Sampling method with proportional probability to variable areas in stands of Eucalyptus dunnii Maiden in Brazil

Many forest companies in Brazil started using plots of varying sizes in their continuous forest inventories (CFI), mainly to avoid the effects of marginal trees and to obtain more consistent estimates in the CFI. Consequently it was quite important to present a comparison between the results of applying a sampling method in which trees are selected with probability proportional to a variable average area with the method in which the trees are selected with probability proportional to a fixed area plot. In this study, the theory for a sampling method was developed, in which a selection of a group of trees in a plot is done with probability proportional to a variable average area of occupation (PPVA) per 60 trees in a plot. The data is composed by 41 permanent plots from a forest inventory of Eucalyptus dunnii Maiden in Santa Catarina, Brazil. In order to compare the PPVA method with the traditional fixed area method (PPA), the area was imaged with an unmanned aerial vehicle (UAV), allowing to obtain a full list of living trees (census). For this fraction, 7 plots for PPVA method and 7 for the PPA method were also measured, allowing the comparison of the total volume obtained in the census with PPVA and PPA results. The total volume of the plot obtained as a function of the census, when compared to the PPVA method, did not present significant differences at the 99% probability level, while the fixed area method was statistically different at the 95% probability level, resulting in an overestimation of 7.5% higher than that found in the census. The PPAV sampling method provides parameter estimates for characterization of the forest population at a lower cost; it becomes more effective than sampling with fixed area plots, because they are practical and operationally easy for delimitation in the field, providing appropriate estimates and more accurate average occupational area for each individual tree (m2) in the population.