@PhDThesis{Palharini:2021:AnExPr,
author = "Palharini, Rayana Santos Araujo",
title = "Analysis of extreme precipitation events estimated by satellite
and its relationship with mesoscale convective systems over South
America",
school = "Instituto Nacional de Pesquisas Espaciais (INPE)",
year = "2021",
address = "S{\~a}o Jos{\'e} dos Campos",
month = "2021-02-24",
keywords = "satellite, extreme rainfall, estimates, MCS, sat{\'e}lite, chuva
extrema, estimativas.",
abstract = "Climate change is increasing the intensity and frequency of
extreme events around the world and our society is vulnerable to
the dangers of natural disasters. According to the Brazilian Atlas
of Natural Disasters, a total of 38,996 disasters were recorded
during the period 1991-2012. According to this database,
approximately 40% of hydrometeorological events were caused by
floods, landslides, hail, local storms and windstorms. One of the
main meteorological variables associated with natural disasters is
precipitation. Understanding the behavior and improving the
prediction of these events is of fundamental importance as heavy
rainfall causes irreparable damage and causes great economic
losses for a country. With the objective of improve the
understanding about extreme rainfall of Brazil a daily 1°x1°
gridded precipitation database was used to assess the performance
of different precipitation products to retrieval extreme rainfall
at different regions of Brazil, as well as an analysis of the
Mesoscale Convective Systems and their influence on extreme rain.
The products evaluated in this investigation were 3B42 RT v7.0,
3B42 RT v7.0 uncalibrated, CMORPH V1.0 RAW, CMORPH V1.0 CRT,
GSMAP-NRT-no gauge v6.0, GSMAP-NRT- gauge v6.0, CHIRP V2.0, CHIRPS
V2.0, PERSIANN CDR v1 r1, CoSch and TAPEER v1.5 from Frequent
Rainfall Observations on GridS (FROGS) database. Some products
considered in this investigation are adjusted with rain gauge
values and others only with satellite information. In this study,
these two sets of products were considered. In addition,
gauge-based daily precipitation data, provided by Brazils National
Institute for Space Research, were used as reference in the
analyses. In order to compare gauge-based daily precipitation and
satellite-based data for extreme values, statistical techniques
were used to evaluate the performance the selected satellite
products over the tropical region of South America. According to
the results, the threshold for rain to be considered an extreme
event in South America presented high variability, ranging from 20
to 150 mm/day, depending on the region and the percentile
threshold chosen for analysis. In addition, the results showed
that the ability of the satellite estimates to retrieve rainfall
extremes depends on the geographical location and large-scale
rainfall regimes. Each region of Brazil is characterized by
extremes of rain with different intensities. The regions with the
highest values are south and north regions of Brazil with values
around 125.0 mm/day. In both regions, the GSMAP product (with and
without rain gauges adjustments) have a better performance. On the
other hand, the regions with the lowest intensities are the
northeastern region (inland and coast) with more frequent extreme
values around the 35.0 mm/day. In those regions 3B42RT v7.0 and
3B42RT v7.0 uncalibrated demonstrated a better performance
respectively. It is worth mentioning that the precipitation values
found in this work do not necessarily cause disasters or generate
impacts in the analyzed regions, they were considered extreme from
a statistical point of view, considering the analyzed database. In
order to describe the morphological characteristics of the MCS and
identify the influence of these systems on extreme rain during the
period 2012-2016 in the tropical region of South America, the
dataset used in this investigation was the CACATOES dataset. It is
a level-3 product derived from the Tracking Of Organized
Convection Algorithm through 3D segmentatioN (TOOCAN). According
to results, small systems with a duration smaller than 12 hours
are the ones that occurred with a higher frequency. However,
systems that have duration above 12 hours are the ones that most
contributed to the extreme rain. A significant influence of the
MCS was identified over a large part of the South America regions.
In addition, the influence of the MCS over the investigated region
presented a significant variability. In order to analyze five case
studies associated to extreme rain which caused natural disaster
in five different regions of Brazil was analysed. The regions were
defined based on previous studies according to the climatological
distribution of rainfall in each region. To be considered
statistically extreme, the cases were analyzed considering rain
values above the 99th percentile during the period 2012-2016.
