A Stochastic Differential Equation Model for A Stochastic Differential Equation Model for the Height Growth of Forest Stands

Oscar García

Forest Research Institute, Rotorua, New Zealand

Summary

A model and an estimating procedure for predicting the height growth of even-aged forest stands was developed as part of a methodology for modelling stand growth in forest plantations [García, 1979, in Mensuration and Management Planning of Exotic Forest Plantations, FRI Symposium No. 20, D. A. Elliot (ed.), 315-353. New Zealand Forest Research Institute]. The data consists of heights measured at several ages in a number of sample plots. The ages and the number of measurements may differ among plots, and the measurements may not be evenly spaced in time. The height-growth model is assumed to have some parameters that are common to all plots and others that are specific to individual plots. In addition to random environmental variation affecting the growth, there are random measurement errors.

The height growth is modelled by a stochastic differential equation in which the determinisic part is equivalent to the Bertalanffy-Richards model (von Bertalanffy, 1949, Nature 163, 156-158; 1957, Quaterly Review of Biology 32, 217-231; Richards, 1959, Journal of Experimental Botany 10, 290-300). The model also includes a component representing the measurement errors. Explicit expressions for the likelihood function are obtained. All the parameters are estimated simultaneously by a maximum likelihood procedure. A modified Newton method that exploits the special structure of the problem is used.

Key words: Richards model; Von Bertalanffy model; Maximum likelihood estimation; Time series; Statistical computing; Optimization.


File translated from TEX by TTH, version 2.00.
On 12 Mar 1999, 22:25.