MÉTODOS DE AJUSTE E PROCEDIMENTOS DE SELEÇÃO DE FUNÇÕES PROBABILÍSTICAS PARA MODELAR A DISTRIBUIÇÃO DIAMÉTRICA EM FLORESTA NATIVA DE ARAUCÁRIA

Abstract

The aim of this work was to evaluate the performance of the probability density functions with different fitting procedures and statistical evaluation to describe the diameter distribution of a Mixed Ombrophyllous Forest fragment. The study area is part of the Irati National Forest (FLONA), in Parana State. Data used were derived from 25 permanent 1-ha plots (100 m x 100m) established and measured in 2002 and re-measured in 2005 and 2008. The fittings were done considering all the species sampled on the re-measurement in 2008. Beta, Weibull 2 and 3 Parameters and Exponential Meyer I and II functions were tested, employing Moments and Maximum Likelihood methods for Beta and Percentiles and Maximum Likelihood for Weibull 2 and 3 Parameters. The Nonlinear Programming was used as an attempt to improve the fittings, except for Meyer model. To evaluate the fittings, the goodness-of-fit tests Kolmogorov-Smirnov (K-S) and Hollander-Proschan (H-P) were used, as well as, the Standard Error (%), Reynolds Index (IR) and Residual Dispersion. The results showed that Weibull 3P function was the best to describe the diameter distribution for the forest as a whole, however the Beta function showed satisfactory results, and it can also be used to evaluate the diameter distribution for the studied area. About the statistics used, it was observed that Reynolds Index presented good results to evaluate the performance to probability density functions and, for the interval class used, the Kolmogorov-Smirnov test presented a higher number of good fitness values when compared with Hollander-Proschan test, nevertheless the K-S is sensible when the frequency is high, causing a non-adherence.