Iterative Method of Discriminant Analysis to Classify Beech (<i>Fagus sylvatica</i> L.) Forest

Abstract

We present a new method for the classification of beech (Fagus sylvatica L.) forest plots based on discriminant and frequency analysis. This method can be used as a tool to allow experts to stratify beech forests in a simple and precise way. The method is based on discriminant analysis with cross-validation of 13 variables measured in 142 plots from the 2005 Second National Forest Inventory and 63 plots from an inventory installed in specific locations together with a frequency analysis of the qualifying variables. In the first stage, the method uses the results of a frequency analysis fitted with an iterative discriminant analysis that allows improving the subsequent classifications taking into account the results of the analysis and the correctly- and wrong-classified plots. This method is applied to beech forest in Burgos (Spain) where six structural groups were described. The discriminant functions show that forest structure depends basically on diameter distribution and almost 94% of the plots are correctly classified using this methodology. The high level of correctly assigned plots indicates an accurate classification of structure that can be used to stratify beech forests with only the diameter at breast height measurement.

Keywords

• discriminant analysis
• beech forest
• diameter class
• forest structure
• forest management
• multivariate statistics