ABSTRACT
The chapter discusses statistical aspects related to taper curves (stem curves), which describe how diameter of tree stem depends on the height. The chapter is not intended as a review of taper curve models. Owing to noncircularity of tree stems, standard ways to measure diameters contain measurement errors. If a taper curve model formulated for diameters is used to predict volumes, then predictions are biased. There are four main approaches for describing diameters: i) assuming a parametric functional form for diameters, ii) using a multivariate formulation for discrete set of dimensions, iii) formulating the stem curve in terms of continuous basis functions, and iv) using interpolating splines. Most published models predict stem curves when breast height diameter and total height are measured. Multivariate models describing stem curves in polar coordinates can be used when tree height is not measured. It is argued that standard autocorrelation functions are not valid in stem curve models based on measured breast height diameter and height. Variance of volume prediction error is presented.
