Sigmoid Functions in Tree Growth Modelling: Limitation of Constant Estimation and Simulation Approach as an Alternative

Abstract

Sigmoid function, in the context of forestry/agroforestry research, has been widely used for rnodelling yield and growth of trees. These functions have a lot of potentialities which reflect their suitability in tree modelling viz their parameters have biological interpretation/physical meaning and they usually lead to unbiased and efficient models. The present study reveals that in spite of these merits, sigmoid functions result in impractical prediction of response variable exclusively for higher extrapolative range of the explanatory variate. This limitation of sigmoid functions has been addressed as 'constant estimation of size. Simulation modeling approach, a combination of curvilinear/allometric and sigmoid functions, has been proposed to overcome constant estimation intricacy. An actual data example was used as an illustration of the proposed modelling approach.