Three databases were used: Precipitation from (i) rain gauges
stations and (ii) different satellite-based estimates and (iii)
Mesoscale convective tracking data. The methodology was based in
identifying events, analyzing the performance of satellite
precipitation estimates to detect the observed extreme rain and
finally quantifying the influence of convective systems on the
extreme rain that occurred. Although all regions of Brazil are
subject to the occurrence of natural disasters caused by extreme
rains, the results suggest that the impacts caused in each region
have different magnitudes. It was noticed that the convective
systems influenced above 90.0 % of the extreme rains in the case
analysed in South region of Brazil while it influenced about 60.0
% to 90.0 % of the extreme rains in the case analysed in Northeast
region of Brazil. In general, satellite products have identified
rain events, however, in the southern region of Brazil, products
have tended to overestimate rainfall, while other regions have
tended to underestimate extreme rain values. It can be seen then
that it is still a challenge for the methods used in the satellite
precipitation estimation products to accurately identify specific
extreme rain events. RESUMO: As mudan{\c{c}}as clim{\'a}ticas
est{\~a}o aumentando a intensidade e a frequ{\^e}ncia de eventos
extremos em todo o mundo. Cada vez mais, a sociedade est{\'a}
vulner{\'a}vel aos perigos dos desastres naturais. De acordo com
o Atlas Brasileiro de Desastres Naturais, um total de 38.996
desastres foram registrados no per{\'{\i}}odo de 1991 a 2012. De
acordo com esta base de dados, aproximadamente 40% dos eventos
hidrometeorol{\'o}gicos foram causados por
inunda{\c{c}}{\~o}es, deslizamentos de terra, granizo,
tempestades locais e vendavais. Uma das principais vari{\'a}veis
meteorol{\'o}gicas associadas aos desastres naturais {\'e} a
precipita{\c{c}}{\~a}o. Entender o comportamento e melhorar a
previs{\~a}o desses eventos {\'e} de fundamental
import{\^a}ncia, pois chuvas intensas causam danos
irrepar{\'a}veis e grandes perdas econ{\^o}micas para um
pa{\'{\i}}s. Com o objetivo de melhorar o entendimento sobre as
chuvas extremas do Brasil, um banco de dados di{\'a}rio de
precipita{\c{c}}{\~a}o em grade de 1°x1° foi usado para avaliar
a habilidade de diferentes produtos de estimativas de
precipita{\c{c}}{\~a}o por sat{\'e}lite em detectar as chuvas
extremas em diferentes regi{\~o}es do Brasil, bem como a
an{\'a}lise da sistemas convectivos de mesoescala e sua
influ{\^e}ncia nas chuvas extremas. Os produtos avaliados nesta
investiga{\c{c}}{\~a}o foram 3B42 RT v7.0, 3B42 RT v7.0 n{\~a}o
calibrado, CMORPH V1.0 RAW, CMORPH V1.0 CRT, GSMAP-NRT-sem
pluviometro v6.0, GSMAP-NRT-com pluviometro v6.0 , CHIRP V2.0,
CHIRPS V2.0, PERSIANN CDR v1 r1, CoSch e TAPEER v1.5 do banco de
dados Frequent Rainfall Observations on GridS (FROGS). Alguns
produtos considerados nesta investiga{\c{c}}{\~a}o s{\~a}o
ajustados com valores de pluvi{\^o}metro e outros apenas com
informa{\c{c}}{\~o}es de sat{\'e}lite. Neste estudo, esses dois
conjuntos de produtos foram considerados. Al{\'e}m disso, dados
de precipita{\c{c}}{\~a}o di{\'a}ria baseados em indicadores,
fornecidos pelo Instituto Nacional de Pesquisas Espaciais do
Brasil, foram usados como refer{\^e}ncia nas an{\'a}lises. A fim
de comparar a precipita{\c{c}}{\~a}o di{\'a}ria baseada em
pluvi{\^o}metro e os dados de sat{\'e}lite para valores
extremos, t{\'e}cnicas estat{\'{\i}}sticas foram usadas para
avaliar o desempenho dos produtos de sat{\'e}lite selecionados na
regi{\~a}o tropical da Am{\'e}rica do Sul. De acordo com os
resultados, o limiar para chuva ser considerada um evento extremo
na Am{\'e}rica do Sul apresentou grande variabilidade, variando
de 20,0 a 150,0 mm/dia, dependendo da regi{\~a}o e do limiar de
percentil escolhido para an{\'a}lise. Al{\'e}m disso, os
resultados mostraram que a capacidade das estimativas de
sat{\'e}lite de recuperar os extremos de chuva depende da
localiza{\c{c}}{\~a}o geogr{\'a}fica e dos regimes de chuva em
grande escala. Cada regi{\~a}o do Brasil {\'e} caracterizada por
extremos de chuva com intensidades diferentes. As regi{\~o}es com
os maiores valores s{\~a}o as regi{\~o}es Sul e Norte do Brasil
com valores em torno de 125,0 mm / dia. Em ambas as regi{\~o}es,
o produto GSMAP (com e sem ajustes de pluvi{\^o}metros) tem
melhor desempenho. Por outro lado, as regi{\~o}es com as menores
intensidades s{\~a}o a regi{\~a}o Nordeste (interior e litoral)
com valores extremos mais frequentes em torno dos 35,0 mm/dia.
