Comparação de modelos estocásticos nas estimativas volumétricas e de biomassa em Eucalyptus spp. da Chapada do Araripe - PE.

Abstract

The objective of this work was to compare and evaluate stochastic models in volumetric and biomass estimates in Eucalyptus spp. with 5.5 years of age and in its second rotation, in Chapada do Araripe - PE. The study was carried out from data on volume and the total biomass of the tree compartments (stem, bark, branch and leaf), collected in experimental plantations of Eucalyptus spp. at the Experimental Station of the Pernambuco Agronomic Institute – IPA – in Araripina-PE, where 150 individuals were randomly sampled for cubage and modeling. Adjustments were made to nine volumetric models, using the SYSTAT (12.0) software. The Stepwise (backward method) was applied to the models and the following criteria were established to select the best equation: significance of the regression coefficients (bi), standard error of the estimate (Sxy), Akaike Information Criterion (AIC), Corrected Fit Index (AIc), graphical analysis of residuals and the weighted value of the statistical scores (VP). Based on the results obtained, it was observed significancy at the level of 1% of probability, for all the coefficients of the equations. However, it was observed that, when using the regression analysis by Stepwise (backward method), some volumetric and biomass models underwent important modifications within their format, resulting in the suppression of independent variables and equality among some equations. After applying the analysis techniques, it was observed that all the proposed models showed a good precision between the volumetric and total biomass estimates, as a function of the DBH and height of the tree individuals. For the volumetric models the Naslund, Naslund (Modified), Schumacher and Hall, Meyer, Meyer (Modified), Prodan and Stoate showed the best fit and precision among the evaluated models, while for the total biomass, the Prodan, Schumacher e Hall, Naslund (Modified), Meyer, Stoate, Spurr (Combined Variable), Naslund e Meyer (Modified) models showed the best fit, being, therefore, the most viable models for future forest inventories in Eucalyptus spp. in Chapada do Araripe.