Abstract Site index equations for old-field plantations and natural stands of loblolly pine were produced from the integrated form of height increment models fitted by segmented regression techniques. After transforming the stem analysis data as the natural logarithm of height and the inverse of age, height increment submodels were fitted for various combinations of the independent variables height, age, and age squared for two segments with a join point being a function of age. Comparisons of the resultant curves indicated that the model having all three variables in both segments was superior to other forms and to published equations in predicting site index of trees both from the data used in parameter estimation and from an independent source. Forest Sci. 28:544-555.
Devan et al. (1982) studied this question.