Estimating tree component biomass using variable probability sampling methods

As a signatory to the Kyoto Protocol, Australia is obliged to stabilize greenhouse gas emissions at 8% above 1990 levels by 2008–2012. To demonstrate achievement of this goal, Australia requires national annual estimates of changes in vegetation biomass as greenhouse gas emissions from land use change. These emission estimates are, however, uncertain due largely to the scarcity of existing allometric equations for calculating biomass. The large investment of time and funding required for harvesting, particularly using traditional techniques such as double regression and ratio sampling, also precludes the generation of new equations. Alternative techniques for rapid, cost-effective, and reliable estimation of biomass therefore require investigation. This study, conducted in central Queensland, compared estimates of component biomass that were generated for seven trees of the woodland species Eucalyptus populnea (poplar box) using ratio sampling and variable probability sampling techniques, namely randomized branch sampling (RBS) and RBS with importance sampling (IS). Application of randomized branch sampling consistently underestimated the biomass of leaf and small branches (<1 cm in diameter) and produced weak prediction equations. In contrast, results suggest that RBS with IS is particularly useful in predicting woody (trunk and branches >1 cm in diameter) biomass, and prediction equations agreed with existing equations for this species. However, this method tended to overestimate individual tree woody biomass. The study concluded that RBS with IS was a viable alternative to current methods.