The production prognosis is performed through growth and production modeling,
where volume estimates are obtained as a function of variables such as DBH and height, for
example. The process of collecting height data in the field demands greater time and cost,
and hypsometric equations are often used and in recent years artificial neural networks to
obtain height estimates. Thus, the objective of this work was to evaluate the application of
RNA in the height estimation of a clonal Eucalyptus urophylla S. T. Blake planting in
Almeirim, Pará, comparing it with the estimates obtained by the hypsometric equation. For
the training of the networks and adjustment of the equations, data from 540 eucalyptus
individuals, aged 24 to 72 months, were used annually between 2013 and 2017. The
Gompertz model was used, and the multilayer percepton network, and the estimates
obtained were evaluated by means of the correlation coefficient between observed and
estimated heights, square error of the mean error (RQEM), Bias, and graphical analysis of
the waste distribution. The two procedures evaluated were efficient for estimating the
height of Eucalyptus urophylla S. T. Blake trees, however, the RNA was higher in the
statistical indicators RQEM and Bias in relation to the regression.
|
