Crown Profile Simulation For Natural Dahurian Larch Based On Two Methods

The sample data was based on 2306 branch samples of 36 trees from natural dahurian larch(Larix gmelinii) located in Songling Forest Bureau in Daxinganling. Two methods were used to simulate growth changes and predict rationally in a crown. Method one was to develop branch diameter, branch length, and branch angle equation systems using the error-in-variable simultaneous equation approach and compare four parameter estimation approaches. The results showed that Seemingly Unrelated Regression (SUR) approach accounted for the correlation of error terms across the equations and decreased standard error of parameter estimation. Equation systems of branch characteristics established in this study show better precision and can better describe variation of branch characteristics and crown profile. In addition, this study also developed branch diameter, branch length, and branch angle equation systems with stand density variables using nonlinear mixed-effects modeling techniques. Stand density has a significant impact on branch diameter, branch length, branch angle and crown profile based on significant parameter estimates and better precision. Method two was that four growth equations were established including Logistic, Weibull,Richard and Mitcherlich functions, and Overall fit and validation statistics were compared with four growth equations. In the four growth equations, the results show that fitting and validation statistics of multiple-factor crown radius function established by Mitcherlich performed well for simulating crown profile and were more suitable for predicting changes of crown radius reasonably. In addition, stand density variables were included in the equation using correlation between random parameters and stand density variables and established a crown curve model. The model reflected the fluence of crown radius not only on growth changes of branches in a crown but also on stand changes for crown growth. By fitting and validation results in the equation, Mitcherlich function with stand variables was selected as the best model and described crown shape curves. Two methods both had advantages and disadvantages,so should consider the actual situation if practical application.