Prediction of Stem Characteristics for Pinus sylvestris
The main aim of this study was to develop an easy and simple prediction method for stem characteristics of Scots pine (Pinus sylvestris L.) which can be implemented on a modern, computer-based forest harvester. The aim was also to investigate whether it is possible to improve results as more data on the stem being processed become available. The minimum mean square error linear predictor was applied to the prediction of stem characteristics. This predictor utilizes information on the nearest trees and on some stem characteristics of the stem to be predicted. The performance of predictors was tested on real data on 669 Scots pine stems. It was observed that the values of some important stem variables can be predicted with moderate accuracy. The highest accuracy was obtained for the diameter at dead branch height. It was further observed that the best performance of predictors is obtained when the predictor is estimated using only a relatively small number of nearest trees. The updating of predictors did not appear to improve the results in this study.