@PhDThesis{JucáOliveira:2017:ChErMo,
author = "Juc{\'a} Oliveira, R{\^o}mulo Augusto",
title = "Characteristics and error modeling of GPM satellite rainfall
estimates during CHUVA campaign in Brazil",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2017",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2017-03-24",
keywords = "estimativa de precipita{\c{c}}{\~a}o por sat{\'e}lite,
modelagem de erro, quantifica{\c{c}}{\~a}o de incertezas,
estimativa de precipita{\c{c}}{\~a}o por radar,
valida{\c{c}}{\~a}o, satellite rainfall estimation, error
modeling, uncertainty quantification, radar rainfall estimation,
validation.",
abstract = "Studies that investigate and evaluate the quality, limitations and
uncertainties of satellite rainfall estimates are fundamental to
assure the correct and successful use of these products in
applications, such as climate studies, hydrological modeling and
natural hazard monitoring. Over regions of the globe that lack in
situ observations, such studies are only possible through
intensive field measurement campaigns, which provide a range of
high quality ground measurements, e.g., CHUVA (Cloud processes of
tHe main precipitation systems in Brazil: A contribUtion to cloud
resolVing modeling and to the GlobAl Precipitation Measurement)
and GoAmazon (Observations and Modeling of the Green Ocean Amazon)
over the Brazilian Amazon during 2014/2015. This study aims to
assess the uncertainty of the Global Precipitation Measurement
(GPM) satellite constellation in representing the main
characteristics of precipitation over different regions of Brazil.
The Integrated Multi-satellitE Retrievals for GPM (IMERG)
(level-3) and the Goddard Profiling Algorithm (GPROF) (level-2)
algorithms are evaluated against ground-based radar observations,
specifically, the S-band weather radar from the Amazon Protection
National System (SIPAM) and the X-band dual polarization weather
radar (X-band CHUVA radar) as references. The space-based rainfall
estimates, based on active microwave (e.g., TRMM-PR and GPM-DPR
[at Ku-band] radars) are also used as references. The results for
the CHUVA-Vale campaign suggest that GPROF has relatively good
agreement (spatial distribution and accumulated rainfall),
especially for convective rain cases, due the significant presence
of ice scattering. However, the intensity and volume of
light/moderate rains is overestimated and performance related to
light/heavy rains (underestimated) are intrinsically linked to
convectivestratiform rainfall occurrences over the study region.
For the study over the Central Amazon Region (CHUVA-GoAmazon),
results showed that during the wet season, IMERG, which uses the
GPROF2014 rainfall retrieval from the GPM Microwave Imager (GMI)
sensor, significantly overestimates the frequency of heavy
rainfall volumes at around 00:0004:00 UTC and 15:0018:00 UTC. This
overestimation is particularly evident over the Negro,
Solim{\~o}es and Amazon rivers due to the poorlycalibrated
algorithm over water surfaces. On the other hand, during the dry
season, the IMERG product underestimates mean precipitation in
comparison to the S-band SIPAM radar, mainly due to the fact that
isolated convective rain cells in the afternoon are not detected
by the satellite precipitation algorithm. The study based on
verification of GPM level 2 by traditional and object-based
analysis shows that volume and occurrence of heavy rainfall are
underestimated, a good agreement of GPROF2014 for TMI and GMI
versus TRMM PR and GPM DPR (Ku band) rainfall retrievals,
respectively, was noted. Such most evident good performances were
found through continuous and categorical analyses, especially
during the wet season, where the number of objects and larger
areas were observed. The larger object area seen by GPROF2014(GMI)
compared to DPR (Ku band) was directly linked to the structure of
vertical profiles of the precipitanting systems and the presence
of bright band was the main source of uncertainty on the
estimation of precipitation area and intensity. The results via
error modeling, through the Precipitation Uncertainties for
Satellite Hydrology (PUSH) framework, demonstrated that the PUSH
model was suitable for characterizing the error from the IMERG
algorithm when applied to S-band SIPAM radar estimates. PUSH could
efficiently predict the error distribution in terms of spatial and
intensity distributions. However, an underestimation
(overestimation) of light satellite rain rates was observed during
the dry (wet) period, mainly over the river. Although the
estimated error showed a lower standard deviation than the
observed error, they exhibited good correlations to other,
especially in capturing the systematic error along the Negro,
Solim{\~o}es and Amazon rivers, especially during the wet season.
