Modelos lineares de efeito misto na utilização de funções taper para espécies nativas

Abstract

Big challenges appear when people talk about forest management and production planning on native stands where it wishes to take a wood multiproducts approach, we can cite limits at data availability because the difficulty of collecting them, high costs associated with obtainment of these explanatory variables, and the law’s limits. Looking up to get around these problems, the idea of mixed effects models’ utilization, that distribute a part of the stochastic error by the insertion of randomic part on the model that will be utilized, appears. This way, it turns easy the fit and calibration of taper functions to the volume’s estimative of assortment at native forests. This work fits taper equations by the likelyhood and find the best taper equation to the available data by the Akaike Information Criterion (AIC). After the calibration it were done by the cross validation technique and the parameters of the model’s randomic part was determined by the Best Linear Unbiased Predictor (BLUP). The data is arising from a stand of native species that under management. This work associates the randomic effect to the group “specie”. The results were not really good because some species did not show less root mean squared error with the calibration. However other species showed it, being clear the need of more studies about this point and the possibility of success in this kind of analysis.