Improving harvester estimates of bark thickness for radiata pine (Pinus radiata D.Don)
Strandgard, M and Walsh, DJ, Improving harvester estimates of bark thickness for radiata pine (Pinus radiata D.Don), Southern Forests: a journal of forest science, 73, (2) pp. 101-108. ISSN 2070-2620 (2011) [Refereed Article]
Harvesters estimate bark thickness primarily from diameter over bark (DOB) and use it to estimate log volume and value. Harvester bark thickness estimation errors can reduce returns through poor bucking optimisation and out-of-specification logs. Radiata pine bark thickness data from harvested logs and permanent sample plots (PSP) were analysed to determine best-fit coefficients for current and potential future harvester bark thickness models. The most suitable current harvester bark thickness model for radiata pine is: double bark thickness = b0 + b1*DOB (the ‘DOB model’). This describes a straight line, whereas radiata pine bark relative to DOB is thicker near the stem base and consistent over the remainder. PSP data set coefficients for this model overestimated upper stem bark thickness and underestimated lower stem values. The harvester model for Scots pine bark fitted better than the DOB model for diameters <400 mm but performed poorly for larger logs, as it is restricted to trees with a diameter at breast height <590 mm. Two better models were identified: (1) a model using relative height to account for bark thickness changes with height, and (2) implementing separate DOB models for the bottom 10% and top 90% of the stem split by relative height. These approaches require approval by the StanForD committee to be implemented. Prior to approval, the upper 90% model coefficients could be applied to the entire stem to improve upper 90% stem volume and value predictions at the expense of the lower 10%. Further research will determine if this approach produces acceptable results when optimising bucking. Increasing variability of bark thickness estimates with increasing DOB in the PSP data set may reflect manual measurement errors with thicker bark or may show that additional explanatory variables are needed. Efforts to identify new variables would need to be weighed against additional returns and probable reductions in numbers of larger trees by as the clearfell age is reduced.