RESUMO: Estudos que investigam e avaliam a qualidade,
limita{\c{c}}{\~o}es e incertezas das estimativas de
precipita{\c{c}}{\~a}o de sat{\'e}lites s{\~a}o fundamentais
para assegurar o uso correto e bem-sucedido desses produtos em
aplica{\c{c}}{\~o}es, como estudos clim{\'a}ticos, modelagem
hidrol{\'o}gica e monitoramento de desastres naturais. Em
regi{\~o}es do globo que n{\~a}o possuem observa{\c{c}}{\~o}es
in situ, esses estudos apenas s{\~a}o poss{\'{\i}}veis
atrav{\'e}s de campanhas intensivas de medi{\c{c}}{\~a}o de
campo, que oferecem uma gama de medi{\c{c}}{\~o}es de
superf{\'{\i}}cie de alta qualidade, por exemplo, CHUVA
(Cloudprocesses of tHe main precipitation systems in Brazil: A
contribUtion to cloud re-solVing modeling and to the GlobAl
Precipitation Measurement) e GoAmazon (Observations and Modeling
of the Green Ocean Amazon) sobre a Amaz{\^o}nia Brasileira
durante 2014/2015. Este estudo tem como objetivo avaliar as
incertezas provenientes da constela{\c{c}}{\~a}o de
sat{\'e}lites do Global Precipitation Measurement (GPM) em
representar as principais caracter{\'{\i}}sticas da
precipita{\c{c}}{\~a}o em diferentes regi{\~o}es do Brasil. Os
algoritmos Integrated Multi-satellitE Retrievals for GPM (IMERG)
(level-3) e Goddard Profiling Algorithm (GPROF) (level-2) s{\~a}o
avaliados em contraste as observa{\c{c}}{\~o}es de radares
meteorol{\'o}gicos, especificamente, do Sistema Nacional de
Prote{\c{c}}{\~a}o da Amaz{\^o}nia (SIPAM) e o radar
meteorol{\'o}gico banda X de dupla polariza{\c{c}}{\~a}o
(X-band CHUVA radar) como refer{\^e}ncia. As estimativas de
precipita{\c{c}}{\~a}o, baseadas em radares de microondas ativos
(por exemplo, radares TRMM-PR e GPM-DPR [na banda Ku]) tamb{\'e}m
s{\~a}o utilizadas como refer{\^e}ncia. Os resultados da
campanha CHUVA-Vale sugerem que o GPROF possui uma boa
concord{\^a}ncia (distribui{\c{c}}{\~a}o espacial e
precipita{\c{c}}{\~a}o acumulada), especialmente para casos de
chuva convectiva, devido {\`a} presen{\c{c}}a significativa de
espalhamento por gelo. No entanto, a intensidade e volume de
chuvas leves/moderadas {\'e} superestimada e um desempenho
(subestimado) relacionado {\`a}s chuvas fracas/intensas
diretamente ligado {\`a}s ocorr{\^e}ncias de chuvas
convectivasestratiformes na regi{\~a}o do estudo. Para o estudo
da regi{\~a}o da Amaz{\^o}nia Central (CHUVA-GoAmazon), os
resultados mostraram que, durante a esta{\c{c}}{\~a}o chuvosa, o
IMERG, que utiliza as estimativas de precipita{\c{c}}{\~a}o do
GPROF2014 a partir do sensor GPM Microwave Imager (GMI),
superestima significativamente a freq{\"u}{\^e}ncia de chuvas
intensas em torno de 00:00-04:00 UTC e 15:00-18:00 UTC. Essa
superestimativa {\'e} particularmente evidente nos rios Negro,
Solim{\~o}es e Amazonas devido ao algoritmo apresentasse
erroneamente calibrado sobre as superf{\'{\i}}cies de {\'a}gua.
Por outro lado, durante a esta{\c{c}}{\~a}o seca, o produto
IMERG subestima a precipita{\c{c}}{\~a}o m{\'e}dia em
compara{\c{c}}{\~a}o com o radar banda-s do SIPAM,
principalmente devido ao fato de que c{\'e}lulas convectivas
isoladas {\`a} tarde n{\~a}o s{\~a}o detectadas por tal
algoritmo. O estudo baseado na verifica{\c{c}}{\~a}o das
estimativas do GPM Level 2 por abordagens tradicional e baseada em
objeto mostra que, embora a subestimiativa do volume e
ocorr{\^e}ncia de chuvas intensas, foi observada uma boa
concord{\^a}ncia do GPROF2014 (TMI e GMI) versus TRMM PR e GPM
DPR (Ku band), Respectivamente. Tais evidentes melhores
desempenhos foram encontrados atrav{\'e}s de an{\'a}lises
cont{\'{\i}}nua e categ{\'o}rica, especialmente durante a
esta{\c{c}}{\~a}o chuvosa, onde o maior n{\'u}mero e maiores
{\'a}reas de objetos foram observados. As maiores {\'a}reas,
observadas pelo GPROF2014 (GMI) comparada ao DPR (banda Ku) esteve
diretamente ligada {\`a} estrutura de perfis verticais dos
sistemas de precipitantes e a presen{\c{c}}a de banda brilhante
foi a principal fonte de incerteza na estimativa da {\'a}rea e
intensidade de precipita{\c{c}}{\~a}o. Os resultados referentes
{\`a} modelagem do erro, atrav{\'e}s da ferramenta Precipitation
Uncertainties for Satellite Hydrology (PUSH), as an{\'a}lises
demonstraram que o modelo PUSH foi adequado para caracterizar o
erro do algoritmo IMERG quando aplicado {\`a}s estimativas de
radar banda S do SIPAM. O modelo PUSH p{\^o}de prever
eficientemente a distribui{\c{c}}{\~a}o de erro em termos
espaciais e de intensidade. No entanto, observou-se uma
subestimativa (superestimativa) das taxas de chuva fracas do
sat{\'e}lite durante o per{\'{\i}}odo seco (chuvoso),
especialmente ao longo do rio. Embora o erro estimado tenha
apresentado menor desvio padr{\~a}o do que o erro observado, eles
apresentaram boas correla{\c{c}}{\~o}es entre si, especialmente
na captura do erro sistem{\'a}tico ao longo dos rios Negro,
Solim{\~o}es e Amazonas, especialmente durante a
esta{\c{c}}{\~a}o chuvosa.",
committee = "Herdies, Dirceu Luis (presidente) and Vila, Daniel Alejandro
(orientador) and Sapucci, Luiz Fernando and Maggioni, Viviana and
Rodriguez, Carlos Augusto Morales",
englishtitle = "Caracter{\'{\i}}sticas e modelagem de erro nas estimativas de
precipita{\c{c}}{\~a}o do sat{\'e}lite gpm durante as campanhas
do chuva no brasil",
language = "en",
pages = "161",
ibi = "8JMKD3MGP3W34P/3NU3598",
url = "http://urlib.net/ibi/8JMKD3MGP3W34P/3NU3598",
targetfile = "publicacao.pdf",
urlaccessdate = "27 abr. 2024"
}