Fake Forests and Quantifying Uncertainty in Allometric Equations for the ‘Real World'.

In calculating tree biomass from diameter using allometric models, often there are choices of what allometric model to use for a particular tree species. However, the uncertainty associated from this choice is rarely included in error estimates. To examine the impacts of model selection on levels of uncertainty, we simulated a forest with 100,000 individuals having characteristics similar to Mexican forests in terms of diameter distribution and then calculated a ‘known’ biomass for each tree.  Optimal sample size and distribution of diameters to generate a precise allometric model for such a forest were calculated using standard sample theory. Many different combinations of number of trees sampled, the diameter distribution range of samples, and equation form were used to understand better the effects on the uncertainty estimates of biomass for the forest as a whole. This work illuminates the effects of allometric model choice on uncertainty for forest inventories such as that of Mexico where a decision tree algorithm has been created to select the most appropriate model from their database in an effort to reduce overall uncertainty in carbon estimates for Mexican forests.