@MastersThesis{Nobre:2021:ImAsDa,
author = "Nobre, Jo{\~a}o Pedro Gon{\c{c}}alves",
title = "Impacto da assimila{\c{c}}{\~a}o de dados conjunto-variacional
na previs{\~a}o de epis{\'o}dios de chuvas intensas no Nordeste
brasileiro",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2021",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2021-02-22",
keywords = "3DVar, 3DEnVar, sistemas convectivos de mesoescala,
assimila{\c{c}}{\~a}o de dados, gridpoint statistical
interpolation, mesoscale convective systems, data assimilation.",
abstract = "O Nordeste Brasileiro (NEB) encontra-se na regi{\~a}o tropical do
Brasil, limitado pelo oceano Atl{\^a}ntico e com um clima e
vegeta{\c{c}}{\~a}o fortemente influenciados pelo Planalto da
Borborema. A presen{\c{c}}a desse planalto marca
significativamente o clima da regi{\~a}o ao manter a leste massas
de ar com caracter{\'{\i}}sticas mais {\'u}midas e
respons{\'a}veis pela ocorr{\^e}ncia de chuvas e o lado oeste,
massas de ar predominantemente secas. Este trabalho visa avaliar a
performance dos diferentes sistemas de assimila{\c{c}}{\~a}o
dados; 3DVar (Three-Dimensional Variational), EnKF (Ensemble
Kalman Filter) e o h{\'{\i}}brido, 3DEnVar (Three-Dimensional
Ensemble-Variational), no estudo de epis{\'o}dios de SCM
(Sistemas Convectivos de Mesoescala), utilizando previs{\~o}es do
modelo meteorol{\'o}gico de mesoescala WRF (Weather Research and
Forecasting), em compara{\c{c}}{\~a}o com previs{\~o}es do WRF
inicializadas com dados do GEFS (Global Ensemble Forecast System),
para dois eventos de SCM ocorridos nos dias 14 e 24 de janeiro de
2017. Para isso, ser{\'a} utilizada a vers{\~a}o V3.0.0 do SMR
(Sistema de Modelagem Regional) do CPTEC (Centro de Previs{\~a}o
de Tempo e Estudos Clim{\'a}ticos) constitu{\'{\i}}do de dois
componentes: o modelo WRF e o sistema de assimila{\c{c}}{\~a}o
de dados GSI (Gridpoint Statistical Interpolation). Atualmente, o
SMR encontra-se configurado para fornecer condi{\c{c}}{\~o}es
iniciais ao modelo WRF atualizadas pelo 3DVar, que utiliza uma
matriz de covari{\^a}ncia dos erros de previs{\~a}o
climatol{\'o}gica, para ponderar os erros do modelo no processo
de minimiza{\c{c}}{\~a}o da fun{\c{c}}{\~a}o custo. No
presente trabalho, utilizou-se o 3DEnVar no SMR, que consiste de
um sistema 3DVar, cuja matriz de covari{\^a}ncia dos erros de
previs{\~a}o {\'e} calculada atrav{\'e}s da
combina{\c{c}}{\~a}o linear dos membros de um conjunto de
previs{\~o}es que servir{\~a}o para atualizar a matriz
climatol{\'o}gica do SMR, com os erros do dia. Desse modo, o
presente trabalho visa melhorar a detec{\c{c}}{\~a}o e
estimativa da quantidade de chuva dos casos de SCM sobre o NEB ao
utilizar a an{\'a}lise do 3DEnVar na previs{\~a}o de chuva
acumulada em 24 h. Resultados obtidos ilustram que o sistema de
assimila{\c{c}}{\~a}o de dados h{\'{\i}}brido (3DEnVar) foi
capaz de gerar melhores an{\'a}lises, se comparado a um sistema
variacional puro (3DVar), para os campos de press{\~a}o
superficial e umidade ao analisar estatisticamente o desempenho
dos sistemas variacionais atrav{\'e}s do BIAS e RMSE (Root Mean
Square Error). O melhoramento obtido na representa{\c{c}}{\~a}o
dos campos de umidade atrav{\'e}s do 3DEnVar foi essencial para
obten{\c{c}}{\~a}o de boas previs{\~o}es de chuva acumulada em
24 horas, com o modelo WRF, ao ser comparado com a
precipita{\c{c}}{\~a}o registrada por esta{\c{c}}{\~o}es
meteorol{\'o}gicas em superf{\'{\i}}cie, do Instituto Nacional
de Meteorologia (INMET), para os dias 14 e 24 de janeiro de 2017
sobre o NEB. ABSTRACT: The Brazilian Northeast (BNE) is located in
the tropical region of Brazil, it is bounded by the Atlantic
Ocean, and its climate and vegetation is strongly affected by the
Borborema Plateau. The presence of the plateau significantly
defines the climate region. It keeps the humid air masses to the
east, which is responsible for the rain episodes, and at the west
side (northeastern hinterland) predominantly dry air masses are
observed. This work evaluates the performance obtained from
different data assimilation methods, 3DVar (Three-Dimensional
Variational), EnKF (Ensemble Kalman Filter), and 3DEnVar
(Three-Dimensional Ensemble Variational), in the study of
Mesoscale Convective Systems (MCS) episodes. The deterministic
predictions was used from the GEFS (Global Ensemble Forecast
System) model to compare with the WRF (Weather Research and
Forecasting) numerical weather forecast model analysis updated by
different data assimilation methods for January 14th and 24th,
2017 MCS episodes. For that purpose, the RMS (Regional Modeling
System) 3.0.0 version from the Center for Weather Forecasting and
Climate Studies was used with two components: the WRF mesoscale
model and the GSI (Gridpoint Statistical Interpolation) data
assimilation system. Currently, the SMR provides the WRF initial
conditions using the 3DVar data assimilation methodology that uses
a climatological forecast error covariance matrix to weight the
model errors in the cost function minimization process. At this
work, the 3DEnVar was used in the SMR, and it updates the SMR
climatological covariance matrix through the forecast ensemble
members with the errors of the day. To summarise, the present work
studied the improvements in the detection and estimation of 24
hours rain accumulated precipitation quality in MCS cases over
BNE. The statistics indexes BIAS and RMSE (Root Mean Square Error)
show that the hybrid data assimilation system (3DEnVar) is the
best variational system in producing better analyses for the
surface pressure and humidity fields. The best humidity
performances with 3DEn- Var were essential in forecasting 24 hours
accumulated precipitation compared with observational data from
the Brazilian National Institute of Meteorology (INMET) stations,
during convective storms over BNE on January 14th and 24th,
2017.",
committee = "Gon{\c{c}}alves, Lu{\'{\i}}s Gustavo Gon{\c{c}}alves de
(presidente) and Herdies, Dirceu Luis (orientador) and Vendrasco,
{\'E}der Paulo (orientador) and Bastarz, Carlos Frederico and
Harter, Fabricio Pereira",
englishtitle = "Ensemble-variational data assimilation impact in heavy rain
forecasting episodes in brazilian Northeast",
language = "pt",
pages = "143",
ibi = "8JMKD3MGP3W34R/448DGFE",
url = "http://urlib.net/ibi/8JMKD3MGP3W34R/448DGFE",
targetfile = "publicacao.pdf",
urlaccessdate = "09 maio 2024"
}