Generalized Nonlinear Mixed-Effect Model of Individual TREE Height to Crown Base for Larix Olgensis Henry in Northeast China

Abstract

ABSTRACT The height to crown base (HCB) of a tree is a necessary variable that includes many growth and yield models as a predictor, and it is important to develop an HCB model due to its contribution to forest management. In this study, we developed an individual tree HCB model for Larix olgensis using a generalized nonlinear mixed-effects model with 2510 Larix olgensis trees on 40 sample plots located on the Dongzhelenghe Forest Farm in northeastern China. According to the evaluated base model, a logistic model that most suited our data was selected as the base model. In addition to height and diameter at breast height, tree and stand level variables that represent tree size, site quality and competition, such as dominant height, basal area of trees larger than subject tree and relative spacing index, were added to the HCB model. The sample plot was set as the random effect, and the parameters describing sample plot-level random effects were included in the HCB model through the mixed-effects model. The variance heteroscedasticity in the residuals was reduced by including a constant plus power variance function in the HCB model. The results showed that the mixed-effects model described a larger part of the HCB variations ( = 0.7642, RMSE = 1.7225) than did the ordinary least square model ( = 0.7063, RMSE = 1.9224).