The Study Of The Distribution And Prediction Models Of Knot Volume Inside The Stem For Korean Pine Plantation

Based on 1207 knots of 49 sample trees of 29 standard plots of Korean pine plantations in Linkou and Dongjingcheng Forest Bureau of Heilongjiang Province,this study extracted longitudinal sections of knots using the image processing software Digimizer and represented the shape of knots using two-dimensional scatter plots.According to the the two-dimensional scatter plots,knots of Korean pine plantation were divided into the following three types:1.Alive knots(The whole knot contains only sound knot portion);2.Nonoccluded dead knots(The whole knot contains both sound and loose knot portions);3.Occluded dead knots(the sound and loose portion of the knot are partially occluded by the bark).For the three types of knots,the volume of sound knot was calculated by mathematical integral of the sound knot shape equation;the volume of loose knot was calculated using the volume equation of a cylinder.The total volume of knots are the sum of sound and loose knot volume.Through the obtained knot volume data,the influence of knot factors(knot diameter,knot length,knot angle),tree factors(relative height of knot,different wood grade),plot factors(stand age,stand density)on the sound knot volume,loose knot volume and total knot volume was analyzed.The results showed that the volume of sound knot,the volume of loose knot and the total volume of knot have a positive correlation with knot diameter and knot length.The volume of sound knot and the total volume of knot increased with the increase of the knot angle,and the volume of loose knot first increased and then decreased with the increase of knot angle.The volume of the knots increases first and then decreases with the increase of the relative height of the knots.At the same height of the knot,with the increase of the sample wood grade,the knot volume increased greatly.In plots with the same stand age and different stand density,stand density have a negative correlation with the volume of sound knot;the volume of loose knot and the total volume of knot are less affected by different stand density.In plots with the same forest density and different forest ages,the knots of the older stands are larger.These showed that knot diameter,knot length,knot angle,relative height of knot,tree grade,stand density and stand age all have significant effects on knot volume.Finally,Based on knot variables(diameter,relative height and total length of knots)and tree variables(diameter at breast height),the prediction models for sound knot volume,loose knot volume,and total volume of knot were established by using the linear mixed model with plot level and tree level.The prediction model of knot volume used AIC,BIC,log-likelihood,R2 and LRT test to evaluate the goodness of fit of different mixed models.The results showed that adding random effects based on tree level and plot level and random parameter variance-covariance structure(UN)can improve the fitting accuracy of the mixed model.Comparing with fixed-effects model,the mixed effect models of the volume of sound knot,loose knot,and total knot have smaller AIC and BIC values,larger log likelihood values,and absolute errors(bias),root mean square error(RMSE)is smaller,and the coefficient of determination R2 has been improved accordingly,indicating that the mixed effect models provided more accurate parameter estimates,more uniform residual distribution,and higher model fitting precision.From validation results,it showed that mixed model has improved relative to the basic model in terms of the average absolute residual and the absolute value of the average relative deviation;in terms of prediction accuracy,the mixed model has improved by 2～4%compared with the basic model.And both reached more than 93%,indicating that these models can be used to describe the change of the volume of sound knot,the volume of loose knot and the total volume of knot,and to predict the size of future knots.