Banca de QUALIFICAÇÃO: ALEXANDRE HENRIQUE QUADROS GRAMOSA

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
STUDENT : ALEXANDRE HENRIQUE QUADROS GRAMOSA
DATE: 31/08/2022
TIME: 14:00
LOCAL: https://www.meet.google.com/hxi-yqib-mak
TITLE:

Seasonal Models for Zero Inflated Extreme Data Analysis

KEY WORDS:

Extreme Events, Zero Inflation, Bayesian Analysis, Precipitation, MCMC, Seasonality.

PAGES: 58
BIG AREA: Ciências Agrárias
AREA: Agronomia
SUMMARY:

The generalized extreme value distribution (GEV) is the limiting result for modeling maximum blocks of size n, which is used in modeling extreme events. However, the situation where data have an excessive number of zeros can also occur in extreme data, making it challenging to analyze and estimate these events by the usual GEV distribution. The Generalized Distribution of Extreme Values Inflated by zeros (ZIGEV) was created recently to solve this problem, with the help of its inflator parameter ω. The main objective of this work is an application of the new ZIGEV distribution in daily rainfall data, transformed into blocks of monthly maximums, where there may be months in which there was no precipitation, which is computed as zero. Time series of the mesoregions of the state of Pernambuco, the northeast region of Brazil, were analyzed. Some of them with a predominance of non-rainy months. Also, in this work, a seasonal model of ZIGEV, the SZIGEV, was created to analyze the seasonality of data with extreme values inflated by zeros. In this case, rainfall data from the cities of Recife and Petrolina in the state of Pernambuco and Sao João do Piauí and Teresina in the state of Piauí were used. In both analyses, inferences were made on the Bayesian paradigm, with parameter estimations being made by numerical approximations of the a posteriori distribution using MCMC. The results of these applications reinforce the need to use the ZIGEV distribution for the analysis of extreme values, especially when inflated from zero, compared to the GEV distribution, as it obtained a more accurate result and with a better quality of fit. However, when comparing ZIGEV with SZIGEV, it was noticed that the seasonal model was better and more adjusted than the ZIGEV, especially for data with more significant seasonality.

BANKING MEMBERS:
Externo à Instituição - FERNANDO FERRAZ DO NASCIMENTO - UFPI
Externo à Instituição - FIDEL ERNESTO CASTRO MORAES - UFRJ
Interno - 2279488 - ANTONIO SAMUEL ALVES DA SILVA
Presidente - 1756995 - FRANK SINATRA GOMES DA SILVA
Interno - 2718900 - JOSIMAR MENDES DE VASCONCELOS