Estimating the volume of Australian hardwoods using centroid sampling

Centroid sampling was tested on 18 sample trees of Flindersia brayleyana and 124 sample trees of five species of Eucalyptus all of which were measured in detail. Volumes of the main bole of each tree and of logs (length approx. 6 m) derived from each bole were estimated using both Huber's formula and centroid sampling. These estimates were compared with ‘true’ volume obtained by summing the volumes derived using Smalian's formula of all the measured short sections comprising the main bole and each log respectively.

The mean error of the centroid estimate of volume of the main bole was not significant for Eucalyptus sieberi and E. diversicolor and was less than that derived from Huber's formula for all species except E. regnans. When all species were combined, the centroid estimate was clearly more accurate, its mean error (-2%) being just significant (p=0.046). Centroid sampling also gave more accurate estimates of the volume of the component logs.

Results of this study and earlier work suggest that centroid sampling is applicable to tree species of both excurrent and deliquescent growth habit.