Uma nova abordagem para otimizar o design de inventários florestais na eucaliptocultura no Brasil

• Ano de defesa: 2025
• Autor(a) principal: Fonseca, Guilherme Rodrigues
• Outros Autores: https://orcid.org/0000-0002-9549-9750
• Orientador(a): Rocha, Samuel José Silva Soares da
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Lavras; Escola de Ciências Agrárias de Lavras (ESAL)
• Programa de Pós-Graduação: Programa de Pós-Graduação em Engenharia Florestal
• Departamento: Departamento de Ciências Florestais
• País: brasil
• Palavras-chave em Português: Inventário florestal; Eucalipto; Manejo florestal; Modelagem de crescimento; Amostragem estatística; Forest inventory; Eucalyptus; Forest management; Statistical sampling; Growth modelling
• Área do conhecimento CNPq: Ciências Agrárias
• Link de acesso: https://repositorio.ufla.br/handle/1/60253

The Brazilian planted forest sector is one of the key drivers of the country's economy, occupying a significant area and generating millions of direct and indirect jobs. Given its importance, continuous monitoring of these forests is crucial to providing essential data for short, medium, and long-term planning. In Brazil, monitoring is primarily conducted through a sampling scheme known as Continuous Forest Inventory (CFI). CFI relies on permanent plots installed when the forests aretwo years old, which are then remeasured annually until the year before harvest. This method entails considerable costs associated with plot installation, maintenance, and measurement, in addition to the growing challenges related to labor shortages.Given this context, this dissertation proposes two distinct yet complementary approaches to optimizing the sampling effort and the number of stands measured in the CFI of commercial eucalyptus plantations in Brazil. The first paper presents a methodology to optimize the sampling effort in the CFI starting when the forest reaches four years of age. The results demonstrated that the developed methodology enabled an average reduction of 63% in the number of plots across the four studied regions, with reductions ranging from 61% to 65%, depending on the region. This reduction remained stable even under different forest conditions, including site classes, silvicultural systems, and genetic materials. Furthermore, applying an Integer Linear Programming model allowed for the optimal selection of plots within each stand, minimizing the impacts of sampling reduction while ensuring the accuracy of volume estimates.In the second paper, a new CFI design was developed, integrating Double Sampling techniques, simulation algorithms, metaheuristics, and growth and yield models to reduce the number of stands measured throughout the forest cycle. The proposed methodology enabled a reduction of approximately 50% in the number of stands measured from the age of four until harvest while maintaining the accuracy and precision of volume. Additionally, a method was developed to estimate other inventory variables for unmeasured stands, ensuring their compatibility with the estimated volume.These approaches not only contribute to reducing the operational costs of the inventory but also make the process more efficient and manageable, decreasing the demand for field labor—a critical factor given the growing forestry sector in Brazil. By proposing these optimization strategies for the CFI, this dissertation aims to advance and enhance forest management in Brazil, benefiting industry stakeholders and promoting a more rational use of available resources.