EVALUATION OF FOREST STAND GROWTH AND YIELD MODELS

Six well known stand growth and yield models are modified to incorporate the thinning effect variable. To determine the "best" model from the chosen six, the models are first classified into one of the following categories: (1) Stand Volume Formulae Models which use stand basal area and stand height as variables to estimate stand growth and yield. (2) Age, Site, Density Models which utilize stand parameters like age, site index and some measure of density as estimators. (3) Biomathematical Models which employ von Bertalanffy's concept of growth which treats growth as the net result of constructive and destructive metabolisms taking place in the stand.; The models are then tested by estimating the volume growth rate and yield for young thinned red pine (Pinus resinosa Ait.) plantations, ranging from 10 to 25 years, using data gathered from three-and-four-year remeasurements of 21 thinned permanent sample plots located in Drury forest in southern Ontario.; Comparison of models is carried out in two stages; stage I involves a comparison within each of the three categories while under stage II the "best" three models from stage I are then compared. Selection of the "best" models in both stages is governed by the results of seven criteria used for comparison. As a safeguard against inferences arising from the use of a rare sample, 24 sample data sets are created through random sampling procedure (without replacement).; Knowing the desirable characteristics of the comparison criterion, results for the 24 sample and test data sets are compared and summarized in comparison tables which basically present a visual display of comparison for various criteria. In the final selection if two models seem to qualify as the "best" model, the choice rests with the user.; On the basis of the data used in this study, analyses and comparison tests reveal that Stoate's model from category 1 is the "best" predictor of growth. However, for yield, Stoate's model and Clutter's model from category 2 produce equally accurate estimates. The closeness of the predictive power of Stoate's and Clutter's yield models coupled with the fact that the categories are collapsible calls for development of a "Hybrid" model which embodies the characteristic features of categories 1 and 2. By including age (raised to an undefined power) as an additional variable in Stoate's yield model, a non-linear "Hybrid" yield function is proposed for estimation of yield. Examination of stand basal area growth rate data reveals existence of a reverse logistic relationship between basal area growth rate and stand parameters such as stand basal area, stand age, stand height and product of stand basal area and stand height. This leads to a new third degree basal area growth rate polynomial which produces a standard error of 0.0645 sq.ft./acre/year and accounts for 97% variation in basal area growth rate (on logarithmic scale). Comparison tests show that for the available data set the "Hybrid" model estimates both growth and yield more accurately than the stand models originally investigated.