Abstract
Long-term growth data usually show autocorrelation among the residuals of successive measurements. Implementation of Generalized Least Squares (GLS) while estimating parameters at the individual tree level brought down the autocorrelation in majority of the cases and revealed the true variances of the estimates which were generally 50% higher than that obtained through Ordinary Least Squares (OLS). However, at the aggregate level (site class means), the differences in the variance estimates obtained through GLS and OLS, turned out nonsignificant as the variance of estimates at the aggregrate level were more affected by variation in parameter values between trees. The comparison of site quality levels with respect to the status of parameters remained unaffected by the use of OLS. These results could be of value to foresters from practical point of view.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
