Banca de DEFESA: GUTIERRE PEREIRA MACIEL

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
STUDENT : GUTIERRE PEREIRA MACIEL
DATE: 28/02/2025
TIME: 14:00
LOCAL: Por videoconferência
TITLE:

"AI-DRIVEN GREEN TECH FOR ANALYZING SOIL, SEDIMENTS AND SUBSTRATES IN MINING AREAS OF THE EASTERN AMAZON"

KEY WORDS:

pXRF; Nix Pro™; soil quality; machine learning; Serra Pelada; Carajás Mineral Province.

PAGES: 102
BIG AREA: Ciências Agrárias
AREA: Agronomia
SUBÁREA: Ciência do Solo
SPECIALTY: Química do Solo
SUMMARY:

The assessment of areas affected by mining activities is crucial for understanding their chemical and physical attributes, which influence plant growth during the revegetation process, as well as for monitoring potentially toxic elements (PTEs) that may pose environmental and human health risks. This thesis evaluated the use of proximal sensors for the chemical and physical characterization of soils, sediments, and substrates in areas directly or indirectly impacted by mining activities in the Eastern Amazon. Additionally, machine learning algorithms were applied to predict the physical and chemical attributes of these matrices. The predictions were performed using machine learning algorithms, including Random Forest and Support Vector Machine, based on data from proximal sensors. The first chapter explored the use of portable X-ray fluorescence (pXRF) for determining the total and available contents of PTEs in areas affected by artisanal gold mining in the Serra Pelada region. The best regression model fits for total PTE contents were achieved for elements such as Cu, Fe, Mn, and Pb, while the prediction models for available Cu and Zn stood out with high performance. The second chapter, in turn, focuses on the characterization of iron mining pit substrates in the Carajás Mineral Province using pXRF and the Nix Pro™ color sensor, which provide total elemental content and color parameter data, respectively. The results indicated that pXRF and Nix Pro™ can effectively characterize open-pit iron mining substrates by providing valuable information on their elemental composition and color properties. pXRF allowed for the quantification of total elemental contents, particularly Fe, Al, and Si, which are key components of these substrates. Meanwhile, the Nix Pro™ color sensor captured color variations related to mineralogical differences, which are associated with soil properties. Furthermore, data fusion provided slightly improved predictive models for various soil attributes compared to the isolated use of each sensor’s data. Overall, the findings of this thesis highlight the great potential of proximal sensors, specifically pXRF and Nix Pro™, for the characterization of the chemical and physical attributes of areas impacted by mining activities in the Eastern Amazon in a practical, rapid, and sustainable manner.

COMMITTEE MEMBERS:
Externo à Instituição - ERLI PINTO DOS SANTOS - UFRRJ
Interno - 3600530 - JOAO FERNANDES DA SILVA JUNIOR
Externo à Instituição - RAFAEL SILVA GUEDES - UNIFESSPA
Externa à Instituição - RENATA ANDRADE - UFLA
Presidente - ***.493.086-** - SILVIO JUNIO RAMOS - ITV