dc.description.abstract |
Taper is the rate of change of diameter over a specified length along the tree stem, which varies
not only by species but also by age, diameter at breast height, and tree height. Taper is important
for the forester to predict the upper stem diameters especially in volume calculations. Although
it is difficult to find taper studies in Sri Lanka, many studies have been carried out in Canada
and New Zealand.
The prime objective of the current study is to develop a taper equation for Pinus caribaea to
predict upper stem diameters for a selected plantation. 25-year-old P. caribaea (Morelet)
plantation in the Yagirala Forest Reserve was selected for this purpose. The selection of sample
plots were carried out randomly after dividing the population into three primary strata i.e.,
valley, slope and ridge. Data were collected from nine 0.05 ha circular plots and 1053 tree
sectional measurements have been collected from these plots. Diameter at breast height and total
height of individual trees were used as the explanatory variables. These were used to estimate
the parameters for the selected equation originally constructed for Douglas fir in Coastal Central
Colombia by Kozak et al (1969). In this study, for different sites, three separate models were
constructed with different parameter sets with the similar model. Due to the difficulty using
separate models, a possibility of using one model for all sites were tested with pooled data using
multiple linear regression. Using the common model with new parameters, normal residuals
were calculated separately for each site type test the bias using one-way ANOV A. This test
indicated the non-significance of the residuals and therefore, it was decided to use the common
model for the prediction of tree taper for the selected area.
For testing the sensitivity of the estimated parameters of the common model data were fitted to
the original model constructed for Douglas fir without changing its parameters and the residual
distribution was tested. The residual distribution indicated that the low sensitivity of the model
proving the ability of using in many site types. |
|