Modelos biomÃtricos florestais com a inclusÃo de covariantes / Biometric forest models with the inclusion of covariants

AUTOR(ES)
DATA DE PUBLICAÇÃO

2008

RESUMO

One of the troubles in biometrical forest modeling is the great number of variables which influence the characteristics of interest in povoamento (stand) and which reflect on their yield. In order to obtain precise estimates of the forestâs lumber stock, the control of those variables becomes a fundamental point to the forest planning of the undertaking. Nevertheless, the heterogeneity of plantings and relationship among the factors which influence its yield, make the modeling process difficult and demand further time. Thus, techniques of decomposing the parameters of the biometrical forest parameters with the inclusion of covariants related to the single trees and with povoamento (stand), can improve the precision and also create a model with a more practical characteristic of utilization. The objective of this study was evaluating the improvement of the quality of the biometrical forest models by including covariants. The data of the present study are coming from commercial plantings of Eucalyptus sp. situated in the southern region of the state of Bahia. Two reduced models were adjusted, one linear and other non-linear, for hypsometry and volumetrics. After the adjustment of the parameters of the selected reduced model, the complete formulation was adjusted with the inclusion of the variables relative to the single tree. There was a 20% reduction in the value of the residual standard error when compared the reduced model and the hypsometric complete model with the inclusion of the covariants; for volume, this reduction was of 7%%. Thus, the inclusion of covariants in the selected models, associated singly to the parameters, contributed significantly to the increased of the precision of models.

ASSUNTO(S)

eucalyptus, growth, complete models. recursos florestais e engenharia florestal eucalipto. crescimento. modelo completo

Documentos Relacionados