An assessment of the structural method of deriving a black spruce site equation
To date, methods of deriving site index (S) equations assume that stochastic error is only present in the regressor. This paper develops a method, termed the "structural method," which recognizes that both dominant stand height (H) and S measurements contain stochastic error. To achieve this, the structural method utilizes the structural relationship that exists between H and S to derive an S equation. S equations are derived for black spruce, Piceamariana (Mill.) B.S.P., using the structural method and various other methods, with linear and nonlinear models that are currently in use. Data used in the study consist of 56 black spruce permanent sample plots, containing a total of 382 observations, from north central Ontario and the Clay Belt Region of northern Ontario. This data set is split into 36 plots (260 observations) for deriving S equations and 20 plots (122 observations) for testing the equations for accuracy in predicting H, S, and future H. The equations are also examined for bias over stand age. Results show that height development of black spruce is not asymptotic and is best described by a linear model. Overall, the structural method provides the most accurate S equation within the range of the data. It predicted 90% of the H test observations with an error of 0.4 m or less, 89% of the S test observations with an error of 0.4 m or less, and 90% of the future H test observations with an error of 0.7 m or less. The structural method also has the advantage of producing only one equation for predicting both H and S. This enables estimates of both H and S to be made from one graph of H over age by S classes.