Modelling Crown Profile And Characteristic Attributes Of Coniferous Tree Species In Northeast China

Forest is the main body of forest ecosystem and the foundation to the existence of society and human.Korean pine(Pinus koraiensis),Olgan Larch(Larix olgensis)and Mongolian pine(Pinus sylvestris var.mongolica)are the main coniferious tree species of plantation construction in northeast China.Modelling the outer crown profile and crown attributes variables quantitative is the basis to study the crown biomass,competition level among individual trees and forest productivity and gives further references to study spatial features of crown(crown volume and crown surface area)and forest dynamic growth.The study will lay foundation to make scientific and reasonable management policy.Data of branch analysis from northeast China was collected and used in our study,including 5123 branches from 83 sample trees,12605 branches from 139 sample trees and 6421 branches from 138 sample trees.The study developed and used segmented polynomial equation,segmented power equation,segmented mitscherlich equation,modified Kozak equation and modified Weibull function to model the outer crown profile models.One level nonlinear mixed effect model of the outer crown profile for three tree species and one level linear mixed effect model of inner crown profile model for Olgan Larch were developed.Nonlinear quantile regression models with different qutiles were developed and compared with the marginal regression model developed by different propotion of dataset.The range of quantiles to describle the outer crown profile models and outermost crown profile models were selected and the largest crown radius was predicted by the models.The combination of mixed effect model and linear quantile regression model provided the basis to study on effect of the random effect parameters on the crown profile model.The differences of outer crown profile between the different directions were studied and the common crown volume and crown surface area models for the three tree species were developed.The main contributes and conclusions were as follows:(1)The development and application of the continuous model with biological reasoning enlarged the application of the crown profile models,such as segmented polynomial function,segmented power equation,segmented mitscherlich equation,modified Kozak equation and modified Weibull function.The modified Kozak function and segmented polynomial equation performed best compared to the other models.(2)The random effect parameters were introduced by the development of nonlinear mixed effect model for outer crown profile of the three tree species and the linear mixed effect mode for the inner crown profile model of Olgan Larch.The accuracy of the models were increased.(3)Based on the nonlinear quantile regression model,the crown radius predicted increased with the increase of q from 0.50 to 0.99 that the crown radius increased with q when RDINC was the same.Based on the largest branches selected from each whorl of individual tree for three tree species,the first marginal regression model was developed.For Korean pine plantation,the quantile regression model with quantile 0.70 performed similar to the the first marginal model;for Olgan larch,the quantile regression model with quantile of 0.75～0.80 performed similar to the first marginal regression model;for Mongolian pine,the quantile regression model with the quantile of 0.75 performed similar to the first marginal regression model.(4)For Korean pine,the nonlinear quantile regression model of q=0.90 was suitable to describe the outer crown profile and q=0.95 was suitable to describe the outermost crown profile;for Olgan larch,q=0.95 was suitable to describe the outer crown profile and q=0.99 was suitable to describe the outermost crown profile model;for Mongolian pine,q=0.90 was suitable to describe the outer crown profile and q=0.95 was suitable to describe the outermost crown profile.The selection of quantile to describe the outer or outermost crown profile was closely related to the branches number per whorl for the specific tree species.(5)Nonlinear quantile model of q=0.90 performed well in predicting the maximum crown radius for the three tree species.The R2 based on the observed values and predicted values were about 0.73,0.60,0.60.(6)The combination of mixed effect model and linear quantile model was an example to solve the problem of random effect explaination of variance-covariance structure of the outer crown profile models.(7)The outer crown profiles from the different directions was asymmetercal.Light source was the essential factor in affecting the crown profiles.The class of the trees in the forest and the competition level from other neighbor trees affected the crown profiles of the different direction.The competition showed stronger effect on the crown profile of the suppressed trees compared to the intermediate trees and dominant trees.(8)Using DBH,CL as the two independent variables,the common crown volume and crown surface area models for three tree species were developed by using dummy variable regression approach.