Nonlinear fixed and random generalized height–diameter models for Portuguese cork oak stands

摘要: The objective of the research was to develop a generalized height–diameter model for Quercus suber L. in Portugal, which can be applied both undebarked and debarked trees, with diameter at breast height over cork larger than 2.5 cm. A nonlinear fixed effects (NLFEM) mixed (NLMEM) approaches were used. Parameters estimates obtained using SAS macro NLINMIX, uses linear approximation marginal likelihood function by expanding it first-order Taylor series on random effects. option their current empirical best unbiased predictors (EBLUP) fitted models evaluated an independent data set, together existing specific trees. To obtain subject predictions NLMEM, conventional improved calibration procedures applied, considering four different tree sub-sampling designs. Both proposed included dominant stand density as covariates explain plot variability. Validation indicated that, even situations where NLMEM is not possible, this should preferred. differences between validated models, more evident young stands, considered. No large predictive accuracy found calibrated or procedures, all considered