Predicting knot properties of Picea abies and Pinus sylvestris from generic tree descriptors

Tree models were developed whereby stem shape and internal knot structure of Scots pine and Norway spruce stems could be predicted using site, stand and tree variables. The intention was to develop models that could be applied, using these tree descriptors, to simulate individual knots for sawmill conversion simulation studies. The following independent descriptor variables were used in the models: tree age, crown length, crown ratio, diameter at breast height (DBH), relative DBH, height to the lowest living branch, total height, height above ground, height increment, ring width from the pith to the first 20 years, site index (dominant height at 100 years) and temperature sum (as calculated from altitude and latitude). Knot properties included were maximum and mean knot diameter, individual knot diameter, number of knots per whorl, soundknot length, loose-knot length and vertical knot inclination. The modelling techniques applied in regression analysis included mixed, general linear, non-linear and segmented models in accordance with the data structure, and in an effort to mimic underlying biological processes (e.g. stand dynamics).