Nessas regi{\~o}es, o 3B42RT v7.0 e o 3B42RT v7.0 n{\~a}o
calibrado demonstraram um melhor desempenho, respectivamente. Vale
ressaltar que os valores de precipita{\c{c}}{\~a}o encontrados
neste trabalho n{\~a}o necessariamente causam desastres ou geram
impactos nas regi{\~o}es analisadas, foram considerados extremos
do ponto de vista estat{\'{\i}}stico, considerando a base de
dados analisada. Com o objetivo de descrever as
caracter{\'{\i}}sticas morfol{\'o}gicas do MCS e identificar a
influ{\^e}ncia desses sistemas nas chuvas extremas durante o
per{\'{\i}}odo de 2012-2016 na regi{\~a}o tropical da
Am{\'e}rica do Sul, o conjunto de dados utilizado nesta
investiga{\c{c}}{\~a}o foi o conjunto de dados CACATOES. {\'E}
um produto de n{\'{\i}}vel 3 derivado do Algoritmo de
Rastreamento de Convec{\c{c}}{\~a}o Organizada por meio da
segmenta{\c{c}}{\~a}o 3D (TOOCAN). De acordo com os resultados,
pequenos sistemas com dura{\c{c}}{\~a}o inferior a 12 horas
s{\~a}o os que ocorreram com maior frequ{\^e}ncia. Por{\'e}m,
os sistemas que t{\^e}m dura{\c{c}}{\~a}o acima de 12 horas
s{\~a}o os que mais contribuem para as chuvas extremas. Foi
identificada uma influ{\^e}ncia significativa do MCS em grande
parte das regi{\~o}es sul-americanas. Al{\'e}m disso, a
influ{\^e}ncia do MCS sobre a regi{\~a}o investigada apresentou
uma variabilidade significativa. Com o objetivo de analisar cinco
estudos de caso associados {\`a}s chuvas extremas que causaram
desastres naturais em cinco diferentes regi{\~o}es do Brasil
foram analisados. As regi{\~o}es foram definidas com base em
estudos anteriores de acordo com a distribui{\c{c}}{\~a}o
climatol{\'o}gica das chuvas em cada regi{\~a}o. Para serem
considerados estatisticamente extremos, os casos foram analisados
considerando-se valores de chuva acima do percentil 99 durante o
per{\'{\i}}odo de 2012-2016. Tr{\^e}s bancos de dados foram
usados: Precipita{\c{c}}{\~a}o de (i) esta{\c{c}}{\~o}es
pluviom{\'e}tricas e (ii) diferentes estimativas baseadas em
sat{\'e}lite e (iii) dados de rastreamento convectivo de
mesoescala. A metodologia baseou-se na identifica{\c{c}}{\~a}o
de eventos, na an{\'a}lise do desempenho das estimativas de
precipita{\c{c}}{\~a}o por sat{\'e}lite para detectar as chuvas
extremas observadas e, por fim, quantificar a influ{\^e}ncia dos
sistemas convectivos nas chuvas extremas ocorridas. Embora todas
as regi{\~o}es do Brasil estejam sujeitas {\`a} ocorr{\^e}ncia
de desastres naturais causados por chuvas extremas, os resultados
sugerem que os impactos causados em cada regi{\~a}o t{\^e}m
magnitudes diferentes. Percebeu-se que o sistema convectivo
influenciou acima de 90,0 % das chuvas extremas no caso analisado
na regi{\~a}o Sul do Brasil enquanto influenciou cerca de 60,0 %
a 90,0 % das chuvas extremas no caso analisado na regi{\~a}o
Nordeste do Brasil. Em geral, os produtos de sat{\'e}lite
identificam eventos de chuva, no entanto, na regi{\~a}o sul do
Brasil, os produtos tendem a superestimar as chuvas, enquanto
outras regi{\~o}es tendem a subestimar os valores extremos de
chuva. Pode-se ver ent{\~a}o que ainda {\'e} um desafio para os
m{\'e}todos usados nos produtos de estimativa de
precipita{\c{c}}{\~a}o por sat{\'e}lite identificar com
precis{\~a}o eventos espec{\'{\i}}ficos de chuva extrema.",
committee = "Gan, Manoel Alonso (presidente) and Vila, Daniel Alejandro
(orientador) and Ferreira, Nelson Jesuz and Rodrigues, Daniele
T{\^o}rres and Mattos, Enrique Vieira",
englishtitle = "An{\'a}lise de eventos extremos de precipita{\c{c}}{\~a}o
estimados por sat{\'e}lite e sua rela{\c{c}}{\~a}o com sistemas
convectivos de mesoescala sobre a Am{\'e}rica do Sul",
language = "en",
pages = "168",
ibi = "8JMKD3MGP3W34R/4465CAB",
url = "http://urlib.net/ibi/8JMKD3MGP3W34R/4465CAB",
targetfile = "publicacao.pdf",
urlaccessdate = "01 maio 2024"
}