Three methods for avoiding the impacts of incompatible site index and height prediction models demonstrated on jack pine curves for Ontario

Abstract

Fixed base-age site indices are commonly used as a covariate in height prediction models, whereby separate site index prediction equations are used with measured age and height to predict the site index when it is unknown. In such systems, a bias may result in the height prediction if the site index equation is incompatible with the height equation. We demonstrated such bias using as an example recently published models for jack pine in northern Ontario with incompatible site index and height equations. Then we offered solutions that reduce the bias in height predictions assuming that the primary objective was to predict height. First, we re-estimated the site index equation parameters using both the site index and height equations as a common prediction system and holding the published height equation parameters constant while minimizing errors in height predictions. This substantially reduced the incompatibility between the site index and the height equations. Second, we demonstrated the use of two dynamic equations as alternatives to the fixed base-age equations. Even using an irrelevant dynamic equation for another species substantially reduced the bias in short-term jack pine height projections. However, the dynamic equation fit to the jack pine height model was the most effective in reducing the bias for height projections relative to all other considered solutions and produced the least biased, most parsimonious, and most flexible solution.