J. D. Sauerländer's Verlag, Bad Orb

The k-tree plot is a plot design in which the k trees nearest to a sample point are taken as sample trees. While field implementation is often stated to be fast and straightforward, the challenge of an unbiased estimator remains. A design-unbiased estimator has been developed but is not operational for practical application yet and it is common practice to employ empirical estimators. Applications can be found in ecological surveys above all, but also in some forest inventories even if it is known since long that the empirical estimators carry an unknown bias, depending on the spatial pattern of the stands.
In this study we summarize recent findings on the sampling statistics of the k-tree plot design. This includes an overview of the inclusion zone approach as illustrated in Figures 1, 2 and 3. We carry out and describe a first illustrating sampling simulation on an artificially generated small stand (as of Figure 1 and 2). Objective is to demonstrate the performance of the first design-unbiased estimator and establish a comparison to fixed area plots and Bitterlich plots. From this study results that the design unbiased estimator yields more precise results for both the estimation of basal area and stem density than fixed area plots or Bitterlich sampling. However, the field effort to implement that estimator is prohibitive. If estimation without bias is a serious issue in a particular forest inventory, the authors recommend resorting to more simply to handle plot designs such as of fixed area plots, thus avoiding the estimator challenge of k-tree plots.

Download