In building forest simulation models it is common to use data containing repeated measurements from individual sample plots. It is inappropriate to use ordinary least squares (OLS) regression to estimate the parameters of a model with such data, since this method will tend to underestimate the variances of the parameter estimates. Ferguson and Leech (1978) developed theory to use generalized least squares (GLS) regression to attempt to solve this problem. They used GLS to construct a stand-volume yield function for unthinned Pinus radiata D. Don in southeast South Australia. This paper corrects an error in their theory of GLS and shows the corresponding changes in their numerical results. There are theoretical and practical difficulties in applying the theory, and great care is needed with its use. Forest Sci. 27:233-239.
Research Scientist, Division of Forest Research, CSIRO, Stowell Ave., Hobart, Tas. 7000, Australia
Publication date: June 1, 1981
More about this publication?
Forest Science is a peer-reviewed journal publishing fundamental and applied research that explores all aspects of natural and social sciences as they apply to the function and management of the forested ecosystems of the world. Topics include silviculture, forest management, biometrics, economics, entomology & pathology, fire & fuels management, forest ecology, genetics & tree improvement, geospatial technologies, harvesting & utilization, landscape ecology, operations research, forest policy, physiology, recreation, social sciences, soils & hydrology, and wildlife management.