2, 1,0 and -1,0 MJ m\$^{-2}\$ and for RMSE of 1,6, 2,0
and 2,0 MJ m\$^{-2}\$. In general, the results reported were
consistent, generating in a natural way, two new versions with
great potential for research and operational purposes.",
committee = "Coelho, Simone Marilene Sievert da Costa (presidente) and
Ceballos, Juan Carlos (orientador) and Pereira, Enio Bueno and
Martins, Fernando Ramos and Ros{\'a}rio, Nilton Manuel {\'E}vora
do",
englishtitle = "A contribution to aerosol and solar radiation components modeling
on GL model",
language = "pt",
pages = "134",
ibi = "8JMKD3MGP3W34P/3Q4PAHH",
url = "http://urlib.net/ibi/8JMKD3MGP3W34P/3Q4PAHH",
targetfile = "publicacao.pdf",
urlaccessdate = "27 abr. 2024"
}