Sistema preditivo para multiprodutos de povoamentos de Tectona grandis Linn. f., no Mato Grosso - Brasil

Abstract

Precision forest management of high-value species, such as Tectona grandis, requires robust tools for growth and yield forecasting. This dissertation aimed to develop and evaluate a predictive system for diameter distribution and multiproduct volume in clonal and seed-based plantations of Tectona grandis in the state of Mato Grosso, Brazil. The study tested the hypothesis that parametric projection methods provide accurate future estimates. Diameter distribution was modeled using a two-parameter Weibull probability density function. Two projection methods were compared: Parameter Estimation Models (PEM), based on regression relationships with stand variables, and Parameter Recovery Models (PRM), which rely on distribution moments. The dataset, collected between 2006 and 2024, included stands subjected to different thinning regimes. Results confirmed the high goodness-of-fit of the Weibull function. Comparative analysis demonstrated the superiority of the PEM method, which exhibited lower bias and reduced errors (MAE and MSE), proving more precise and stable in capturing post-intervention structural dynamics. It is concluded that the integrated system, based on PEM and complemented by height-diameter and taper relationships, constitutes a reliable tool for growth forecasting, wood production by assortments, and supporting the management planning of Tectona grandis in the study region.