Evaluation of variable-exponent taper functions for Eucalyptus spp. trees.

Variable-expoent taper functions were fitted to Eucalyptus spp. data, using different approaches for estimating inflection point (p). The precision and the prediction abilities of the taper functions were evaluated for diameter along the stem and total volume (inside bark and outside bark). A new approach was presented for estimating the p value and a modified version of Kozak's model was developed. The geometrical methods EDE (Extremum Distance Estimator), ESE (Extremum Surface Estimator), BEDE (Bisection - EDE) e BESE (Bisection - ESE) can be applied as an alternative to estimate p. The fitted equations performed well, with good precisions (S-yx% de 6.35% a 7.20%) and good ability to predict diameter along the stem and total volume (outside and inside bark). It was found that the different approches tested to determine p had little influence on the precision of the equations. It was observed that the relative height at the inflection point (p) vanes from tree to tree and the results indicated that DBH and H have little or no effect in this variation. The modified variable exponent was similar to Kozak's equations, but it is advantagious because it is a simpler exponent.