ObjectiveTo improve the quality of wood, the mixed effect model of the number of missing-rings was established based on knot analysis data in order to predict the number of missing-rings and provide a theoretical basis for artificial pruning.
MethodIn this paper, the Larix olgensis plantation in Mengjiagang Forest Farm of Heilongjiang Province of northeastern China was used as the research object. Using 1 434 knot data of 50 Larix olgensis trees, based on the Poisson distribution, the glimmix module in SAS9.4 software was used to establish the generalized linear mixed model of the number of missing-rings in the knot. The best hybrid model was selected by calculating the corresponding indicators.
ResultConsidering the tree effects, the optimal mixed effect model is a model that adds the random effect parameters to the intercept, knot height and knot relative height. Considering the rank effect, the optimal mixed effect model is a model that adds random effect parameters to the knot relative height and the diameter of the knot. Comprehensive comparison showed that the fitting effect of the two hybrid models was better than the basic model, and the fitting effect of considering the tree effect was the best. The fitting results of the model showed that the number of missing-rings was closely related to the height of the knot and the diameter of the knot. The knots with lower position and larger diameter were inhibited by competition, but the survival ability was strong, so the number of missing-rings was large. The higher the birth position of the knot is , the better the living condition is, and the less the number of missing-rings is.
ConclusionThrough the establishment of the mixed model of the number of missing-rings in the Larix olgensis plantation, the prediction effect of the model is tested. The test results show that the hybrid model can predict the number of missing-rings and the deviation is small. In the next study, it can be further improved to provide a theoretical basis for artificial pruning.
DownLoad: