Proposta de um modelo não linear de crescimento e produtividade florestal

Abstract

The growth and productivity estimatives are indispensable tools for the decision making at the florestal studies. In this work is highlighted the importance of the usage of non-linear models to estimate those variables, with the proposal of utilization of a variation of the exponential model, that is compared with the three parameters logistic model. The curvature values are used to validate the adjustes of the models, through the evaluation of the "linear proximity". With the aim of acquire a higher acuracy, those models were adjusted and compared without adding co-variables at the respective assinthotic parameter. It is also used the methodology of Monte Carlo's simulation, to generate 10000 samples, what makes possible to produce a study of the parameters distribution. All those models comparations were done considering three classes of site was observed that the proposed model presents a higher quality of linear proximity at all classes of site evaluated.