Estimation of Tree Volume Models Using Weighted Covariance Analysis with Dummy Variables

Abstract

In view of augmenting precision of tree volume estimation models for forestry inventory, potential of relatively recent statistical aspects have been explored in this study. Tree volume data regarding five eucalypt species were collected from Victoria. Different functions were fitted separately to diameter overbark and tree height and diameter overbark and merchantable height measurements. Error variance was found to be heterogeneous and weighting of selected models were performed by exponential error variance function developed. Grouping of species for volume estimation was studied by using covariance analysis with dummy variables. Smooth-barked species could be successfully grouped and so could the rough-barked ones. Finally two different models were selected which provided greatest gain in terms of precision.