Base-age invariant site index models from a generalized algebraic parameter prediction approach

Abstract

A simple idea is proposed to develop polymorphic base-age invariant models with multiple asymptotes: the asymptotic parameter is taken as the site-specific parameter (χ) and one of the other parameters is taken as dependent on χ as a simple power function. This approach is a constrained form of the generalized algebraic difference approach (GADA) and eliminates the requirement in previous GADA models to obtain an explicit solution for χ: χ just has to be estimated along with other parameters of the model. The approach may be used to adapt base-age specific models, developed in terms of the "parameter prediction" method, to be "base-age invariant." The new approach and the models it produces are evaluated on data sets for Chinese fir, red alder, and birch, with models being fitted using the "dummy variable" method in which each observed data series has its own site-specific χ parameter. The base-age invariant models developed are superior to their comparable base-age specific models. Copyright © 2008 by the Society of American Foresters.