Height-diameter curves with random intercepts and slopes for trees growing on drained peatlands

Abstract Models for predicting tree height were constructed for Scots pine (Pinus sylvestris), Norway spruce (Picea abies) and pubescent birch (Betula pubescens). The data consisted of two separate sets of permanent sample plots forming a representative sample of drained peatland stands in the whole country. A logarithmic height-diameter curve with one nonlinear parameter specific to each tree species was applied. It was assumed that the intercept and slope of the curve would vary randomly from stand to stand. Stand characteristics were used to predict the mean intercept and slope. A nonhomogeneous variance of the residual error was modelled as a function of tree diameter. A mixed linear model technique was applied to fit the models. The diameter of the tree of the median basal area, stand basal area, geographical location of the stand, and site quality were used as fixed independent variables in explaining the variation in the intercept. The diameter of the tree of the median basal area and the stand basal area were used in explaining the variation in the slope.