Implications of random variation in the stand prognosis model

Abstract

Although the Stand Prognosis Model has several stochastic components, features have been included in the model in an attempt to minimize run-to-run variation attributable to these stochastic components. This has led many users to assume that comparisons of management alternatives could be made based on a single run of the model for each alternative. Recent analyses have demonstrated that this assumption may often be incorrect. Several possible solutions are given, and the author recommends that in almost all applications of the Stand Prognosis Model it would be wise to make at least two or three projections for each alternative. The number of replications required for any specific application must be determined by evaluating the tradeoff between the added costs of additional replications and the need for additional precision based on intended uses of model output.