Research On Stand Growth Models For Platycladus Orientails And Pinus Tabulaeformis In Beijing

Stand growth and yield model is the basic method to study the dynamic change of forest growth,which has been paid much attention by domestic and international forestry workers for a long time.The stand basal area and volume is an important variable to estimate the stand yield,which has high stability and predictability,and can reflect the status and level of forest management.In this study,we regard Platycladus orientalis and Pinus tabulaeformis as the object of research in Beijing.Then,we construct the Platycladus orientalis forest growth models with different stand origin,and compatible forest growth models of Platycladus orientalis with different stand origin were built through introducing the dummy variable in the model.Pinus tabulaeformis stands were divided into five levels by different initial densities:level 1(ID<400 strain/hm2),level II(400 ID<800 strain/hm2),level ?(800 ID<1200 strain/hm2,level IV(1200)ID<1600 strain/hm2),level V(ID=1600 strain/hm2).Then we construct stand basel area and volume basic model using traditional methods.Consider density level effect or plot effect of Pinus tabulaeformis based on basic model,and use the function of R language NLME,Pinus tabulaeformis stand mixed effects model was constructed.The results show:(1)Using Schumacher and Richards as basic model and parameterizing the model,we construct the growth model of Platycladus orientalis plantation and natural forest stand,the constructed models has a good fitting effect.With the independent test of established models,the coefficient of determination(R2)is higher than 0.97,and the fitting precision is above 95%.The precision of plantation was higher than that of natural forest,whether fitting stand basal area or stand volume.In addition,when using stand density index as the density index of the model,we can get a better accuracy than that of using the number of tree per hectare,which could reflect the variation of the Platycladus orientalis stand basal area.However,when fitting stand volume,the model regarding stand basal area as the density index can get better fitting effect.(2)By introducing a dummy variable representing the origin of the forest,we establish the Platycladus orientalis plantation and natural forest growth model based on dummy variables.The fitting effect and the prediction precision of the model are very good,and the fitting effect of the stand volume is slightly better than that of the stand basal area.The constructed model can be applied to both plantations and natural forests,which will reduce the workload and solve the problems of incompatibility of different stand origin.(3)We construct stand basal area model of Pinus tabulaeformis using the Schumacher and Richards as the basic model,the results shows the Richards model,of which independent variables including status index,stand age and number of tree per hectare,get a high precision.Then nonlinear mixed model is constructed based on that Richards model.The results shows that the precision of both single level mixed model and two-level mixed model are better than that of basic model,The two-level mixed model has a better fitting effect than basic model and single level mixed effect.Considering the variance covariance structure,the power function structure variance structures could effectively remove the heteroscedasticity in the data,[ARMA(1,1)]time series correlation structure can express the error correlation between plots.(4)We construct stand volume linear basic model of Pinus tabulaeformis using basal area and the average height of dominant trees as independent variables,the results shows that the R2 of the basic model of Pinus tabulaeformis is 0.9818,and the fitting effect is good.Consider density level effect or plot effect of Pinus tabulaeformis based on basic model,the mixed effect model is constructed respectively.Compared with the basic model,the fitting effect of the mixed model is improved greatly.When considering the heteroscedasticity,the exponential function structure remove the heteroscedasticity best.When the time series autocorrelation structure is considered,the fitting effect of the[ARMA(1,1)]autocorrelation structure is the best.