@MastersThesis{HuamánChinchay:2018:UsMuGO,
author = "Huam{\'a}n Chinchay, Joao Henry",
title = "Uso de multi-canais do GOES-16 para previs{\~a}o imediata de
densidade de descargas el{\'e}tricas",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2018",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2018-05-29",
keywords = "nowcasting, descarga el{\'e}trica, sensoriamento remoto, ABI,
GLM, lightning, remote sensing.",
abstract = "Este trabalho emprega os multi-canais do sat{\'e}lite GOES-16
para a previs{\~a}o imediata de densidades de descargas
el{\'e}tricas. Para isto se utilizaram imagens do Advanced
Baseline Imager (ABI) e as coordenadas dos flashes registrados
pelo Geostationary Lightning Mapper (GLM). A {\'a}rea de estudo
se localiza na regi{\~a}o norte do Brasil, sobre a cidade do
Manaus. A metodologia empregada consistiu em identificar as
respostas radiativas que apresentam as nuvens de tempestades com
respeito aos atributos f{\'{\i}}sicos: tamanho das
part{\'{\i}}culas, intensidade do fluxo ascendente, profundidade
da nuvem e glacia{\c{c}}{\~a}o no seu topo; mediante campos de
interesse (singulares bandas e suas diferen{\c{c}}as), e
relaciona-los com sete categorias de densidade de flash,
acumulados entre 0-5, 5-10 e 10-15 minutos posteriores ao
hor{\'a}rio das imagens do ABI. Desta rela{\c{c}}{\~a}o se
elaboraram histogramas de frequ{\^e}ncia relativa, que permitiu
identificar aos campos de interesse que apresentam a maior
sensibilidade com respeito ao incremento dos flashes. Por meio das
frequ{\^e}ncias relativas acumuladas e da curva com a m{\'a}xima
derivada foi poss{\'{\i}}vel determinar os campos de interesse
(associado a cada atributo f{\'{\i}}sico) e seus limiares, os
quais foram empregados como preditores da densidade de descargas
el{\'e}tricas. A partir destes preditores, foram elaborados sete
modelos de previs{\~a}o para o per{\'{\i}}odo diurno, noturno e
para as 24 horas do dia. As avalia{\c{c}}{\~o}es das
previs{\~o}es dos preditores e dos modelos mostraram que no
intervalo de tempo de 5-10 minutos se observam os menores valores
de false alarme (FAR) e a maior probabilidade de
detec{\c{c}}{\~a}o (POD). Tomando como crit{\'e}rio baixos
valores de FAR e altos valores de POD, determinou-se que a banda
de 10.35 \μm {\'e} o melhor preditor. No caso dos modelos,
o modelo-05, formado pelas bandas 10.35 \μm e 3.9 \μm
- 10.35 \μm, foi o que apresentou o melhor resultado para o
per{\'{\i}}odo noturno, enquanto que para os outros dois
per{\'{\i}}odos do dia, o modelo-07, formado pela banda de 10.35
\μm e tend{\^e}ncia temporal em 30 minutos, foi o melhor.
ABSTRACT: This study employs the GOES-16 satellite multi-channel
for the immediate prediction of electric discharge densities. For
this we used images from the Advanced Baseline Imager (ABI) and
the coordinates of the flashes recorded by the Geostationary
Lightning Mapper (GLM). The study area is located in the northern
region of Brazil, in the city of Manaus. The methodology used
consisted in identifying the radiative responses of storm clouds
with respect to physical attributes: particle size, updraft
strength, cloud depth and cloud-top glaciation; using interest
fields (singular bands and their differences), and relates them to
seven categories of flash density, accumulated between 0-5, 5-10
and 10-15 minutes after the time of ABI images. Relative frequency
histograms were elaborated from this relation, which allowed to
identify to the interest fields that present the greater
sensitivity with regard to the increase of the flashes. By means
of the relative accumulated -frequencies and the maximum
derivative curve it was possible to determine the interest fields
(associated with each physical attribute) and their thresholds,
which were used as predictors of the flash density. From these
predictors, seven prediction models were developed for the daytime
and night periods and 24 hour of the day. Predictor and model
predictor evaluations showed that the lowest false alarms (FAR)
and the highest probability of detection (POD) were observed in
the 5-10 minute interval. Based on low FAR values and high POD
values, the 10.35 \μm band was determined to be the best
predictor. In the case of the models, the model-05, formed by the
bands 10.35 \μm and 3.9 \μm - 10.35 \μm, was the
best for the night period, whereas for the other periods of the
day, the model-07, formed by the 10.35 \μm band and its
temporary trend of 30 minutes, was the best.",
committee = "Vila, Daniel Alejandro (presidente) and Machado, Luiz Augusto
Toledo (orientador) and Ceballos, Juan Carlos and Kummerow,
Christian",
englishtitle = "Use of the GOES-16 satellite multi-channel for the immediate
prediction of electric discharge densities",
language = "pt",
pages = "163",
ibi = "8JMKD3MGP3W34R/3R5M85H",
url = "http://urlib.net/ibi/8JMKD3MGP3W34R/3R5M85H",
targetfile = "publicacao.pdf",
urlaccessdate = "27 abr. 2024"
}