A state-space growth model for Eucalyptus pilularis in subtropical Australia, fitted with and without seemingly unrelated regression

Abstract

ABSTRACT Inventory data from even-aged, monoculture blackbutt (Eucalyptus pilularis) forests in subtropical eastern Australia were used with the algebraic difference approach to construct a state-space model that predicted stand growth with age for each of quadratic mean diameter, basal area and dominant height, given some initial measurement of the stand. Conventional regression analysis was used to fit each part of the model separately and seemingly unrelated regression (SUR) was used to fit them jointly. Although using SUR resulted in slightly more efficient parameter estimators than when the models were fitted separately, there was no evidence that its use provided any appreciably better fit to the data or precision of estimates of future sizes. Other work in forest science that has used SUR is reviewed briefly and appears to have gained little advantage from its use.