Banca de QUALIFICAÇÃO: KEMUEL MACIEL FREITAS

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : KEMUEL MACIEL FREITAS
DATE: 31/08/2023
TIME: 10:00
LOCAL: Auditório NTB, UFOPA/Campus Tapajós
TITLE:

FOREST FIRE SUSCEPTIBILITY PREDICTION IN THE APA TRIUNFO DO XINGU WITH ARTIFICIAL INTELLIGENCE TECHNIQUES

KEY WORDS:

Remote Sensing. Conservation Area. Machine Learning. Random Forest; Amazon

PAGES: 55
BIG AREA: Outra
AREA: Ciências Ambientais
SUMMARY:

The use of fire, despite being a common practice in the Amazon, can result in a series of environmental, economic, and social damages. The behavior and characteristics of fire are determined by a variety of factors, including topography, vegetation, and climate, rendering the task of predicting where to allocate resources for combatting and preventing forest fires complex. To address this, artificial intelligence tools such as Neural Networks, Random Forests, and Classification and Regression Trees have demonstrated promise in the field of fire prediction, leading to the development of various models on a global scale, with applications also tailored to the Brazilian context. However, the management of an area as vast as the Amazon demands a specific and optimized utilization of these resources, along with enhancements in decision-making strategies pertaining to the prevention and mitigation of environmental risks. This is especially critical in conservation areas facing pressures from both recurrent fires and deforestation, as exemplified by the Triunfo do Xingu Environmental Protection Area in the state of Pará. This research aims to conduct the mapping of areas susceptible to forest fires within the Triunfo do Xingu Environmental Protection Area, employing machine learning algorithms to ascertain the influence of environmental, topographical, socioeconomic, and vegetation factors on fire occurrences. The Normalized Burn Ratio (NBR) is intended to be employed as the response variable, alongside a set of 12 predictor variables encompassing physical, environmental, and socioeconomic dimensions, spanning the period from 2005 to 2020. It is expected that by assessing the performance of different algorithms in the regression process through three metrics, R², RMSE, and MAE, the optimal fire prediction model can be derived. Subsequently, this model can yield a representative map of areas most susceptible to fires within the Triunfo do Xingu Environmental Protection Area.

COMMITTEE MEMBERS:
Presidente - 1963026 - ANDERSON ALVARENGA DE MOURA MENESES
Externo à Instituição - MARLA TERESINHA BARBOSA GELLER - ULBRA
Externa à Instituição - MILENA MARILIA NOGUEIRA DE ANDRADE - UFRA