Height growth rate of Douglas-fir: a comparison of model forms

Five model forms were evaluated for their ability to predict height growth rate of individual Douglas-firs (Pseudotsuga menziesii [Mirb.] Franco) growing in even or uneven-aged stands of southwest Oregon. Three models had been previously used for Douglas-fir; the fourth was a simple modification of one of these, and the fifth was developed in this study by means of multidimensional graphing and modeling. Two forms were age-dependent, in that they used transformations of stand age, and three were age-independent. The model developed by multidimensional techniques provided the lowest mean squared error for both an even-aged data set (1,763 observations) and a combined even- and uneven-aged data set (2,242 observations). The five models were verified on a randomly drawn subset of the data consisting of 241 observations. Again the model developed with multidimensional techniques had the lowest mean residual and the lowest mean squared error. Independent variables in this model are stand crown closure at the top of the tree, crown ratio, and potential height growth rate as predicted by an existing dominant height equation. The resulting model is age-independent and can be applied to trees in both even- and uneven-aged stands. Multidimensional techniques were most useful in modeling this complex nonlinear surface. For. Sci. 34(1):165-175.