REDES NEURAIS PARA ESTIMATIVA VOLUMÉTRICA DE CLONES DE EUCALYPTUS SPP. NO PÓLO GESSEIRO DO ARARIPE

Abstract

The objective of this study was to estimate the volume of Eucalyptus spp clones (genus of rapid growth) in  the Araripe Gypsum Pole, responsible for 97% of the national production of gypsum, employing the methodology of Artificial Neural Networks (ANNs) and comparing it with the volumetric models of schumacher and Hall and Spurr and also verify the efficiency of the estimation using different sample sizes and evaluate the contribution of a categorical variable in the estimation. Data came from an experiment implanted in the Experimental Station of the Agronomic Institute of Pernambuco, where was tested 15 clones of Eucalyptus spp planted in 2002, with final cut in 2009. It was also valued the adjustment of the best models for sample size. The goodness of fit of the models was evaluated based on: the adjusted coefficient of determination (R²_aj), square root of the percent mean error (RMSE%), standard error estimate (Syx%) and an analysis graphic of the residues. The results obtained in the study showed that all modeling was adequate and it was observed that the efficiency of the adjustments depends not only on the sample size, but also on the variance, and that the addition of a categorical variable in the ANNs does not show any perceptible differences, necessary for volume estimation.e sample size.