Eficiência das distribuições diamétricas para prognose da produção de Eucalyptus camaldulensis

Abstract

Chapter 1 had as main objective to test the efficiency of six diameter distribution functions by different methods for estimating present number of trees, and volume by diameter class, for different ages and site condictions for Eucalyptus camldulensis. The utilized data came from commercial stands of the “Manesmann Florestal LTDA”, located at João Pinheiro country MG. The sampling was carried out on stands with ages ranging from 14 to 64 months, by using variable size sample plots every one being measured at least twice. Several percentual values of minimum diameter were tested for obtaintion of location parameter of the Weibull and SB of Johnson distributions by the methods of maximum likelyhood and moments. The fitting consystency obtained by each method and for each distribution was verified by the KolmogorovSmirnov test. The analysis of the theoretical frequency by diameter class, in relation to the observed one , and the analysis of the volumetric estimate from the theorical frequency, in relation to the estimated volume from the observed frequency were facilitaded by adoption of a ranking among distributions. It was verified that the best location parameter of the weibull distribution was zero, and that the ξ value of the distribution was equal to 5% of the minimum diameter. The distributions SB and Beta, fitted by the method of moments were the most efficients for estimating the theorical frequency and the respectives volumetric estimates. The distributions log-normal and gamma presented the worse performance for both, theorotical frequency and volumetric estimates. Chapter 2 had as objective to test the efficiency of the Beta and SB de Johnson distributions for the volume yield prediction for different ages and sites, and also to select and to develope functions to represent the stand atributes. Data from 52 permanent sample plots, remeasured at least 2 times, and at most 5 times, were utilized. Formulation of the SB and Beta distributions, submited to 2 fitting methods was presented. The investigation of the prognosis was perfomed through: “t” test, prognosis deviation and the correlation among prognosticated and observed plot volumes. As results, it was verified that the prognosis system was factible to be utilized with efficiency, basing on similar trends among prognosticated and observed plot volumes. In general way, the prognosis errors remained below 20%, mainly if considered the SB of Johnson distribution fitted by any fitting method. It was still verified that prognosis obtained from stand characteristics at very young ages were very efficients, independenthy of age of prognosis.