Predicting basal area of Scots pine branches

Abstract

The aim of this study was to assess and to develop models for the variation of basal area of branches in Scots pine at different heights within the live crown. The models had to be applicable as submodels in growth simulators in order to predict the effects of silvicultural treatments on wood quality. Models were developed for predicting the basal area of the largest and smallest live branch in a whorl (method I), relative basal area distribution on the basis of the thickest branch (method II), and basal area of branches and its variance (method III). Random variation of the dependent variables was divided into variance components at the stand, plot, tree, whorl and branch level. Furthermore, the mutual correlation of the dependent variables in the different models was taken into account by using multivariate models. Method I predicted the basal areas of the largest and smallest branch unbiased when the total basal area of branches and number of live branches in each whorl were known. Because methods II and III do not require that the total basal area of branches be predicted with other models, they can be applied more generally. However, methods based on total or mean basal area of branches (methods I and III) were more accurate than models relying on basal area of the largest branches (method II). Even though there was bias in predicting some branch properties, the behaviour of the models was logical and they provide a framework for predicting basal areas of branches within the live crown on the basis of routine stand and tree measurements.