Modelagem da necromassa e do crescimento de árvores individuais em uma área manejada de floresta tropical úmida na Amazônia

Abstract

A good planning for the use of forest resources is directly linked to knowledge of the influence of management on local mechanisms, with this, it is possible to make better decisions regarding continuous production continuous production. This work aimed to associate the evaluation of two components still little explored in the Amazonian context: necromass stock and individual growth of trees of commercial interest. Three papers were developed, where ahe modeling of forest variables was a central theme, The first paper associated recent advances in machine learning techniques to model the volume of fallen trunks in order to develop a tool to estimate the necromass stock in forest inventories. The best fits were observed for modeling with nonlinear regression. The second paper aimed to evaluate the necromass stock in a managed forest 20 years after logging, including fallen trunks and branches, and standing dead trees with a diameter from 10 cm. It was observed that the forest under analysis has a tendency to necromass decomposition, with its stock being concentrated with high decomposition. The third paper aimed to model the individual growth rate of commercial species and evaluate the influence of categorical variables on growth. The modeling was carried out applying concepts of Generalized Additive Models for Location, Scale and Shape. The canopy shape and lighting characteristics were the ones that showed the best response for understanding the growth rate. Keywords: machine learning; dead wood; categorical variables.