Abstract

The aim of this study is to develop a method for estimating the accuracy of site index curves which is independent of the method used for the construction of the curves. Site index curves for Scots pine in Sweden are used to illustrate the method. The investigation deals mainly with prediction errors, i.e. errors caused by actual height development not following the site index curves. Permanent plot data is used to construct a stochastic, autoregressive simulation model, by means of which it is possible to study the variation of site index over age within stands over long periods. A number of simulations are performed and used for estimating prediction errors. Earlier investigations provide figures on sampling and measurement errors which, together with estimatedprediction errors, are used to obtain recommendations on the number of plots required within a stand to reach a predefined level of accuracy at site index estimation.