INFLUÊNCIA DE DIFERENTES NÍVEIS DE INTENSIDADE AMOSTRAL NO AJUSTE DE MODELOS HIPSOMÉTRICOS

Main Article Content

Rebecca Araújo GARCIA Thelma Shirlen SOARES Marcelo Marques COSTA Edmilson Santos CRUZ

Abstract

This study aimed to test different levels of sampling intensity of diameter-height pairs per plot in the fit of hypsometric models with data of clonal planting of Eucalyptus grandisand Eucalyptus urophylla. Four traditional hypsometric models were fitted considering five intensities: 5, 10, 15, 20 and, 25 trees per plot. The  selection of the best fitted model was based on the adjusted coefficient of determination (R²aj), residual standard error (Sx), relative residual standard error (Sy.x%), test F, corrected Akaike information criterion (AICc) and, graphical analysis of residuals. To verify the predictive capacity of the selected model in the height estimation, the validation was perform using the Mean Absolute Error (MAE), Mean Absolute Percent Error (MAPE) and Residual Standard Deviation (RSD). The Z test (a = 0.05) was perform to select the best level of sample intensity. The four models tested can be used to estimate the tree heights, with superiority for the Stoffels model which was validated with satisfactory results. All levels of sample intensity showed average heights statistically equal to the average height of the stand. Therefore, in order to reduce inventory costs, fits with 5 trees per plot is the most adequate to estimate tree heights.

Article Details

How to Cite
GARCIA, Rebecca Araújo et al. INFLUÊNCIA DE DIFERENTES NÍVEIS DE INTENSIDADE AMOSTRAL NO AJUSTE DE MODELOS HIPSOMÉTRICOS. REVISTA BRASILEIRA DE BIOMETRIA, [S.l.], v. 38, n. 3, p. 353-368, sep. 2020. ISSN 1983-0823. Available at: <http://www.biometria.ufla.br/index.php/BBJ/article/view/454>. Date accessed: 20 oct. 2020. doi: https://doi.org/10.28951/rbb.v38i3.454.
Section
Articles