Comparison of estimation methods for fitting weibull distribution to the natural stand of Oluwa Forest Reserve, Ondo State, Nigeria

Abstract

The relative performance of any distribution function truly depends on the estimation methods and where this is wrongly chosen poor fit is inevitable. This may mislead forest managers and thus thwart effort towards  sustainable forest management. This study therefore compared estimation methods for fitting 3-parameter  Weibull distribution to the natural stand of Oluwa Forest Reserve, Ondo State, Nigeria with a view to  enhancing sustainable management of the tree resources. Systematic sampling technique was used in the  laying of eight (8) temporary sample plots (TSPs) of size 50m x 50m in the natural forest. Three fitting  methods were used that based on maximum likelihood, moments and percentile. Comparison was based on  Kolmogorov-Smirnov statistic (K-S), bias, mean absolute error (MAE) and mean square error (MSE). The  result revealed that maximum likelihood method was more accurate in fitting the Weibull distribution to the  natural stand. It had the smallest mean bias and MSE values of 0.00009 and 0.00021, respectively. Maximum  likelihood method is therefore recommended for fitting the 3-parameter Weibull distribution to natural stand of  the reserve.

Keywords: Weibull distribution, maximum likelihood, moments, percentile, natural stand