Acessibilidade / Reportar erro

Regression analysis of the profile of the stef of Corymbia citriodora formed in savanas area

Abstract

This work aimed to carry out a regression analysis of the tapering of the stem of Corymbia citriodora. For this, 24 felled sample trees were cubed, providing the necessary data for the adjustment 46 fixed-effect (FE) taper models, which were evaluated by means of a detailed regression analysis based on statistical criteria: graphical analysis of residuals, residual standard error, adjusted coefficient of determination, randomized block design in a split-plot scheme with subsequent application of the Dunnett test, mean absolute deviation, square root of mean error, sum of squares of relative residuals, standard deviation of differences, Akaike's information criterion and sum of squares of residuals. The EF model that most stood out among the 46 evaluated was adjusted by including the random effects of the diametric class and the H/D ratio (CHD), resulting in a mixed-effect equation (ME). Thus, the equations obtained FE and ME were evaluated by also adopting the statistical criteria referring to the significance of the estimation of the regression coefficients, regression assumptions regarding normality, homoscedasticity and independence of residues and maximum likelihood ratio test. It was concluded that a model with the best performance was the nonlinear exponent-form type with the inclusion of the CHD random effect, with expressive superiority in both the precision and the accuracy of the EM equation compared to the FE equation.

Keywords:
Wood multiproducts; Forest Biometrics; Mixed models; Eucalyptus

Universidade Federal de Santa Maria Av. Roraima, 1.000, 97105-900 Santa Maria RS Brasil, Tel. : (55 55)3220-8444 r.37, Fax: (55 55)3220-8444 r.22 - Santa Maria - RS - Brazil
E-mail: cienciaflorestal@ufsm.br