Random coefficient mixed models were developed to predict vertical and horizontal density variations in stems of Scots pine and Norway spruce. Modeling and validation were based on investigations of 132 spruce and 60 pine trees in southern Norway. Density was measured on discs and small clearwood samples that were taken from 4 different tree heights (stump, 25%, 50%, and 75%). In pine, the common vertical trend in disc density increased with approx. 100 kg m-3 from the stump to the top; the level of this trend, however, varied strongly among trees. The horizontal density trend was less clear than the vertical one and considerably differed between pine trees. Within trees, it was strongly dependent on the tree height position. The most accurate model for density in pine included the following explanatory variables: the ratio of tree height to diameter at breast height, diameter of stem disc with bark, bark volume on disc, and ring width. Validation on independent material gave R2 of 0.45 for density of clearwood samples.
In spruce, the vertical difference of density was low as compared to pine. The average disc density was approx. 15 kg m-3 higher at 75% tree height than at the stump. In vertical direction, density varied considerably between spruce trees. Silviscan measurements revealed that density rapidly decreased by approx. 100 kg m-3 within the first 5 annual rings. Differences between the average densities of clearwood samples were not statistically significant at any tree height level. The most accurate model gave an R2 of 31% in the validation procedure. The model included sample position, site index, diameter at breast height, and bark volume as input variables.
Keywords: density, random coefficient mixed models, wood quality prediction
Authors
Hundhausen U.
Norsk Treteknisk Institutt (Norwegian Institute of Wood Technology). Oslo. Norway
Høibø O. A.
Department of Ecology and Natural Resource Management. Norwegian University of Life Sciences. Ås. Norway
Login to download the PDF