Banca de DEFESA: LUANA SOUSA COSTA

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
DISCENTE: LUANA SOUSA COSTA
DATA: 22/02/2024
HORA: 08:00
LOCAL: DCS02
TÍTULO:

Soil-environment digital information to provide solutions for solid waste disposal and coffee yield modeling

PALAVRAS-CHAVES:

Digital soil mapping; environmental covariate; machine learning; coffee system; land suitability.

PÁGINAS: 100
GRANDE ÁREA: Ciências Agrárias
ÁREA: Agronomia
SUBÁREA: Ciência do Solo
ESPECIALIDADE: Gênese, Morfologia e Classificação dos Solos
RESUMO:

Multiple digital information sources, field observations, or laboratory routines to characterize the soil and environment features coupled with powerful predictive models are vital for solving problems, offering new insights, discovering complex patterns, and optimizing land management and use. This dissertation presents two chapters involving soil and its occurrence environment as a fundamental basis. The first chapter proposed a suitability system for solid construction waste disposal with further application in a digital soil map of Nepomuceno municipality as a study case. Despite extensive pedological knowledge, current environmental legislation and regulations have simplified aspects that may pose risks for solid construction waste disposal activities and human health. A digital soil mapping framework was developed to map soils and to provide environment characterization. In addition, a suitability system was established based on interpretations of the soil-landscape relationship through attributes listed in a guide table, discussing the potentialities and limitations associated with its position in the landscape. The second chapter aimed to develop farm-scale coffee yield predictive models from machine learning algorithms and digital terrain models, soil fertility information, magnetic susceptibility, airborne gamma-ray espectroccopy, monthly precipitation, and satellite vegetation indices. The random forest, gradient boosting machine (GBM), and radial support vector machine (SVM) algorithms retrieved a proper accuracy of predictions for soil classes (chapter one, only random forest applied) and coffee yield (chapter two). The attributes in the first chapter proposed as criteria for the suitability system complement the current state legislation. Topography and soil depth were the most limiting factors of the areas in the case study. A total of 236 ha closer to the urban perimeter connected by roads in good condition were classified as suitable for managing medium- and small-scale daily volume, whose destination might reduce transportation and installation costs in the study area. Although different predictive models of plant yield were developed considering the coffee biennial state (positive and negative biennially), the model containing the four seasons outperformed the others, probably due to the larger number of input information to adjust models along their excellent capability to predict yield containing different ranges of values.

MEMBROS DA BANCA:
Interno - ADELIA AZIZ ALEXANDRE POZZA (Suplente)
Externo ao Programa - DALYSE TOLEDO CASTANHEIRA - DAG/ESAL (Suplente)
Externo à Instituição - ELVIO GIASSON - UFRGS (Membro)
Presidente - MICHELE DUARTE DE MENEZES (Membro)
Interno - RENATA ANDRADE (Membro)
Interno - SERGIO HENRIQUE GODINHO SILVA (Membro)
Externo ao Programa - TIAGO TERUEL REZENDE - DAG/ESAL (Membro)