Construction Of The Height To Crown Base Mixed Model For Korean Pine

Tree height to crown base(HCB)is an important variable commonly included as one of the predictors in growth and yield models that are the decision-support tools in forest management.Constructing a model of height to crown base to provide a theoretical basis for forest management.Based on the data of 2 972 Korean pine plantations in 63 plots in Maor mountains,the model of height to crown base of Korean pine plantations was constructed by nonlinear mixed model and linear mixed model with stand variables.Firstly,R2,RMSE and MAE were used to select the optimal nonlinear basic height to crown base models.The above evaluation indexes were used to determine the combination of stand variables which had a greater effect on the height to crown base,and then the nonlinear generalized model of the height to crown base was constructed.There may be a high degree of autocorrelation due to the same different observations,the plot random effect was added,and the evaluation indexes such as AIC,BIC and LOGLIK were used to determine the combinations of random effect variables needed to be added to construct the optimal nonlinear base mixed model and the optimal nonlinear generalized mixed model.In order to compare the linear model and the nonlinear model which is more suitable for describing the height to crown base model,based on the height to crown base nonlinear model,the height to crown base linear base model,the height to crown base linear generalized model,the height to crown base linear base mixed model and the height to crown base linear generalized mixed model are constructed according to the corresponding variables.The fitted effect and prediction ability of 8 models were compared based on the construction of height to crown base,and then the optimal model of height to crown base was selected.The study results show that:(1)Introducing the dominant tree height(HD),crown width(CW)and basal area sum larger than subject tree(BAL)into the model can significantly improve the model’s fitting accuracy and predictability.(2)Regardless of linear or nonlinear height to crown base model,adding stand variables and random effects can improve the fitting accuracy and predictive ability of the model;(3)Considering the predictive ability and sampling cost,it was recommended to select 4 strains for the correction of the mixed model.For the nonlinear generalized height to crown base mixed model,random sampling is the best.(4)The fitted effect and prediction ability of the nonlinear model are better than the linear model except for the basic model,which indicates that the nonlinear model is the most suitable method for the construction of the height to crown base.The nonlinear generalized mixed model was superior to the other seven models in terms of fitting accuracy and prediction ability,so it was recommended as the model of height to crown base of Korean pine plantation.